Hey, that's wierd.

Uncle Steve · May 16, 2013

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

Jim Thompson · May 16, 2013

On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com>
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

From the "body arrow" direction I assume the MOSFET is NMOS?

Then the gate drive needs to be well above the +18V to turn it on.

The "weirdness" you are seeing is probably due to not having
everything tied to a common ground. Though you might need a tin-foil
hat >:-}

Or go to your local community college and take a few courses.

...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice

Uncle Steve · May 16, 2013

On Thu, May 16, 2013 at 10:28:00AM -0700, Jim Thompson wrote:

On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

From the "body arrow" direction I assume the MOSFET is NMOS?

No, P-channel, but I may have hooked it up backwards since I
mistakenly soldered in a known broken n-channel MOSFET yesterday by
accident. Nope. Was wired correctly, and it seems the gate is
broken.

Then the gate drive needs to be well above the +18V to turn it on.

The "weirdness" you are seeing is probably due to not having
everything tied to a common ground. Though you might need a tin-foil
hat >:-}

If I unhook everything from the board except the DVM, the voltage does
descend to ~-.3 V spontaneously. Perhaps it's a miracle. Or the
Orbital Mind-Control Laser Cannon.

Or go to your local community college and take a few courses.

That would be ideal.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

Uncle Steve · May 16, 2013

On Thu, May 16, 2013 at 10:39:47AM -0700, George Herold wrote:

On May 16, 1:02 pm, Uncle Steve <stevet...@gmail.com> wrote:
I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

Maybe it the big cap? What happens if you pull that.

Get this:

4mV on my coffee table; 30mV on the balcony; 70mV 4" from a 15W CFL;
450mV with the coil touching the CFL spiral tube. 50mv in the shade,
moving 8" into direct sunlight max 1.5V, although it varies depending
on the orientation of the coil. It can't be sensitive to visible
light. It just can't.

What I did was disconnect the +terminal of the cap from the circuit.
Removing the cap entirely... Same thing.

I guess it's possible for an AC magnetic field to be turning on the
1n4004 diode during part of the cycle... and charging up (or down)
the output.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I'm guessing the AC magnetic fields. (Which you can't see with a
compass)

The metal box seems to work as a Faraday Cage now, which pretty much
means this circuit is RF sensitive. Why the magnitude of the signal
varies so much in sunlight vs. shade is odd. Using my scope, which
maxes out at a 72MHz sampling rate doesn't show a recognizable
waveform.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

George Herold · May 16, 2013

On May 16, 1:02 pm, Uncle Steve <stevet...@gmail.com> wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

Maybe it the big cap? What happens if you pull that.

I guess it's possible for an AC magnetic field to be turning on the
1n4004 diode during part of the cycle... and chargeing up (or down)
the output.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I'm guessing the AC magnetic fields. (Which you can't see with a
compass)

George H.

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

John Larkin · May 16, 2013

On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com>
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

--

John Larkin Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation

Uncle Steve · May 16, 2013

On Thu, May 16, 2013 at 01:12:24PM -0700, John Larkin wrote:

On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

1Hz

It appears the gate was borken on the P-Channel MOSFET. I have
purchased a replacement and will install it tomorrow.

While walking downtown to the local electronics shop, I wondered if the
voltage level I'm seeing is related to the diode capacitance. I'm
sure if I removed all the resistors and reinstalled the cap that the
sun would charge the capacitor quite a lot. Vf of the diode is
slightly less than a volt at moderate power levels.

The question in my mind is in determining what frequency of the EM
spectrum it is that is exciting this circuit.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

John Larkin · May 17, 2013

On Thu, 16 May 2013 18:46:08 -0400, Uncle Steve <stevet810@gmail.com> wrote:

On Thu, May 16, 2013 at 01:12:24PM -0700, John Larkin wrote:
On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

1Hz

It appears the gate was borken on the P-Channel MOSFET. I have
purchased a replacement and will install it tomorrow.

If it's a p-fet, the drain should be on the right.

And 4.7K is a really big (ie, slow) gate resistor.

--

John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

ABLE1 · May 17, 2013

"George Herold" <gherold@teachspin.com> wrote in message
news:a92a2867-c6da-43f7-86c7-77fc88d90af4@v14g2000yqm.googlegroups.com...
On May 16, 2:30 pm, Uncle Steve <stevet...@gmail.com> wrote:

On Thu, May 16, 2013 at 10:39:47AM -0700, George Herold wrote:
On May 16, 1:02 pm, Uncle Steve <stevet...@gmail.com> wrote:
I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

Maybe it the big cap? What happens if you pull that.

Get this:

4mV on my coffee table; 30mV on the balcony; 70mV 4" from a 15W CFL;
450mV with the coil touching the CFL spiral tube. 50mv in the shade,
moving 8" into direct sunlight max 1.5V, although it varies depending
on the orientation of the coil. It can't be sensitive to visible
light. It just can't.

What I did was disconnect the +terminal of the cap from the circuit.
Removing the cap entirely... Same thing.

