Another discrepancy (LTspice/switchcad3)

[Robert Baer]

Mike Engelhardt wrote:

Helmut,


I did some more simulations with some components
changed to get more reliable results. Maybe I
see now what you mean with wrong pulses.
My simulation has shown double pulses at the gate
output of the LT1619 where I hadn't expected it.

I have sent my results to Mike.
Let's see what he will tell me.


That's what happens with really poorly designed
SMPS's. There's lots of ways you can get
double/triple pulses that are due to design
flaws(s). You have to deal with min and max
on/off times, slope compensation(which nearly
swaps Robert's current slope with the 1mH
inductor moving the low frequency compensation
pretty close to voltage mode), and loop
instablies near the switching frequency. SMPS
have all sorts of non-linear oscillation modes.

But now it looks like the problem is that Robert
is using a different MOSFET on the bench than
in the schematic.

--Mike


Not relevant; see my other recent responses.

 

[Jim Thompson]

On Wed, 01 Jun 2005 01:37:35 GMT, Robert Baer
<robertbaer@earthlink.net> wrote:

Mike Engelhardt wrote:

Robert,


The results are way off.

The simulation is probably right around correct.
That FET and diode have capacitance.


Adding a small capacitance has no effect.

But that doesn't remove the capacitance that's
already swamping the circuit.

And a pF or two cannot "swamp" anything here.

Not a pF or two. A 1.5nF for the MOSFET and 126pF
for the diode. It's hard to run a power SMPS down to
the tiny currents you're trying to do.

I had thought that i made it clear that i have


That wasn't clear to me because the circuit you
posted clearly does not work.


The models show *square waves* !!
The models show *AMPS* of current !!
The model does not show anything near reality!


The difference is that you are using a different
MOSFET in simulation then the bench. The simulation
results would agree with using a the MOSFET you have
on that schematic you posted. Now you're here talking
about different MOSFETs.

LTspice is showing what reality would look like for the
parts you have in the schematic. Your bench circuit is
entirely different. The schematic is totally shorted
out with capacitance, your bench circuit apparently
not. See the corrected schematic I posted with
capacitance-free switches.


The classic definition of an inductor is a (linear)
component that opposes a change of current.


And the simulation you posted does show that. But you
have to look at the current in the inductor and
the voltage across it. Neither the sense resistor
nor the switch current will follow that ramp, because
capacitive displacement currents swamp the inductor
current ramp.

--Mike


Oh, then no matter what inductor i use, no matter what the time scale
i use, that is the excuse?
Put a resistor in series with the inductor and see what that current
is; you will get the *same* square wave as every where else.
BTW, i originally used a FET from those available in the software
package, and not an ideal one - same result.
An ideal switch VS a slow, high capacitance FET like the IRFBG20
makes little difference on the workbench; current thru an inductor will
be a ramp - pure and simple.
My tests show the "slow" turnoff, wider flyback pulse with no "flat"
top when using the IRFBG20.
The same test fixture, with everything else the same - but using an
IRLZ24N shows a reasonably fast turnoff, a reasonably narrow flyback
pulse, and the classic flat top.
The model shows only square waves.
No ramp, no flyback pulse!
With an ideal switch, one cannot say that the inductor energy "got
swampped" to explain away the loss of a flyback pulse.
Be at least a little realistic...

Robert, Please send me copies of all the circuits you've tried; with a
short narrative for each.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[analog]

Robert Baer wrote:

Errr...I have reduced the problem to the most simple possible; a
FET driving an inductor. *No* SMPS, just a basic circuit.
*Same* problem.

Robert, I looked at the text in your asc file that you posted to
start off this misbegotten thread and the simulation appears to
run for only 25us. Most of Linear Tech's models accurately model
start up delays, soft start, etc. I suggest you increase the run
time to several milliseconds and click "Skip initial operating point
solution" (set the UIC flag) in the simulation dialog box. Watch
the start up. It should be just like the real circuit if you've
done your job correctly.

I guess for this, i will have to scrap LTspice/switchcad3.

Your loss. I've *never* found a circuit LTspice couldn't simulate
realistically (and as well or better than any other simulator).

Please post your simplified LTspice asc file. I'm sure Helmut,
Mike, I or others could get it to work for you.

Regards -- analog

 

[legg]

On Wed, 01 Jun 2005 01:23:00 GMT, Robert Baer
<robertbaer@earthlink.net> wrote:

Errr...I have reduced the problem to the most simple possible; a FET
driving an inductor. *No* SMPS, just a basic circuit.
*Same* problem.
See an earlier response with that simple circuit *working* in
TopSpice (1mH, 20uSec, 10V), virtually identical to practice, but
garbage square waves in LTspice/switchcad3.
As i mentioned earlier, i used my simple test fixture (to emulate the
"1619" example) using a 10uH inductor and 3.3V supply and got a *ramp*
of current to a few hundred mA - just like theory (and the *definition*
of an inductor). But LTspice/switchcad3 gives square waves.
Different inductance and time scales, but the same result for practice.
Modelling with a given SPICE program is consistent for that program,
but TopSpice echos practice and the other does not.
I guess for this, i will have to scrap LTspice/switchcad3.

Your problem appears to be that you haven't sussessfully simulated the
circuit.

You've recieved some very useful suggestions on how to make your
simulation approximate the real-life case.

Why don't you use this advice and learn from it?

Simulation is a tool. Once you've been able to duplicate real behavior
with this tool, you are in a position to predict behavior of other
unbuilt circuits, with some confidence. This is the main purpose of
the simulation tool - where testing of the real circuit may be
impractical.

RL

 

[Mike Engelhardt]

Robert,

The results are way off.

