I suspect the input 100k is just for DC bias, not related to source impedance.
The heating issue is likely oscillation, and layout is important as I mentioned before so that might be worth tightening up. The fact higher voltages lead to instability is normal, more voltage, more gain-bandwidth product usually, so the phase- or gain-margins reduce.
However one other thing that might be worth considering is whether there is a ground-loop once the power supply is involved. Batteries are an ideal floating supply but if the PSU ground is not the same as the signal ground there could be an issue there. I'd consider some RF filtering on
the input as if RF is coming in that could also look like oscillation. 1k/1nF RC filting on the input might be worth a try.
The heating issue is likely oscillation, and layout is important as I mentioned before so that might be worth tightening up. The fact higher voltages lead to instability is normal, more voltage, more gain-bandwidth product usually, so the phase- or gain-margins reduce.
However one other thing that might be worth considering is whether there is a ground-loop once the power supply is involved. Batteries are an ideal floating supply but if the PSU ground is not the same as the signal ground there could be an issue there. I'd consider some RF filtering on
the input as if RF is coming in that could also look like oscillation. 1k/1nF RC filting on the input might be worth a try.
@EmuMannen I will try adding the 100nf across the supply.
@Mooly no apologies needed, especially given my knowledge level! The supply is dual output. I also tried the SilentSwitcher and a +-12v virtual ground circuit I made on a 24v battery, same results.
@jschwender you’re right. I don’t really need anything more than say 10k input impedance. Should I just lower that resistor to 10k?I need a low output impedance because the e1da adc has only 225 ohms impedance at the lowest voltage setting in mono mode. I was therefore concerned about increasing the series output resistor past 10-20 ohms.
@Mooly no apologies needed, especially given my knowledge level! The supply is dual output. I also tried the SilentSwitcher and a +-12v virtual ground circuit I made on a 24v battery, same results.
@jschwender you’re right. I don’t really need anything more than say 10k input impedance. Should I just lower that resistor to 10k?I need a low output impedance because the e1da adc has only 225 ohms impedance at the lowest voltage setting in mono mode. I was therefore concerned about increasing the series output resistor past 10-20 ohms.
OK.Soto further the oscillation hypothesis, I did add the 100nf cap directly across pins 4 and 8 but it had no effect. I then retried first an ne5532. Not great harmonic performance but stone cold at +-15v. Then an lm4562 which got just the tiniest bit warm, almost undetectable but there and performance better than the ne5532.
Tried the opa1612 and opa1656 and much better performance( thd+n. -111) but too hot to touch.
So re comments above about the increasing speed of the op amp more likely to cause oscillation this seems consistent.
This is a one off temporary build so I’ll just continue with the point to point build. Any other specific recommendations to reduce likelihood of oscillation other than designing and ordering a pcb?
Tried the opa1612 and opa1656 and much better performance( thd+n. -111) but too hot to touch.
So re comments above about the increasing speed of the op amp more likely to cause oscillation this seems consistent.
This is a one off temporary build so I’ll just continue with the point to point build. Any other specific recommendations to reduce likelihood of oscillation other than designing and ordering a pcb?
A 5532 should not be stone cold at -/+15 simply because of the quiescent current draw. There should be detectable warmth, not hot but very definitely warm. I would suggest you pop DVM on its current range in series with one of the rails and see what current it really draws. The data sheet says 8 milliamps and so that is 0.24 watt at 30 volt supply.
I would also say that subjectively (and certainly on a breadboard layout) that you would not hear any differences in THD between different opamps.
Fwiw although I'm acutely aware that breadboard layouts can suffer stability issues (and so its easy to say to you that its layout issues) I've used opamps like the ones you mention with zero issues at all in some real lashups and not had anything like you describe happen.
I suspect something else is going on with your setup. I would check the supply current to each different opamp you use and also check the basics that the input and output pins all actually measure close to zero volts DC when operating. A meter won't show oscillation as such but lets see what you have.
In+ and - pins are less than .05 ma when operating. I will check currents. After all these additions I borrowed a thermistor temp meter and measured 50 deg C on the body of the opa1656. This is still “hot” but within normal operating temp(25 deg above ambient).
Do I just leave it at that? Sound normal for 2v into a 350ohm load( .01 w). I tried doing the temp calculations based on 3.9ma per channel (from the data sheet ) quiescent current with+-15v and 2v out. Seemed like it was in the ballpark temp wise when adding up the components but I don’t really know what I’m doing.
stone cold was probably a bad description on the ne5532( I didn’t go back and measure that with thermistor).
Do I just leave it at that? Sound normal for 2v into a 350ohm load( .01 w). I tried doing the temp calculations based on 3.9ma per channel (from the data sheet ) quiescent current with+-15v and 2v out. Seemed like it was in the ballpark temp wise when adding up the components but I don’t really know what I’m doing.
stone cold was probably a bad description on the ne5532( I didn’t go back and measure that with thermistor).
Last edited:
The input pins should show absolutely minimal DC currents and the exact value depends on the opamp. A 5532 should be in the 100 nano amp region (0.0001 milliamp) with a 4562 about 10 times lower still. FET opamps are many thousands of times lower still.
Opamps can get noticeably warm when delivering highish currents into a load but that is not a common scenario with music sources.
Opamps can get noticeably warm when delivering highish currents into a load but that is not a common scenario with music sources.
@Mooly Given the dc voltages I measured can I calculate the amps at the input pins?
@jan.didden there are currently 10uf caps and .1uf caps directly from each power supply pin to ground and 100nf cap directly across the power supply pins. I tried it also with just the 10uf’s and also no 10uf’s but .1 and .01 uf in parallel from each power supply pin to ground. This is point to point soldered on a small board.
Thereason I used 10 ohms on the series resistor is the the load can be in the 200 ohm range and the ouput voltage starts to drop with higher values affecting snr. In some cases I’m measuring very low thd and wanted to avoid that. In any case the heating comes with no load connected as well so that series resistor not in play.
I measured that 50 deg C @25 deg ambient with a 2v 1 khz signal running through it so that would mean added temp factors, right?
I will remeasure with no signal applied.
@jan.didden there are currently 10uf caps and .1uf caps directly from each power supply pin to ground and 100nf cap directly across the power supply pins. I tried it also with just the 10uf’s and also no 10uf’s but .1 and .01 uf in parallel from each power supply pin to ground. This is point to point soldered on a small board.
Thereason I used 10 ohms on the series resistor is the the load can be in the 200 ohm range and the ouput voltage starts to drop with higher values affecting snr. In some cases I’m measuring very low thd and wanted to avoid that. In any case the heating comes with no load connected as well so that series resistor not in play.
I measured that 50 deg C @25 deg ambient with a 2v 1 khz signal running through it so that would mean added temp factors, right?
I will remeasure with no signal applied.
I was asking for the ground connections, the physical layout/routing.
Gnd is not everywhere the same.
Jan