Very Strange Problem

[dscottguitars]

Hi,  I am having a very strange problem.  I'm building a ~100 watt power amp, based on the Fender Twin?  See schematic.  The problem is when I turn the bias control pots for more current I am getting a strange pop-pop-pop-pop noise.  The speed of the pops gets faster as the voltage gets less negative, starting at -44v on the grids.  I can only get down to -37v which in itself is an issue as it should be around -27v on the other end of the pot.

Other effects on the speed of the pops are using the meter to check voltages, sometimes it's faster and other times slower.  But it's not consistent either. 

More strange effects are checking the plate and screen voltages:  At -44v for bias, the plates are at about 430v AND the amp sounds pretty good with my guitar plugged straight into it not using a preamp.  Also there is no popping or other strange noises at this -44v setting.  It's only when I start turning either pot it makes this noise.  When I check the plate voltages with a lower bias voltage, the meter is reading all different values ranging from 600 volts to over a 1000??  The values change constantly without doing anything but holding the probes tight to the ground and plate pin.  How can this be??  The voltage at the output transformer is steady at 435 or so. 

I have on the schematic a switching rectifier which is not used yet and high/low voltage switch which also is not completed.  I did have a switchable bias for fixed and cathode but I took that completely out thinking there was a problem there-nope.

I also had 100 ohm pot on the 6.3v heater circuit and found trouble there, so I made a center tap with two 200 ohm resistors going to ground instead.  No difference.  HOWEVER- when I disconnected the pot and held the twisted wires in my hand the noise went away.  I couldn't check the voltages doing this so I decided to eliminate it and that didn't help, but it leads me to believe the problem could be related to the heater wiring.  I cannot find any layout problems or loose connections. 

Plate current reads 28mA at the -44 bias voltage; for two tubes so that's about 14 mA each tube.
Screen voltage is 428v
Plate voltage is 430v
4 GE 6L6 tube
1 Raytheon 12AT7 tube
Transformers are used and quite large

At one time I had the preamp built in this chassis but had too much hum, thinking it was due to the cramped components so I decided to make them separate.  I had no problems like this then.  The current and voltages were fine and the bias circuit worked. 

I am really confused!!  Any experience like this out there???

Thanks,

Daniel

[PRR]

What effect does the Feedback pot have?

[dscottguitars]

If you mean effect on the noise problem, not too much, it just makes it quieter but doesn't take it away or change the popping much that I can tell.  I should also add that I played it with the popping noise-at around the -37v for bias and it's much quieter/less gain than having the bias voltage at -44v.

[HotBluePlates]

Your schematic shows the feedback wire to the center lug of a SPDT switch to select speaker impedance. What happens if you unsolder that wire at the switch?

If the put-put goes away, swap primary wires, and reconnect that feedback wire directly to the lug the 4Ω tap is on.

Feedback wires are most often connected to a specific tap of the OT, not to the speaker output when using an impedance select switch. Although I understand why you connected it the way you did, choosing one set tap for the feedback wire attachment yields a consistent amount of feedback; the way you have it wired changes the amount of feedback when a different load is used.

Since you're effectively copying a Twin output section, that feedback wire would correctly connect to the 4Ω tap, as in the Twin.

[dscottguitars]

I will try that.  I can't see why it wouldn't work this way but it is different than when I had it working before without the put put noise.  The only other thing I changed was using one 1.5K resistor going to the grids of the output tubes.  Before I had each tube with a resistor on it.  Could that make a difference?

Thanks for the replies...

[sluckey]

...but it is different than when I had it working before without the put put noise.  The only other thing I changed was using one 1.5K resistor going to the grids of the output tubes.  Before I had each tube with a resistor on it.

So, it once worked, then you changed some things, and now it put, puts? Undo your changes and see if it improves. I highly recommend a separate 1.5K grid stopper for each of the 4 output tubes.

[dscottguitars]

Yep, I should know better than to not undo changes...  I changed the resistor to grid setup because Marshall's '59 was like that.  Oh well, thanks for the lesson.  I guess I needed someone else to tell me that...

Daniel

[HotBluePlates]

I changed the resistor to grid setup because Marshall's '59 was like that.

You'd be surprised. I had an amp that was unconditionally stable with 1.5k grid stoppers and installed a set of Groove Tubes KT66's. The amp then exhibited a tendency towards oscillation until I upped the value of the grid stoppers.

I will try that.  I can't see why it wouldn't work this way but it is different than when I had it working before without the put put noise. 

I used to not understand how the feedback should be sourced from the OT. I assumed (because I mainly used amps without an impedance select switch) that the feedback wire always came from the speaker hot connection.