The simulation is probably right around correct.
That FET and diode have capacitance.

Adding a small capacitance has no effect.

But that doesn't remove the capacitance that's
already swamping the circuit.

And a pF or two cannot "swamp" anything here.

Not a pF or two. A 1.5nF for the MOSFET and 126pF
for the diode. It's hard to run a power SMPS down to
the tiny currents you're trying to do.

I had thought that i made it clear that i have

That wasn't clear to me because the circuit you
posted clearly does not work.

The models show *square waves* !!
The models show *AMPS* of current !!
The model does not show anything near reality!

The difference is that you are using a different
MOSFET in simulation then the bench. The simulation
results would agree with using a the MOSFET you have
on that schematic you posted. Now you're here talking
about different MOSFETs.

LTspice is showing what reality would look like for the
parts you have in the schematic. Your bench circuit is
entirely different. The schematic is totally shorted
out with capacitance, your bench circuit apparently
not. See the corrected schematic I posted with
capacitance-free switches.

The classic definition of an inductor is a (linear)
component that opposes a change of current.

And the simulation you posted does show that. But you
have to look at the current in the inductor and
the voltage across it. Neither the sense resistor
nor the switch current will follow that ramp, because
capacitive displacement currents swamp the inductor
current ramp.

Oh, then no matter what inductor i use, no matter
what the time scale i use, that is the excuse?

No excuses, but you might apologize so getting so excited
about having the errors in hour schematic pointed out.
Anyway, the time scale is hard to vary because the 1619
wants to switch at 300KHz. But if you remove the
capacitance, then you get want you expect, like the
other schematic I posted.

Put a resistor in series with the inductor and see what that current is;
you will get the *same* square wave as every where else.

Eh, just click on the inductor to plot its current. There's
nothing wrong there. Drag the mouse across the inductor to
plot it's waveform. Corresponds to the dI/dt. Duh.

BTW, i originally used a FET from those available in the software package,
and not an ideal one - same result.

OK, the FET from the software package has capacitance modeled and
that that keeps your design as posted from working. It's just
telling you your design doesn't work. The ideal switch does
make the corrected version of your design convert power and
regulate.

An ideal switch VS a slow, high capacitance FET like the IRFBG20 makes
little difference on the workbench; current thru an inductor will be a
ramp - pure and simple.

Only if there's a square wave across it. With the 300KHz 1619,
freq, you're a bit pressed to see the ramp in a milliHenery
for the current spike of the highly non-linear Miller cap
of the FET in your posted schematic.

My tests show the "slow" turnoff, wider flyback pulse with no "flat" top
when using the IRFBG20.
The same test fixture, with everything else the same - but using an
IRLZ24N shows a reasonably fast turnoff, a reasonably narrow flyback
pulse, and the classic flat top.
The model shows only square waves.
No ramp, no flyback pulse!

It's a mistake to confuse flybacks with boost converters.

With an ideal switch, one cannot say that the inductor energy
"got swampped" to explain away the loss of a flyback pulse.
Be at least a little realistic...

I never said inductor energy gets swapped. I said the circuit
was swapped with capacitance as in AC shorted out. If you use
the capacitance-free switch you get square wave across the
inductor, a ramp in current through the sense resistor that
the 1619 can servo to regulate the output. You haven't run
the corrected version of your circuit that I posted. But with
the circuit as you posted, with a 1nF MOSFET and 100pF diode,
you won't see the ramp in the sense resistor because the current
is swamped by displacement currents.

The simulation is correct, your design as posted is the problem.

I can only tell what the problem is, you have to do a bit of effort,
say, running the corrected circuit as I posted.

--Mike

 

[Robert Baer]

Jim Thompson wrote:

I'm busy, so it may take a few days, but I'll give it a try in PSpice.

...Jim Thompson
I found the problem in the LTspice circuit...the "generic" FET isn't

complete enough to work.
I added the IRFBG20 as best as i could to the FET library, using:
..model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

Now that a *specific* FET is there the simple model works!

 

[Robert Baer]

Jim Thompson wrote:

On Wed, 01 Jun 2005 01:37:35 GMT, Robert Baer
robertbaer@earthlink.net> wrote:


Mike Engelhardt wrote:


Robert,



The results are way off.

The simulation is probably right around correct.
That FET and diode have capacitance.


Adding a small capacitance has no effect.

But that doesn't remove the capacitance that's
already swamping the circuit.

And a pF or two cannot "swamp" anything here.

Not a pF or two. A 1.5nF for the MOSFET and 126pF
for the diode. It's hard to run a power SMPS down to
the tiny currents you're trying to do.

I had thought that i made it clear that i have


That wasn't clear to me because the circuit you
posted clearly does not work.



The models show *square waves* !!
The models show *AMPS* of current !!
The model does not show anything near reality!


The difference is that you are using a different
MOSFET in simulation then the bench. The simulation
results would agree with using a the MOSFET you have
on that schematic you posted. Now you're here talking
about different MOSFETs.

LTspice is showing what reality would look like for the
parts you have in the schematic. Your bench circuit is
entirely different. The schematic is totally shorted
out with capacitance, your bench circuit apparently
not. See the corrected schematic I posted with
capacitance-free switches.



The classic definition of an inductor is a (linear)
component that opposes a change of current.


And the simulation you posted does show that. But you
have to look at the current in the inductor and
the voltage across it. Neither the sense resistor
nor the switch current will follow that ramp, because
capacitive displacement currents swamp the inductor
current ramp.

