Simulation of AD284/AD-Effect of open loop on gain

[stanislavb]

PSpice

Hi,
I tried simulate OP AD284/AD and got strange results. For investigation to GND and got output Vout ~ Vs.
Do you have explanation for it?
Thanks

 

[keith1200rs]

PSpice

Not strange at all. Any tiny offset will be amplified by the open loop gain & smack the output to one supply rail or the other. What were you trying to find out?

Keith

 

[stanislavb]

Re: PSpice real OP

Hi,
I expected this answer. Now I built invertor based on this OP. For compare I also built invertor on ideal OP. Of course, ideal work fine(see attached file). However real is not. I presume I miss something but don't undersatnd what.
Thank you

 

[keith1200rs]

PSpice

I will check later when I am near a computer, but you are trying to get a lot of current out of the opamp - are you sure it can deliver it?

Try increasing all you 1k resistors to 100k.

Keith

 

[LvW]

Re: PSpice real OP

Hi,
I expected this answer. Now I built invertor based on this OP. For compare I also built invertor on ideal OP. Of course, ideal work fine(see attached file). However real is not. I presume I miss something but don't undersatnd what.
Thank you

Recommendation: Check the polarities of the input pins from the real opamp.

 

[stanislavb]

Re: PSpice

Keith,
I did it. The graph changed but it isn't what expected.
Thank you

Added after 6 minutes:

LvW,
I reversed polarity and got expected result-inversed pulse. But what is the idea behind it?For ideal OP the result is the same for both polarities. Morover I built invertor "by the book"
Thank you

 

[LvW]

Re: PSpice

LvW,
I reversed polarity and got expected result-inversed pulse. But what is the idea behind it?For ideal OP the result is the same for both polarities. Morover I built invertor "by the book"
Thank you

* The idea behind what? (Your first schematic has shown positive instead of negative feedback!)
* Recommendation: Remove the single 1Meg resistor, or lower it! What's its purpose?
(For bias current compensation choose app. the same order as feedback resistors)
* An input step with 10volts is too high using 10 volt supply voltages.
*Be careful with ideal opamp models! They tell you not always the truth!

 

[keith1200rs]

PSpice

A DC sweep is a better way to start than a transient analysis.

Keith

 

[stanislavb]

Re: PSpice

Keith,
Attached sweep analysis. It's looks fine except region near 9.5V
But I tried pulse with amplitude below 5V and got the same negative result
Thank you

 

[stanislavb]

Re: PSpice

LvW,
-You are right about feedback. However I wanted to check real op with schematics based on theoretical basis of Sedra(Microelectronics circuit)-see attached file Fig 2.5.
-I tried without 1M resistor and use the similar value as feedback resistor without success
- I used lower(even less than 1 volt) amplitude without success
- I agree with you about ideal OP. However theory "by the book "show invertor should work!
Thank you

 

[stanislavb]

Re: PSpice

Hi,
Additional information. When I changed rise and fall time of the pulses I got inversion. See attached files. Especially it concerning rise time.
Any idea?
Thank you

 

[keith1200rs]

Re: PSpice

OK, I have managed to get a quick look at the circuit. I think there is something wrong with your circuit. A DC sweep of your circuit is attached and gives the results I would expect. Could you post your netlist? It should look something like the one attached. It may have a .cir or .net extension (.cir is more normal for Pspice, I think).

Another point is that while I suggested 100k rather than 1k due to loading the opamp output, 100k is high enough to cause instability due to the opamp input impedance, so practical values may need to be smaller. I have also attached the DC simulation with a 1k load - again it is what I would expect - not quite reaching the rails on the output and sinks better than it sources.

Keith.

 

[stanislavb]

Re: PSpice

Keith,
Attached netlist I combined from netlist file and simulation output file.
I found if I change pulse parameters it lead to expected result. From my opinion or this is problem of OP model or PSpice rev.9.2.3, or PSpice configuration(may be):
What do you think about?
About 100k you are right. But it doesn't fix the problem for transient simualtion.
By the way I did not see yours transient simulation result. As you remember my sweep simulation was correct also. Try to use short pulses TR=TF=1ms and PW=3ms
Thank you

 

[keith1200rs]

PSpice

I did a pulse as well and it produced a square wave output but very ringy. That is how I discovered the instability with 100k. Dropping the 100k got it down to a nice square wave. I will look when I am back at my computer. One difference I noticed with your simulation is you are using the E version of the opamp. I will check later.

Keith

Added after 1 minutes:

Edit:

I used very fast rise & fall - yours are very slow.

Keith

 

[LvW]

Re: PSpice

LvW,
-You are right about feedback. However I wanted to check real op with schematics based on theoretical basis of Sedra(Microelectronics circuit)-see attached file Fig 2.5.

- I agree with you about ideal OP. However theory "by the book "show invertor should work!
Thank you

Hello stanislavb !

Obviously, the left Fig. in SEDRA's book has a wrong polarity!
Don`t believe blindly everything you see or read - even if the author is Sedra.
Certainly, it's a printing error as it is mentioned that an "inverting" amplifier is shown. This cannot be true when the positive input pin is connected to the signal source.

 

[keith1200rs]

PSpice

LvW - well spotted. I never noticed the reversed +/- inputs. I drew out the circuit and simulated it and it worked, but I automatically put the feedback as negative!

Keith.

 

[stanislavb]

Re: PSpice

Hi,
I pointed out it(succeed if feedback connected to - input) in one of my mails. However interesting know what is the difference. From theoretical point of view it should be the same. Simulation of ideal OP confirm it. Also basic analysis don't see difference between - and + input(assumption that input current much lower then feedbak):
Vin/R1=-Vout/R2=> Vout=-Vin*R2/R1; R1=R2=> Vout=-Vin. And it should not depends from + or - input. (Ideal OP shows it)However for real OP eventually this is not always true. For me interesting, what is the difference. (Even if Sedra did mistakes )

 

[LvW]

Re: PSpice

stanislavb, you picked up a very important and interesting point!

And you have learnt a lot. What? Don't trust simulation results blindly!

And you are right, in some cases the use of an ideal opamp model can lead to simulation results - especially in case of an ac analysis - which are not realistic.
I hesitate to say "wrong", because the simulator has calculated correctly. And you can verify this by hand calculations.

Explanation: For an ideal opamp (without internal delay elements) the simulator finds an operating point and can calculate the gain - even when there positive feedback.
Why? Because the machine assumes that there is no power switch-on transient and there is absolutely no noise. That is something like balancing a small ball on a larger one. But, as soon as there is a tiny disturbance......(like noise or switch-on).
Do a transient analysis with a simple opamp model which incorporates at least one pole - and you will see that the circuit goes into saturation.

Believe me, in reality you always must apply NEGATIVE feedback for amplifiers.
If you see a circuit with positive feedback (without any additional negative feedback path) it is no amplifier!

 

[stanislavb]

Re: PSpice

LvW,
So if I understand right, you means feedback on + input aciting in the same way as positive feedback(unwilling of course) in amplifier theory? If yes, the phase signal from feedback should be the same as input signal- 0° or 360° (if I remeber well feedback theory), isn't it?
Thank you

 

[LvW]

Re: PSpice

No, not in general. It depends on the kind of simulation.
In detail: An ac analysis for a circuit with pos. feedback will show the correct magnitude response, but the phase response will show a rising characteristic.
In contrary, for neg. feedback it will always show a phase which goes more negativ.
Thus, you have the chance to see if a circuit has - without intention - positive feedback.