squaring up a sinusoidal signal

M

Michael Robinson

Guest
Let's say I have a sinusoidal signal that I put through an RC low pass
filter with the primary goal of shifting the phase.

sine in ---R---+---- out
|
|
=
|
|
gnd

The frequency isn't set in stone but I know what it will be within say a
factor of two. I choose my resistor and capacitor values so that the RC
constant is an order of magnitude greater than the reciprocal of what I
estimate will be my signal's angular frequency. This will give me a phase
delay a little less than 90 degrees and attenuate my signal by a factor
between 5 and 20. Correct?
Now I want to square up the sinusoidal output of the RC filter. What's the
best way to do this? I looked at datasheets for 4046 PLL chips. Those
chips can be used with sinusoidal or digital inputs, but only with a
digital input can you get exact phase locking. If you're following a
sinusoidal signal the phase can be off as much as 90 degrees in either
direction. At least that's how I interpreted the datasheets. And I don't
want that.
So I'm looking for a way to square up the output of the RC filter in some
way that won't introduce a delay (nor load the RC filter). If there's a
chip that will do this, off the top of my head I'd say a few tens of
nanoseconds propagation time might be acceptable. This signal I'm dressing
up is likely going to be somewhere on the order of 100 kHz.
If I can get it squared up I won't need a PLL. The signal will be just the
phase I need and go directly to a half-bridge driver.
That's the story, anyway.

---------------------------------------
Posted through http://www.Electronics-Related.com
 
Snip garbled text
Comaprator? LM393 as long as things aren't too fast.

Oh, just read your 10nS time... You might want one that's a bit
faster.

The all-pass opamp circuit is also nice for making phase shifts.

George H.

A comparator is just what I need, but something much faster than the 393.
Click to expand...
Doesn't that thing have a delay something like a microsecond?
I would use Digikey's parametric search but you can't sort comparators b
propagation or switching time, so I'm down to asking for suggestions.
Something well under 100 nS.

---------------------------------------
Posted through http://www.Electronics-Related.com
 
On Aug 27, 10:38 am, "Michael Robinson"
<kellrobinson@n_o_s_p_a_m.n_o_s_p_a_m.yahoo.com> wrote:
Let's say I have a sinusoidal signal that I put through an RC low pass
filter with the primary goal of shifting the phase.

sine in ---R---+---- out
               |
               |
               >                |
               |
              gnd

The frequency isn't set in stone but I know what it will be within say a
factor of two.  I choose my resistor and capacitor values so that the RC
constant is an order of magnitude greater than the reciprocal of what I
estimate will be my signal's angular frequency.  This will give me a phase
delay a little less than 90 degrees and attenuate my signal by a factor
between 5 and 20.  Correct?
Now I want to square up the sinusoidal output of the RC filter.  What's the
best way to do this?  I looked at datasheets for 4046 PLL chips.  Those
chips can be used with sinusoidal or digital inputs, but only with a
digital input can you get exact phase locking.  If you're following a
sinusoidal signal the phase can be off as much as 90 degrees in either
direction.  At least that's how I interpreted the datasheets.  And I don't
want that.
So I'm looking for a way to square up the output of the RC filter in some
way that won't introduce a delay (nor load the RC filter).  If there's a
chip that will do this, off the top of my head I'd say a few tens of
nanoseconds propagation time might be acceptable.  This signal I'm dressing
up is likely going to be somewhere on the order of 100 kHz.
If I can get it squared up I won't need a PLL.  The signal will be just the
phase I need and go directly to a half-bridge driver.
That's the story, anyway.          

---------------------------------------        
Posted throughhttp://www.Electronics-Related.com
Click to expand...
Comaprator? LM393 as long as things aren't too fast.

Oh, just read your 10nS time... You might want one that's a bit
faster.

The all-pass opamp circuit is also nice for making phase shifts.

George H.
 
On 08/27/2010 07:38 AM, Michael Robinson wrote:
Let's say I have a sinusoidal signal that I put through an RC low pass
filter with the primary goal of shifting the phase.

sine in ---R---+---- out
|
|
=
|
|
gnd

The frequency isn't set in stone but I know what it will be within say a
factor of two. I choose my resistor and capacitor values so that the RC
constant is an order of magnitude greater than the reciprocal of what I
estimate will be my signal's angular frequency. This will give me a phase
delay a little less than 90 degrees and attenuate my signal by a factor
between 5 and 20. Correct?
Now I want to square up the sinusoidal output of the RC filter. What's the
best way to do this? I looked at datasheets for 4046 PLL chips. Those
chips can be used with sinusoidal or digital inputs, but only with a
digital input can you get exact phase locking. If you're following a
sinusoidal signal the phase can be off as much as 90 degrees in either
direction. At least that's how I interpreted the datasheets. And I don't
want that.
So I'm looking for a way to square up the output of the RC filter in some
way that won't introduce a delay (nor load the RC filter). If there's a
chip that will do this, off the top of my head I'd say a few tens of
nanoseconds propagation time might be acceptable. This signal I'm dressing
up is likely going to be somewhere on the order of 100 kHz.
If I can get it squared up I won't need a PLL. The signal will be just the
phase I need and go directly to a half-bridge driver.
That's the story, anyway.