--Mike



Oh, then no matter what inductor i use, no matter what the time scale
i use, that is the excuse?
Put a resistor in series with the inductor and see what that current
is; you will get the *same* square wave as every where else.
BTW, i originally used a FET from those available in the software
package, and not an ideal one - same result.
An ideal switch VS a slow, high capacitance FET like the IRFBG20
makes little difference on the workbench; current thru an inductor will
be a ramp - pure and simple.
My tests show the "slow" turnoff, wider flyback pulse with no "flat"
top when using the IRFBG20.
The same test fixture, with everything else the same - but using an
IRLZ24N shows a reasonably fast turnoff, a reasonably narrow flyback
pulse, and the classic flat top.
The model shows only square waves.
No ramp, no flyback pulse!
With an ideal switch, one cannot say that the inductor energy "got
swampped" to explain away the loss of a flyback pulse.
Be at least a little realistic...



Robert, Please send me copies of all the circuits you've tried; with a
short narrative for each.

...Jim Thompson
Gack!

May not be needed; see my earlier posting (to your earlier one)
labeled similar to "sank one sub".

..model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

 

[Robert Baer]

analog wrote:

Robert Baer wrote:


Errr...I have reduced the problem to the most simple possible; a
FET driving an inductor. *No* SMPS, just a basic circuit.
*Same* problem.


Robert, I looked at the text in your asc file that you posted to
start off this misbegotten thread and the simulation appears to
run for only 25us. Most of Linear Tech's models accurately model
start up delays, soft start, etc. I suggest you increase the run
time to several milliseconds and click "Skip initial operating point
solution" (set the UIC flag) in the simulation dialog box. Watch
the start up. It should be just like the real circuit if you've
done your job correctly.


I guess for this, i will have to scrap LTspice/switchcad3.


Your loss. I've *never* found a circuit LTspice couldn't simulate
realistically (and as well or better than any other simulator).

Please post your simplified LTspice asc file. I'm sure Helmut,
Mike, I or others could get it to work for you.

Regards -- analog
My simplified model uses no LT controllers, so that start-up business

is not present.
Just a simple pulse generator driving a simple FEt to switch an inductor.
Modified the generic FET, as i found that it does not work.

..model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

 

[Robert Baer]

Mike Engelhardt wrote:

Robert,


The results are way off.

The simulation is probably right around correct.
That FET and diode have capacitance.

Adding a small capacitance has no effect.

But that doesn't remove the capacitance that's
already swamping the circuit.

And a pF or two cannot "swamp" anything here.

Not a pF or two. A 1.5nF for the MOSFET and 126pF
for the diode. It's hard to run a power SMPS down to
the tiny currents you're trying to do.

I had thought that i made it clear that i have

That wasn't clear to me because the circuit you
posted clearly does not work.


The models show *square waves* !!
The models show *AMPS* of current !!
The model does not show anything near reality!

The difference is that you are using a different
MOSFET in simulation then the bench. The simulation
results would agree with using a the MOSFET you have
on that schematic you posted. Now you're here talking
about different MOSFETs.

LTspice is showing what reality would look like for the
parts you have in the schematic. Your bench circuit is
entirely different. The schematic is totally shorted
out with capacitance, your bench circuit apparently
not. See the corrected schematic I posted with
capacitance-free switches.


The classic definition of an inductor is a (linear)
component that opposes a change of current.

And the simulation you posted does show that. But you
have to look at the current in the inductor and
the voltage across it. Neither the sense resistor
nor the switch current will follow that ramp, because
capacitive displacement currents swamp the inductor
current ramp.


Oh, then no matter what inductor i use, no matter
what the time scale i use, that is the excuse?


No excuses, but you might apologize so getting so excited
about having the errors in hour schematic pointed out.
Anyway, the time scale is hard to vary because the 1619
wants to switch at 300KHz. But if you remove the
capacitance, then you get want you expect, like the
other schematic I posted.


Put a resistor in series with the inductor and see what that current is;
you will get the *same* square wave as every where else.


Eh, just click on the inductor to plot its current. There's
nothing wrong there. Drag the mouse across the inductor to
plot it's waveform. Corresponds to the dI/dt. Duh.


BTW, i originally used a FET from those available in the software package,
and not an ideal one - same result.


OK, the FET from the software package has capacitance modeled and
that that keeps your design as posted from working. It's just
telling you your design doesn't work. The ideal switch does
make the corrected version of your design convert power and
regulate.


An ideal switch VS a slow, high capacitance FET like the IRFBG20 makes
little difference on the workbench; current thru an inductor will be a
ramp - pure and simple.


Only if there's a square wave across it. With the 300KHz 1619,
freq, you're a bit pressed to see the ramp in a milliHenery
for the current spike of the highly non-linear Miller cap
of the FET in your posted schematic.


My tests show the "slow" turnoff, wider flyback pulse with no "flat" top
when using the IRFBG20.
The same test fixture, with everything else the same - but using an
IRLZ24N shows a reasonably fast turnoff, a reasonably narrow flyback
pulse, and the classic flat top.
The model shows only square waves.
No ramp, no flyback pulse!


It's a mistake to confuse flybacks with boost converters.


With an ideal switch, one cannot say that the inductor energy
"got swampped" to explain away the loss of a flyback pulse.
Be at least a little realistic...


I never said inductor energy gets swapped. I said the circuit
was swapped with capacitance as in AC shorted out. If you use
the capacitance-free switch you get square wave across the
inductor, a ramp in current through the sense resistor that
the 1619 can servo to regulate the output. You haven't run
the corrected version of your circuit that I posted. But with
the circuit as you posted, with a 1nF MOSFET and 100pF diode,
you won't see the ramp in the sense resistor because the current
is swamped by displacement currents.

The simulation is correct, your design as posted is the problem.

I can only tell what the problem is, you have to do a bit of effort,
say, running the corrected circuit as I posted.

