Voltages idle and under load

[surfsup]

I'm trying to track down a nagging very slight oscillation. I can't seem to find it, so I decided to put a dummy load on and watch voltages on the amp at idle and at full signal to see if I notice anything strange. here's a chart I'm referencing with voltages at all the stages at idle (columns A/B), with signal (C/D) and swapping the PI tube for a different one (F/G and H/I)

[img:658:404]http://chicagocadcam.com/ChrisHahn/voltages.jpg[/img]

observations:
1) the current draw through the plate and cathode on the preamp tubes seems about equal which I would expect
2) the current draw on the PI is not adding up (0.9+1.0=1.9 not ~1.5) - I assume some current runs through the 1M resistors?
3) Sending in a signal I noticed:
- the 3rd stage cathode voltage goes up about 50%
- 1st stage goes down slightly
- PI indirect side decreases as well

I thought I would find a stage drawing more current than normal to identify the oscillation, but I find this instead. Is there anything in this chart that jumps out at anyone?

[Colossal]

Surfsup,

I noticed you are using 82k/100k plate resistors for your PI. IIRC, the X10/LP/Voxy PI tail resistor is 47k so this typically negates the need to correct the PI imbalance by lowering one plate lead to 82-91k, hence in Voxy amps, the PI plate resistors are matched at 100k/100k. Did you significantly lower your tail resistor value in choosing to offset the PI plate resistors to 82k/100k?

[surfsup]

Colossal, yea I know the bigger voltage helps with uneven gain in the PI. Even with the 47k in there, the amp has a slight oscillation still. The 47k is in there still I just have another 47k currently parallel.

The reason I messed with the PI biasing resistances is by changing them, I got rid of some harsh overdrive tones that were plaguing me as well. Even with cleans, a single string note would first sound a slight crunch/fizz, then the note would ring clear. Changing these fixed that. So I wanted to keep them. I was going to swap the 82 for a 100k but since the oscillation was there before with the 47k/1k2 I figured i would leave it until I got the oscillation fixed.

So the voltages seem normal in behavior to you?

One thing i forgot to mention. When I turn up the MV (PPIMV), turning from 0-7 results in more signal, PI voltges dropping, etc. Then from 7-10, the signal starts dropping. I assume this is a slight impedance problem but could that contribute to an oscillation? The 220k/220k PI grid leaks and 10k/10k grid stops are pretty standard stuff for EL84 so has every amp in the past had this impedance issue, as well? Or does that explain the oscillation?

[Colossal]

Even with the 47k in there, the amp has a slight oscillation still. The 47k is in there still I just have another 47k currently parallel.

Ok, so there's 1k2/23k5 as the tail resistor with the second 47k tacked in.

The reason I messed with the PI biasing resistances is by changing them, I got rid of some harsh overdrive tones that were plaguing me as well. Even with cleans, a single string note would first sound a slight crunch/fizz, then the note would ring clear. Changing these fixed that. So I wanted to keep them. I was going to swap the 82 for a 100k but since the oscillation was there before with the 47k/1k2 I figured i would leave it until I got the oscillation fixed.

So the voltages seem normal in behavior to you?

The voltages seem reasonable but under signal conditions you see the imbalance in the PI plate voltages worsen so I'm wonder if that spread is increasing even more under real conditions (not just stationary 1kHz input conditions). Have you tried adding say a 10k grid resistor at the entrance of the PI?

One thing i forgot to mention. When I turn up the MV (PPIMV), turning from 0-7 results in more signal, PI voltges dropping, etc. Then from 7-10, the signal starts dropping. I assume this is a slight impedance problem but could that contribute to an oscillation? The 220k/220k PI grid leaks and 10k/10k grid stops are pretty standard stuff for EL84 so has every amp in the past had this impedance issue, as well? Or does that explain the oscillation?

It is interesting because in the X10 thread, we all discussed at length that there is a fixed set of grid leaks at 220k/220k but also the PPIMV in parallel after the cut control. This presents a different impedance to the PI. I did something similar with my Express in that I tacked 470k in parallel with the 250kA-dual pot for a max value of 160k instead of the usual 2M2 resistors. I like the crunch it gives and god knows you can't open up an Express without an attenuator. But in an offline discussion, Martin Manning modeled the effect of that change to the PPIMV and it does produce a different effect than a PPIMV with 250k/2M2. Most amps have either a PPIMV substituting for the usual fixed grid leaks, but not both fixed grid leaks AND a PPIMV in parallel. The 10k grid resistors are a sensible idea. The X10 seems pretty well hedged against blocking distortion with very small coupling cap values.

Have you tried temporarily lifting the cathode bypass cap from ground on the output tubes and seeing if the oscillation is there? This would provide more local feedback but also just test if that cap might be bad. Also have you tested to see if your choke is ringing? You might tack an 0.047uF cap just in front of it just to see if the artifact goes away. I'm just grasping at straws here. Your layout is great as is the quality of the build so this issue is quite vexing.

[surfsup]

Have you tried adding say a 10k grid resistor at the entrance of the PI?

I had a 100k there before. Not now. I will put a 10k or something in there to see.

The X10 had a 250k dual pot with 1M strapped across the wiper to ground. I have a 1M pot with 220ks across the wiper to ground. I assume this is electrically the same.