If you're willing to use a 4046 anyway, why not square up your input
Click to expand...
sine wave with a comparator then lock to it using the XOR phase
comparator? That'll give you a nice dependable 90 degree phase shift,
and if your frequency ratio really isn't more than 2:1 you should be
able to set it up for dependable locking, too.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html
 
On Aug 27, 11:53 am, "Michael Robinson"
<kellrobinson@n_o_s_p_a_m.n_o_s_p_a_m.yahoo.com> wrote:
Snip garbled text

Comaprator?  LM393 as long as things aren't too fast.

Oh, just read your 10nS time... You might want one that's a bit
faster.

The all-pass opamp circuit is also nice for making phase shifts.

George H.

A comparator is just what I need, but something much faster than the 393.
Doesn't that thing have a delay something like a microsecond?
I would use Digikey's parametric search but you can't sort comparators by
propagation or switching time, so I'm down to asking for suggestions.
Something well under 100 nS.      

---------------------------------------        
Posted throughhttp://www.Electronics-Related.com
Click to expand...
Yeah the LM393 is a bit slow. (Sometimes that's a good thing.) How
'bout the LT1016. I'm using it to look at ~10ns pulses. Someone
(John L.?) mentioned using video amps as fast swtichers.

George H.
 
"Tim Wescott" <tim@seemywebsite.com> wrote in message
news:qMqdneq0jNeNYerRnZ2dnUVZ_g6dnZ2d@web-ster.com...
On 08/27/2010 07:38 AM, Michael Robinson wrote:
Let's say I have a sinusoidal signal that I put through an RC low pass
filter with the primary goal of shifting the phase.

sine in ---R---+---- out
|
|
=
|
|
gnd

The frequency isn't set in stone but I know what it will be within say a
factor of two. I choose my resistor and capacitor values so that the RC
constant is an order of magnitude greater than the reciprocal of what I
estimate will be my signal's angular frequency. This will give me a
phase
delay a little less than 90 degrees and attenuate my signal by a factor
between 5 and 20. Correct?
Now I want to square up the sinusoidal output of the RC filter. What's
the
best way to do this? I looked at datasheets for 4046 PLL chips. Those
chips can be used with sinusoidal or digital inputs, but only with a
digital input can you get exact phase locking. If you're following a
sinusoidal signal the phase can be off as much as 90 degrees in either
direction. At least that's how I interpreted the datasheets. And I
don't
want that.
So I'm looking for a way to square up the output of the RC filter in some
way that won't introduce a delay (nor load the RC filter). If there's a
chip that will do this, off the top of my head I'd say a few tens of
nanoseconds propagation time might be acceptable. This signal I'm
dressing
up is likely going to be somewhere on the order of 100 kHz.
If I can get it squared up I won't need a PLL. The signal will be just
the
phase I need and go directly to a half-bridge driver.
That's the story, anyway.
If you're willing to use a 4046 anyway, why not square up your input sine
wave with a comparator then lock to it using the XOR phase comparator?
That'll give you a nice dependable 90 degree phase shift, and if your
frequency ratio really isn't more than 2:1 you should be able to set it up
for dependable locking, too.

--

Tim Wescott
Click to expand...
Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater running on 12 volt power, to
heat steel tubing. The induction coil will be wound permanently around the
steel tubing workpiece, and I intend to keep the tubing temperature below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor. If
my thinking is correct, for the inverter drive to be in phase with the tank
current, it needs to be 90 degrees out of phase with the capacitor voltage.
And it should really switch a few degrees before that to keep the load on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is like
picking my way through heavy underbrush. I know guys have built kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.
 
On Fri, 27 Aug 2010 10:02:43 -0700 (PDT), George Herold
<ggherold@gmail.com> wrote:

On Aug 27, 11:53 am, "Michael Robinson"
kellrobinson@n_o_s_p_a_m.n_o_s_p_a_m.yahoo.com> wrote:
Snip garbled text

Comaprator?  LM393 as long as things aren't too fast.

Oh, just read your 10nS time... You might want one that's a bit
faster.

The all-pass opamp circuit is also nice for making phase shifts.

George H.

A comparator is just what I need, but something much faster than the 393.
Doesn't that thing have a delay something like a microsecond?
I would use Digikey's parametric search but you can't sort comparators by
propagation or switching time, so I'm down to asking for suggestions.
Something well under 100 nS.      

---------------------------------------        
Posted throughhttp://www.Electronics-Related.com

Yeah the LM393 is a bit slow. (Sometimes that's a good thing.) How
'bout the LT1016. I'm using it to look at ~10ns pulses. Someone
(John L.?) mentioned using video amps as fast swtichers.

George H.
Click to expand...
No, my favorite trick is to use LVDS receivers as comparators. They
are cheap, fast, and have low-C, hi-Z inputs. But they do have large
poorly-specified DC offsets. That can be fixed with feedback,
sometimes.