--Mike


I have run every circuit you gave.

As i mentioned, i also used a simple test fixture with a 10uH
inductor, 3.3V supply, and "slow" FET and saw a ramp and a good-sized
flyback pulse - not the square wave stuff in the "1619" circuit.
A faster FET made for some speed improvements, but the basic ramp and
flyback pulse was still there.
And would still be there even with a mercury wetted relay as a switch.

The simple FET + inductor circuit needed to have the "generic" FET
specified; the "generic" does not work.
Used:

..model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

One sub sank.

 

[analog]

Robert Baer wrote:

My simplified model uses no LT controllers, so that start-up
business is not present. Just a simple pulse generator driving
a simple FEt to switch an inductor. Modified the generic FET,
as i found that it does not work.

.model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

Robert, I tried your model in your pulse circuit (posted previously
to JT higher in the thread) and it works exactly as expected. The
current ramps up to about 190ma at turn off after which the voltage
rings up to about 900 volts and back down to zero again. Perfect.

By the way (not that it makes any difference to your test case),
your model has unrealistically small resistances throughout (you
have milliohms where you probably meant to have ohms).

Here is your test schematic (slightly redrawn - mind the word
wrap in the model statement at the end):
~~~~~~~~~~~~~~~~~~~~~~~~ Bad_Baer.asc ~~~~~~~~~~~~~~~~~~~~~~~~~~~
Version 4
SHEET 1 1184 680
WIRE 0 128 0 112
WIRE 0 224 0 208
WIRE 0 320 0 304
WIRE 0 416 0 400
WIRE 64 112 0 112
WIRE 64 304 0 304
WIRE 80 304 64 304
WIRE 128 112 64 112
WIRE 128 128 128 112
WIRE 128 224 128 208
WIRE 128 240 128 224
WIRE 144 224 128 224
FLAG 0 224 0
FLAG 128 320 0
FLAG 0 416 0
FLAG 64 112 in
FLAG 144 224 D
FLAG 64 304 G
SYMBOL voltage 0 112 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 33 32 Left 0
WINDOW 3 33 80 Left 0
SYMATTR InstName V1
SYMATTR Value 10V
SYMBOL ind 112 112 R0
SYMATTR InstName L1
SYMATTR Value 1000ľH
SYMBOL nmos 80 224 R0
SYMATTR InstName M1
SYMATTR Value IRFBG20
SYMBOL voltage 0 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 34 32 Left 0
WINDOW 3 31 82 Left 0
SYMATTR InstName V2
SYMATTR Value PULSE(0 10 0 10n 10n 19u98 100u)
TEXT 262 154 Left 0 !.tran 0 25u 0 .1u uic
TEXT 264 208 Left 0 ;IRFBG20: Vds=1000 Ron=6 Qg=25n
TEXT 264 256 Left 0 !.model IRFBG20 vdmos(Rg=3 Vto=3.8 Rd=3.6 Rs=2.7 Kp=120 \n+ Cgdmax=0n7 Cgdmin=30p Cgs=0n9 Cjo=0n2 Is=68p Rb=5)

 

[Mike Engelhardt]

Robert,

I can only tell what the problem is, you have
to do a bit of effort, say, running the
corrected circuit as I posted.

I have run every circuit you gave.

And still don't see it working and still have trouble
with LTspice's solution.

Look, you've not been posting your schematic "work"
lately, so I can't fix any new problems you're introducing,
but going back to your original circuit, the one that
bothers you that there's no ramp in the inductor, and
assert that LTspice giving the wrong answer, just plot
the voltage across the inductor, I think V(n008,n009)
are the node numbers, but just drag the mouse across
the ends of the inductor to plot the voltage across it.
Then plot the inductor dI/dt with D(I(L1)). The waveforms
overlay, except for some number plotting issues from
LTspice plotting finite differences instead of the
derivative. LTspice is computing the correct solution
there. I can make it look like a ramp, but to do that,
I have to remove the capacitance that you've shorted
the thing out with. I have you two examples showing
ramps in the inductor.

Here's the deal: LTspice has corrected computed
the solution of every circuit posted in this thread.

You might need to get the guy who designed the circuit
you've got working on the bench to explain it to you.
I'm not getting through.

--Mike

 

[Mike Engelhardt]

Robert,

I'm busy, so it may take a few days, but I'll give it a
try in PSpice.

I found the problem in the LTspice circuit...the
"generic" FET isn't complete enough to work.
I added the IRFBG20 as best as i could to the FET library,
using: .model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m
Rb=5m Kp=120 Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p
Is=68p mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)

Now that a *specific* FET is there the simple model works!

Oh, that's a mistake I've seen before. It's not a "generic"
FET but the default FET. LTspice uses the same default FET
as other SPICE's. They're conventions to follow or you'll