I guess it's possible for an AC magnetic field to be turning on the
1n4004 diode during part of the cycle... and charging up (or down)
the output.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I'm guessing the AC magnetic fields. (Which you can't see with a
compass)

The metal box seems to work as a Faraday Cage now, which pretty much
means this circuit is RF sensitive. Why the magnitude of the signal
varies so much in sunlight vs. shade is odd. Using my scope, which
maxes out at a 72MHz sampling rate doesn't show a recognizable
waveform.

OK as a WAG maybe you are picking up some radio station or other
RF. Does you 'scope do an FFT? Maybe you can see something there.

George H.

Regards,

Uncle Steve

When all rational thoughts fail to find a solution the only thing left to
investigate would be aliens. You might want to consider looking
for the Mother Ship.

George Herold · May 17, 2013

On May 16, 2:30 pm, Uncle Steve <stevet...@gmail.com> wrote:

On Thu, May 16, 2013 at 10:39:47AM -0700, George Herold wrote:
On May 16, 1:02 pm, Uncle Steve <stevet...@gmail.com> wrote:
I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

Maybe it the big cap? What happens if you pull that.

Get this:

4mV on my coffee table; 30mV on the balcony; 70mV 4" from a 15W CFL;
450mV with the coil touching the CFL spiral tube. 50mv in the shade,
moving 8" into direct sunlight max 1.5V, although it varies depending
on the orientation of the coil. It can't be sensitive to visible
light. It just can't.

What I did was disconnect the +terminal of the cap from the circuit.
Removing the cap entirely... Same thing.

I guess it's possible for an AC magnetic field to be turning on the
1n4004 diode during part of the cycle... and charging up (or down)
the output.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I'm guessing the AC magnetic fields. (Which you can't see with a
compass)

The metal box seems to work as a Faraday Cage now, which pretty much
means this circuit is RF sensitive. Why the magnitude of the signal
varies so much in sunlight vs. shade is odd. Using my scope, which
maxes out at a 72MHz sampling rate doesn't show a recognizable
waveform.

OK as a WAG maybe you are picking up some radio station or other
RF. Does you 'scope do an FFT? Maybe you can see something there.

George H.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.- Hide quoted text -

- Show quoted text -

Uncle Steve · May 17, 2013

On Thu, May 16, 2013 at 09:46:02PM -0700, John Larkin wrote:

On Thu, 16 May 2013 18:46:08 -0400, Uncle Steve <stevet810@gmail.com> wrote:

On Thu, May 16, 2013 at 01:12:24PM -0700, John Larkin wrote:
On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

1Hz

It appears the gate was borken on the P-Channel MOSFET. I have
purchased a replacement and will install it tomorrow.

If it's a p-fet, the drain should be on the right.

And 4.7K is a really big (ie, slow) gate resistor.

Yeah, I kind of figured out what I did wrong. First was mixing up my
parts; the IRF1140pbF is N-channel, and slightly overkill anyways for
this purpose. The part I shoulda soldered in there is a FQP27P60,
which is in fact P-channel. I'm driving the gate with a
resistor/divider network that is connected to the collecter of a
2n2222, to be driven by TTL. Dunno how fast that will be, but it
works at 1Hz.

Still have no clue what the coil is sensing, but I made another one so
I can investigate just that.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

Uncle Steve · May 17, 2013

On Fri, May 17, 2013 at 05:57:48AM -0700, George Herold wrote:

On May 16, 2:30 pm, Uncle Steve <stevet...@gmail.com> wrote:
On Thu, May 16, 2013 at 10:39:47AM -0700, George Herold wrote:
On May 16, 1:02 pm, Uncle Steve <stevet...@gmail.com> wrote:
I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

Maybe it the big cap? What happens if you pull that.

Get this:

4mV on my coffee table; 30mV on the balcony; 70mV 4" from a 15W CFL;
450mV with the coil touching the CFL spiral tube. 50mv in the shade,
moving 8" into direct sunlight max 1.5V, although it varies depending
on the orientation of the coil. It can't be sensitive to visible
light. It just can't.

What I did was disconnect the +terminal of the cap from the circuit.
Removing the cap entirely... Same thing.

I guess it's possible for an AC magnetic field to be turning on the
1n4004 diode during part of the cycle... and charging up (or down)
the output.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I'm guessing the AC magnetic fields. (Which you can't see with a
compass)

The metal box seems to work as a Faraday Cage now, which pretty much
means this circuit is RF sensitive. Why the magnitude of the signal
varies so much in sunlight vs. shade is odd. Using my scope, which
maxes out at a 72MHz sampling rate doesn't show a recognizable
waveform.

OK as a WAG maybe you are picking up some radio station or other
RF. Does you 'scope do an FFT? Maybe you can see something there.

No FFT. It's a fairly inexpensive DSO, 1/2F is 35Mhz, but the display
probably wouldn't quite be enough to resolve that. It's just a tiny
handheld about the size of a smartphone.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

John Larkin · May 18, 2013

On Thu, 16 May 2013 18:46:08 -0400, Uncle Steve <stevet810@gmail.com> wrote:

On Thu, May 16, 2013 at 01:12:24PM -0700, John Larkin wrote:
On Thu, 16 May 2013 13:02:38 -0400, Uncle Steve <stevet810@gmail.com
wrote:

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Regards,

Uncle Steve

Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

1Hz

Yikes! The inductor is a short at 1 Hz. It's not a switching regulator.

