I mean, gnd lines have resistance and inductance. Mess it up and you create feedbacks that can cause oscillations.
Jan
Jan
Just measured temp with no signal applied. It went up 19-20 deg C from ambient. When I applied a signal it went up another 4 deg. So I still don’t know if that’s normal/acceptable or an indication of oscillation. Every few minutes the REW rta display does this,
If you measure the DC voltage across R1 (100k) in your diagram then you can calculate the current. 100 nanoamps for example would give a DC voltage of 10 millivolts across the resistor. The NE5532 if genuine will be the one with the highest current out of the devices you list. Those current should be negligible in the scheme of things though.
@jan.didden sorry I misunderstood. Can you suggest a “best practice” on a point to point wiring situation on a small solderable breadboard?
So measured dc voltage across what is now a 10k resistor from + input to ground. It wanders very quickly from .02 to .07 mvdc. So say .05 would then yield .000005 ma
0.05 millivolts across 10k is just 0.005 microamps. What was the opamp used for that test?
That current is so small that I think it has to be a FET device you are using.
That current is so small that I think it has to be a FET device you are using.
@Mooly wanted to clarify. DC Voltage across +input resistor to ground, powered on;
Nothing connected —— 1.2v
DAC only connected —— 0v (20 ohms output)
ADC only connected —— 1.2v (350 ohms input)
DAC and ADC connected —— .65mv
DAC and ADC connected and 2vrms 1khz sine wave applied —— .05mv
And yes the opa1656 is fet
Nothing connected —— 1.2v
DAC only connected —— 0v (20 ohms output)
ADC only connected —— 1.2v (350 ohms input)
DAC and ADC connected —— .65mv
DAC and ADC connected and 2vrms 1khz sine wave applied —— .05mv
And yes the opa1656 is fet
How about a resistor in series with the input, maybe 1k. Also try more resistance in series with the output.
@Nelson Pass I’ll try 1k. What about a “best” value for the in+ to ground. I’ve tried 10k, 100k, and 1M. Assuming the 1k series resistor, what might be the best value for that bias resistor?
My problem on output is that the e1da Cosmos input is only a little more than 200 ohms at the 1.7vrms mono setting. I’m trying to leave the voltage high to keep a better snr without running the opa1656 at any gain. Perhaps better to add the larger series output resistor and add some gain? Although that risks adding some noise with additional resistors in the feedback loop etc.
I’m open to trying anything.
My problem on output is that the e1da Cosmos input is only a little more than 200 ohms at the 1.7vrms mono setting. I’m trying to leave the voltage high to keep a better snr without running the opa1656 at any gain. Perhaps better to add the larger series output resistor and add some gain? Although that risks adding some noise with additional resistors in the feedback loop etc.
I’m open to trying anything.
1.2 volts?
Something is very much amiss with that result. No opamp whether bjt or FET should do that. Either it is totally unstable as built and the voltage is a consequence of HF oscillation or something really fundamental is going wrong somewhere.
Please post pictures of the breadboard and of the connections to the supply, to the source and to the load. Maybe someone will see something wrong somewhere.
@Mooly I was using a 12v battery and dual supply/ virtual ground circuit to power it.
That’s what caused the 1.2v.
I went back to the bench supply I had been using for the rest of the tests and got;
Nothing connected —— .01mv
DAC only connected —— 0v (20 ohms output)
ADC only connected —— 4.8mv (350 ohms input)
DAC and ADC connected —— .65mv
DAC and ADC connected and 2vrms 1khz sine wave applied —— .09mv
Sorry for the screw up.
That’s what caused the 1.2v.
I went back to the bench supply I had been using for the rest of the tests and got;
Nothing connected —— .01mv
DAC only connected —— 0v (20 ohms output)
ADC only connected —— 4.8mv (350 ohms input)
DAC and ADC connected —— .65mv
DAC and ADC connected and 2vrms 1khz sine wave applied —— .09mv
Sorry for the screw up.
I was in the final stages of assembling an opamp-based active filter on a solder protoboard when I saw this, so had the opportunity for a little experimentation. Supply was connected using rather thin probe leads (sorry) from a good quality Philips dual bench supply.
Each opamp is locally-ish decoupled with 100nF rail-rail. The first stage is configured as a unity gain buffer as in your case.
Before adding the red rail-ground electrolytics, I could indeed get some HF oscillation at its output, dependant on the rail voltage; at +-5v, all was quiet, but above about +-8v a ~10Mhz oscillation would occur at the buffer output, with an amplitude of about 5-10% of the absolute supply voltage. Waving the supply leads around a bit slightly altered the frequency and amplitude of the oscillation, so maybe the (mutual) inductance of the leads was a contributing factor.
Introducing the electrolytics completely cured the issue at any supply voltage. So this definitely appears to back up the concerns/ideas of all the helpful suggestions already received.
Each opamp is locally-ish decoupled with 100nF rail-rail. The first stage is configured as a unity gain buffer as in your case.
Before adding the red rail-ground electrolytics, I could indeed get some HF oscillation at its output, dependant on the rail voltage; at +-5v, all was quiet, but above about +-8v a ~10Mhz oscillation would occur at the buffer output, with an amplitude of about 5-10% of the absolute supply voltage. Waving the supply leads around a bit slightly altered the frequency and amplitude of the oscillation, so maybe the (mutual) inductance of the leads was a contributing factor.
Introducing the electrolytics completely cured the issue at any supply voltage. So this definitely appears to back up the concerns/ideas of all the helpful suggestions already received.
This is most likely a Buffer and/or system activity issue. Are you using ASIO?
If so, in the ASIO Control Panel, increase your buffer size to maximum and re-check whether REW still does this. Another thing you could do is to deactivate as many processes as possible when measuring, e.g. turn off your Wi-Fi.
BTW, if this happens, click on the Reset Averaging button and REW will re-start the calculation afresh.
Is there any reason you aren't showing a picture of your breadboard circuit?
Last edited:
The LM4562 is unity gain stable according to spec so it should not be such a hassle to keep it from oscillating, right? Anything could happen on the workbench if the setup is to blame but lets assume its OK.
I found this old thread: https://www.diyaudio.com/community/threads/lm4562-oscillation.212835/
Loads of information in that thread, maybe worth browsing through?
I get the problem with the low input impedance of the E1DA Cosmos but is it really 350Ω @ 2.7Vrms and what about input sensitivity? Do I read the correct specs?
I found this old thread: https://www.diyaudio.com/community/threads/lm4562-oscillation.212835/
Loads of information in that thread, maybe worth browsing through?
I get the problem with the low input impedance of the E1DA Cosmos but is it really 350Ω @ 2.7Vrms and what about input sensitivity? Do I read the correct specs?
Up to 10 Vrms balanced @ 3.48kΩ and 30% less unbalanced, 1kΩ @ 2.7Vrms, is that right? How much voltage do you plan to feed it? Why not apply some gain (seems like there is headroom) if its hard to keep it stable at unity gain? I would personally not worry about a reasonable amount of gain from a "noise from resistors in the feedback loop" perspective. But I don't think I would be able to hear the difference between a LM4562 and a bog standard NE5532 either so please take my words with a grain of salt...
A totally-unscientific-what-first-came-to-hand 470uF 25v. Load is 3xOPA2134 + 1xOPA134