mess up the people that know what they're doing. The default
SPICE FET is what you might have for one of those monolithic
FET's lithographed on an IC, not a power MOSFET.

But you never posted this error so we couldn't help you with
that. The circuit you posted has shorted out the circuit
with capacitance. The design simply did not work and LTspice
faithfully let you know it in no uncertain terms.

Regards,

--Mike

 

[Genome]

"Robert Baer" <robertbaer@earthlink.net> wrote in message
news:d8Lme.3891$MI4.26@newsread2.news.pas.earthlink.net...

The results are way off.
The sense resistor is for a peak current of 240mA, the time when the
gate is supposed to turn off.
The current in the 1mH inductor is supposed to be a linear ramp from
zero to 240mA, and take about 2.3uSec for an inductor supply of 100V
(8.9uSec for 30V, 5.0uSec for 50V and 1.0uSec for 200V). These values
are from a real circuit, and seem to be reasonably close to calculated
values using E=-L*(dI/dT).
The current never gets into the ampere region(!!).
Furthermore, when the FET turns off, there should be a large, narrow
voltage pulse (not some variable wierd-shaped and slow rise waveform).
What is wrong? Code follows:

You are an idiot, that's what is wrong.

What does C3 and D2 do in your circuit......?

You think it's some clever sort of voltage doubler type thing, don't you?

Ask yourself what happens when C3 has charged up to some voltage and you
switch the mosfet on.... That's right, D2 becomes forward biased and C3 gets
shorted out through D2 and the mosfet. You get a BIG spike on your current
sense resistor.

Fortunately the LT1619 ignores this because it has leading edge blanking.

So..... that pisses that one off.

You get the drain waveform you get because at start up you are charging the
output capacitors. Do you need a slap around the head to understand that
one.

Another one pissed off.

If you run your circuit for a while so things come into regulation you get
double pulses because you haven't bothered to compensate things properly,
you just pulled values off the data sheet. I'm sure you don't know how.
Probably won't work without a reasonable load anyway.

It's unstable.....

Your last one pissed off.

Let's face it, you design shit because you don't know shit. You also draw
shit circuits. I don't care if you've got something on the bench that works,
it might do but....... it's shit.

I'm the sort of poor sod who gets the job after you've been fucked off and
then I have to fix your shit which is in production and top bosswank, your
bum chum, cannot believe that there is more to it than changing the value of
RZZZ111A.

Tell me where you work so I don't have to go through it again.

And stop blaming your inadequacies on other people. That's my tactic.


Mike and Helmut have missed the bit about your concept of a voltage doubler
but the rest of the advice they are giving you is good. These are two
exceptionally tolerant (and clever) people.

I am not so...... Fuck You.

Here's my contribution...... I will warn you that I have waved my wet finger
in the air to get a result. If I was pushed then I'd do the job right and
provide supporting information.

Unfortunately you come across as a bleating wanker so I'm not going to waste
my time.

Have a crap day.

DNA

Version 4
SHEET 1 1416 740
WIRE -464 432 -464 -48
WIRE -464 544 -464 512
WIRE -464 576 -464 544
WIRE -272 0 -320 0
WIRE -272 32 -272 0
WIRE -272 192 -320 192
WIRE -272 192 -272 112
WIRE -272 224 -272 192
WIRE -272 432 -272 224
WIRE -272 544 -464 544
WIRE -272 544 -272 512
WIRE -160 -48 -464 -48
WIRE -160 32 -160 -48
WIRE -160 144 -160 112
WIRE -144 304 -320 304
WIRE -144 336 -144 304
WIRE -144 432 -144 400
WIRE -144 544 -272 544
WIRE -144 544 -144 512
WIRE -80 144 -160 144
WIRE -80 224 -272 224
WIRE -80 304 -144 304
WIRE 80 -48 -160 -48
WIRE 80 80 80 -48
WIRE 80 544 -144 544
WIRE 80 544 80 368
WIRE 304 -48 80 -48
WIRE 304 144 240 144
WIRE 304 144 304 -48
WIRE 352 432 352 -48
WIRE 352 544 80 544
WIRE 352 544 352 512
WIRE 528 -48 352 -48
WIRE 576 192 512 192
WIRE 576 224 240 224
WIRE 576 224 576 192
WIRE 576 272 512 272
WIRE 576 304 240 304
WIRE 576 304 576 272
WIRE 608 224 576 224
WIRE 720 224 688 224
WIRE 768 -48 608 -48
WIRE 768 96 512 96
WIRE 768 96 768 -48
WIRE 768 144 768 96
WIRE 768 304 576 304
WIRE 768 304 768 240
WIRE 768 432 768 304
WIRE 768 544 352 544
WIRE 768 544 768 512
WIRE 800 -48 768 -48
WIRE 896 -48 864 -48
WIRE 896 432 896 -48
WIRE 896 544 768 544
WIRE 896 544 896 496
WIRE 992 -48 896 -48
WIRE 992 432 992 -48
WIRE 992 544 896 544
WIRE 992 544 992 512