--

John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

Uncle Steve · May 18, 2013

On Sat, May 18, 2013 at 08:18:10AM -0700, John Larkin wrote:

On Thu, 16 May 2013 18:46:08 -0400, Uncle Steve <stevet810@gmail.com> wrote:
Electrolytic caps are dissimilar plates in an acid bath, which is a
lot like a battery. Local magnetic fields won't charge a cap.

Hey, I have a Digikey bag-o assorted electrolytic caps over there on
the floor... time to clean up maybe? They all measure positive, 10 to
maybe 80 millivolts. If you short one, it starts recharging right
away. The COE issues are interesting here.

That 1N4004 is slow to use as the catch diode in a switcher. Better to
a schottky, 1N5819 maybe.

You're using an n-channel fet as a follower, so the gate drive had
better swing from ground to +25 maybe.

What's your switching frequency?

1Hz

Yikes! The inductor is a short at 1 Hz. It's not a switching regulator.

I think I have viable firmware for a test run with the uC, so I'll see
what happens when I run PCM through it. Prior tests were only to
verify that it worked, which it didn't until I realized I'd have to
discharge the gate of the MOSFET before it would turn off...

In other news, the second coil I wound should be something like 168uH
since I used smaller wire and a tighter winding. With just a diode
and the 3300uF cap in any orientation possible, the cap develops
only a negative charge, and more slowly than with the 10uH part (which
has 3 1.1M ohm resistors parallel with the coil). Peak voltage was
-120mV after about 1/2hour and didn't seem to be sensitive to direct
sunlight as with the buck-converter circuit previously described.

A 150nF polypropylene cap substituted does not reproduce the
phenomenon. I'll do a random walk with a few different parts and see
if I can knock loose some additional information. The electrolytic
does not spontaneously develop charge in a way that would explain what
is occuring.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

P E Schoen · May 18, 2013

"Uncle Steve" wrote in message news:dfe9ae5f13b3de2-b857a@gmail.com...

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Well, if that is a P-FET, then the D-S need to be reversed, so your SW will
turn it ON when connected to ground. 18V might be too high for the G-S
voltage. And you should have a resistor from G-S so it turns off. The gate
resistors should be more like 100 ohms if you pan to use 10 kHz or more.

When the MOSFET turns ON, it will light the LED, and the 18V will be
essentially applied to to the 3300 uF capacitor after the short time delay
of the 10 uH coil. Then the current will be limited only by the RdsON of the
MOSFET and the ESR of the capacitor, and a few milliohms in the inductor and
wiring. So probably no more than 0.1 ohms or 180 amps from the 18V supply.
It very likely is the current limiting factor.

When (or if) the MOSFET turns off, the inductance of the 10 uH coil will try
to maintain the current by an inductive "kick" that causes the voltage at
the cathode of the 1N4004 to go negative, which is limited by its 700 mV
forward drop (although the actual voltage waveform will be affected by its
capacitance and switching characteristics). But if there is a 180 amp
charging current, that current will flow through the diode, and may damage
or destroy it.

What you need to do is monitor the current through the inductor and switch
it off when it reaches a reasonable value that is within the ratings of the
power supply and the MOSFET. Then allow the energy in the inductor to be
transferred to the capacitor before starting another cycle. For a reasonable
value of 10 amps, the time may be determined by:

I = dV*dT/L or T = L*I/V = 10*10/18 = 5.5 uSec

So your switching frequency needs to be in the order of 200 KHz, which
requires a fast MOSFET drive. The IRF1404 is N-channel, so it might work as
shown if you drive it with an isolated supply from G-S. The maximum Vgs is
20V so you are pretty close to catastrophe with 18V. The gate capacitance is
5600 pF, so for a rough 5.5 uSec PWM with 10% switching losses, the TC for
the gate drive resistance should be in the order of

R = T / C = 550 nS / 5.6 nF = 100 ohms

You really want to switch even faster to keep out of the linear region. If
you look at the rise/fall/delay times you will see that they are in the
order of 30-200 nSec, so you need to use a switching frequency of about 500
kHz maximum. Even 200 KHz is tricky. This circuit will work better if you
use a much larger inductor such as 100 uH.

Check my math. These are just quick estimates, but I think they are at least
:in the ballpark".

Go O's!

Paul

Uncle Steve · May 18, 2013

On Sat, May 18, 2013 at 05:09:03PM -0400, P E Schoen wrote:

"Uncle Steve" wrote in message news:dfe9ae5f13b3de2-b857a@gmail.com...

I'm seeing something really interesting with the coil I built. I
wired up a buck converter circuit yesterday and tried it out, but just
with a switch and resistor tied to the gate of an IRF-1404PbF, which
lists a Total Gate Charge of 131nC, which I take to mean nano-
coulombs. When the circuit is 'rested', power applied does what is
expected.

