• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Oscillation in tube amps

scott17,

I look forward to your schematic, if and when you find the time.

In my Post # 299, I looked at the schematic in Post # 290.
It was the last schematic posted, and when someone after a posted schematic talks about their amplifier but does not say which schematic, I automatically assume the last one to be what their amplifier is like. That is often the wrong assumption.

You said you changed from triode mode to UL mode. That is easy for me to look at another schematic, and make the change of the screen connection in my mind.
You also said that you had a 133kHz oscillation.
My post 299, section B. 3. was to answer your questions.

Even though it was only in one channel, and swapping the output tubes did not fix the oscillation; and the output transformers should be fairly identical, the right channel could be just beyond good stability, and the left channel just inside of good stability.
Perhaps the wiring dress is different for the right channel, than the left channel.

It would be nice if we are able to figure out the cause of your 130kHz oscillation, even though you seem to have a fix for it.

I always use a non-inductive 8 Ohm load resistor for testing my amplifiers.
Just because one of my amplifiers is stable on that load, with a loudspeaker if may not be stable.
So after testing with the load resistor, I test with one or more loudspeakers.
Different loudspeakers have different impedance versus frequency.
I put an impulse through the amp and test the broadband spectrum; If that looks good, then I load the CD player and listen to music (with the scope / spectrum analyzer still monitoring the amplifier.


The other parts of my post were for other posters, based on what they said (of course you might find those additional answers I gave useful too).

In earlier posts there were comments about the feedback, and about the cathode bypass cap.
My post 299, sections A., 1. 2. were to give my thoughts on those earlier comments.

NickCool2000 mentioned hum and buzz. My post 299, sections B. 1. and B. 2 were to address that.

My post 299 section 4. was to warn newbies that the secondary must be grounded. The schematic in post 290, sort of hid the ground symbol there, easy to overlook, I did too.
 
Last edited:
  • Like
Reactions: 1 user
Thank you 6A3sUMMER for the explanation.

A) About C2 and C1 (22nf) the idea was to increase the treble using this by-pass for the high frequencies, that's why it has this weird value. I will remove C2 and increase C1 to 100nf and let you know.

B) I think I will avoid the open loop at DC by putting a resistor in parallel with C3. The idea of this loop was to extend the bass by decreasing the feedback. And in fact it sound with very good bass.

C) I tried to lower R4 from 180k to 100k, with no luck, the amp started to oscillate. I know that ECL84 is problematic regarding oscillations. :(

D) Hum and buzz, much better with the new regulated PS, but still need improvements.

This is the new schematic (not tested yet):

Screen Shot 2022-12-24 at 18.50.23.jpg
 
B) I think I will avoid the open loop at DC by putting a resistor in parallel with C3. The idea of this loop was to extend the bass by decreasing the feedback. And in fact it sound with very good bass.

C) I tried to lower R4 from 180k to 100k, with no luck, the amp started to oscillate. I know that ECL84 is problematic regarding oscillations. :(

B) It‘s open loop at DC anyway - C22 is responsible for that. Putting the bypass on C3 results in a mA and a half of DC leakage into R101 - the bias point of V1 will shift. Ok if that’s what you want, but it might not be. Just be aware that this will happen - R101 might need to change.

C) The dominant pole is formed by the 180k an the total input capacitance of V2P. It obviously doesn’t handle going up by a factor of 1.8x (Makes GBW too high and runs out of phase margin). Putting the dominant pole there can be problematic since it’s frequency does depend on the loading in the plate circuit (Which affects gain and the amount of Miller multiplication). It might not be stable for all loads. It is more customary to make R4 small and put a cap across R22 to make the dominant pole, because it is more weakly dependent on what load you hang off the transformer.
 
  • Like
Reactions: 1 user
Xmas are happier when the projects start to run... :cool:

After debugging a bad soldering and some experimentation, this is the new schematic (not fully tested yet).

Without the C1 (to cut the highs) and with R11 at 470k (dropped to 150k to cut the lows) the BW exceeds 20hz-20khz. No oscillations.
R4 still has to be dropped (to 100k?)
Hum and Buzz were not tested yet, I was using a resistor as load and the oscilloscope.