WIRE 1088 -48 992 -48
FLAG -464 576 0
FLAG 512 192 DRV
IOPIN 512 192 In
FLAG 512 96 DRAIN
IOPIN 512 96 Out
FLAG 512 272 ISNS
IOPIN 512 272 Out
FLAG 1088 -48 VOUT
IOPIN 1088 -48 Out
FLAG -320 0 VOUT
IOPIN -320 0 In
FLAG -320 192 VFB
IOPIN -320 192 Out
FLAG -320 304 VERR
IOPIN -320 304 Out
SYMBOL voltage -464 416 R0
WINDOW 0 -120 48 Left 0
WINDOW 3 -120 68 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName VSUPP
SYMATTR Value 15V
SYMBOL res 752 416 R0
WINDOW 0 -72 45 Left 0
WINDOW 3 -69 70 Left 0
SYMATTR InstName RSNS
SYMATTR Value 0R22
SYMBOL ind 512 -32 R270
WINDOW 3 70 65 VBottom 0
WINDOW 0 65 58 VTop 0
SYMATTR Value 1mH
SYMATTR InstName L1
SYMBOL voltage 352 416 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 38 41 Left 0
WINDOW 3 40 66 Left 0
SYMATTR InstName VBUS
SYMATTR Value 100V
SYMBOL cap 880 432 R0
WINDOW 0 42 24 Left 0
WINDOW 3 41 47 Left 0
SYMATTR InstName C2
SYMATTR Value 4n7
SYMBOL diode 800 -32 R270
WINDOW 0 59 30 VTop 0
WINDOW 3 62 37 VBottom 0
SYMATTR InstName D1
SYMATTR Value MUR460
SYMBOL PowerProducts\\LT1619 80 224 R0
WINDOW 0 -159 -162 Left 0
SYMATTR InstName U1
SYMBOL res -160 416 R0
WINDOW 0 -56 49 Left 0
WINDOW 3 -58 76 Left 0
SYMATTR InstName R3
SYMATTR Value 100K
SYMBOL cap -160 336 R0
WINDOW 0 -57 20 Left 0
WINDOW 3 -58 42 Left 0
SYMATTR InstName C1
SYMATTR Value 220p
SYMBOL res -288 416 R0
WINDOW 0 -74 47 Left 0
WINDOW 3 -76 71 Left 0
SYMATTR InstName R2
SYMATTR Value 12.4K
SYMBOL res -288 16 R0
WINDOW 0 -77 45 Left 0
WINDOW 3 -77 71 Left 0
SYMATTR InstName R1
SYMATTR Value 5E6
SYMBOL res -176 16 R0
WINDOW 0 -47 48 Left 0
WINDOW 3 -47 74 Left 0
SYMATTR InstName R4
SYMATTR Value 10K
SYMBOL res 976 416 R0
WINDOW 0 39 56 Left 0
SYMATTR InstName RLOAD
SYMATTR Value 100K
SYMBOL nmos 720 144 R0
WINDOW 3 56 61 Left 0
SYMATTR Value Si9420DY
SYMATTR InstName M1
SYMBOL res 704 208 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R5
SYMATTR Value 10R
TEXT -472 616 Left 0 !.tran 0 250u 10n 10n

 

[Mike Engelhardt]

Genome,

Mike and Helmut have missed the bit about your concept
of a voltage doubler but the rest of the advice they
are giving you is good. These are two
exceptionally tolerant (and clever) people.

I didn't miss it, I ignored it and changed his
circuit to straight boost in the interest of
simplicity(He had trouble with the relation between
dI/dt and inductance). But yes, you can put a
doubler there on a topology otherwise boost.

Thanks the post!

--Mike

 

[Robert Baer]

analog wrote:

Robert Baer wrote:


My simplified model uses no LT controllers, so that start-up
business is not present. Just a simple pulse generator driving
a simple FEt to switch an inductor. Modified the generic FET,
as i found that it does not work.

.model IRFBG20 VDMOS(Rg=3 Vto=3.8 Rd=3.6m Rs=2.7m Rb=5m Kp=120
Cgdmax=700p Cgdmin=30p Cgs=900p Cjo=200p Is=68p
mfg=International_Rectifier Vds=1000 Ron=6 Qg=25n)


Robert, I tried your model in your pulse circuit (posted previously
to JT higher in the thread) and it works exactly as expected. The
current ramps up to about 190ma at turn off after which the voltage
rings up to about 900 volts and back down to zero again. Perfect.

By the way (not that it makes any difference to your test case),
your model has unrealistically small resistances throughout (you
have milliohms where you probably meant to have ohms).

Here is your test schematic (slightly redrawn - mind the word
wrap in the model statement at the end):
~~~~~~~~~~~~~~~~~~~~~~~~ Bad_Baer.asc ~~~~~~~~~~~~~~~~~~~~~~~~~~~
Version 4
SHEET 1 1184 680
WIRE 0 128 0 112
WIRE 0 224 0 208
WIRE 0 320 0 304
WIRE 0 416 0 400
WIRE 64 112 0 112
WIRE 64 304 0 304
WIRE 80 304 64 304
WIRE 128 112 64 112
WIRE 128 128 128 112
WIRE 128 224 128 208
WIRE 128 240 128 224
WIRE 144 224 128 224
FLAG 0 224 0
FLAG 128 320 0
FLAG 0 416 0
FLAG 64 112 in
FLAG 144 224 D
FLAG 64 304 G
SYMBOL voltage 0 112 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 33 32 Left 0
WINDOW 3 33 80 Left 0
SYMATTR InstName V1
SYMATTR Value 10V
SYMBOL ind 112 112 R0
SYMATTR InstName L1
SYMATTR Value 1000ľH
SYMBOL nmos 80 224 R0
SYMATTR InstName M1
SYMATTR Value IRFBG20
SYMBOL voltage 0 304 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 34 32 Left 0
WINDOW 3 31 82 Left 0
SYMATTR InstName V2
SYMATTR Value PULSE(0 10 0 10n 10n 19u98 100u)
TEXT 262 154 Left 0 !.tran 0 25u 0 .1u uic
TEXT 264 208 Left 0 ;IRFBG20: Vds=1000 Ron=6 Qg=25n
TEXT 264 256 Left 0 !.model IRFBG20 vdmos(Rg=3 Vto=3.8 Rd=3.6 Rs=2.7 Kp=120 \n+ Cgdmax=0n7 Cgdmin=30p Cgs=0n9 Cjo=0n2 Is=68p Rb=5)
In "badbaer.asc", i see that the delay in turnoff of the FET (due to