10uH 0R1
18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
D| _/S | | | | |
| v | | | | | |
- - - \ | \ \ |
-+--- / 2.2K | / 1M 1M / |
| \ | \ \ |
\ / 1N --- / / |
/ 4.7k | 4004/-\ | | ---
\ | | \ \ --- 3300uF
/ \_/ LED | / 100K 100k / |
to SW | -+- | \ \ |
-------+ | | / / |
| | | | |
| | | | |
GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,
but I may have damaged the MOSFET. When it worked, the MOSFET
wouldn't shut off. I either the coil is generating a field that
induces current in the gate wire, or something similar. No big deal,
I'll work out what's wrong because I had something similar set up on a
breadboard (without the inductor) and things switched on and off as
expected.

The weird thing is that a -.3.06V charge develops across Vout and GND
even if the circuit is completely disconnected from supply and GND.
So, sitting out in the open with a meter attached to the output,
voltage climbs over a period of minutes. -2mV per second or so, but
it slows down as the voltage approaches the -.2V level.

My compass shows that there are varying magnetic fields throughout my
apartment, but even in locations where the needle does not appear to
be deflected from Magnetic North, the voltage climbs as described.
If I shorted the capacitor (at ~1V) to drain accumulated charge, would
I have induced some sort of charge on the negative plate that would
cause this to occur? Or is this, as it appears, a phenomenon related
to the ambient magnetic fields occurring here where I am working?

I tried putting the circuit in a metal box, but obviously I had to
allow the DVM leads to exit the box so I could watch the DVM, and
continued to observe the negative voltage climb. Weird, eh?

Well, if that is a P-FET, then the D-S need to be reversed, so your SW will
turn it ON when connected to ground. 18V might be too high for the G-S
voltage. And you should have a resistor from G-S so it turns off. The gate
resistors should be more like 100 ohms if you pan to use 10 kHz or more.

I already corrected this part of the circuit. Delete the 4.7K and
imagine 18V - 220K - GATE - 1M - 2n2222 collector - 2n2222 emitter to
ground. Works fine.

When the MOSFET turns ON, it will light the LED, and the 18V will be
essentially applied to to the 3300 uF capacitor after the short time delay
of the 10 uH coil. Then the current will be limited only by the RdsON of
the MOSFET and the ESR of the capacitor, and a few milliohms in the
inductor and wiring. So probably no more than 0.1 ohms or 180 amps from the
18V supply. It very likely is the current limiting factor.

When (or if) the MOSFET turns off, the inductance of the 10 uH coil will
try to maintain the current by an inductive "kick" that causes the voltage
at the cathode of the 1N4004 to go negative, which is limited by its 700 mV

More like 900mV.

forward drop (although the actual voltage waveform will be affected by its
capacitance and switching characteristics). But if there is a 180 amp
charging current, that current will flow through the diode, and may damage
or destroy it.

The power supply is nowhere near capable of 180 amps. 2.5 on a good
day.

What you need to do is monitor the current through the inductor and switch
it off when it reaches a reasonable value that is within the ratings of the
power supply and the MOSFET. Then allow the energy in the inductor to be
transferred to the capacitor before starting another cycle. For a
reasonable value of 10 amps, the time may be determined by:

I = dV*dT/L or T = L*I/V = 10*10/18 = 5.5 uSec

Aha, math. I was anticipating using the sense-resistor to measure
power, but I note that the coil reads .5ohms according to my meter,
but how accurate is that going to be with the other stuff in
parallel?

So your switching frequency needs to be in the order of 200 KHz, which
requires a fast MOSFET drive. The IRF1404 is N-channel, so it might work
as shown if you drive it with an isolated supply from G-S. The maximum Vgs
is 20V so you are pretty close to catastrophe with 18V. The gate
capacitance is 5600 pF, so for a rough 5.5 uSec PWM with 10% switching
losses, the TC for the gate drive resistance should be in the order of

Dammit. I mixed up my parts and installed the IRF1404 thinking it was
the P-channel part. Installed a FQP27P06, which is p-channel, and
everything works as described with the 220K/1M divider as described
above. The FQP27P06 gate capacitance may be 4400pF, if I'm
translating correctly: while the IRF part shows gate capacitance, the
FQP part datasheet only described gate charge of 43nC, but 55nC for
the IRF part.

R = T / C = 550 nS / 5.6 nF = 100 ohms

You really want to switch even faster to keep out of the linear region. If
you look at the rise/fall/delay times you will see that they are in the
order of 30-200 nSec, so you need to use a switching frequency of about 500
kHz maximum. Even 200 KHz is tricky. This circuit will work better if you
use a much larger inductor such as 100 uH.

