ST120 power section oscillation using Pavel's AB-Qi ST70/120 autobias board

System: VTA ST-120, KT88's, Pavel's autobias board AB Qi ST70/120, VTA driver board version 16.

I'm experiencing an approximately 200 KHz oscillation, that runs away with increasing input. Both channels are affected. At onset the oscillation looks like this (measured at the output into 8 ohms.

With slightly more input it runs away:

Zoomed in it looks like this:

Has anyone else run into this? Voltage regulator instability?

To follow up: I've isolated the input/ driver circuitry and can run it up to clipping without any sign of the oscillation. I assume its occurring in the power section, including the AB board.

I had a similar problem some time ago but it was on my preamp. The standard way to chase this down is to understand the schematic and probe with your scope specific locations in the signal path. There are also some places in the power path as well. You have to be very careful and it can be a lengthy process on a workbench because you power down to change to many of the probe points. I have even soldered a short 16ga piece of wire on some location to be able to probe them. It is a good idea to have your scope powered on a lifted ground so to minimize risk.
ONE MORE WARNING: This is a lot of volts so do not move probes and grounds with the power on and make sure you take the time lt let the caps bleed down. I am seasoned with more than 40 years working with this stuff and I still turn off the power to move probes.

Some add 47Ω wrapped with bell wire which is effectively 0Ω at audio frequencies.



These create an additional pole that suppresses the lead / element RC pole

See Red Plate

I installed the auto bias kit some years ago (fantastic board) and started using KT88 drivers. To protect the auto bias board and my rectifier, I also added the micro-fuses on each driver. That saved me headaches twice now. Much easier to change out a socketed fuse than buying new rectifier tubes.

Now, I am considering the dual rectifier kit. Using solid state rectifiers on this design IMHO is foolish.
I have been doing fine with my Telefunken rectifier tube.

HDroller wrote:System: VTA ST-120, KT88's, Pavel's autobias board AB Qi ST70/120, VTA driver board version 16.

I'm experiencing an approximately 200 KHz oscillation, that runs away with increasing input. Both channels are affected. At onset the oscillation looks like this (measured at the output into 8 ohms.

With slightly more input it runs away:

Zoomed in it looks like this:

Has anyone else run into this? Voltage regulator instability?

I think it would help if you checked the ground connections.
The connection of the driver and AB to ground should be as short as possible and with thicker wire. Also, the earth point is very important - this needs to be tested and the occurrence of oscillations monitored.

Pavel

"Using solid state rectifiers on this design IMHO is foolish."

Tube rectifiers are great for musical instrument amps.
They have no place in a HiFi system.

HiFi power supplies should be as stiff as possible, not drooping when the going gets going.

AND since tube rectifiers wear just like audio tubes, amp performance is doubly compromised over time.

HDroller wrote:System: VTA ST-120, KT88's, Pavel's autobias board AB Qi ST70/120, VTA driver board version 16.

I'm experiencing an approximately 200 KHz oscillation, that runs away with increasing input. Both channels are affected. At onset the oscillation looks like this (measured at the output into 8 ohms.

a nice description of finding wobbling oscillations with a spectrum analyzer can be found at
https://www.diyaudio.com/community/threads/oscillation-in-tube-amps.145103/

Interesting information Seamus. Thank you. Those resistor chokes are in series with the OT Plate and screen taps? I'm trying to learn more about oscillation. The more I read it looks like there are many ways to produce oscillation. In the case with the resistor chokes, was it an interaction between the OT and the KT88? A resonance?

I need to look more at the ielogical site..

HDroller wrote:Interesting information Seamus. Thank you. Those resistor chokes are in series with the OT Plate and screen taps? I'm trying to learn more about oscillation. The more I read it looks like there are many ways to produce oscillation. In the case with the resistor chokes, was it an interaction between the OT and the KT88? A resonance?

I need to look more at the ielogical site..

Those resistor chokes are tend in series with the circuit heating  tube in series with the circuit heating  tube. Prevent signal transmission along the path of the heating.

The snubbers are in series with the plate and screen to swamp Miller Capacitance.

This is an excellent article @ Aiken Amplifiers on Miller Capacitance

I've not experienced any problems with heater contamination on tube amps, but that's not to say it can't happen with some valves and circuits.

The grounding, and lead dress, has been cleaned up to the best of my ability. The oscillation remains though, with the same behavior--clean at small (100 mV) input, Oscillation onset at 200mV, then blows up at 300 mV input. Below is the time and frequency domain responses at 200 and 300 mV, measured at the grid on V6

At 100 mV input nothing untoward shows up in the spectrum.

