Ultrapath

[J M Fahey]

Would that be .030, .045 or .060?

Well, it depends on your fuel mixture, compression and how oily/carbonized is your spark plug.

Back to Electronics, this is what I meant, a typically German solution to transformer killing overvoltage.
Fully pre-adjusted at Herr Von Dynacord's Laboratorium, of course:

[Malcolm Irving]

There was a discussion over on diyaudio about a similar circuit (in which there was a bypass cap from the cathode to ground as well as the cap to B+):
Ultrapath with cathode bypass cap: which value of C is correct? - diyAudio
The general idea there seemed to be that if you chose the correct ratio between the two caps, ripple voltage in the B+ can be cancelled (in the audio) by injecting an attenuated version of it at the cathode. (The two caps acting as an AC voltage divider.)

[Dave H]

M S said

"How can this not act like any other input and get amplified?"

Ahead of the output tube there are two inverting (12ax7) stages, which also have the imprint of the p.s. noise, so the the noise is impressed on the signal at the grid of the output tube, as well as on the anode and cathode, all in the same phase.

Dan

I think I understand the reasoning behind the ultrapath connection now. It makes sense if you believe that having electrolytic capacitors in the signal path degrades the signal. With the standard connection the signal current flows through the output transformer, tube, cathode bypass capacitor and reservoir capacitor but with the ultrapath connection signal current only flows in the transformer, tube and ultrapath capacitor (non-electrolytic). The ultrapath cap is now effectively the reservoir cap as little signal current flows in the original reservoir cap (first filter cap).

What I don't understand is why Dan doesn't hear the ripple injected at the power tube cathode. If the preamp ripple is somehow bucking the ripple at the cathode why doesn't he hear the bucking ripple when he switches back to the standard connection? My simulation circuit also had two 12AX7 inverting stages and it had lots of ripple on the output with the ultrapath connection. Can you post your schematic Dan?

[nickb]

There was a discussion over on diyaudio about a similar circuit (in which there was a bypass cap from the cathode to ground as well as the cap to B+):
Ultrapath with cathode bypass cap: which value of C is correct? - diyAudio
The general idea there seemed to be that if you chose the correct ratio between the two caps, ripple voltage in the B+ can be cancelled (in the audio) by injecting an attenuated version of it at the cathode. (The two caps acting as an AC voltage divider.)

I don't think that will work. A signal injected into the cathode will have the same phase at the anode. Therefore the output ripple will be increased rather than cancelled. Note the phases in the image below to get cancellation at the anode.

Red=B+, Blue=Anode, Green= Cathode

[Malcolm Irving]

I don't think that will work. A signal injected into the cathode will have the same phase at the anode. Therefore the output ripple will be increased rather than cancelled....

EDIT:

If the B+ goes more positive, plate voltage also goes more positive. If the cathode is made more positive, Vgk becomes more negative, plate current is reduced and again the plate voltages goes more positive.

But what we want to do is to get the same magnitude and phase ripple at the top and the bottom of the load (100k resistor in your example) so that there is no net ripple voltage across the load. So we need to increase the amount of ripple that is usually on the plate, so that it equals the ripple on the B+.

[nickb]

EDIT:

If the B+ goes more positive, plate voltage also goes more positive. If the cathode is made more positive, Vgk becomes more negative, plate current is reduced and again the plate voltages goes more positive.

But what we want to do is to get the same magnitude and phase ripple at the top and the bottom of the load (100k resistor in your example) so that there is no net ripple voltage across the load. So we need to increase the amount of ripple that is usually on the plate, so that it equals the ripple on the B+.

Yes, I stand corrected. For the Ultrapath in question Its the current in the load that matters. Thanks Funny how even the simplest things can trip you up.

EDIT: On the other hand, if the load in not a transformer and the output is referred to ground as in my sim then it won't work, as I had stated.

[Malcolm Irving]

.... Its the current in the load that matters....

Well yes, if the load in question is the primary of a single-ended output transformer - the current through the primary (and the voltage across it) is what matters.
But if the plate load is a resistor (like in your example) and we couple the voltage at the plate via a coupling cap to a subsequent stage grid, then I guess we want the voltage (relative to ground) at the plate to be ripple free! In which case your point was absolutely right.