I'm not really at the point yet where I feel I can characterize the
circuit requirements that accurately, and I haven't seen R = T/C
before. Let me do it my way: 1C = 1VA. 4.4nF, so 3.2V / 220k =
14.5uA. Which means i need lower resistors to drive the gate since
I'm limited to less than 2.9kHz. OK. Let's say I'm working with
62500Hz, which is the fastest I can go with this part unless I majorly
refactor the software. If I want to use the ADC to sample at a
particular phase angle, I have to use timer0, which is limited to the
PCLK, which is basically max 16MHz. The timer is 8 bits, so thats
16MHz/256. Maybe I could double that by fiddling with the timer
registers on the fly, within the interrupt, but I'd rather not and
keep the 8-bit resolution and save the uC cycles. After all there are
only 256 cpu cycles available each PCM period, and I'm eating some by
not relying entirely on the timer to frob the GPIO. (I use the timer
to turn it off, but the rising edge is set by the ISR, and steals
about 10% of the CPU to do so.) I'm sampling the ADC at 0, 90, 180,
and 270 degrees with a 128kHz ADCCLK by playing games with the timer
comparator, and thats with PCM at 31250Hz. The ADC gets more
inaccurate as its clock goes up, and it takes 2 ADCCLKs to sample-and-
hold so you can see how ugly this will get if I have to move to a PWM
frequency of 200KHz.

So anyways, I need 2-4V drop at the gate with two resistors having a
ratio of about 1:5, and it needs to come on really quickly so I don't
waste all of the PWM phase waiting for the thing to turn on. So,
20ohms/100ohms gives a 3V drop at 160mA. That's 10ns to charge the
gate by my calculations, which I have no real confidence in

Check my math. These are just quick estimates, but I think they are at
least :in the ballpark".

I'll run over the numbers again, but I think I get the idea from what
you've written. I'm going to test the circuit without a capacitor and
with a simulated load of some kind to see what happens when the uC pin
goes high with my DSO watching the MOSFET output and triggering on the
GPIO rising edge. Plus I'm gathering lots of ADC measurements which I
can capture via serial output from the uC.

Go O's!

Who? Oakland?

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

P E Schoen · May 19, 2013

"Uncle Steve" wrote in message news:a94389325ae9c69-f0989@gmail.com...

I already corrected this part of the circuit. Delete the 4.7K and
imagine 18V - 220K - GATE - 1M - 2n2222 collector - 2n2222 emitter to
ground. Works fine.

I'm not sure why you are using such high resistor values. I think I
recreated your circuit with LTSpice, but to get the MOSFET to switch
properly even at 10 kHz I had to use much lower values. Here is what seems
to be what you are doing, and this circuit works as a buck converter with
18V in and about 13V out into 10 ohms and 100 uF.

http://enginuitysystems.com/pix/UncleSteve2.png

The power supply is nowhere near capable of 180 amps. 2.5 on a
good day.

You need to consider how much capacitance is across the 18V supply. You can
easily get more than 100 amps for a few milliseconds, which is enough to
damage some components.

Aha, math. I was anticipating using the sense-resistor to measure
power, but I note that the coil reads .5ohms according to my meter,
but how accurate is that going to be with the other stuff in
parallel?

You can read the current on the high side with an IC made for that purpose.
A simple differential amplifier will work.

Dammit. I mixed up my parts and installed the IRF1404 thinking it was
the P-channel part. Installed a FQP27P06, which is p-channel, and
everything works as described with the 220K/1M divider as described
above. The FQP27P06 gate capacitance may be 4400pF, if I'm
translating correctly: while the IRF part shows gate capacitance, the
FQP part datasheet only described gate charge of 43nC, but 55nC for
the IRF part.

The 220k/1M divider might turn the MOSFET on and off, but with 4400 pF and
220K you have a TC of nearly 1 mSec. That would be just barely adequate for
100 Hz switching. If you are trying to reduce power this is NOT the way to
do it. During the linear portion of the switching you may have 3 amps at 10
volts or more which is 30 watts.

I'm not really at the point yet where I feel I can characterize the
circuit requirements that accurately, and I haven't seen R = T/C
before. Let me do it my way: 1C = 1VA. 4.4nF, so 3.2V / 220k > 14.5uA. Which means i need lower resistors to drive the gate since
I'm limited to less than 2.9kHz. OK. Let's say I'm working with
62500Hz, which is the fastest I can go with this part unless I majorly
refactor the software.

You can use as low as 10 kHz for the PWM as shown in my schematic. If you go
higher you can get better efficiency, but you will need to use a gate
driver. A smaller MOSFET will also be OK and will have lower gate
capacitance.

If I want to use the ADC to sample at a
particular phase angle, I have to use timer0, which is limited to the
PCLK, which is basically max 16MHz. The timer is 8 bits, so thats
16MHz/256. Maybe I could double that by fiddling with the timer
registers on the fly, within the interrupt, but I'd rather not and
keep the 8-bit resolution and save the uC cycles. After all there are
only 256 cpu cycles available each PCM period, and I'm eating some by
not relying entirely on the timer to frob the GPIO. (I use the timer
to turn it off, but the rising edge is set by the ISR, and steals
about 10% of the CPU to do so.) I'm sampling the ADC at 0, 90, 180,
and 270 degrees with a 128kHz ADCCLK by playing games with the timer
comparator, and thats with PCM at 31250Hz. The ADC gets more
inaccurate as its clock goes up, and it takes 2 ADCCLKs to sample-and-
hold so you can see how ugly this will get if I have to move to a PWM
frequency of 200KHz.