I went around with a sniffer and the oscillation shows up everywhere at 300 mV input: grids, cathode resistors, coupling caps, even the B+ rail.

Seamus, thank you for clarifying the resistor/choke placement, I'll check out Aiken.

Pavil, do you have any thoughts? Or points of measurement?

To any and all, a general question on the VTA ST120 power supply: What is the purpose of the 1 uf ESL cap?

HDroller wrote:The grounding, and lead dress, has been cleaned up to the best of my ability. The oscillation remains though, with the same behavior--clean at small (100 mV) input, Oscillation onset at 200mV, then blows up at 300 mV input. Below is the time and frequency domain responses at 200 and 300 mV, measured at the grid on V6
At 100 mV input nothing untoward shows up in the spectrum.

I went around with a sniffer and the oscillation shows up everywhere at 300 mV input: grids, cathode resistors, coupling caps, even the B+ rail.

Seamus, thank you for clarifying the resistor/choke placement, I'll check out Aiken.

Pavil, do you have any thoughts? Or points of measurement?

To any and all, a general question on the VTA ST120 power supply: What is the purpose of the 1 uf ESL cap?

I would check the feedback circuit from sec. OT, R7 (R8) C11 (12) and grounding of the OT secondary side.
Pavel

HDroller wrote:To any and all, a general question on the VTA ST120 power supply: What is the purpose of the 1 uf ESL cap?

The ESL bypass cap is supposed to clean up HF garbage on B+. However, it is a bit more complex than simply hanging a film cap across the main filter caps. Capacitors have L&R properties as well as C. Too much L may cause more problems than the C removes. You could try lifting it and see what happens.

It's not surprising the driver alone doesn't oscillate. It uses almost no current, so the B+ is very stable.

What tap are you loading with the speakers / load resistance?
If you are using a load resistor, is it non-inductive. Some load resistors claimed to be non-inductive are very inductive.

What rectifier are you using?

What happens with one channel driven and the other input back-shorted?

What size are C13 & C15?

You could increase C12 to 330pF to lower bandwidth. This will be inaudible.

You could add 220pF across R4 to clean incoming garbage.

How clean / robust is your AC power?

Any big cell towers or radio operation nearby?

Thanks for the insight on the ESL cap Seamus. I suspected it was an attempt to shunt off higher frequency hash.

I was able to tame the oscillations a bit by decreasing NFB (larger resistor). With the larger resistor the onset of oscillation happened at higher input levels, but would still oscillate at higher input. I then stumbled across grounding the OT secondaries to the chassis, and with both of those things the oscillations stopped. Many thanks to Pavel for steering me to look at the NFB--he also suggested grounding the secondaries. Morgan Jones (Building Valve Amplifiers) talks about oscillations occurring in the 30kHz-300KHz range with too much NFB.

I'm still left with the initial problem of intermittent bias on V6. It appears as an under biased signal (cross over distortion), that will snap to full clean sinusiod, then drop out entirely, or it will sit at an under biased level. Sounds distorted. So far it appears to effect one half of one channel. The other three tubes are biasing OK.

I've been using the 8 ohm tap for speakers and testing. The power resistors are wirewound. I don't think those are specifically non-inductive.

The rectifiers are UF5408 (2 in series for ea leg), full wave configuration w/ PT CT.

I haven't tried shorting the input on one channel while driving the other. Is back shorting the same as shorting the input? I have RCA shorting plugs for shorting the inputs.

C13 & 15 are 100uF, C12 is 220 pF. For C12, it looks like it bypasses R7 to ostensibly cancel HF oscillation?  The caps across R3 & R4 would be to shunt RF to ground at the inputs? Would you suggest ~500KHz for roll off frequency? I'm thinking low end of AM radio. I need to do more work...

No cell towers, or radio for that matter. We've got the empty ocean to west and mountains to the east.



AC power is ~120-121. Clean and reliable?...meh.

After much more testing, I'm arriving at the thinking the AB board is not the problem. I think I got fooled by the cross over distortion, and fluctuating amplitude I observed at the output. If I understand the AB board correctly, all it does is provide VDC to the grids of the output tubes. However, I'm less confident I know what is going on with the cathode current monitoring. That said, when I monitor the VDC at the grids during the fluctuations the VDC is steady. So it seems the problem is on the AC (signal) side...or at least elsewhere. Also, both sides of the right channel, V6&V7, exhibit the problem, not just V6 as I previously reported.

Pavel, thank you for enduring my suspicions about the AB board.

