Aleph-phase lag with non-inverting input

[jupiterjune]

Allan-
Do you mean a cap to ground? I tried putting a cap across across the 4.3kohm resistor (I had it in place of the zener to give a higher bias current--and that is what was causing the hum) and.... the hum got louder (I had that channel conneceted to a speaker) I had the 'scope on the CCS mosfet gate, and the ripple was completely squashed.

Which sort of lead me to think, maybe the ripple is being canceled out as I described in the post above.

Or maybe I have a fundamental misunderstanding of how the CCS works. Do you need identical ripple on the source and the gate of the CCS mosfet to get constant current?

Off to do that search you suggested, maybe I will find the answer there.

[Blues]

jupiter,

Across the zener (Z5) would be between its cathode & anode leads. This will reduce some of its intrinsic noise and the ripple going through it from the +PS. Recommended value is from 1uF to 4.7uF, so your favored film cap will do. Ripple and zener noise at the Gate of the CCS fet will find its way into the input diff pair.

Constant current is set by the zener's 9V-4V(fet G-S drop)=5V across the 221R which is about 22mA.

[wuffwaff]

Hi,

to get a higher bias through the diff pair don´t change the zener but the resistor that defines the current. Just lower it to get more bias current.
Using a resistor in place of the zener you´ve made yourself a nice little amplifier for the power supply ripple.

For desoldering 3 pin devices you can use some solder wick. It´s very easy that way.

William

[Babowana]

Originally posted by jupiterjune


Or maybe I have a fundamental misunderstanding of how the CCS works. Do you need identical ripple on the source and the gate of the CCS mosfet to get constant current?

To get the constant current, you should secure a constant Vgs.
If you have the 9.1V zener, the zener bootstraps the potential
difference between the gate and the +rail so that the CCS
mosfet can not see the ripple voltage of the +rail.

So, you are right. You need identical ripple on the source, the
gate and "the +rail" to get the constaant current. If you have a
resistor (voltage divider) instead of the zener, you hardly get the
identical ripple on the gate.

Hope this info will be useful.

[jupiterjune]

Thank you for the CCS tutoring -- now it is crystal clear. In hindsight, it is sooo simple, but I was not seeing it.

Now to un-threadjack this thread...

I finally popped the input mosfets out of the board to test Vgs (RS never has de-solder braid in stock, but I found if I use my giant soldering iron and contact all 3 pins at once, they pop out in about 1.29 seconds)

And the Vgs are: Good (right channel):
T2 (the inverting input) Vgs=3.076
T3 (the non-inverting input) Vgs=3.083

Bad (left channel):
T2 Vgs =3.067
T3 Vgs =3.064

I had a lot of trouble at first getting consistent repeatable readings. Just touching the metal tab causes the Vgs to drop...
So, I ended up soldering the test wires to the mosfet pins, and pressing the mosfets firmly to a large chunk of heatsink.

These Vgs readings don't seem so horrible -- so if you've read this entire thread, you might be wondering about the very different Vgs I got when I tried to evaluate the mosfets in place.
Well--the didn't have heatsinks on them. They got hot. Moreover, the T2's drain is connected to the negative supply rail, T3's drain is connected to a 390 ohm resistor, then to the negative supply reil. T2 sees more wattage, heats up, and becomes even more of a current hog. But this was NOT causing the phase lag.

Someone posted (I can't find it now) that I should swap out the mosfets between the left and right board, which should give the answer real quick as to where the problem is...

and -- the phase lag stayed with the left (bad) board.

Next I noticed the problem exists with the amp off -- going into the input mosfets! I should have checked this much sooner. I was always looking at the amp output, never using the scope to compare the input signal at the input mosfet's gates.

It gets worse from there, but I have to sign off for now.

[cviller]

Originally posted by jupiterjune

Someone posted (I can't find it now) that I should swap out the mosfets between the left and right board, which should give the answer real quick as to where the problem is...

and -- the phase lag stayed with the left (bad) board.

Next I noticed the problem exists with the amp off -- going into the input mosfets! I should have checked this much sooner. I was always looking at the amp output, never using the scope to compare the input signal at the input mosfet's gates.

It gets worse from there, but I have to sign off for now.

Glad to hear you finally followed my advise.

So is it correct that you have the signal going from the input terminal through a

series resistor -> zener/resistor shunt -> resistor in series = phase lag?

If that is the case, I think you should look at the zeners and shunt resistor. Try removing the zeners and check solder joints and resistor values...

[jupiterjune]

my thoughts exactly. I removed the zeners. Then I replaced the three resistors with new ones. The phase lag actually looked worse at the mosfet input. When I turned the amp on -- the left channel was "dead silent" in the bad sense--not the cviller fishamp good sense.

I think the output gain mosfet died, but I haven't had a chance to look into it.

I cut the trace to the input mosfet input, once I get the channel working again, I will try a point-to-point wired input network.

[cviller]

Sounds really strange...

Could you take some high quality pictures of your setup?
I know it's unlikely, but perhaps someone can spot something you have missed...

How about drawing a schematics with details on where you can see the phase lag?

Why do you think the mosfet died?

[jupiterjune]

Cviller-

I will post some pix and schematics as promised -- might be a few days though.
I think the problem is close to being solved, so don't spend too much time thinking about it.
If I can't solve this last little problem real quick, I will be back begging for help (with good pix and clear schematics).

I will post results of what I find either way.

Careful inspection of the PCB traces with a flashlight, both through the board and on the bottom, has revealed nothing.

I think the output gain mosfet died, but I haven't had a chance to look into it.

--I wan't to retract that statement, thinking it through a bit more, it seems unlikely that it is the output mosfet.