You may want to consider a microcontroller with a built-in S-R flip flop and
a comparator so you can use hardware to sense the current rise and turn off
the MOSFET drive. Or you can determine a PWM frequency and a range of duty
cycles that gives you the output you need for all loads, and then just
monitor the DC level of the output voltage and current to tweak the PWM
parameters. As long as you stay in discontinuous mode the inductor current
will always reach a certain peak and then returns to zero, as you can see in
the waveforms of my simulation. If you increase the duty cycle the inductor
current will alternate between a minimum and a maximum, which is continuous
mode, and it can be prone to runaway unless you monitor the peak inductor
current and shut down the drive before it reaches saturation (or an
excessively high current since an air core choke does not saturate.

So anyways, I need 2-4V drop at the gate with two resistors having
a ratio of about 1:5, and it needs to come on really quickly so I don't
waste all of the PWM phase waiting for the thing to turn on. So,
20ohms/100ohms gives a 3V drop at 160mA. That's 10ns to charge the
gate by my calculations, which I have no real confidence in

You really need more like 10V on the gate to get lowest RdsON, and you need
to get there pretty quick or you will get into linear mode which is a power
hog. I think I see how your circuit is constructed now. But you would do
better with two 100 ohm resistors for about 9V gate drive. That is 90 mA.
And the RC time constant will be about 100 * 0.0044 uF or 440 nSec. I
generally use the RC time constant for rough calculations. Actually there
are other considerations that affect turn-on and turn-off time, such as the
Miller plateau which is caused by the gate-drain capacitance conducting
current into or out of the gate as the Vds changes.

I'll run over the numbers again, but I think I get the idea from what
you've written. I'm going to test the circuit without a capacitor and
with a simulated load of some kind to see what happens when the uC pin
goes high with my DSO watching the MOSFET output and triggering on the
GPIO rising edge. Plus I'm gathering lots of ADC measurements which I
can capture via serial output from the uC.

You may want to design a really low speed buck converter that switches
slowly enough for you to take enough samples to see what's going on. A
digital storage scope is good for that. But you can also run the circuit at
more normal speeds like the 10 kHz I show, and you can use a cheap 10 MHz
scope to see the waveforms well enough.

Go O's!

Who? Oakland?

No, those are the "A's". I'm from Baltimore, and the team is the Orioles!

Paul

Uncle Steve · May 19, 2013

On Sun, May 19, 2013 at 12:08:57AM -0400, P E Schoen wrote:

"Uncle Steve" wrote in message news:a94389325ae9c69-f0989@gmail.com...

I already corrected this part of the circuit. Delete the 4.7K and
imagine 18V - 220K - GATE - 1M - 2n2222 collector - 2n2222 emitter to
ground. Works fine.

I'm not sure why you are using such high resistor values. I think I
recreated your circuit with LTSpice, but to get the MOSFET to switch
properly even at 10 kHz I had to use much lower values. Here is what seems
to be what you are doing, and this circuit works as a buck converter with
18V in and about 13V out into 10 ohms and 100 uF.

http://enginuitysystems.com/pix/UncleSteve2.png

Neat. The high-resistor values are clearly a mistake, a result of my
misconceptions about how these devices work, though as I work through
the math the data-sheet parameters are becoming clearer.

The power supply is nowhere near capable of 180 amps. 2.5 on a
good day.

You need to consider how much capacitance is across the 18V supply. You can
easily get more than 100 amps for a few milliseconds, which is enough to
damage some components.

There are no capacitors filtering the 18V supply, ATM.

Aha, math. I was anticipating using the sense-resistor to measure
power, but I note that the coil reads .5ohms according to my meter,
but how accurate is that going to be with the other stuff in
parallel?

You can read the current on the high side with an IC made for that purpose.
A simple differential amplifier will work.

For my purposes, this is about learning as much as it is about
charging a battery, so I'm interested mostly in keeping things as
simple as possible. Plus, I'm using what I have on-hand as much as
possible, so the capacitor on the output is 3300uF because I have
one. I suppose that's a risk since it may source a lot of current as
it charges up, though the MOSFET I'm using is rated at -108A for
repeated pulses of short duration so perhaps it will be O.K.

Dammit. I mixed up my parts and installed the IRF1404 thinking it was
the P-channel part. Installed a FQP27P06, which is p-channel, and
everything works as described with the 220K/1M divider as described
above. The FQP27P06 gate capacitance may be 4400pF, if I'm
translating correctly: while the IRF part shows gate capacitance, the
FQP part datasheet only described gate charge of 43nC, but 55nC for
the IRF part.

The 220k/1M divider might turn the MOSFET on and off, but with 4400 pF and
220K you have a TC of nearly 1 mSec. That would be just barely adequate for
100 Hz switching. If you are trying to reduce power this is NOT the way to
do it. During the linear portion of the switching you may have 3 amps at 10
volts or more which is 30 watts.

Yeah, I read that the MOSFETS dissipate most power during state
transitions so lowering the switching time is the way to go.