Analog Devices makes bipolar comparators with delays in the
picoseconds. Just keep the resistor low enough that the input bias
current doesn't make trouble.

LTC has some medium-fast comparators with TTL outputs.

John
 
On Aug 27, 7:38 am, "Michael Robinson"
<kellrobinson@n_o_s_p_a_m.n_o_s_p_a_m.yahoo.com> wrote:
Let's say I have a sinusoidal signal that I put through an RC low pass
filter with the primary goal of shifting the phase.
[which causes phase shift and attenuation]
So I'm looking for a way to square up the output of the RC filter
Click to expand...
to make a square wave at 90 degrees to the input.

As Tim Wescott says, a '4046 (using XOR type phase detector)
does the whole thing in one step, no RC low pass required.

Also of interest, though, is the making of a square wave from
a sine. To make the square wave sharp, a comparator
would be programmed with hysteresis (but that causes your
phase to be changed). Leave out the hysteresis and the
comparator jitters at the zero crossing (high gain and
shot noise guarantee this with a broadband amplifier).

Sinewave input is repetitive, the desired squarewave will
be symmetric (50% duty cycle); you can use that.

Better might be a one or two transistor AC amplifier, suitably
Baker-clamped, with a simple integrator (operational amplifier style)
setting the operating point to ensure the output duty cycle stabilizes
at 50%.
I've also done this with one transistor (a slow negative DC feedback,
AC coupled input signal), it works fine.
 
On Aug 27, 1:18 pm, "Michael Robinson" <nos...@billburg.com> wrote:

Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater ... I'm lookng to use a series
(voltage-driven) tank.  Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor.
Click to expand...
Why not just put a transformer (could be air-core) sense element
in series with the work coil? It's a current-transformer, you'd
want the output shorted or nearly so (at AC, this just means
a transistor with capacitor from collector to base).
 
"whit3rd" <whit3rd@gmail.com> wrote in message
news:41c9f583-e528-47ee-ad99-a8a6215eb3d6@h19g2000yqb.googlegroups.com...
On Aug 27, 1:18 pm, "Michael Robinson" <nos...@billburg.com> wrote:

Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater ... I'm lookng to use a
series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor.
Click to expand...
Why not just put a transformer (could be air-core) sense element
in series with the work coil? It's a current-transformer, you'd
want the output shorted or nearly so (at AC, this just means
a transistor with capacitor from collector to base).

I know about using a CT, but usually people put a burden resistor on it. I
don't picture this transistor load you're talking about, or how it wouuld be
connected to the CT.
 
On Fri, 27 Aug 2010 16:18:25 -0400, "Michael Robinson"
<nospam@billburg.com> wrote:


Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater running on 12 volt power, to
heat steel tubing. The induction coil will be wound permanently around the
steel tubing workpiece, and I intend to keep the tubing temperature below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor. If
my thinking is correct, for the inverter drive to be in phase with the tank
current, it needs to be 90 degrees out of phase with the capacitor voltage.
And it should really switch a few degrees before that to keep the load on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is like
picking my way through heavy underbrush. I know guys have built kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.
Click to expand...
---
Two questions:

1. Can you use dual supplies?

2. Can the thing be self-oscillating or does it have to be driven by
an external source?


---
Click to expand...
JF
 
On Aug 28, 6:30 pm, "Michael Robinson" <nos...@billburg.com> wrote:
"whit3rd" <whit...@gmail.com> wrote in message

news:41c9f583-e528-47ee-ad99-a8a6215eb3d6@h19g2000yqb.googlegroups.com...
On Aug 27, 1:18 pm, "Michael Robinson" <nos...@billburg.com> wrote:

Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater ... I'm lookng to use a
series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor.

Why not just put a transformer (could be air-core) sense element
in series with the work coil?  It's a current-transformer, you'd
want the output shorted or nearly so (at AC, this just means
a transistor with capacitor from collector to base).

I know about using a CT, but usually people put a burden resistor on it.
Click to expand...
The burden resistor is to ensure low-enough impedance on the
secondary (so the primary effective impedance is near-short-circuit).
A low-Z amplifier input at high frequency is just as easy to arrange.

[warning: bad ASCII art follows]
I've left out the base bias resistors, but this is the general idea

|
Rc
|
+------>out
+-----------+
| /
= |/
| |
in>-----+-------|
|\
in>--+ V
| |
| +--+
+--Vb | |
| = Re
= | |
| | |
+------------+--+--GND
 
"John Fields" <jfields@austininstruments.com> wrote in message
news:8i1l7697li0fouesu23qmpkk8fcumageug@4ax.com...
On Fri, 27 Aug 2010 16:18:25 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater running on 12 volt power,
to
heat steel tubing. The induction coil will be wound permanently around
the
steel tubing workpiece, and I intend to keep the tubing temperature below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor. If
my thinking is correct, for the inverter drive to be in phase with the
tank
current, it needs to be 90 degrees out of phase with the capacitor
voltage.
And it should really switch a few degrees before that to keep the load on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is
like
picking my way through heavy underbrush. I know guys have built kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.