Screen Shot 2022-12-25 at 21.33.08.jpg
 
6A3sUMMER,

I have attached the schematic of the amplifier with the 133KHz oscillation on one channel. The output section is from a design by diyaudio member famousmockingbird. The input section I calculated using the Mu Stage Philosophy article by Alan Kimmel.
I attached a scope image of the right-channel 7.5usec/133kHz oscillation as connected to a 8 ohm non-inductive load. Also a scope image with the RC network. The left channel did not have the problem. It was easier to wire the RC from UL to ground than across the plate and UL connections, and it seems to have the same effect. The amp wiring is more or less symmetrical from left to right channel. The RC did fix the issue. Any thoughts as to what might have caused the oscillation? Just a chance issue with one of the output transformers? Thank you for looking at this.

with_rcnet.jpg 7.5usec.jpg 6HJ5SEUL1.png
 
Do the 100V zeners have plenty of zener current going to the screens? Not just a quiescent condition, but during the maximum negative going signal at the output tubes g1 grid?
Does the screen current drop below the minimum rated zener current?
Is one zener is better than the other?

If you disconnect the coupling capacitor from the input tube to the output tube g1, does the oscillation stop?
(even if you remove the RC on the UL tap)?

Perhaps you have a ground loop on one channel only, and the other channel ground loop is marginally stable.

I am not familiar with the input tube, and the output tube.
 
Last edited:
Hi,

I have a new 5F2A princeton based on MojoTone's kit that I'm retrofitting into an old empty 8" tube combo cabinet (an ex 120v 1960s widowmaker) and I'm going bonkers debugging a nasty howl that kicks in at full gain.

The amp stays clean, if under-powered, then starts acting like a proper 5W tube amp loud yet busted fuzz pedal before it goes into oscillation.

My oscilloscope is dead and I'm off work or I'd borrow one from there, but testing shows the issue is 100% coming from the gain stage (ignore my comment in the video, I've never noticed a power tube flicker like that, but then I've never looked for it)

Voltages are within reasonable tolerances, I've traced and retraced the circuit, done chopstick and hot meter testing. I've also swapped tubes, touched up nearly every solder point, re-routed wires, no joy. I've refurbished a number of tube amps, including a vintage super reverb, I'm educated and work in electrical/electronic engineering. This is my first scratch build and I've spent 20+ hours trying to sort this out, but I have to admit defeat and defer to those with more experience.

So, power tube out, AC volt meter on the input pin of the power tube holder and I see voltage when the preamp oscillates. Oddly it seems tied to the tone knob as well. If I have to trace the circuit again I'm going to pull whats left of my hair out. I'm ready to rip it apart and start from scratch, any tips on troubleshooting my howling gain section because I am stumped.
 

Attachments

  • 20221228_205252.mp4
    4.6 MB
The high-gain reissue Fenders have problems with preamp instability. The 22µF 500V preamp supply caps age rapidly, and the amps sometimes fail after only three years or so: they make varying noises, from scrreeching to howling. Replacing the 22µF parts restores them.

Use new electrolytics from authorized distis. Bad capacitors are easily encountered. Caps from ebay or salvaged from old gear are not suitable in this position. You could try adding a little more capacitance (though it may alter the overdrive sound), but don't use low-impedance electrolytics, as the ESR is likely to act as useful damping. Nichicon VR (or its axial variant) Samwha SD or RD, and Panasonic M or AM are known-good.

The wiring-length to the capacitor should be minimised, and the ground side wire should not carry current from other stages.

It could be caused by numerous other problems, but it answers very well to this one!
 
Apparently we are talking about 2 or more completely different amplifiers.

Recently, we have schematics in Post # 302, 305, and 306. None of those have tone controls.

Post # 311 talks about tone rolled off. (tone control?)

It is hard to keep track of which Posts are about what in this thread.

Please list what schematic we are talking about for Post # 311.
Thanks!
 
And here we are now, the amp works well but only with a source in the input.

Let's call this amp the ECL84 to avoid confusions.

If I remove the source (an MP3 player in this case) it starts to oscillate at around 12khz.
When plugged again the oscillations disappear.

You will see a 330p cap in parallel with R22, and the gid stoppers R2 and R4 reduced to 10k and 100k. I also reduced R3 to 180k to have -3db at 15hz and reduce subsonic stuff. All these was done to avoid oscillations, that were mitigated but not killed.

The PS is regulated and well filtered, the heaters are working with DC, all is over PCBs but still on the bench, not in a cabinet.

Any ideas?

Screen Shot 2022-12-30 at 09.44.21.jpg
 
Last edited:
Connecting C1 in parallel with R22 isn't right:

B= 1/R + jwC

PHI= arctan(wRC) (After some complex number arithmetic)

That means that the phase shift approaches 90deg with increasing frequency, and that's not what you want. To improve phase margin, you need to parallel the plate load resistor with a Zobel -- a series capacitor and resistor. The resistance is typically:

RZ= 0.1RP

CZ is a compromise between high frequency roll off and phase margin. You could keep the 330pF and see how that works. The series resistor serves to keep the phase from going to 90deg -- a value likely to cause enough phase shift to promote high frequency oscillation.
 