capacitance) is fairly closed to the observed amount.

 

[Robert Baer]

Genome wrote:

"Robert Baer" <robertbaer@earthlink.net> wrote in message
news:d8Lme.3891$MI4.26@newsread2.news.pas.earthlink.net...

The results are way off.
The sense resistor is for a peak current of 240mA, the time when the
gate is supposed to turn off.
The current in the 1mH inductor is supposed to be a linear ramp from
zero to 240mA, and take about 2.3uSec for an inductor supply of 100V
(8.9uSec for 30V, 5.0uSec for 50V and 1.0uSec for 200V). These values
are from a real circuit, and seem to be reasonably close to calculated
values using E=-L*(dI/dT).
The current never gets into the ampere region(!!).
Furthermore, when the FET turns off, there should be a large, narrow
voltage pulse (not some variable wierd-shaped and slow rise waveform).
What is wrong? Code follows:



You are an idiot, that's what is wrong.

What does C3 and D2 do in your circuit......?

You think it's some clever sort of voltage doubler type thing, don't you?

Ask yourself what happens when C3 has charged up to some voltage and you
switch the mosfet on.... That's right, D2 becomes forward biased and C3 gets
shorted out through D2 and the mosfet. You get a BIG spike on your current
sense resistor.

Fortunately the LT1619 ignores this because it has leading edge blanking.

So..... that pisses that one off.

You get the drain waveform you get because at start up you are charging the
output capacitors. Do you need a slap around the head to understand that
one.

Another one pissed off.

If you run your circuit for a while so things come into regulation you get
double pulses because you haven't bothered to compensate things properly,
you just pulled values off the data sheet. I'm sure you don't know how.
Probably won't work without a reasonable load anyway.

It's unstable.....

Your last one pissed off.

Let's face it, you design shit because you don't know shit. You also draw
shit circuits. I don't care if you've got something on the bench that works,
it might do but....... it's shit.

I'm the sort of poor sod who gets the job after you've been fucked off and
then I have to fix your shit which is in production and top bosswank, your
bum chum, cannot believe that there is more to it than changing the value of
RZZZ111A.

Tell me where you work so I don't have to go through it again.

And stop blaming your inadequacies on other people. That's my tactic.


Mike and Helmut have missed the bit about your concept of a voltage doubler
but the rest of the advice they are giving you is good. These are two
exceptionally tolerant (and clever) people.

I am not so...... Fuck You.

Here's my contribution...... I will warn you that I have waved my wet finger
in the air to get a result. If I was pushed then I'd do the job right and
provide supporting information.

Unfortunately you come across as a bleating wanker so I'm not going to waste
my time.

Have a crap day.

DNA

Version 4
SHEET 1 1416 740
WIRE -464 432 -464 -48
WIRE -464 544 -464 512
WIRE -464 576 -464 544
WIRE -272 0 -320 0
WIRE -272 32 -272 0
WIRE -272 192 -320 192
WIRE -272 192 -272 112
WIRE -272 224 -272 192
WIRE -272 432 -272 224
WIRE -272 544 -464 544
WIRE -272 544 -272 512
WIRE -160 -48 -464 -48
WIRE -160 32 -160 -48
WIRE -160 144 -160 112
WIRE -144 304 -320 304
WIRE -144 336 -144 304
WIRE -144 432 -144 400
WIRE -144 544 -272 544
WIRE -144 544 -144 512
WIRE -80 144 -160 144
WIRE -80 224 -272 224
WIRE -80 304 -144 304
WIRE 80 -48 -160 -48
WIRE 80 80 80 -48
WIRE 80 544 -144 544
WIRE 80 544 80 368
WIRE 304 -48 80 -48
WIRE 304 144 240 144
WIRE 304 144 304 -48
WIRE 352 432 352 -48
WIRE 352 544 80 544
WIRE 352 544 352 512
WIRE 528 -48 352 -48
WIRE 576 192 512 192
WIRE 576 224 240 224
WIRE 576 224 576 192
WIRE 576 272 512 272
WIRE 576 304 240 304
WIRE 576 304 576 272
WIRE 608 224 576 224
WIRE 720 224 688 224
WIRE 768 -48 608 -48
WIRE 768 96 512 96
WIRE 768 96 768 -48
WIRE 768 144 768 96
WIRE 768 304 576 304
WIRE 768 304 768 240
WIRE 768 432 768 304
WIRE 768 544 352 544
WIRE 768 544 768 512
WIRE 800 -48 768 -48
WIRE 896 -48 864 -48
WIRE 896 432 896 -48
WIRE 896 544 768 544
WIRE 896 544 896 496
WIRE 992 -48 896 -48
WIRE 992 432 992 -48
WIRE 992 544 896 544
WIRE 992 544 992 512
WIRE 1088 -48 992 -48
FLAG -464 576 0
FLAG 512 192 DRV
IOPIN 512 192 In
FLAG 512 96 DRAIN
IOPIN 512 96 Out
FLAG 512 272 ISNS
IOPIN 512 272 Out
FLAG 1088 -48 VOUT
IOPIN 1088 -48 Out
FLAG -320 0 VOUT
IOPIN -320 0 In
FLAG -320 192 VFB
IOPIN -320 192 Out
FLAG -320 304 VERR
IOPIN -320 304 Out
SYMBOL voltage -464 416 R0
WINDOW 0 -120 48 Left 0
WINDOW 3 -120 68 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName VSUPP
SYMATTR Value 15V
SYMBOL res 752 416 R0
WINDOW 0 -72 45 Left 0
WINDOW 3 -69 70 Left 0
SYMATTR InstName RSNS
SYMATTR Value 0R22
SYMBOL ind 512 -32 R270
WINDOW 3 70 65 VBottom 0
WINDOW 0 65 58 VTop 0
SYMATTR Value 1mH
SYMATTR InstName L1
SYMBOL voltage 352 416 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
WINDOW 0 38 41 Left 0
WINDOW 3 40 66 Left 0
SYMATTR InstName VBUS
SYMATTR Value 100V
SYMBOL cap 880 432 R0
WINDOW 0 42 24 Left 0
WINDOW 3 41 47 Left 0
SYMATTR InstName C2
SYMATTR Value 4n7
SYMBOL diode 800 -32 R270
WINDOW 0 59 30 VTop 0
WINDOW 3 62 37 VBottom 0
SYMATTR InstName D1
SYMATTR Value MUR460
SYMBOL PowerProducts\\LT1619 80 224 R0
WINDOW 0 -159 -162 Left 0
SYMATTR InstName U1
SYMBOL res -160 416 R0
WINDOW 0 -56 49 Left 0
WINDOW 3 -58 76 Left 0
SYMATTR InstName R3
SYMATTR Value 100K
SYMBOL cap -160 336 R0
WINDOW 0 -57 20 Left 0
WINDOW 3 -58 42 Left 0
SYMATTR InstName C1
SYMATTR Value 220p
SYMBOL res -288 416 R0
WINDOW 0 -74 47 Left 0
WINDOW 3 -76 71 Left 0
SYMATTR InstName R2
SYMATTR Value 12.4K
SYMBOL res -288 16 R0
WINDOW 0 -77 45 Left 0
WINDOW 3 -77 71 Left 0
SYMATTR InstName R1
SYMATTR Value 5E6
SYMBOL res -176 16 R0
WINDOW 0 -47 48 Left 0
WINDOW 3 -47 74 Left 0
SYMATTR InstName R4
SYMATTR Value 10K
SYMBOL res 976 416 R0
WINDOW 0 39 56 Left 0
SYMATTR InstName RLOAD
SYMATTR Value 100K
SYMBOL nmos 720 144 R0
WINDOW 3 56 61 Left 0
SYMATTR Value Si9420DY
SYMATTR InstName M1
SYMBOL res 704 208 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R5
SYMATTR Value 10R
TEXT -472 616 Left 0 !.tran 0 250u 10n 10n