I'm not really at the point yet where I feel I can characterize the
circuit requirements that accurately, and I haven't seen R = T/C
before. Let me do it my way: 1C = 1VA. 4.4nF, so 3.2V / 220k =
14.5uA. Which means i need lower resistors to drive the gate since
I'm limited to less than 2.9kHz. OK. Let's say I'm working with
62500Hz, which is the fastest I can go with this part unless I majorly
refactor the software.

You can use as low as 10 kHz for the PWM as shown in my schematic. If you
go higher you can get better efficiency, but you will need to use a gate
driver. A smaller MOSFET will also be OK and will have lower gate
capacitance.

It is encouraging to see your simulation working at 10kHz, which tells
me that I won't have to do a major rework of the software. By gate
driver do you mean something like a darlington transistor to replace
the 2n2222? I have a BC517 I could use instead, and which will handle
1A, which means 20ohms to the gate should work.

If I want to use the ADC to sample at a
particular phase angle, I have to use timer0, which is limited to the
PCLK, which is basically max 16MHz. The timer is 8 bits, so thats
16MHz/256. Maybe I could double that by fiddling with the timer
registers on the fly, within the interrupt, but I'd rather not and
keep the 8-bit resolution and save the uC cycles. After all there are
only 256 cpu cycles available each PCM period, and I'm eating some by
not relying entirely on the timer to frob the GPIO. (I use the timer
to turn it off, but the rising edge is set by the ISR, and steals
about 10% of the CPU to do so.) I'm sampling the ADC at 0, 90, 180,
and 270 degrees with a 128kHz ADCCLK by playing games with the timer
comparator, and thats with PCM at 31250Hz. The ADC gets more
inaccurate as its clock goes up, and it takes 2 ADCCLKs to sample-and-
hold so you can see how ugly this will get if I have to move to a PWM
frequency of 200KHz.

You may want to consider a microcontroller with a built-in S-R flip flop
and a comparator so you can use hardware to sense the current rise and turn
off the MOSFET drive. Or you can determine a PWM frequency and a range of
duty cycles that gives you the output you need for all loads, and then just
monitor the DC level of the output voltage and current to tweak the PWM
parameters. As long as you stay in discontinuous mode the inductor current
will always reach a certain peak and then returns to zero, as you can see
in the waveforms of my simulation. If you increase the duty cycle the
inductor current will alternate between a minimum and a maximum, which is
continuous mode, and it can be prone to runaway unless you monitor the peak
inductor current and shut down the drive before it reaches saturation (or
an excessively high current since an air core choke does not saturate.

The ATtiny85 has a flexible timer that does all the work, albeit at a
lower frequency, with a mode called phase-correct PCM. I'm using the
timer overflow interrupt to turn on the GPIO, and a timer compare
register to bring it low, which gives me slightly less than 31.25kHz
with a 8MHz system clock. This frees up the second timer compare
register to co-ordinate ADC measurements. The second timer in the
part will do 250kHz PWM with 8-bits resolution for the duty-cycle,
but doesn't interface with the ADC, so I'm using it to do the
serial-output for data logging.

I sort of figured I would have to derive the output characteristics
empirically, but that's the fun part. I hope to be able to code the
thing so it adapts to dynamic loading, but mainly I have two or three
voltage targets which should be current-limited as per the recommended
charge-curve for lead-acid batteries.

So anyways, I need 2-4V drop at the gate with two resistors having
a ratio of about 1:5, and it needs to come on really quickly so I don't
waste all of the PWM phase waiting for the thing to turn on. So,
20ohms/100ohms gives a 3V drop at 160mA. That's 10ns to charge the
gate by my calculations, which I have no real confidence in

You really need more like 10V on the gate to get lowest RdsON, and you need
to get there pretty quick or you will get into linear mode which is a power
hog. I think I see how your circuit is constructed now. But you would do
better with two 100 ohm resistors for about 9V gate drive. That is 90 mA.
And the RC time constant will be about 100 * 0.0044 uF or 440 nSec. I
generally use the RC time constant for rough calculations. Actually there
are other considerations that affect turn-on and turn-off time, such as the
Miller plateau which is caused by the gate-drain capacitance conducting
current into or out of the gate as the Vds changes.

Hmm. That implies it is safe to drive the gate with 10V, whereas I
was assuming I should drive it close to the gate threshold voltage.
Ok, I can work with that, too. The FQP27P06 datasheet lists ~200ns
rise time and so forth, so I can estimate that the maximum usable
frequency it will operate at is about 1MHz, which is well in excess of
my requirements.

I'll run over the numbers again, but I think I get the idea from what
you've written. I'm going to test the circuit without a capacitor and
with a simulated load of some kind to see what happens when the uC pin
goes high with my DSO watching the MOSFET output and triggering on the
GPIO rising edge. Plus I'm gathering lots of ADC measurements which I
can capture via serial output from the uC.

You may want to design a really low speed buck converter that switches
slowly enough for you to take enough samples to see what's going on. A
digital storage scope is good for that. But you can also run the circuit at
more normal speeds like the 10 kHz I show, and you can use a cheap 10 MHz
scope to see the waveforms well enough.

