What transformer trickery is this? (RG?)

[nuke]

This video by Rift Amps popped up in my YT feed.

The amplifier is a Divided-by-13 FT37. Nice amp, I've honestly only heard them from a distance and have yet to set hands upon one. Not a lot of reliable info on the 'net about the circuit.

Nice views of the guts of the amp for the curious. I did some careful over the shoulder viewing of the circuit.

The output section is a quad of 6V6's, in partial-fixed/partial-cathode bias, with a half-power switch enabling the second pair. The 6V6's are grouped on the chassis in phases and there's nothing weird about the wiring or circuit on the sockets or board. <-- this will be interesting in a moment...

But skip ahead to 25 minutes in and watch the oscilloscope waveforms. That's some interesting stuff going on there, and it is not from the pre-amp side.




I'm puzzled. The gears in my head are turning with ideas though. Interesting, whatever it is. The outputs are driven with clean signal from the PI, but the speaker outputs and the plate waveforms show that really interesting artifact.

[TUBEDUDE]

The anodes looked clean individually earlier in the video, so replacing the quad might not show anything.
I'd say it's being overdriven with an impedance mismatch on the OT primary.

[nuke]

No the clean waveforms were the 6V6 grid inputs from the phase inverter plates. You can tell by the amplitude if you zoom in on the scope's display.

The "notched" waveform was from the secondary (speaker side) of the output transformer and from the plates of the 6V6's.

What I notice from the scope is that it is one phase only, that the notch is exactly the same voltage point on the y-axis of the scope, that phase has reduced amplitude versus the other and the notched phase reaches a lower total amplitude when read from the plates of the 6V6's.

I'm thinking there's some trickery buried in that custom-made output transformer.

Been pondering how one might do such a thing. Came up with some ideas. Pretty sure, that's an "on-purpose" going on there.

[martin manning]

Also its not present unless the volume is turned up, maybe to where it transitions from A to AB1? Sequence starts at 25:00. Is it possible that there is magnetic coupling from the OT to the output tubes?

[R.G.]

About the only way inductors (and transformers, which are specialized inductors) can cause any voltage glitches is when something external tries to change their winding currents quickly. Then they definitely cause voltage glitches.

Like everything with transformers, this has a bunch of footnotes and asterisks.

In this case, I suspect that the real cause has been mentioned - the changeover from one to two pairs of outputs at a certain signal level. Another possibility is one of the cathode capacitors doing a minor breakover at some voltage level. That might cause a voltage notch.

Transformers are tricky because the input voltage "charges" the core with ampere-turns of magnetic field. Secondary windings can "steal" some of this magnetic field charge, but the physics then allows the primary to recharge the core instantly too. So power conceptually flows in from the primary and out through the secondary without affecting the field charged into the core. If you set up something to try to cram in or take out the magnetic field in the core that's not compensated for in the instant flow of energy into the core and out through the secondary, the field change jerks a voltage change on all the windings.

[nuke]

I don’t think it is that complicated manipulation of magnetics.

How about the primary having a partial tap, sort of like an ultra-linear but not as far away from the anode lead.

Now stick a zener diode across that tap, so that as the out tube plate conducts more current, the voltage across the zener increases until it breaks over the zener voltage, effectively shorting part of the winding and reducing the wind count.

Might be back to back zeners, but that’s how I’d do it.

I’ll draw a schematic later.

But that’s cool trick that no one would figure out by the schematic. Just hidden inside the transformer. Their customers wound so they could do whatever they wanted.

[TUBEDUDE]

So possibly simulated with a UL tranny and a zener.

[nworbetan]

Also its not present unless the volume is turned up, maybe to where it transitions from A to AB1? Sequence starts at 25:00.

Could a problem with one of the cathode bias resistors or capacitors cause that kind of issue? Half of the "push" side and half of the "pull" side operating at a different bias or unbypassed?

[Helmholtz]

I tend to think it's a fault, not a feature (whatever the clueless guy in the vid says about sound).
This amp will not produce a good clean output. Why didn't he measure output power?
One halfwave looks fine while some threshold seems to prevent a neat takeover to the other half.
First thing I'd check would be tube bias currents.
Of course, the voltages on both sides of the CT must be mirror images with a well coupled OT.

[nuke]

The notch only kicks in above a certain amplitude.

The cathode capacitors are shared across each push-pull pair. The first p-p pair share a cathode resistor and bypass cap. The second p-p pair share the other resistor/capacitor via the full/half power switch. So, it ain’t that.

[nuke]

So possibly simulated with a UL tranny and a zener.

Exactly.

All hidden away inside the transformer.

Just don’t know how many winds in the tap is and what voltage the Zener might be.

If I had a UL transformer handy, I’d try it.

[pdf64]

I got bored, did he ever try a known good set of output valves?

[martin manning]

No. From 34:30 you can hear what was checked before the owner picked it up.

[martin manning]

In this case, I suspect that the real cause has been mentioned - the changeover from one to two pairs of outputs at a certain signal level.

Not the cause but I think that’s when it happens. The kink only shows up in the positive swing of the output signal (25:10), so it’s only happening on one side of the power stage, I believe when that side hits cut-off. And, as the narrator points out there is already significant distortion before the kink appears.

[nworbetan]

In this case, I suspect that the real cause has been mentioned - the changeover from one to two pairs of outputs at a certain signal level.

Not the cause but I think that’s when it happens. The kink only shows up in the positive swing of the output signal (25:10), so it’s only happening on one side of the power stage, I believe when that side hits cut-off. And, as the narrator points out there is already significant distortion before the kink appears.

If you re-watch the video, at the 27 minute mark he shows the anode voltages at both ends of the OT primary and explains that "if we go to the other side, the other pair of 6V6s, you can see that kink is there but it is obviously inverted."

I haven't used LTspice lately, but I just updated that and dug up my project files and I'm about to get started running some simulations to see if I can come up with something that looks similar. I'm expecting that to take a couple/few hours to knock the rust off and set up a good hybrid fixed+cathode bias power amp. I won't be surprised if someone can do it faster than me.