I think I was also fooled by evaluating the input/driver circuitry with the output tubes removed. With the output tubes in place, V1-3 exhibit the fluctuations w/ crossover distortion, dropouts, etc. At the same time, the VDC (B+ voltages, plates (V1, V2, V3)), cathode voltages, are steady and where they should be for V1-3.

Grasping at straws...the VDC at the power tube plates is 500-515 VDC thanks to 121V line voltage, and SS rectification. Is it possible there is leakage, or arcing occurring in the OT? I don't know what the OT's are rated for. I'm guessing the symptoms would be catastrophic though. It is interesting that during playing music things drop out more at lower input levels, and I can cause things to "catch", and momentarily function by increasing the input levels.

Here is the errant behavior caught on the scope. Taken at the right channel output.
The link is to a 30 sec video I'm attempting to share from dropbox. I hope it works...

https://www.dropbox.com/scl/fi/d77lhsx5zqfr6osaa65pm/IMG_3580.MOV?rlkey=x991cver9w1irbhp9zsygmyl1&dl=0

What's a tale without a gutshot?

https://i.servimg.com/u/f76/20/52/15/08/gut_sh12.jpg

C12 makes a LP feedback.

Input LP could be 50k-100k, not 500kHz as the amp doesn't get anywhere near that. It's invisible on square wave response.

============================ 

I'm not a fan of grounding on the chassis. IMO, a single star point ZERO volts and the chassis as a Faraday cage ONLY isolated from 0vA by an Earth Safety cct.

AB boards adjust bias to setpoint under no signal conditions. The have a <1-2Hz[?] LP filter to prevent them from adjusting bias during playback.

============================ 

If the plate voltages remain constant on a scope on the problem channel, the OPT is probably OK.

In the video, the noise at the start is wavy? Is that with an osc connected? What does noise look like w input shorted?

You can probably run w/o feedback connected to see where problem occurs.
It looks like the bias cuts out on both halves of the waveform.

Follow the signal from input w/o feedback.

Also check the outboard ends of the two diodes between the four capacitors on the AB boar near the input transformer. Failure of the input power is about the only thing that could cause the symmetrical underbias.

Which channel is which in the inside view?

WAG:
Are the Heater polarities the same on both channels?
Heater CT correct and caps OK?
What happens if you change AB power channel / polarity.

All talk and no data? No way...

I spent the better part of yesterday taking measurements. Here is a link to the data which includes: VDC points, Performance data so far, and if you're interested in the ST-120 OT, I took data for three samples.

I appears to be making power, but the phase splitters operating points seem off (one side clipping), hard clipping of the signal from Rk on the AB board. The initial problem of Right channel drop out and under biasing still exists, and is unsolved.

https://www.dropbox.com/scl/fi/7ik7krh2s72etf4dnjxyi/ST120-Measrements-100323.xlsx?rlkey=7bd9o3sag7cikyz7rdyzfgnye&dl=0

This is the left channel looking at the plate of V5 (yellow), and across the 8 Ohm load (blue), just before clipping starts. It's at least making power. I missed adjusting the range on the output and it's clipped, so actual is higher.  The right came in at a little over 20V at the output.



This is looking at the grid of V5 (blue), and the cathode resistor (red) on the AB board. It's a .pdf so click to view. This was taken at the same time as the above. All the cathodes looked like this--aside clean grid signal. Clipping occurred earlier than the max input I tested. Seemed like it was running into a limiter on the AB board.

https://www.dropbox.com/scl/fi/f7q9nl994kkqfcg7it48i/Rk-AB-V7-1.42V-in.pdf?rlkey=5ya0ke59tzy0g562yjglb1sx9&dl=0

HDroller wrote:This is the left channel looking at the plate of V5 (yellow), and across the 8 Ohm load (blue), just before clipping starts. It's at least making power. I missed adjusting the range on the output and it's clipped, so actual is higher.  The right came in at a little over 20V at the output.



This is looking at the grid of V5 (blue), and the cathode resistor (red) on the AB board. It's a .pdf so click to view. This was taken at the same time as the above. All the cathodes looked like this--aside clean grid signal. Clipping occurred earlier than the max input I tested. Seemed like it was running into a limiter on the AB board.

https://www.dropbox.com/scl/fi/f7q9nl994kkqfcg7it48i/Rk-AB-V7-1.42V-in.pdf?rlkey=5ya0ke59tzy0g562yjglb1sx9&dl=0

Limiter in cathode circuit of power tube - varistor on AB - switches at 15V pp. Cathode resistor on AB is 10 Ohm / 2W. At 15V this resistor would burn up. So the limiter does not cause signal limiting.
Pavel