But in an offline discussion, Martin Manning modeled the effect of that change to the PPIMV and it does produce a different effect than a PPIMV with 250k/2M2

What effect did you see? I currently lifted the cc caps at the 220ks and put a 220k in series with the cap, connected to the other 220k, effectively reducing the gain into the MV by half. Still doing it. I also tried to parallel 220ks with the 220k grid leaks going down to 110k in parallel with the MV and still the same.

Have you tried temporarily lifting the cathode bypass cap from ground on the output tubes and seeing if the oscillation is there?

No but I did parallel a 1000u/50v with it for an effective 1050u and still had the problem. I guess that wouldn't help if the cap was bad. I will try that. I was also thinking about splitting the cathode into two separate resistors I have room off to the side I think.

Also have you tested to see if your choke is ringing? You might tack an 0.047uF cap just in front of it just to see if the artifact goes away.

We chatted a while back and I put a 1k in place of the choke. Though I didn't test this at the time for oscillation. I'll try to sub a R for the Choke and what do you mean by a 0.047 cap in front of the choke? in series? or to ground parallel with the filter cap?

I'm open to all suggestions...!

[surfsup]

Put a 100k to the PI, no change
Disconnected one leg of the cathode cap on the EL84s. No change aside from a tad bit less gain
Put metal shields over the tubes which were all rolled a few times. no change
I shorted the choke with a jumper between each cap (didn't desolder the choke)...no change
went with 440k/440k grid leaks after PI (2-220k in series ea side) no change

Sigh...I admit I'm ready to chuck this thing. One thing that helped marginally more is lowering the tail resistor some more. I had 47k||47k, then 33k||47k and now 22k||47k and the oscillation is very close to remedied but not quite. It is still there.

Also, why does the plate voltage increase with more signal on the one side of the PI?

[Colossal]

Also, why does the plate voltage increase with more signal on the one side of the PI?

This might be a dumb guess, but are you sure you have 1M grid leaks on both sides of the PI? Have measured or calculated the grid voltages on each PI triode at idle?

The underlying theme in all of this is that your diligent analysis, substitution, and even our potshots, suggestions, and guesswork have not remedied the issue. I'm beginning to suspect a bad part (bad coupling cap? wrong part?). One more dumb idea, have you tried applying feedback just to see its effects?

[surfsup]

Both are IRC 1M resistors. I suppose one could be bad but the voltages are fairly symmetrical. I replaced all the caps except rail caps over the week. I'll measure the 1M later today when I get home.

I haven't tried feedback since i wanted to first get this resolved but I guess it won't hurt. I'll do that. I'm sure I have all the parts...maybe even try that presonance circuit.

Stupid question but is there a maximum signal you shouldn't exceed into an el84?

[surfsup]

Sorry for being a pest here.

- tried feedback using circuit from liverpool schemo on 8R tap. turning the pot from 0-10 doesn't seem to really affect anything, including the amp sound. The amp does seem a slight bit clearer, but there is still oscillation or phasing or whatever this is.

- 1M grid leaks measure at 1.002M and 1.004M.
- 580R bias and 15k3 tail are their measurements

Questions:

1) is there a maximum signal to send into the el84s? even at low volumes though, there is still oscillation which tells me it is in the preamp
2) no one answered my question about voltages. In the chart in black you can see the current is roughly the same through the plate and cathode and both idle and signal. Current also increases. In red, on the PI, you can see that current increases on one side but decreases on the other. Also the currents don't match up to the cathode or tail resistor currents. they are off by like 0.5mA!

[img:658:404]http://chicagocadcam.com/ChrisHahn/voltages2.jpg[/img]

[surfsup]

More info. I moved my PT primaries going to the mains switch. Now when I put my DMM on the plate side of the PI with the 82k, the voltage on this plate behaves strangely. With guitar plugged in, if I don't mute the strings, the moment I touch the plate lead with my DMM, the amp emits a high pitched whine. almost like a 1kHz or 2kHz sine wave, I think. The voltage INCREASES from 191vdc to 215vdc the moment the whine starts. If I put my hand on the guitar strings, the whine stops.

On the 100k side there is a slight audible whine. Voltage shifts from 188 to 190 if memory serves, there was only a 2v difference. There was no high pitched whine prior to this moment!

So there's something up with the PI. I checked the socket. It looks brand new (and kinda is, only about a year old).

[Structo]

Positive feedback?

[surfsup]

PI voltages going up with signal: I think this is due to the impedance issue. I pulled out the 220k/220k leaks that were || with the MV pot like on the 18w schemo. At 7 on the MV, the signal coming out of the PI starts dropping.

I'm about to give up and tear this amp apart. I have been trying to fix this for three solid weeks every evening. I think I'm done. I can't do this any anymore!

[Structo]

Is this the MV you tried?

[NickC]

........

I'm about to give up and tear this amp apart. I have been trying to fix this for three solid weeks every evening. I think I'm done. I can't do this any anymore!

Sorry I can't offer a ready solution for the problem.

It's natural to get frustrated after spending a lot of time and effort without success. Sometimes you have to walk away ...... set it aside and come back to it another day ...... or another month ...... or another year. Keep learning and moving ahead. When you next work on it, the answer (based on learning and moving ahead) may seem obvious and easy to fix.

[Reeltarded]

This thread is too technical for me, but swapped primaries makes a nasty noise and messing with PPIMV is the only time I have done that.