Very low frequency noise/motorboat?

Take a look and see what you think this might be. (Schematic attached).

https://youtu.be/-PJO9xfy4Ps

1. The low freq ripple/noise is still apparent even when quickly disconnecting the PSU cable as the unit cools off for a few seconds. This makes me believe it is not the PSU itself, which is a basic RCRCRC design.

2. It increases and decreases with gain rotary switch.

Try upping C20, you could bridge it with another 10uF to try it out.

Does it alter if you disconnect the first stage cathode bypass cap?

Cheers

Ian

Tried bridging the 10uF with another 10uf and a 100uF. No change.

Also tried adding a 10uf to the first gain stage. Nothing.

Disconnecting the first bypass cap does make it increase slightly.

Decouple the 1st & 2nd stage with another RC.
Duke

Maybe even two rc series stages extra for the first valve .

Not enough data. Is this one channel or all four connected. What does the B+ output look like, specs.

Also context is missing and it's hard to follow with all the tweaking since you have spread the troubleshooting topics to 10-15 different threads.

All four channels are spitting out this same identical low frequency oscillation. Have also isolated and disconnected all but a single channel, and it’s still there.

Kingston said:

... you have spread the troubleshooting topics to 10-15 different threads.

Click to expand...

Not 15, surely?

PRR said:

Not 15, surely?

Click to expand...

Trying to become 10-15 times smarter than I was this time last week.

With just one channel connected, isolate stages. Starting from input stage moving on to the output connect each grid to ground. Testing just the output is easiest by disconnecting V1 completely so you don't have to detach stage 2 plate.

This will narrow down the problem.

Motorboating was often associated with power supply softness providing an alternate path for output to corrupt inputs.... or not.

JR

Just noticed R29 2M2 C21 0.1uF filter seems like a slightly unhealthy pairing and their corner frequency is 0.72hz. Looks like the motorboat almost dances to this tune.

For a quick debug, replace R29 2M2  with far higher resistance, like 10k. If that helps, then simply replace this feedback path with 1uF and higher resistance that still gives you flat response.

The only thing that seems to work is lowering the value of the output C22 and essentially creating a HP filter. The low frequency wobble is all through the circuit.

Isolated down to one channel. Bypass v1 entirely and it's there. Bypass v2 entirely and it's there. Added another RC to the B+. Doesn't fix it. Added a few RC stages between the tube sections with no luck.

I'm stumped.

BluegrassDan said:

The only thing that seems to work is lowering the value of the output C22 and essentially creating a HP filter. The low frequency wobble is all through the circuit.

Click to expand...

Are you measuring with a 600 ohm load at the output or into a higher impedance? As it stands, with a 600 ohm load the -3dB point is about 3.4Hz but it could be a lot lower into a 10K load for examples.

The other thing that comes to mind is the resonance between the 4.7uF output cap and the primary inductance of the output transformer. This can often cause a significant hump in the very low frequency response which might expalin why extra desoupling makes no difference.

Cheers

Ian

I think we’re narrowing it down to the output. Sweeps indicate an up and down ripple below 20Hz. Looks kinda like transformer ringing, but low frequency.

Also realized that the oscillation itself is nothing more than regularly expected  noise being amplified by the frequency bumps.

Test unbalanced with output transformer replaced with 10k resistor.

Kingston said:

Test unbalanced with output transformer replaced with 10k resistor.

Click to expand...

Good idea. Also try with 150 ohm resistor in series with 4u7 op cap. The size of the hump depends on the Q of the LC circuit which in turn depends on the circuit resistance. The White follower op Z is very low so the the main resistive element in determining Q is the dc resistance of the transformer primary. Adding 150 ohms in series should dampen the resonance somewhat and reduce the hunp.

Cheers

Ian

Do you also recommend strapping 620 ohms across the output?

BluegrassDan said:

Do you also recommend strapping 620 ohms across the output?

Click to expand...

With the transformer, sure. It's is a good frequency response test compared to the usual high impedance A/D inputs. I certainly have output stage and transformer combos that prefer loaded secondaries (read: flatter frequency response). But I doubt this removes the motorboat.

Unbalanced test needs to see that 10k load. This configuration probably can't drive 600ohm. 2-4k perhaps just looking at it (I didn't calculate)