I encountered an odd-ball this morning. I'm working on my first attempt at a tube bass amp using the tubes and iron from a Baldwin 46C organ amp chassis. The power amp is a quad of 6BQ5's. I have the power section cathode-biased at roughly 65% for safety, and the PI is a JCM800 circuit but using a 12AU7 tube and a 27K tail instead of 10K. Coupling caps are all .1uf since this is for a bass. The 6BQ5 screen sets are fed through a pair 10K resistors from the choke side of the power supply with 4.7uf screen decoupling to the preamp ground. The cathode resistors on each pair of 6BQ5's are a 1-ohm 1% resistor (for plate current measurement) in series with a 470 ohm 5% sand-block resistor. Grids have 150K to main chassis ground.
What I'm seeing is a bit of what looks like soft clipping at the outputs of the PI and the PA when the input goes past a certain point. This by itself is normal, since all amps clip at some point when the input increases. What is strange is that the clipping disappears if I put my hand near the output jack of the signal generator. I've attached extra grounds from the scope chassis to the amp chassis, and the signal generator is chassis-grounded via the output cable. I didn't notice the clipping change at all with my hand near any other cable or near the amp chassis itself (don't worry, I always observe safety precautions when working with high voltage). I have never seen this phenomenon with any other amp build.
My assumption is that somehow the 'scope measurement is somehow corrupt, since I don't see how my hand being near the jack (NOT touching, just near) would eliminate the clipping. Oh, the clipping isn't eliminated with the measured signal being reduced at all; instead, its as though the clipped off part is added back on to the sine wave tip.
Does anyone have any clue as to what I'm seeing here?
What I'm seeing is a bit of what looks like soft clipping at the outputs of the PI and the PA when the input goes past a certain point. This by itself is normal, since all amps clip at some point when the input increases. What is strange is that the clipping disappears if I put my hand near the output jack of the signal generator. I've attached extra grounds from the scope chassis to the amp chassis, and the signal generator is chassis-grounded via the output cable. I didn't notice the clipping change at all with my hand near any other cable or near the amp chassis itself (don't worry, I always observe safety precautions when working with high voltage). I have never seen this phenomenon with any other amp build.
My assumption is that somehow the 'scope measurement is somehow corrupt, since I don't see how my hand being near the jack (NOT touching, just near) would eliminate the clipping. Oh, the clipping isn't eliminated with the measured signal being reduced at all; instead, its as though the clipped off part is added back on to the sine wave tip.
Does anyone have any clue as to what I'm seeing here?
Last edited:
Is the output changing? 'Scopes have lied to me due to stray pickup on the probe, but not at audio frequencies.
On another subject... fixed bias would be better if you want to bias it cold, and the screen resistors should be lower (maybe 1K) and there should be four of them. Likewise wise four grid stoppers and you may need two or four plate stoppers (100 Ohm in parallel with 5-8 turns of wire wrapped on the body). VHF oscillation is responsible for some mysterious behavior - why chance it?
On another subject... fixed bias would be better if you want to bias it cold, and the screen resistors should be lower (maybe 1K) and there should be four of them. Likewise wise four grid stoppers and you may need two or four plate stoppers (100 Ohm in parallel with 5-8 turns of wire wrapped on the body). VHF oscillation is responsible for some mysterious behavior - why chance it?
It is a possibility, haven't had time to pursue it further. I'd say oscillation is more likely than picking up a stray RF signal, if one of those is the cause. This is my first time working with the 6BQ5, so its a learning curve for sure.
I can't actually answer that. I was not looking at the 'scope's digital voltage measurement, just the waveform. The waveform only changed with regards to the clipped portion coming and going.
As for the biasing, I had it fixed bias, but my POT's seem to be intermittent even after multiple cleaning, and I don't want my BQ's burning up because of faulty POT's. I will be going back to fixed once I get some reliable POT's.
With regards to screen resistors, I went with the 10K's after a ton of research and reading. I originally was going to go with triode-connected screens, but screen current always seemed to be too high. I may try lower resistors if I can find some here with a high enough power rating.
Grid stopper resistors I left out because I've read in a few places to only use them if absolutely needed. They can introduce a risk of oscillation themselves according to several of the posts I've seen. The plate stoppers could make sense, although I've never seen them in any of the guitar amp circuits I've seen online. I have seen them many times in HF amplifiers, however.
Your words & music description of the circuit is nice but a bit short of ideal.
A Quad of toobz adds up to two sets of parallel connexions. Paralleled toobz are well known to take off & oscillate.
The cure in the past has been to use CC resistors on all the plate & grid leads.
On the plates something like 47R while on the grids One K. Physically small, skin effect can sink you.
All the Rs must be physically attached as close as possible to the socket. The resisters must be CC. 👍
A Quad of toobz adds up to two sets of parallel connexions. Paralleled toobz are well known to take off & oscillate.
The cure in the past has been to use CC resistors on all the plate & grid leads.
On the plates something like 47R while on the grids One K. Physically small, skin effect can sink you.
All the Rs must be physically attached as close as possible to the socket. The resisters must be CC. 👍
Carbon Composition resisters still look like resisters at Radio Frequencies.
Many other resisters often used by Audiofiles do not. If your amp is oscillating at 500 KHz you may not know that.
The problem is often related to the driving signal level. But the resulting audio program material will be grossly affected.
There is quite a bit of reference to the paralleled tube oscillation problem in earlier publications.
The way out is by 'swamping' the RF paths, one less problem to worry about.
Many other resisters often used by Audiofiles do not. If your amp is oscillating at 500 KHz you may not know that.
The problem is often related to the driving signal level. But the resulting audio program material will be grossly affected.
There is quite a bit of reference to the paralleled tube oscillation problem in earlier publications.
The way out is by 'swamping' the RF paths, one less problem to worry about.
I added plate stoppers using 100-ohm CC w/4 turns, plus changed the screens to 1K (kept the decoupling). Better results, no longer see the odd clipping artifact, and better gain, but had another strange symptom pop up after the amp was on for a bit. I suddenly lost audio output from the power amp section ('scoped across a 15-ohm load on the OPT). I checked the BQ plates, and no output signal there either even with 100V audio on the grids. I still have audio dropping across the BQ cathode resistors, and the DC operating voltage/current on the cathode resistors with no audio drive is still correct, and I still see that current rise with audio input to the power tubes. I put one BQ on a tester, and it checks good which matches up with the DC characteristics still observed in the amp. I set the project to the side for now to do actual gainful-employment-type work here at the shop.