Overshoot fix 8

[itsmoked]

I have a circuit here that essentially minimally amplifies a voltage coughed up by a DAC. It's a non-inverting amp that has a wee bit of gain because the DAC is only good for 4.096V and the output needs to reach 10V. Meanwhile I need to also PWM the output. Hence the FET that yanks the input to 0V without bothering the DAC.

It all works as I desire with the exception of the overshoot seen in my sketch. Does anyone have a simple solution to it?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[electricpete]

I haven't messed with electronics in quite awhile and I'd have to stop and think to figure out the equation for gain of op amp.

To keep my head from hurting thinking about the Op Amp, my very crude thinking is to put a cap on the non-inverting input to create an input low-pass filter with time constant R*C where R = 2.5 kohms.

Just a thought... I'm sure you'll treat it with due grains of salt.

=====================================
(2B)+(2B)' ?

 

[electricpete]

The cap of course would be connected from non-inverting input to ground

=====================================
(2B)+(2B)' ?

 

[electricpete]

Better yet, from the "+" sign to ground. (I'm not too sure of my terminology)

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

There seems to be some charge injection from gate to Drain and from there to the N.I. input of the opamp.

Try reducing the gate drive. Or, better, use a simple 4066 analogue gate. They have very low charge transfer. Four times overkill, but you are in a generous X-mas mode. Aren't you?

Of course, you can slow down the opamp by adding a capacitor parallell to the feed-back resistor. But you want a sharp, clean edge, I think.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

How about a 10 Volt zener across the output??

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

A variable zener would then be needed since voltage range is 0 - 10 V and I guess that Smoked wants that "overshoot" (which it isn't, technically, but still) to go away at all levels.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[VEBill]

"...charge injection from gate to Drain..."

Could this be fixed by a resistor from gate to ground?

 

[waross]

Thanks Gunnar. I misunderstood the problem.
mmm Variable zenner. May be a good project for early in April.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[ijl]

Some guesses, educated or not. You do not mention the frequency or rise time of your signal. But the output looks like as if the feedback was "too slow". The feedback path should be low-inductance and low-capacitance, maybe even low-resistance. Initially, you might try with resistances in the 100 ohm range. Better still, use some high-frequency amplifier, no op-amp at all.

Another question (where I am guessing even more) is what happens, when the output signal approaches the maximum value available, about Vcc-1.5V. Strange things may happen. A higher supply voltage, say 15 V might help, maybe.

 

[Skogsgurra]

ijl
The resistors are fine. No risk that a 358 would show erratic behaviour caused by parasitic inductance (or capacitance). Also, it cannot supply enough current for 100-ohmish resistors.

Bill: Start a write-up on those zeners! Editors need long time to go from e-mail to publishing.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Comcokid]

I think Skogsgurra suggestion of charge injection is it, and if it wasn't for the fact that your op-amp negative rail is ground, you would probably see a similar spike at the negative transision. The correction is to use a analog switch IC, and lots of choices are available. If there is still a small amount of overshoot after you address the switch issue, put a small capacitor across your 7.3K resistor. Maybe between 100pF and 1nF which will kill-off some of your HF gain.

 

[ijl]

Gunnar,
I did not think the behaviour of the amplifier itself, but that the feedback path with the stray capacitances etc. together with the input capacitance may act as a low-pass filter. This together with the delay in the amplifier makes the feedback "slow". If the signal is fast enough, the output may overshoot before the feedback has time to counteract. Ok, 100 ohm is too small, but 1 to 2 kOhm might show some difference in the output.

 

[Skogsgurra]

Sorry. I don't think that computes either. Or, to put it less bluntly: How would that work?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[mikekilroy]

cap around 7.3k gain resistor to slow op amp response to remove the overshoot? too big and rather than overshoot you are underdamped?

 

[mikekilroy]

ur fet pulls input lo and just lets go of it FAST - seems like it is a simple case of your freq response of letting it go is too fast for the skew rate of your op amp. perhaps get a faster (higher bandwidth) op amp that does not saturate causing the overshoot from this high freq input change, or at lo pass filter around the op amp to slow it down to eliminate the overshoot (add cap around 7.3k R)

 

[Skogsgurra]

As you know, Smoked usually tells the truth. And is also true to details.

If this were a bandwith or marginal stability problem, he would have shown that in his waveform sketch with a damped ringing - or at least a classical ramp-up_overshoot_settle waveform. He doesn't.

Instead, he shows a typical response when injected charge causes an overvoltage, which then dies out more or less linearly as the 1+R2/R1 gain of the opamp comes into play. Which is a very clear indicator that this has nothing to do with resistors or parasitic phenomena other than the C_gd capacitance in the transistor.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[itsmoked]

Wow great stuff folks. Many thanks on all your considered contributions.

I built this thing on a perf-board with all short trimmed thru-hole parts flying on the top. Not much parasitic anything.

Sorry I didn't mention the PWM period. tsk tsk bad form -5pts. It's 200Hz. I'll be running it from an unsuspecting microcontroller via interrupts in it's spare time so I don't want the PWM service to be too fast.

The output slews very nicely as is at about 1V/us. This helps with EMI.

A well behaved four channel analog switch is a great idea! Especially since there will be four channels of the above circuit!!

As for the 100ohm output, that isn't the load. It's only to provide rudimentary protection to the opamp's output pin. The load will be about 2mA max(500uA typical). The 0-10V need not be precise. I'll be using 20mA PTC fuses with 100ohms cold value in lue of resistors. The LM358 is good for about 50mA of drive but I have no control of who hooks what to this thing.

I'll let y'all know if a cap across Rf works or less base-drive or an analog switch kills dear Spike. (My money is on the analog switch though the reduced base drive is the easiest to test.

Stay tuned.


Keith Cress
kcress -

http://www.flaminsystems.com

 

[mikekilroy]

itsmoked, sorry for somehow calling you a liar; it was by no means my intention. Guess I best stay in the back room hidden from other people with my apparent stereotypical engineer mentality (and incompetent one of those too with my poor suggestion) that causes what I write to come out as calling folks liars. Sorry.

 

[Skogsgurra]

I don't think that anyone gets the idea that Smoked was lying. He always sez the truth. But that truth may have been misinterpreted on a few hands.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.