Need help with 60hz hum/ground loop. Schematic attached.

[Fusion916]

Well I figured out what was going on with the hum even after removing the opamp. In my 2 transistor current mirror my output transistor (Q4 in your schematic) was going into thermal runaway or something as the current would mirror within about 10% of the diode connected device then would rail off into high current as the device heated up, knocking my voltage regulator out of regulation and sending 60hz to the speaker.

I instead used remitter resistors in my current mirror (47ohm) and adjusted the bias resistor to get about ~400mA Ibias and now the bias current stays rock solid even after the device heats up. The hum is completely eliminated.

Question about your schematic however. Do you split R1 and R2 to provide a signal ground? Is only the input stage required to be referenced to signal ground?

[Fusion916]

forgot to ask, does anyone know where I can get spice models for the BC517 and the NTE261?

[cbdb]

Originally Posted by Fusion916

Question about your schematic however. Do you split R1 and R2 to provide a signal ground? Is only the input stage required to be referenced to signal ground?

No, The cap and those resistors form a low pass filter and are there to isolate the amp input from the negative rail. Although that may also act as an input filter.

[cbdb]

Originally Posted by Fusion916

forgot to ask, does anyone know where I can get spice models for the BC517 and the NTE261?

search this forum for "spice models"

[Fusion916]

Originally Posted by cbdb

No, The cap and those resistors form a low pass filter and are there to isolate the amp input from the negative rail. Although that may also act as an input filter.

Isnt isolating the amp from the negative rail the same thing as providing an ac ground for the input source?

[Mooly]

Great to hear you have solved it, and I think others have answered your question

So yes, the cap that decouples R1 and R2 must be large enough such that no audio is developed across it. In practice we use very large values for the cap such that it shunts away any ripple on the bias voltage.

I would say that for a simple amp like this standard spice models should be fine.

[Mooly]

Actually I was really surprised at the high level of ripple that spice shows for this type of circuit. If anyones interested here is the file with ripple on the rails. I've probably made the amplitude unrealistically high but even at lower levels there is a big problem. Not sure why it simulates like this as you mention it is now silent...

[epicyclic]

He is using what looks like 6600uF ? on each supply rail and we dont know what voltage his regs are dropping as we dont know the transformer specs ( to calculate equivalent series resistance )............however it is only half wave rectified !!!!! so i am mostly with you Mooly.

[Fusion916]

I'm not sure what the big deal of half wave rectified is your caps are big enough. As long as vripple is small there should be no ripple at the output of your regulator. And I'm using 3 6800uf caps per rail.

[Mooly]

Its no big deal if you design for using half wave and get the result you want.

Full wave is more efficent and the higher frequency ripple component and harmonics are easier to remove.