But when you look at a Marshall schematic, the feedback is connected to one tap only, and stays put when load impedance is switched.

The reason is that when you play the amp, an output voltage is generated on the entire secondary winding, with specific voltages split in accordance with the tappings for each output impedance. To have a consistent amount of feedback returned to the phase inverter, you need a consistent source voltage for that feedback. You obtain this by keeping the feedback wire connected at one tap.

Even if the speaker is connected to the 8Ω tap, and the feedback wire is connected to the 4Ω tap, there is still voltage being supplied to the feedback wire. And that voltage is exactly the same as if the speaker were connected to the 4Ω tap.

By contrast, if the feedback source follows the speaker, switching to the 8Ω tap returns a bigger voltage to the feedback circuit than the 4Ω tap. That may be enough to turn a stable amp unstable, depending on the exact nature of the feedback circuit. At a minimum, the two load settings will sound more-different than they should.

[dscottguitars]

PROBLEM SOLVED!!!!  

It was the grid stoppers.  Changing them back to one on each tube solved the issue.  Had I done that in the first place I wouldn't have needed to ask the question.  But I also would not have been able to talk to you guys either...

I left the negative feedback line alone for now...But,

I do have a question for HBP:

You say that putting the negative feedback-NFB line on the 8 ohm tap would give more voltage to the circuit.  That seems opposite to me.  Isn't there more voltage coming out of the 4 ohm tap?  That's why the power goes up when using a 4 ohm speaker setup, yes?

I am looking for the lowest amount of NFB to add into the phase inverter.  That's what I discovered with putting a pot and the resistor setup I have.  I can dial in anywhere from 22%-clean to 0.8%-mean.  The original Bassman uses about 12% and the Marshall 59 uses about 5%. (if my calculations are correct)

To me this was the best discovery I made when researching how to build guitar amps.  With less negative feedback I get much more distortion on the output tubes and even some nice feedback when cranked.  This was a suggestion from Aiken amp pages explaining negative feedback.  I'd sure like to know more, especially from you guys with the experience and knowledge I lack...

Thanks again, the amp sounds great now.  

[sluckey]

Isn't there more voltage coming out of the 4 ohm tap?  That's why the power goes up when using a 4 ohm speaker setup, yes?

Voltage is directly proportional to turns ratio, ie, more turns equals more voltage. There are more turns for the 8Ω tap than for the 4Ω tap.

[dscottguitars]

Thanks, I didn't think of it like that.  I changed it to the 4 ohm tap.  I also got the bias switch back in and the other switches finished.  I plugged in a simple Presonus preamp in it with my guitar and it's really nice.  Now I'm going to make my own preamp with bass, treble and channel switching to go with it...

[PRR]

> the feedback wire ... What happens if you unsolder that...

He has a pot which will turn-down NFB to practically zero (<1:100, when forward gain is probably ~~20). Since that didn't fix the problem, I penciled-out bad NFB (my first guess also).

> one 1.5K resistor going to the grids of the output tubes.  Before I had each tube with a resistor on it.  Could that make a difference?

Yes, and circuit logic does not help.

You think each "side" of the push-pull quad will work together. Like a two-pair of oxen.

But it can be more like a two-pair of greyhounds. A pair which should "pull evenly", one dog gets a little ahead, then the other, very rapidly. Your V4 and V5, which are just "push" for audio, interact with the slightly different lead-lengths and stray capacitances so they act as a supersonic (radio frequency) push-pull oscillator. For a while. After a few thousand cycles the grid voltages has pounded itself negative, and the tubes cut-off. Either the cut-off or the recovery "pops!" This depends also on static voltage and grid resistance (including meter loading).

Yes, sometimes the layout and parts "get lucky" and it doesn't do that. (Until HBP puts in a slightly hotter tube set.)

Here's the conservative thing.

*Separate* resistors for each G1, G2, and plate.

In pentodes, the G1 resistor can be much larger than a K without real audio effect. 10K probably won't hurt. 5K is popular. Use separate 5K per G1.

G2 at 500 ohms is fine. Much smaller won't spoil supersonic troubles well, and much larger will restrict EL34 power.

Most good reliable quad-output amps have a small resistor in each *plate* lead. 47 ohms 2 Watts is common. This has hardly-measurable effect on audio power, but is enough to spoil the typically low-impedance radio resonances.

Yes, total of 12 resistors. You see why musician-market amplifiers try to get away with less. But DIY economics is different. You already spent more time head-banging than 12 resistors are worth.