At 31.25kHz I can still read the ADC at arbitrary phase angles, but
it's slow to switch channels, and I'm thinking I'd be better off with
a multiplexer to route all the measurements to one ADC pin. As it is
sampling four phase angles from three different ADC channels takes
about 1.5mS, but that's just for testing. In practice the most
significant phase angles differ for line, inductor, and sense-resistor
measurement. I should be able to code such that the feedback loop
operates at about 3kHz, perhaps faster if the ADC turns out to be
reasonably accurate at higher frequencies.

Go O's!

Who? Oakland?

No, those are the "A's". I'm from Baltimore, and the team is the Orioles!

This is way more fun than watching baseball on TV.

Regards,

Uncle Steve

--
There should be a special word in the English language to identify
people who create problems and then turn around and offer up their own
tailor-made bogus non-solutions designed to completely avoid the root
causes of the situation under consideration. 'Traitor' might be a
good choice, but lacks the requisite specificity. One of the problems
with contemporary English is it lacks many such words that would
otherwise categorically identify certain kinds of person, place, or
thing -- making it difficult or impossible to think analytically about
such objects. These shortcomings of the English lexicon are
representative of Orwellian linguistics at work in the real world.

Steve Gonedes · May 29, 2013

Uncle Steve <stevet810@gmail.com> writes:

< I'm seeing something really interesting with the coil I built. I
< wired up a buck converter circuit yesterday and tried it out, but just
< with a switch and resistor tied to the gate of an IRF-1404PbF, which
< lists a Total Gate Charge of 131nC, which I take to mean nano-
< coulombs. When the circuit is 'rested', power applied does what is
< expected.

10uH 0R1

< 18V +---- ---------+---------+--^^^^^--+--/\/\/----+----+----+ Vout
< D| _/S | | | | |
< | v | | | | | |
< - - - \ | \ \ |
< -+--- / 2.2K | / 1M 1M / |
< | \ | \ \ |
< \ / 1N --- / / |
< / 4.7k | 4004/-\ | | ---
< \ | | \ \ --- 3300uF
< / \_/ LED | / 100K 100k / |
< to SW | -+- | \ \ |
< -------+ | | / / |
< | | | | |
< | | | | |
< GND +--------------+---------+---------+-----------+----+----+ GND

Note that this test was without a load. It worked a couple of times,

< but I may have damaged the MOSFET. When it worked, the MOSFET
< wouldn't shut off. I either the coil is generating a field that
< induces current in the gate wire, or something similar. No big deal,
< I'll work out what's wrong because I had something similar set up on a
< breadboard (without the inductor) and things switched on and off as
< expected.

I'm just taking a guess; not too sure about the circuit.

The 1N4004 diodes "should" have a PIV of 400 Volts. I have found some
weird diodes that have a PIV of 50 volts - they seem to fragmentate
quite well. It might be an parts issue. I have two IRF510 from radio
shack. The one I can't use anymore has a drain-source voltage of 100v
and a Drain-gate voltage of 100v. The other box has strange acronyms
that I don't understand.

The part number I used for prototyping is rs276-2072A. The new one seems
to be lacking the letter A. These are the only mosfets I have in a box.
The box states clearly that the IRF510 Power MOSFET (N-channel) should
be soldered with a grounded solder iron and be handle with the funny
foam in place.

It seems that finding replacement parts is hard these days.

< The weird thing is that a -.3.06V charge develops across Vout and GND
< even if the circuit is completely disconnected from supply and GND.
< So, sitting out in the open with a meter attached to the output,
< voltage climbs over a period of minutes. -2mV per second or so, but
< it slows down as the voltage approaches the -.2V level.

The MOSFET chip I have in the box states that is should be in the funny
foam ``until'' ...

< My compass shows that there are varying magnetic fields throughout my
< apartment, but even in locations where the needle does not appear to
< be deflected from Magnetic North, the voltage climbs as described.
< If I shorted the capacitor (at ~1V) to drain accumulated charge, would
< I have induced some sort of charge on the negative plate that would
< cause this to occur? Or is this, as it appears, a phenomenon related
< to the ambient magnetic fields occurring here where I am working?

Time to gut the record player yet? Just kidding...

< I tried putting the circuit in a metal box, but obviously I had to
< allow the DVM leads to exit the box so I could watch the DVM, and
< continued to observe the negative voltage climb. Weird, eh?

yeah.

Hey, that's wierd.

Uncle Steve

Guest

Jim Thompson

Guest

Uncle Steve

Guest

Uncle Steve

Guest

George Herold

Guest

John Larkin

Guest

Uncle Steve

Guest

John Larkin

Guest

ABLE1

Guest

George Herold

Guest

Uncle Steve

Guest

Uncle Steve

Guest

John Larkin

Guest

Uncle Steve

Guest

P E Schoen

Guest

Uncle Steve

Guest

P E Schoen

Guest

Uncle Steve

Guest

Steve Gonedes

Guest

Log in

Welcome to EDABoard.com

Sponsor