---
Two questions:

1. Can you use dual supplies?
Click to expand...
No, there's just a single power source available.
2. Can the thing be self-oscillating or does it have to be driven by
an external source?

Whatever works. I think with induction heaters usually there's a loop that
Click to expand...
locks the inverter in phase with the sinusoidal current in the tank (work
coil/capacitor), but with this unvarying load perhaps I could just determine
the correct frequency empirically and set it to run at that frequency, open
loop.
Does that answer your question?
---
JF
Click to expand...
 
"whit3rd" <whit3rd@gmail.com> wrote in message
news:b07d9c63-662b-4ef1-abd8-c61079c4f82b@i13g2000yqe.googlegroups.com...
On Aug 28, 6:30 pm, "Michael Robinson" <nos...@billburg.com> wrote:
"whit3rd" <whit...@gmail.com> wrote in message

news:41c9f583-e528-47ee-ad99-a8a6215eb3d6@h19g2000yqb.googlegroups.com...
On Aug 27, 1:18 pm, "Michael Robinson" <nos...@billburg.com> wrote:

Before I go any further, let me describe the project. I'm designing a
low
power (no more than 200 watts) induction heater ... I'm lookng to use a
series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor.

Why not just put a transformer (could be air-core) sense element
in series with the work coil? It's a current-transformer, you'd
want the output shorted or nearly so (at AC, this just means
a transistor with capacitor from collector to base).

I know about using a CT, but usually people put a burden resistor on it.
Click to expand...
The burden resistor is to ensure low-enough impedance on the
secondary (so the primary effective impedance is near-short-circuit).
A low-Z amplifier input at high frequency is just as easy to arrange.

[warning: bad ASCII art follows]
I've left out the base bias resistors, but this is the general idea

|
Rc
|
+------>out
+-----------+
| /
= |/
| |
in>-----+-------|
|\
in>--+ V
| |
| +--+
+--Vb | |
| = Re
= | |
| | |
+------------+--+--GND

You really have to spell it out for me. Do the inputs for your amplifier
come from either end of the burden resistor on the current transformer
secondary? What is Vb?
 
On Sun, 29 Aug 2010 20:22:04 -0400, "Michael Robinson"
<nospam@billburg.com> wrote:

"John Fields" <jfields@austininstruments.com> wrote in message
news:8i1l7697li0fouesu23qmpkk8fcumageug@4ax.com...
On Fri, 27 Aug 2010 16:18:25 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


Before I go any further, let me describe the project. I'm designing a low
power (no more than 200 watts) induction heater running on 12 volt power,
to
heat steel tubing. The induction coil will be wound permanently around
the
steel tubing workpiece, and I intend to keep the tubing temperature below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in the
work coil, I thought of sensing the voltage across the tank capacitor. If
my thinking is correct, for the inverter drive to be in phase with the
tank
current, it needs to be 90 degrees out of phase with the capacitor
voltage.
And it should really switch a few degrees before that to keep the load on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is
like
picking my way through heavy underbrush. I know guys have built kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.

---
Two questions:

1. Can you use dual supplies?

No, there's just a single power source available.

2. Can the thing be self-oscillating or does it have to be driven by
an external source?

Whatever works. I think with induction heaters usually there's a loop that
locks the inverter in phase with the sinusoidal current in the tank (work
coil/capacitor), but with this unvarying load perhaps I could just determine
the correct frequency empirically and set it to run at that frequency, open
loop.
Does that answer your question?
Click to expand...
---
Yes.

Just a thought:

Version 4
SHEET 1 1476 916
WIRE -224 -336 -496 -336
WIRE 32 -336 -224 -336
WIRE 208 -336 32 -336
WIRE 32 -288 32 -336
WIRE 208 -240 208 -336
WIRE -112 -192 -304 -192
WIRE 32 -160 32 -208
WIRE 160 -160 32 -160
WIRE -224 -144 -224 -336
WIRE -304 -128 -304 -192
WIRE -256 -128 -304 -128
WIRE 32 -128 32 -160
WIRE -112 -112 -112 -192
WIRE -112 -112 -192 -112
WIRE -256 -96 -352 -96
WIRE -224 -48 -224 -80
WIRE 32 -16 32 -64
WIRE -112 96 -112 -112
WIRE -112 96 -304 96
WIRE 32 96 32 48
WIRE -496 128 -496 -336
WIRE -224 128 -496 128
WIRE -224 160 -224 128
WIRE -304 176 -304 96
WIRE -256 176 -304 176
WIRE -112 192 -112 96
WIRE -112 192 -192 192
WIRE 32 192 32 160
WIRE 32 192 -112 192
WIRE 208 192 208 -144
WIRE 352 192 208 192
WIRE -352 208 -352 -96
WIRE -256 208 -352 208
WIRE -352 224 -352 208
WIRE 32 224 32 192
WIRE 352 240 352 192
WIRE -352 320 -352 304
WIRE 352 320 352 304
WIRE 352 320 -352 320
WIRE 32 368 32 288
WIRE 352 384 352 320
WIRE 32 448 32 432
WIRE 208 528 208 192
WIRE 32 544 32 512
WIRE 160 544 32 544
WIRE 352 544 352 464
WIRE -496 576 -496 128
WIRE 32 592 32 544
WIRE -496 720 -496 656
WIRE -224 720 -224 224
WIRE -224 720 -496 720
WIRE 32 720 32 672
WIRE 32 720 -224 720
WIRE 208 720 208 624
WIRE 208 720 32 720
WIRE 352 720 352 624
WIRE 352 720 208 720
WIRE -496 816 -496 720
FLAG -496 816 0
FLAG -224 -48 0
SYMBOL voltage -496 560 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 12
SYMBOL res -368 208 R0
WINDOW 0 -38 37 Left 0
WINDOW 3 -62 80 Left 0
SYMATTR InstName R3
SYMATTR Value 1e6
SYMBOL nmos 160 -240 R0
SYMATTR InstName M1
SYMATTR Value Si7868DP
SYMBOL pmos 160 624 M180
SYMATTR InstName M2
SYMATTR Value FDS4465
SYMBOL res 16 -304 R0
SYMATTR InstName R6
SYMATTR Value 510
SYMBOL res 48 688 R180
WINDOW 0 36 76 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName R8
SYMATTR Value 510
SYMBOL diode 16 -128 R0
WINDOW 3 36 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D3
SYMBOL diode 16 448 R0
WINDOW 3 41 30 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D1
SYMBOL diode 16 96 R0
WINDOW 3 37 34 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D2
SYMBOL diode 16 224 R0
WINDOW 3 41 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D4
SYMBOL diode 16 -16 R0
WINDOW 3 41 31 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D5
SYMBOL diode 16 368 R0
WINDOW 3 43 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D6
SYMBOL ind2 336 368 R0
WINDOW 0 61 31 Left 0
WINDOW 3 39 65 Left 0
SYMATTR InstName L1
SYMATTR Value 1.59e-5
SYMATTR Type ind
SYMBOL cap 336 240 R0
WINDOW 0 -39 5 Left 0
WINDOW 3 -84 59 Left 0
SYMATTR InstName C2
SYMATTR Value 1.59e-7
SYMBOL res 336 528 R0
SYMATTR InstName R1
SYMATTR Value .2
SYMBOL Opamps\\LT1498 -224 -176 R0
SYMATTR InstName U2
SYMBOL Opamps\\LT1498 -224 128 R0
SYMATTR InstName U1
TEXT -480 760 Left 0 !;ac oct 256 20 1000000
TEXT -480 792 Left 0 !.tran .001 startup

I just picked a couple of MOSFETs out of LTC's menu based on Vds(on)
and Rds(on), and an LT1498 because it can source/sink about 30 mA.

There are two in a package, and I'm using them as unity-gain buffers,
so I connected them in parallel instead of sparing one.

---
JF
 
"John Fields" <jfields@austininstruments.com> wrote in message
news:mr6n76h7ps46lil100mbe01vue0ehu2c79@4ax.com...
On Sun, 29 Aug 2010 20:22:04 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


"John Fields" <jfields@austininstruments.com> wrote in message
news:8i1l7697li0fouesu23qmpkk8fcumageug@4ax.com...
On Fri, 27 Aug 2010 16:18:25 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


Before I go any further, let me describe the project. I'm designing a
low
power (no more than 200 watts) induction heater running on 12 volt
power,
to
heat steel tubing. The induction coil will be wound permanently around
the
steel tubing workpiece, and I intend to keep the tubing temperature
below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in
the
work coil, I thought of sensing the voltage across the tank capacitor.
If
my thinking is correct, for the inverter drive to be in phase with the
tank
current, it needs to be 90 degrees out of phase with the capacitor
voltage.
And it should really switch a few degrees before that to keep the load
on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It
will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is
like
picking my way through heavy underbrush. I know guys have built
kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.

---
Two questions:

1. Can you use dual supplies?

No, there's just a single power source available.

2. Can the thing be self-oscillating or does it have to be driven by
an external source?

Whatever works. I think with induction heaters usually there's a loop
that
locks the inverter in phase with the sinusoidal current in the tank (work
coil/capacitor), but with this unvarying load perhaps I could just
determine
the correct frequency empirically and set it to run at that frequency,
open
loop.
Does that answer your question?

---
Yes.

