New trainwreckish Plexi build

[Jana]

I have no experience with the GMX so I can't comment on that. One thing I did hear in the clip is at the end it sounded like a slight tremolo effect (is tremolo the one that wavers the volume?). That would reinforce what John said about an oscillation happening. If there is a high frequency oscillation going on (above the range of hearing) that could be doing something like this.

[Reeltarded]

Good ear, DAWG! (sorry, no randy jackson graphic)

It's sort of plexi. Umm.. I don't see 5k6 output grid resistors.

[pablogt]

I agree, I does look like a stability issue.

Yes, there are 5k6 grid stoppers. you don't see them in the first picture because they were soldered in vertical position at socket. In the last picture they are mounted between the socket and the new three lug terminal

I might increase them, as well as decrease the grid leak resistors, but I doubt that will help, since the instability also happens at low master volume settings.

Pablo

[John_P_WI]

since the instability also happens at low master volume settings.
Pablo

Hmmm, could it be the imp-bootstrapped master? Maybe it is in the PI as KOC questioned?? What would happen if you upped the 68 puff cap across the plates on the PI - just to see if it is a PI stability issue - or put individual caps across the pi plate load resistors?

I know you disconnected the FB wire before, did you disconnect it at the impedance selector or at the presence network (I'm curious as the FB wire can radiate all kinds of noise and cause problems, disconnecting it at the impedance selector would be best for this experiment)?

Sorry for my continual questions, I want to see this thing work and learn from it too...

[pablogt]

I turned the amp into Snubberland this morning. 1000pf between PA grid and cathodes, 250pf extra cap between PI plates and 500pf across both PI plate resistors.

Tried 100pf snubbers between power tube plate and cathodes.

Increased grid stoppers to 10k

Apart from killing the amp's tone, it had no effect on the parasitic oscillation. It is the most stable instability I've seen (does that make sense?)

I've also tried bypassing some parts of the B+ line, with no results.

John, I did disconnect NFB at the impedance selector side. I was aware of the potential noises related to the FB cable, that's why I put the FB wire along with the OT's ground line in a shielded twisted pair. The OT is grounded at the PI's filter ground (s3, from Star 3, in my schematic).

Pablo

[Reeltarded]

Apart from killing the amp's tone, it had no effect on the parasitic oscillation. It is the most stable instability I've seen (does that make sense?)

Pablo

Success! Well, I mean to say that us mostly being Americans, and given the fact that Chevy made the Volt, a car that burns as often while parked as any Fiat ever did while moving..



It sounds like DC, doesn't it? Like a cap discharging into the signal path.. play, charge, discharge.

This is as interesting as it is annoying. I want it to work as much as you do.

[pablogt]

It's more frustrating than anything else... I had to take yesterday off the Plexi, I was sick and tired of it. On the other hand, it is a learning experience, a tough one though

P.

[Reeltarded]

Yeah, I hear that as loud as the problem.

Remove it and find somethig fun to do for a while. When I did Martin's presonance circuit, the amp I first did it to never worked out right, but the two after that worked perfectly. Same amps, basically.

I fixed it by cheating. I measured the resistance on the pot where the problem appeared, and added a resistor slightly larger than that to it, not because I am a smart guy but because I am a smartass, and there wasn't any higher math, just arithmetic.

Your problem is like rocket science to me. If I were building a (space) rocket, I would fail 1000 times, and develop from new ideas, and my *best* failure. It's all I know.

[pablogt]

I'll keep trying. I still have an oscilloscope to buy and learn to use

P.

[Reeltarded]

I wonder if sub-millimeter vision would allow us to see stray signals that cause these problems we never find.

I need a telescope, and use my rocket science failures to launch my problem amps into very high orbits. 10.. 9.. 8.. etc, and etc..

[pablogt]

In the link below are my Plexi's design Visio documents.

In the chassis drill plan, holes for countersunk screws are surrounded by red circles.


Enjoy,
Pablo

[oj]

I turned the amp into Snubberland this morning. 1000pf between PA grid and cathodes, 250pf extra cap between PI plates and 500pf across both PI plate resistors.

Tried 100pf snubbers between power tube plate and cathodes.

Increased grid stoppers to 10k

Apart from killing the amp's tone, it had no effect on the parasitic oscillation. It is the most stable instability I've seen (does that make sense?)

I've also tried bypassing some parts of the B+ line, with no results.

John, I did disconnect NFB at the impedance selector side. I was aware of the potential noises related to the FB cable, that's why I put the FB wire along with the OT's ground line in a shielded twisted pair. The OT is grounded at the PI's filter ground (s3, from Star 3, in my schematic).

Pablo

Ground the 22k in PI bye it self. Precens will worke better and you can balance the PI even better.

OJ /Sweden

[gingertube]

If you have an instability when using GMX then it is probably the GMX MOSFETs going into ultrasonic oscillation. I assume that there are gate stop resistors on that little vertical board next to the MOSFET pins. Try increasing the gate stop resistor value.
I'm not familiar with that particular GMX Circuit but any zener protection diodes from gate to source should be mounted on that same PCB such that gate stops and zener protection diodes are right on the MOSFET pins.

Remember that MOSFETs (like power tubes) are RF devices, if you don't feed them RF they like to make their own. Gate stops are essential.

A bit of a story - I designed a Laser Diode Driver for the day job, I included gate stop resistors for the MOSFETs but passed the PCB layout job to someone else. The gate stops ended up at the end of 2" of PCB track away from the MOSFETs and they oscillated like crazy. I cured that by replacing the gate stop PCB resistors with wire links, cutting the PCB track right at the gate connection and bridging the track cut with surface mount resistors of the same value. Get those gate stops right on the MOSFET gate, the same as you would do for output tube grid stops ONLY MORE CRITICAL because MOSFETs have much more gm than your typical output tube and are therefore more prone to parasitic oscillation.

Cheers,
Ian

[gingertube]

A quick note to those readers not familiar with GMX.

GMX stands for gm (transconductance) multiplication.

The GMX system measures the audio current that the output tubes are putting through the output tranny primary and multiplies it using MOSFETs wired in parallel with the output tubes. That is, it is design to give you tube tone at higher power. You like EL84s or 6V6s but want 100W for more muscle, then GMX is the go. Actually it measures tube cathode current and multiplies that but the theory remains the same.

Cheers,
Ian

[pablogt]

Thanks all... Actually, you can also wire the GMX as a "feel enhancement", which is what I did. That is, the output power is still the same (limited by the OT in this case) but you get the feeling of a bigger amp. And it does work, you get the same feeling as when you go from 18w to 36 in a 4xel84 amp with half power option. It would be great if it wasn't for that parasitic oscillation.

I'll give it a last chance this week. And then decide if GMX stays or goes.

oj: check the diagrams below. If that's what you mean, I might give it a try but I think it would reduce scratching noise on the presence control but have little effect on the oscillation.

Ian: I'm walking on slippery ground here..... three years ago I didn't know the difference between volts and amperes. The zeners are right by the mosfet pins (see photo). I'll try increasing r3 and r5 resistors first (see schematic detail)

Pablo