Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

How to kill HF parasitic oscillation?

7/15/1999 5:43 PM
Leo Padron How to kill HF parasitic oscillation?
Following all your helpful comments regarding my Tweed Pro clone, I've recently concluded that my amp has been stricken with the deadly HF "Parasitic Oscillation" problem.  
This is my first project, and I did it cramping the components top-wise in a 13x5x2 chassis with slightly messy point-to-point wiring using terminal strips, so I'm guessing (a) the proximity of the component leads and (b) the proximity of the components themseves may have much to do with this.  
What are the most likely culprits for HF parasitic oscillation and what can be done to kill it? What are the troublesome points in particular and possible fixes?  
And now, a word from our sponsors:

7/15/1999 6:57 PM
Michael Cameron

You might want to try grid stop resistors mounted as close to the tube pin as possible on the preamp tubes.Try different values and settle on the smallest that does the job. Leave like 1/8" lead on resistor and solder quick so as not to destroy resistor. This is probably the least tone invasive oscillation killer that I know of.  
The 5E5 already has grid stoppers on the output tubes. 12AY7 would have less gain than 12AX7 and might just be enough drop in gain to stop the oscillation. You also might try hookin amp up to speaker and poke around your grid wires with a Sharpie to see if that changes anything. On a couple of my homebrew amps changing the orientation of coupling caps 90degrees has killed and oscillation.  
Thats all I can currently think of with out going to suppression caps that suck the life out of the amp.  
I built a 5D5 a few years ago and think it is great. I go back and forth between 5881 and EL34's .  
Good luck  
7/15/1999 7:20 PM

When I built my first amp an AC30, I didn't have any clue where the wires were supposed to go, so I just threw it together. It turns out I had a HF oscillation, which made the amp extremely noisy, along with the fact that it barely made any guitar sound at all.  
The problem? I had my screen grid resistors for my power tubes too far away from the sockets, this made big loops in the screen suppy wiring. The screen grid and signal grid wires for the power tubes should be as short as possible.  
I wish you luck, as I know how frustrated you are.  
P.S. your most probable next problem.... HUM REDUCTION
7/15/1999 7:49 PM

I had the same thing, but mine turned out to be a coupling cap problem. I was using cheapo ceramics. I wanted to try for a "grainy" sound. Anyway...I took out the ceramic disks and replaced them with Orange Drops. Problem Gone.  
7/15/1999 9:22 PM
ken g
"What are the most likely culprits for HF parasitic oscillation and what can be done to kill it? What are the troublesome points in particular and possible fixes? "
OK, in my experince, there's a few things you can do. Essentially you have too much gain at high frequency for the current layout. You may be able to a) lower the gain, or b) rolloff the HF, or c) change the layout, and remove the oscillation--any one sufficiently done, or a combination of all of them, will do the trick.  
The HF rolloff is the easiest... just shunt a cap to ground from a signal node. This can be across a plate resistor up to B+, or it can be from grid to ground. You can also connect a small ceramic cap from plate to grid on one stage--this will have a profound effect, since the two elements are out of phase, and the gain of the tube will multiply the effective capacitance. All of these techniques create a HF rolloff (a pole in network speak) of 6 dB/octave. This will limit the extent of the HF the amp will pass as well, though, so I usually try to avoid it unless absolutely necessary. It has been necessary in some cases. Usually placing the rolloff at an earlier stage will limit the value of capacitance needed, and reduce the loss of HF. Earliest place would be on the plate of the input stage.  
Decreasing gain: This could be increasing cathode resistors, removing cathode bypass caps, decreasing plate load resistors, switching tube types, increasing the amount of voltage division between stages, etc. Obviously this is not the ideal solution if you're going for a high gain design, but sometimes all it takes is a small amount of change to make the amp stable.  
Changing layout. This is obviously the hardest one to test and impliment. Some things I've noticed--coupling caps usually have quite a large EM field around them. As a result I usually keep my cap leads as long as possible so that I can bend them this way and that to try to put as much air around them as possible. Remember what this oscillation is caused from: a positive feedback at high frequencies. That means you've got to have two nodes that are IN PHASE with one another... the one with the higher voltage, usually the later stage, is coupling with the earlier node, probably capacitively. Putting space between the components that would be in phase is one step in the right direction. Also take a small film cap and bypass your decoupling caps (which are probably electrolytic) to make sure its not leaking through the B+ rail.  
All of these tests should be done--!!VERY CAREFULLY!!--while the amp is on and oscillating. I haven't been zapped doing it yet... just be careful if you use gator clips, since the cap you're touching here and there will probably be swinging around underneath/beside the amp. One (or maybe both) leads may have HV on it, so I usually clamp the cap in the old Panavise to keep it in one place and not brushing my willie.  
Try some of these out. The most ideal is the layout change, since you're keeping the circuit exactly the same. The easiest (and the one Mesa Boogie seems to like) is the cap shunt to ground. You may not even notice a change in tone with that mod--if done properly (and the situation isn't hopeless) you shouldn't.  
Good luck  
7/15/1999 9:44 PM
ken again
Oh yeah, don't forget about the lead dress from the finals to the OPT. Those are the highest AC voltages in the amp, and the OPT is often located far from the PT, which is to say it's close to the input.  
7/15/1999 9:42 PM
1. try the PI "shunt" cap (as seen in the 5F6-A etc) start with a 47pF and work your way up.. (SilvMica cap from plate to plate on the PI)  
2. try bigger grid stoppers on the outputs.. 5.6K to 10K  
3. re-work your output section wiring paying attention to :  
a: try to keep crossing signal leads at 90 deg from each other to minimize interaction  
b: keep speaker leads away from grid leads  
c: get as close to the OEM layout as possible  
d: keep plate leads away from other leads (down  
against the chassis is usually good)  
Be patient, do ONE thing at a time and don't lose  

  Page 1 of 2 Next> Last Page>>