Oscillation on opamp output with P-mosfet

[gavron04]

OPA1611 without C15 1nF - between IN- and OUT of OPAMP, measured in "TP-PRZEWOD"



Schematic1


Schematic2 - supply


OPA1611 + 1nF between IN- and OUT of OPAMP, measured in "TP-PRZEWOD"


PCB layout


Output of OPAMP

I have tested:
 - 3 differents OPAMPs - TL071; OPA277; OPA1611; uA741
 - tested differents capacitor between IN- and OUT of opamp (22pF;33pF;47pF;100pF;1nF;10nF;100nF)
 - tested mosfet driven directly by OPAMP and via NPN transistor - there are no differences between circuits. Always oscillations,
 - tested R2 1k; 10k
 - R5 - tested 47R, 100R

About circuit. Its tool for testing laboratory cables (banana plug).

X1 - its connector for power supply. 5V 16A. Aimted DC/DC 5-12/12 for opamp supply.

C14 - 1000uF

CSR-2.0-R010 its shunt resistor - 10mΩ.

PMosfet - IXTH120P065T.

R_I - selectable resistor 10R;50R;80R;100R - 1A;5A;8A;10A testing current.

PR1 - calibration 1mA across T1.


Its problem with OPAMP, layout or...?

Have a good day


OPA1611 + 1nF between IN- and OUT of opamp. Measured in TP-PRZEWOD


OPA1611 + 1nF between IN- and OUT of opamp. Measured in TP-PRZEWOD

[Wimberleytech]

Maybe the additional phase shift through the emitter follower plus the source follower in a unity gain configuration is enough to cause oscillation.

[Kevin.D]

And does the oscillation stop with a resistive load. ?
If so It's probably stray inductance in the leads you're testing (typically a few uH/meter), current sources tend to have a hard time with primarily inductive loads since
inductance resists any change in current --> creates a phase lag in a current controlled feedback loop --> you have an oscillator .
One solution is a damping network across the inductance as you already have here with C11 and R34. I would say the values are wrong though for the type
of loads your testing (only leads with banana plugs -> low resistance and primarily inductive above some frequency ). Try values around  4u7 for C11 and 2r2 for R34,
for unconditional stability best R and C values depends on MOSFET Cgd and a few other variables (more Cgd means larger C lower R damping values generally)
with those present then R2/C15 can be reduced.

Regards

[xavier60]

Oscillations don't usually look like that. There could be an external source.

[gavron04]

@Kevin.D
Try values around  4u7 for C11 and 2r2 for R34
>> tried, didnt help

And does the oscillation stop with a resistive load. ?
>> never stop, only when I disconnect testing cables from banana connector. Then opamp on output has 12V.

@xavier60 - maybe you are right. See photo.
1) Power supply is turned ON, output - NO, so my pcb has no power
2) On osciloscope you can see same oscillation which I have on output of opamp and on TP-przewod.
Probe is connected on shunt resistor.

On this video you can see what power supply can do - output is off. Probe of osciloscope is connected to TP-przewod:
https://youtu.be/WpYM72BdQmo


I have disassembly aimtec dc/dc converter and pluged linear power supply on 317/337. Its too much better than on aimtec.

Amplitude is now about 400uV, before on aimtec was 7mV!

Also now scale is 1mV/div.

[Vovk_Z]

What is a control voltage looks like, at opamp input, or at T1-R3 connection?

[gavron04]

//now Im using linear power supply for opamp its better than aimtec//

I have changed shunt resistor for 0R22, also for 1A test I have changed RI value

U=I*R 1A*0,22R=0,22V

R=U/I , 0,22V/0,001A = 220R

so for 1A, RI has 220R.


In attachments:
- simulation circuit from LTSpice
- osciloscope probe in TP-Przewod
- osciloscope probe in output of OPAMP
- and T1-R3

[gavron04]

Here you can see effect of: LED light; KSGER T12


(youtube links isnt recognized on forum??)

here - power supply output is off, Im powering on/off AC voltage:


here testing power supply:


I think that is not oscillation, only quality of my electrical supply.

[Vovk_Z]

You have a good possibility to dive into a world of common and differential mode noises.

[gavron04]

But for me it isnt desirable

[homebrew]

Maybe a detail ...
Why do you have a split ground plane like that? Especially with respect to C14. Its current must pass entirely around the split to reach the other pin of X1.

[gavron04]

My friend told me that it will better.

I have tested C14 connected to X1 - nothing change

I have tested to connect GND of banana laboratory cable to X1 GND pin - nothing change.

So its not problem of layout

[xavier60]

A useful test is to see how much noise is seen when the DSO's probe is touch to the same place as the probe's ground lead is connected.

[gavron04]

I have tested this. Probe and probe ground is connected to R26 (shunt resistor) from +5V and aimtec on PCB.

[xavier60]

If that's 1mV/div, there is nothing to worry about.

[homebrew]

I'm just a hobbyist, but I see so many things in that schematic that raise my eyebrows ...

- Why use a precision opamp without its offset compensation and with a questionable reference voltage source?
- Your whole control loop seems to depend on the stability of the supply voltage.
- Why use a P-MOSFET and go though all the troubles with negative voltages?
- What happens when you switch the ranges? Could the wiper loose contact at any point in time? If so, your PMOS transistor would fully open and fry everything! At least it blows the fuse ...
- Read the datasheet of your DC-DC converter (if you REALLY need it). It makes special EMC recommendations.
- What is the reason for R35?
- Why have this transistor inverter around T2?
- Layout - As noted before, you did create yourself an unnecessary huge current loop running the entire width of the board. You do cut ground planes to control where the current is flowing, not to make matters worse.

I think that it is MUCH too complex for that simple task. If I understand correctly, you just want to run 1,2,...,10A through a cable an measure the voltage drop reliably ...

[gavron04]

Thanks for your reply.

1) Thats good question. I have forget to control offset of OPAMP. Why used precision OPAMP? Better performance - better opamp, but as I see - not this time. Now Im using OPA1611.
2) Probably yes.
3) In LTSpice all looks good, but on real layout it isnt. Maybe better option will be BJT transistor like 2SC5200
4) Im using BBM rotary switch - there is no problem when Im changing current (1;5;8;10A). CK1051. Should be MMB,  I fu*k up this. Now I see.
5) http://www.aimtec.com/method-to-reduce-the-output-ripple-noise-of-power-supplies its better shown than datasheet. But im staying with linear power supply for opamp
6) Its my fault - on PCB I havent soldered this resistor.
7) T2 was for test, but situation was same. Mosfet driven by opamp (directly via 47R) vs Mosfet driven by NPN transitor. I have disassembled T2, now Im driving mosfet directly by opamp, via 1k resistor.
I will jumper this cut. But as I wrote. I have connected directly GND banana lead to X1, and nothing happen.

9) you just want to run 1,2,...,10A through a cable an measure the voltage drop reliably ... - in simple words, yes it is.

[Vovk_Z]

It seems to me now I get a part of your problem. We have 'a trap for a young players situation' here:
You should turn on 20 MHz bandwidth limit on your scope.
We usually do things with power supplies with this '20BW' bandwith mode. (And all parameters normally are stated for 20MHz BW). And don't turn on a 'Full BW' mode until you really need a full bandwith.

And you should usually work with a 10:1 divider switched on (otherwise you will not have good bandwidth). Again, work with 1:1 mode onlly when you really need it.
So if a lowest voltage range was 1 mV/div you'll have 10 mV/div, and thus much lower noise on a screen.

[gavron04]

Thats a good point..

Thanks for your reply. Today I will post new measurement.

Have a good day