Just a thought:

Version 4
SHEET 1 1476 916
WIRE -224 -336 -496 -336
WIRE 32 -336 -224 -336
WIRE 208 -336 32 -336
WIRE 32 -288 32 -336
WIRE 208 -240 208 -336
WIRE -112 -192 -304 -192
WIRE 32 -160 32 -208
WIRE 160 -160 32 -160
WIRE -224 -144 -224 -336
WIRE -304 -128 -304 -192
WIRE -256 -128 -304 -128
WIRE 32 -128 32 -160
WIRE -112 -112 -112 -192
WIRE -112 -112 -192 -112
WIRE -256 -96 -352 -96
WIRE -224 -48 -224 -80
WIRE 32 -16 32 -64
WIRE -112 96 -112 -112
WIRE -112 96 -304 96
WIRE 32 96 32 48
WIRE -496 128 -496 -336
WIRE -224 128 -496 128
WIRE -224 160 -224 128
WIRE -304 176 -304 96
WIRE -256 176 -304 176
WIRE -112 192 -112 96
WIRE -112 192 -192 192
WIRE 32 192 32 160
WIRE 32 192 -112 192
WIRE 208 192 208 -144
WIRE 352 192 208 192
WIRE -352 208 -352 -96
WIRE -256 208 -352 208
WIRE -352 224 -352 208
WIRE 32 224 32 192
WIRE 352 240 352 192
WIRE -352 320 -352 304
WIRE 352 320 352 304
WIRE 352 320 -352 320
WIRE 32 368 32 288
WIRE 352 384 352 320
WIRE 32 448 32 432
WIRE 208 528 208 192
WIRE 32 544 32 512
WIRE 160 544 32 544
WIRE 352 544 352 464
WIRE -496 576 -496 128
WIRE 32 592 32 544
WIRE -496 720 -496 656
WIRE -224 720 -224 224
WIRE -224 720 -496 720
WIRE 32 720 32 672
WIRE 32 720 -224 720
WIRE 208 720 208 624
WIRE 208 720 32 720
WIRE 352 720 352 624
WIRE 352 720 208 720
WIRE -496 816 -496 720
FLAG -496 816 0
FLAG -224 -48 0
SYMBOL voltage -496 560 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 12
SYMBOL res -368 208 R0
WINDOW 0 -38 37 Left 0
WINDOW 3 -62 80 Left 0
SYMATTR InstName R3
SYMATTR Value 1e6
SYMBOL nmos 160 -240 R0
SYMATTR InstName M1
SYMATTR Value Si7868DP
SYMBOL pmos 160 624 M180
SYMATTR InstName M2
SYMATTR Value FDS4465
SYMBOL res 16 -304 R0
SYMATTR InstName R6
SYMATTR Value 510
SYMBOL res 48 688 R180
WINDOW 0 36 76 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName R8
SYMATTR Value 510
SYMBOL diode 16 -128 R0
WINDOW 3 36 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D3
SYMBOL diode 16 448 R0
WINDOW 3 41 30 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D1
SYMBOL diode 16 96 R0
WINDOW 3 37 34 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D2
SYMBOL diode 16 224 R0
WINDOW 3 41 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D4
SYMBOL diode 16 -16 R0
WINDOW 3 41 31 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D5
SYMBOL diode 16 368 R0
WINDOW 3 43 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D6
SYMBOL ind2 336 368 R0
WINDOW 0 61 31 Left 0
WINDOW 3 39 65 Left 0
SYMATTR InstName L1
SYMATTR Value 1.59e-5
SYMATTR Type ind
SYMBOL cap 336 240 R0
WINDOW 0 -39 5 Left 0
WINDOW 3 -84 59 Left 0
SYMATTR InstName C2
SYMATTR Value 1.59e-7
SYMBOL res 336 528 R0
SYMATTR InstName R1
SYMATTR Value .2
SYMBOL Opamps\\LT1498 -224 -176 R0
SYMATTR InstName U2
SYMBOL Opamps\\LT1498 -224 128 R0
SYMATTR InstName U1
TEXT -480 760 Left 0 !;ac oct 256 20 1000000
TEXT -480 792 Left 0 !.tran .001 startup

I just picked a couple of MOSFETs out of LTC's menu based on Vds(on)
and Rds(on), and an LT1498 because it can source/sink about 30 mA.

There are two in a package, and I'm using them as unity-gain buffers,
so I connected them in parallel instead of sparing one.

---
JF
Click to expand...
I copied the circuit into LTSpice and took a look. I can see that it will
need something to get the tank started oscillating, which is probably why my
effort to sim it came to nothing. I think to get that circuit to start I
will need to insert a supply above or below the rails and give a kick to one
of the mosfets to get the tank going.
Gives me a good excuse to start learning Spice.
 
On Mon, 30 Aug 2010 12:53:48 -0400, "Michael Robinson"
<nospam@billburg.com> wrote:

"John Fields" <jfields@austininstruments.com> wrote in message
news:mr6n76h7ps46lil100mbe01vue0ehu2c79@4ax.com...
On Sun, 29 Aug 2010 20:22:04 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


"John Fields" <jfields@austininstruments.com> wrote in message
news:8i1l7697li0fouesu23qmpkk8fcumageug@4ax.com...
On Fri, 27 Aug 2010 16:18:25 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