Look at the 80 to 90 microsecond region; I(l1) and Id(M1) look

reasonable, but V(Isns) looks rather trashy and incorrect.
The drain current in the FET should show up in the source....

 

[legg]

On Mon, 06 Jun 2005 08:35:52 GMT, Robert Baer
<robertbaer@earthlink.net> wrote:

Look at the 80 to 90 microsecond region; I(l1) and Id(M1) look
reasonable, but V(Isns) looks rather trashy and incorrect.
The drain current in the FET should show up in the source....

....along with 'some' of the gate current and anything forced in
parallel with DS, when un-enhanced.

It's possible for a model to be more 'accurate' than your observed
scope traces, depending on your selection of model component strays
and the actual physical measurement method used.

RL

 

[xray]

On Mon, 06 Jun 2005 08:35:52 GMT, Robert Baer <robertbaer@earthlink.net>
wrote:

Look at the 80 to 90 microsecond region; I(l1) and Id(M1) look
reasonable, but V(Isns) looks rather trashy and incorrect.
The drain current in the FET should show up in the source....

Robert,

Just curious what caused you to stop posting in this thread?
Did Jim T get back to you about PSpice, or did you have a flash of
insight, or did you just get tired?

Curious minds want to know.

 

[Robert Baer]

xray wrote:

On Mon, 06 Jun 2005 08:35:52 GMT, Robert Baer <robertbaer@earthlink.net
wrote:


Look at the 80 to 90 microsecond region; I(l1) and Id(M1) look
reasonable, but V(Isns) looks rather trashy and incorrect.
The drain current in the FET should show up in the source....



Robert,

Just curious what caused you to stop posting in this thread?
Did Jim T get back to you about PSpice, or did you have a flash of
insight, or did you just get tired?

Curious minds want to know.

Somewhen i mentioned that after i had a "real" FET model, that Spice

and bench measurements were in reasonable agreement.

 

[Robert Baer]

Mike Engelhardt wrote:

Robert,


The results are way off.

The simulation is probably right around correct.
That FET and diode have capacitance. .

Adding a small capacitance has no effect.


But that doesn't remove the capacitance that's already
swamping the circuit. The problem is the circuit design,
not the models.
There is no problem with the design.

Both theory and practice show that when a voltage is presented across
an inductor, that the current starts at zero and increases in the
standard L/R exponential "decay" to maximum, the assumption is a linear
inductor. Classical definition of an inductor!
The problem, is that the model does *not* show that.
And a pF or two cannot "swamp" anything here.

I will try your 1619 circuit later


That one works. Notice that that SMPS topology is
a boost. To get higher output voltage, you might try
a tapped-inductor boost. That will reduce the voltage
rating required of the MOSFET and make the circuit
less sensitive to its capacitance. If you're not really
after a SMPS, but some high impedance high voltage,
you can probably get a chopped square wave working
into a diode voltage multiplier to work. That gets rid
of the problem of getting high impedance energy
storage inductors with little stray capacitance.

-Mike


I think i mentioned the word "autoformer" somewhere; have a

Ferroxcube 14/8-3C81 core with 56 turns pri and 129 total (secondary),
and get about 1800V DC with about 58Meg load.
Inductor supply can run from about 5V to 200V (bypass cap rating);
looks like i could get up to a 400V input.

I tried the square wave idea (variable amplitude, variable frequency)
idea with a CFL transformer.
Worked fine, efficency improved as i drove it from 1/7, to 1/5, to
1/3 and at "resonance".
Those transformers are very lossy at "high" frequencies; resonance
near 300KHz for the Cooper CTX210407 and two other randomly chosen CFL
xfmrs.
Input square wave, output slightly clipped sinewave; re-winding the
primary drive made no major difference, except with fine wire and many
turns, i saw I*I*R losses (efficency went from 30 percent area to 16
percent).

BUT. Again, there was *no* instant current spike in the amps; there
is clearly an error in the modeling, and i do not see it.