[dcoyle]

Here is the schemo and a pic of the physical layout. There is nothing unusual about the amp. The only unique twist is to use a 100 ohm, 5 W wirewound pot for the resistor leg from NFB from the output transdformer secondary to ground at the previous stage. Let's me dial in how much NFB, and the presence control is now redundant.

In looking, I see that there is a 47 ufd non-atom cap in the cathode bypass beneath the adjustable resistor, not a 100 ufd atom, as I had said.

It's not the amp, it's the circuit.

Dan

[g1]

Am I totally lost, or is the the schematic without the "ultrapath" ?

[dcoyle]

Yeah, the schematic is without the ultrapath. I haven't adopted it, yet, it's still on the switch. I describe what I have hooked up in post 35 on p. 1. In ultra path, the 47 ufd electrolytic output cath bp cap is replaced by a 30 ufd film running opt tube cathode to top of the output transformer.

Dan

[dcoyle]

A follow up:

I tried the ultrapath in my push pull 6V6 anp, and I do hear 120 Hz noise.

Also, in checking the PP amp I observed that the ultrapath "in" position of my switch had an effect on the sound even when I was in fixed bias mode. The ultrapath 30UFD film cap is only the value of the cathode resistor off ground, and I conclude some of what I am hearing may be from the additional (film) capacitance being added, as someone mentioned earlier.

I didn't use the ultrapath in the push pull amp, but did add the 30 ufd film cap bypassing the B+ at the output transformer for a hint less distortion, to my ear.

I rechecked my SE amp, and the noise is not there. The SE amp does have an inductor before the power tube B+ .

Dan

[nickb]

Good info - thank you Dan.

In the original JE article, the PSU has a lot of filtering - series resistor + 2 x 100uf + choke. SE designs generally have lower ripple than PP as they are less tolerant of it. I expect that is why you don't have a problem with it.

I did a simulation experiment using a 6SN7. With the cathode bypassed, you can get only 1.8mW out at 0.25% THD which is what JE had in the article. If you remove the bypass cap then you have 0.14% THD at the same power level. Therefore, for maximum ripple rejection and lowest THD, save on the PSU complexity and film bypass cap - just leave the cathode bypass out altogether (at the expense of a bit of gain).

[Dave H]

I rechecked my SE amp, and the noise is not there. The SE amp does have an inductor before the power tube B+

It's the inductor. If I simulate it with just a 47u filter cap it has 2.3V rms of ripple at the speaker. Adding a 5H inductor and second 47u cap to form a pi filter reduces the ripple to 30mV rms.

[Gingertube]

Back to Basics,
The reason for the Ultrapath connection idea was to remove the DC idle current from the output tranny primary.
That meant that an airgap in the core to avoid saturatiion of the core was not necessary, and eliminating the airgap meant that the primary inductance was very much higher.
So you could then get much better bass response from a smaller and cheaper transformer. The whole idea was intended to be applied to HIFI single Ended amps.
When going to push pull output then 95% of the design intentions of the ultrapath go "out the window" as in push pull, the idle currents cancel in the primary and so very small airgap is used anyway (just enough to cater for any imbalance in the idle currents from the push pull sides).
I don't see any real benefits in applying ultrapath to a guitar amp where bass response is not required/wanted anyway, and any reduction in output tranny size/cost will be marginal at best.
Cheers,
Ian

[nickb]

Back to Basics,
The reason for the Ultrapath connection idea was to remove the DC idle current from the output tranny primary.
That meant that an airgap in the core to avoid saturatiion of the core was not necessary, and eliminating the airgap meant that the primary inductance was very much higher.
So you could then get much better bass response from a smaller and cheaper transformer. The whole idea was intended to be applied to HIFI single Ended amps.
When going to push pull output then 95% of the design intentions of the ultrapath go "out the window" as in push pull, the idle currents cancel in the primary and so very small airgap is used anyway (just enough to cater for any imbalance in the idle currents from the push pull sides).
I don't see any real benefits in applying ultrapath to a guitar amp where bass response is not required/wanted anyway, and any reduction in output tranny size/cost will be marginal at best.
Cheers,
Ian

It sounds like you may be thinking of a different topology where the OPT is AC coupled. In the Ultrapath the objective is to eliminate highly non linear electrolytics from the signal path.