Before I go any further, let me describe the project. I'm designing a
low
power (no more than 200 watts) induction heater running on 12 volt
power,
to
heat steel tubing. The induction coil will be wound permanently around
the
steel tubing workpiece, and I intend to keep the tubing temperature
below
200F. So the load is in effect invariant. I'm lookng to use a series
(voltage-driven) tank. Instead of using a CR to sense the current in
the
work coil, I thought of sensing the voltage across the tank capacitor.
If
my thinking is correct, for the inverter drive to be in phase with the
tank
current, it needs to be 90 degrees out of phase with the capacitor
voltage.
And it should really switch a few degrees before that to keep the load
on
the mosfets inductive. The RC filter I described accomplishes that by
providing a phase shift (delay) of slightly less than 90 degrees. It
will
also attenuate the high voltage of the tank. And it's dirt simple.
If I use a PLL to get a phase shift of 90 degrees, then switching may
actually occur slightly late; bad for the mosfets.
I don't know enough about the 4046 to respond very cogently to your
suggestion about the XOR comparator. Just reading those datasheets is
like
picking my way through heavy underbrush. I know guys have built
kilowatt
induction heaters without PLL at all, so I'm not attached to the idea of
PLL, just willing to consider it.

---
Two questions:

1. Can you use dual supplies?

No, there's just a single power source available.

2. Can the thing be self-oscillating or does it have to be driven by
an external source?

Whatever works. I think with induction heaters usually there's a loop
that
locks the inverter in phase with the sinusoidal current in the tank (work
coil/capacitor), but with this unvarying load perhaps I could just
determine
the correct frequency empirically and set it to run at that frequency,
open
loop.
Does that answer your question?

---
Yes.

Just a thought:

Version 4
SHEET 1 1476 916
WIRE -224 -336 -496 -336
WIRE 32 -336 -224 -336
WIRE 208 -336 32 -336
WIRE 32 -288 32 -336
WIRE 208 -240 208 -336
WIRE -112 -192 -304 -192
WIRE 32 -160 32 -208
WIRE 160 -160 32 -160
WIRE -224 -144 -224 -336
WIRE -304 -128 -304 -192
WIRE -256 -128 -304 -128
WIRE 32 -128 32 -160
WIRE -112 -112 -112 -192
WIRE -112 -112 -192 -112
WIRE -256 -96 -352 -96
WIRE -224 -48 -224 -80
WIRE 32 -16 32 -64
WIRE -112 96 -112 -112
WIRE -112 96 -304 96
WIRE 32 96 32 48
WIRE -496 128 -496 -336
WIRE -224 128 -496 128
WIRE -224 160 -224 128
WIRE -304 176 -304 96
WIRE -256 176 -304 176
WIRE -112 192 -112 96
WIRE -112 192 -192 192
WIRE 32 192 32 160
WIRE 32 192 -112 192
WIRE 208 192 208 -144
WIRE 352 192 208 192
WIRE -352 208 -352 -96
WIRE -256 208 -352 208
WIRE -352 224 -352 208
WIRE 32 224 32 192
WIRE 352 240 352 192
WIRE -352 320 -352 304
WIRE 352 320 352 304
WIRE 352 320 -352 320
WIRE 32 368 32 288
WIRE 352 384 352 320
WIRE 32 448 32 432
WIRE 208 528 208 192
WIRE 32 544 32 512
WIRE 160 544 32 544
WIRE 352 544 352 464
WIRE -496 576 -496 128
WIRE 32 592 32 544
WIRE -496 720 -496 656
WIRE -224 720 -224 224
WIRE -224 720 -496 720
WIRE 32 720 32 672
WIRE 32 720 -224 720
WIRE 208 720 208 624
WIRE 208 720 32 720
WIRE 352 720 352 624
WIRE 352 720 208 720
WIRE -496 816 -496 720
FLAG -496 816 0
FLAG -224 -48 0
SYMBOL voltage -496 560 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 12
SYMBOL res -368 208 R0
WINDOW 0 -38 37 Left 0
WINDOW 3 -62 80 Left 0
SYMATTR InstName R3
SYMATTR Value 1e6
SYMBOL nmos 160 -240 R0
SYMATTR InstName M1
SYMATTR Value Si7868DP
SYMBOL pmos 160 624 M180
SYMATTR InstName M2
SYMATTR Value FDS4465
SYMBOL res 16 -304 R0
SYMATTR InstName R6
SYMATTR Value 510
SYMBOL res 48 688 R180
WINDOW 0 36 76 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName R8
SYMATTR Value 510
SYMBOL diode 16 -128 R0
WINDOW 3 36 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D3
SYMBOL diode 16 448 R0
WINDOW 3 41 30 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D1
SYMBOL diode 16 96 R0
WINDOW 3 37 34 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D2
SYMBOL diode 16 224 R0
WINDOW 3 41 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D4
SYMBOL diode 16 -16 R0
WINDOW 3 41 31 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D5
SYMBOL diode 16 368 R0
WINDOW 3 43 32 Left 0
SYMATTR Value 1N4148
SYMATTR InstName D6
SYMBOL ind2 336 368 R0
WINDOW 0 61 31 Left 0
WINDOW 3 39 65 Left 0
SYMATTR InstName L1
SYMATTR Value 1.59e-5
SYMATTR Type ind
SYMBOL cap 336 240 R0
WINDOW 0 -39 5 Left 0
WINDOW 3 -84 59 Left 0
SYMATTR InstName C2
SYMATTR Value 1.59e-7
SYMBOL res 336 528 R0
SYMATTR InstName R1
SYMATTR Value .2
SYMBOL Opamps\\LT1498 -224 -176 R0
SYMATTR InstName U2
SYMBOL Opamps\\LT1498 -224 128 R0
SYMATTR InstName U1
TEXT -480 760 Left 0 !;ac oct 256 20 1000000
TEXT -480 792 Left 0 !.tran .001 startup

I just picked a couple of MOSFETs out of LTC's menu based on Vds(on)
and Rds(on), and an LT1498 because it can source/sink about 30 mA.