  • Like
Reactions: 2 users
Weirdness continues. When and ONLY when I have the gain maxed and the tone rolled off it works like a charm. Other than that it fizzles out or goes into oscillation.

Darn it! I only have 3 colors of wire and even though I traced it out over and over, each time I failed to notice my wire from the gain pot to the 2nd gain stage of the 12ax7 ran under the heater wires, in-between them and the grounded chassis.
(I finally got a new scope and saw the noise was a mix of 60hz plus harmonics so that helped a lot!)

Anyway, I rerouted the signal and everything is good now. 👍
 
  • Like
Reactions: 1 user
Here's an interesting one. Just finished rebuilding an amp (https://www.diyaudio.com/community/threads/crcrcrcrc.401998/#post-7421982). Tested great on the bench. Sounded great on the bench. Put the bottom cover on and hooked it up to my main speakers, and it sounded very weird. The right channel was all wonky. Put it back on the bench and found nothing on the scope. When I turned the volume knob all the way up without a signal, I saw this in the right channel:

Oscillation.jpg


The oscillation disappeared when I touch the RCA input ground. I switched the tubes around and the same thing happened, just in the right channel. With the volume knob turned back down, the oscillation disappeared. I unhooked the RCA cable, and the oscillation returned, regardless of the position of the volume knot. The oscillation disappeared if I keep my finger on the RCA ground, regardless of the volume knob position. I hooked the RCA input back up and noted that the oscillation returned if I touched a screwdriver on the chassis and brought it close to the capacitor.

Right Channel.jpg


In the stable left channel, the ICEL logo was towards the output tube side, so I figured the external foil lead is on the wrong side. I unsoldered the capacitor and tested it on the oscilloscope, but I couldn't tell which side was which as the magnitude of the 60 Hz signal did not change in either orientation. I flipped the capacitor and put it back in, and the oscillation disappeared. The culprit was definitely the capacitor orientation, but I don't understand why the oscillation disappeared when I touched the RCA ground on the outside. Any ideas?
 
Member
Joined 2019
Paid Member
I'm having a very hard time taming the D3a in a single-ended circuit I'm working on, seeking some advice as I'm running out of ideas. Any thoughts on what is or is not helping so far would be appreciated.

Here's a schematic, it's a single-ended headphone amplifier with a transformer coupled D3a input stage, cathode biased. A current source in the power supply rail sets the current for the D3a. Each channel has a Maida style regulator. Output stage is a HY69 directly heated pentode wired as a cathode follower with the OPT in the cathode, biased by the OPT DCR.

HY69 Schematic.png

Here's what the amp looks like, sounds good despite a measurable problem.

PXL_20231224_194016898.NIGHT.jpg

I've tried a number of different configurations of stoppers and ferrites. Here's where I'm at right now:

  • Switched to ceramic sockets with shields
  • 3.9K CC stopper on G1 with a ferrite
  • 100R CC stopper on the plate with a ferrite
  • Two 100R CC stoppers in series connecting G2 to the plate
  • Ferrite on the cathode, no stopper. 1000uF electrolytic bypass cap with a 1nF ceramic bypass cap for HF decoupling.
  • Heaters are AC, DC elevated 40V. I have ferrites on the heater leads and they are bypassed to ground with a pair of 1nF ceramic caps. There are also some 680mOhm metal film voltage dropping resistors, heater voltage is a little high otherwise.

I wired the ceramic sockets keeping leads as short as possible. Center pin is the grounding point for the electrodes, then a single 20AWG wire from the center pin to my star ground.

PXL_20240104_195704005.jpg PXL_20240104_195712521.jpg PXL_20240104_195718319.jpg

Here's a look at my layout on the right channel. Some things are less than ideal - heater leads are quite long from the rear of the chassis. Leads from the IST to the D3a plate are 4-5 inches.

PXL_20240104_211738476.jpg PXL_20240104_211746565.jpg

Here's is what I see on scope at the grid of the HY69 zoomed into a 1kHz square wave. The ringing does actively pulse, telling me the D3a are trying to oscillate. Also a FR sweep showing tendency for HF oscillation.

PXL_20231226_022642434.jpg D3a FR.jpg

Some things I haven't yet tried:
  • Putting a 10R stopper on the cathode (although that will likely mean longer leads
  • Changing the heaters from AC with DC elevation to just DC (could be achieved with a voltage doubler)

I know the D3a is a tricky one, but I thought I would have made some progress by now. Any thoughts are appreciated.