There are two in a package, and I'm using them as unity-gain buffers,
so I connected them in parallel instead of sparing one.

---
JF

I copied the circuit into LTSpice and took a look. I can see that it will
need something to get the tank started oscillating, which is probably why my
effort to sim it came to nothing. I think to get that circuit to start I
will need to insert a supply above or below the rails and give a kick to one
of the mosfets to get the tank going.
Gives me a good excuse to start learning Spice.
Click to expand...
---
Hmmm...

It starts here every time,

You're running a transient analysis, right?

---
JF
 
On Mon, 30 Aug 2010 12:53:48 -0400, "Michael Robinson"
<nospam@billburg.com> wrote:


I copied the circuit into LTSpice and took a look. I can see that it will
need something to get the tank started oscillating, which is probably why my
effort to sim it came to nothing. I think to get that circuit to start I
will need to insert a supply above or below the rails and give a kick to one
of the mosfets to get the tank going.
Gives me a good excuse to start learning Spice.
Click to expand...
---
Just a hunch...

If you're not familiar with LTspice, then maybe it's working but you
haven't picked what to display.

If you can see the schematic, but no waveforms, right-click on an
empty spot in the schematicf pane, and then click "RUN" in thye box
that appears. Then, when everything settles down, left-click the
schematic pane and then mouse around until you see some pencil-like
thingies appear when you get close to wires.

Those are voltage probes, and if you left-click when they're pointing
at a wire, then what's on that wire will appear on the waveform
display pane.

If you see a clamp-on looking thing with an arrow going through it
appear when you're over a component, that's a current probe, and left
clicking will display the current in that component.

Good luck :)

---
JF
 
"John Fields" <jfields@austininstruments.com> wrote in message
news:ev6o76lifeb01nsoj5vlgvcb8qr7f9fj0e@4ax.com...
On Mon, 30 Aug 2010 12:53:48 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


I copied the circuit into LTSpice and took a look. I can see that it will
need something to get the tank started oscillating, which is probably why
my
effort to sim it came to nothing. I think to get that circuit to start I
will need to insert a supply above or below the rails and give a kick to
one
of the mosfets to get the tank going.
Gives me a good excuse to start learning Spice.

---
Just a hunch...

If you're not familiar with LTspice, then maybe it's working but you
haven't picked what to display.

If you can see the schematic, but no waveforms, right-click on an
empty spot in the schematicf pane, and then click "RUN" in thye box
that appears. Then, when everything settles down, left-click the
schematic pane and then mouse around until you see some pencil-like
thingies appear when you get close to wires.

Those are voltage probes, and if you left-click when they're pointing
at a wire, then what's on that wire will appear on the waveform
display pane.

If you see a clamp-on looking thing with an arrow going through it
appear when you're over a component, that's a current probe, and left
clicking will display the current in that component.

Good luck :)

---
JF
Ok, got it. And thanks for posting this elegant little circuit, John!
Click to expand...
 
On Mon, 30 Aug 2010 22:42:52 -0400, "Michael Robinson"
<nospam@billburg.com> wrote:

"John Fields" <jfields@austininstruments.com> wrote in message
news:ev6o76lifeb01nsoj5vlgvcb8qr7f9fj0e@4ax.com...
On Mon, 30 Aug 2010 12:53:48 -0400, "Michael Robinson"
nospam@billburg.com> wrote:


I copied the circuit into LTSpice and took a look. I can see that it will
need something to get the tank started oscillating, which is probably why
my
effort to sim it came to nothing. I think to get that circuit to start I
will need to insert a supply above or below the rails and give a kick to
one
of the mosfets to get the tank going.
Gives me a good excuse to start learning Spice.

---
Just a hunch...

If you're not familiar with LTspice, then maybe it's working but you
haven't picked what to display.

If you can see the schematic, but no waveforms, right-click on an
empty spot in the schematicf pane, and then click "RUN" in thye box
that appears. Then, when everything settles down, left-click the
schematic pane and then mouse around until you see some pencil-like
thingies appear when you get close to wires.

Those are voltage probes, and if you left-click when they're pointing
at a wire, then what's on that wire will appear on the waveform
display pane.

If you see a clamp-on looking thing with an arrow going through it
appear when you're over a component, that's a current probe, and left
clicking will display the current in that component.

Good luck :)

---
JF
Ok, got it. And thanks for posting this elegant little circuit, John!
Click to expand...
---
You're welcome; my pleasure. :)

---
JF
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top