Single Ended Stereo Preamp - Shared power supply

[StarTrucker]

On Power amps, I've seen sharing the B+ directly between channels, but what about on Single ended amps?

This preamp has a mono schematic and I housed 2 of them in the same chassis to run in stereo off the same power supply. They're single ended and I get oscillation/interference between the channels. The problem is in channel 2 but I get junk in channel 1. Disconnect ch 2's B+ and the problem goes away in both channels.

The B+ ends with a cap to ground (last PSU Cap) between the channels, so I figured any junk would go to ground.

This is troubleshooting, but the real question is how to share a power supply between 2 SET amps, so I posted it here in the drawing board forum.

 

[Rusan]

Which tube type are you using? Parallelling two triodes effectively creates a new, single triode with twice the transconductance and half the anode resistance. Which, of course, makes it much more susceptible to funny business like oscillations, on top of what was already a worst-case scenario for power supply rejection. It'll only be aggravated further by low Rp, high Gm types like 6922/6DJ8, 5687, ECC99, etc. If using such types, grid stopper resistors of adequate value are of double importance, with one per grid connection and very short leads.

How well is your B+ filtered and decoupled? How long are the cathode resistor connections to the ground side of the filter cap? Do the grounds run through the chassis? Do the grounds or the signal leads run close by the output transformers, which could be inducing a feedback loop?

In such a simple circuit (if you don't have any hum going on), it's quite possibly a lead dress/wiring issue of some type in Channel 2.

 

[StarTrucker]

Which tube type are you using?

12AY7 with both halves parallel'd. Ua100d I the circuit, and the PSU is in a separate enclosure with a 2foot run to the audio circuit which carries B+, signal and chassis grounds, and heaters.

How well is your B+ filtered?

4 stages RC


Is the schemo I posted correct where you can just link the two output transformers together from the B+ supply? Nothing needed to knock down cross talk?

 

[Rusan]

Is the schemo I posted correct where you can just link the two output transformers together from the B+ supply? Nothing needed to knock down cross talk?

No issue there. However....

....the PSU is in a separate enclosure with a 2foot run to the audio circuit which carries B+, signal and chassis grounds, and heaters.

Is the decoupling cap that the output transformer primaries are connected to housed in the PSU chassis, or the audio chassis? A two-foot long wire from the grid leak and cathode resistors of those two amplifier stages to the ground side of that cap could be an issue.

 

[StarTrucker]

Nope, the B+ enters into the audio chassis and is immediately connected to a 22uf cap. This then goes to the OT primary. The negative side of this cap is the ground for this parallel triode.

 

[dai h.]

hi,

my (novice) understanding was that a separate decoupling cap helps (with crosstalk) since it's a part of the signal path, like this:



source: http://www.nutshellhifi.com/library/Pix-A.gif

 

[ruffrecords]

Is there any decoupling local to the preamp where the HT enters? If not try adding 220uF electrolytic and a 220nF film across the HT at that point..You problem is possibly due to the inductance of the HT cable from the PSU to the preamp.

Cheers

Ian

 

[StarTrucker]

Here's a more complete schemo. Add the highlighted blue cap and increase the size of the electrolytics in the preamp?

How does inductance cause noise? Like an antenna?

 

[ccaudle]

How does inductance cause noise?

Inductance creates an impedance which rises with frequency. The noise referred to is the voltage variation of the power supply with current draw of the load circuit. The voltage across an impedance follows the AC version of Ohms law, so v=iz, where i is the frequency dependent current of the load, z is the impedance at a particular frequency, and v is the AC voltage generated across the impedance. That leaves the original power supply voltage minus the noise voltage at the load (or plus the noise voltage when the current load is decreasing).
An ideal class A stage does not have power supply current variation with signal, but any variation caused by non-ideal behavior is also not rejected by a class-A stage, the noise can show up directly in the output.

 

[Rusan]

The fact that Channel 2 is injecting audible artifacts into Channel 1 means unwanted inductive feedback loops are not only likely happening, but it could well also mean that (as Ian indicated) your B+ connection isn't adequately decoupled at the lowest and highest frequencies. In other words, the circuit's AC ground at those frequency extremes has excessively high impedance. If you still have the issue after following Ian's wise suggestion to more adequately decouple the B+ connection at the low/high frequency extremes, the culprit is possibly an inductive feedback loop of some sort.

I noticed from the schematic you have AC on the heaters, which means 600mA of AC (assuming 6.3V) flowing parallel with the signal, circuit ground, and B+ wires for two feet. Even if the heater wires are twisted tightly to break up the AC fields, that's a loooong run. Mr. Faraday and Mr. Maxwell tell us this is a problem, even under the best of circumstances. AC heater wires should never be paralleled with any other wires, AAMOF they should ideally only cross at 90°, and even then only when absolutely necessary.

Now, consider that the cathodes and plates are in-phase relative to one another, but in anti-phase to the control grid. If you're inductive-crosstalking between the various connections in that two-foot run, you'll be creating induced signals that are in-phase and 180° out-of-phase with respect to the various elements. Any positive feedback created will do what positive feedback does, which is turn your amplifier into an oscillator. There are potentially several such loops at play here.

But, because there are also reactive elements (caps, transformers) present, there's also frequency-dependent phase shifting involved, so "the plot thickens," so to speak. Add it all together with the fact (as ccaudle noted above) that a transformer-loaded single ended stage has virtually zero noise rejection of any kind to begin with, then parallel two triodes to halve the plate resistance (virtually its only guard against artifacts that are riding on the B+) and make it that much worse, and well, you get the idea.

Out of curiosity, what is your raw B+ immediately after after the rectifier diodes, and what are the values of the series CRC filter resistors?

 

[StarTrucker]

but it could well also mean that (as Ian indicated) your B+ connection isn't adequately decoupled at the lowest and highest frequencies

Will fix this in the preamp

. Even if the heater wires are twisted tightly to break up the AC fields, that's a loooong run. Mr. Faraday and Mr. Maxwell tell us this is a problem, even under the best of circumstances.

DC heaters for Mr. Faraday

If you're inductive-crosstalking between the various connections in that two-foot run, you'll be creating induced signals that are in-phase and 180° out-of-phase with respect to the various elements.

First two fixes will knock this down I hope. Add 10pf caps between B+ and ground?

In the case of preamps, I thought separate power supplies were preferable to PSU housed on board in the same chassis. By separating things out, the big PSU magnetic fields get away from sensitive input circuitry. Is this not the case? Is it better to have the PSU and audio circuit in the same chassis?

 

[dai h.]

I thought you needed an R for each 22uF otherwise common impedance coupling can occur through the B+ and travel a long way back towards the power supply C (big loop area doesn't seem ideal and L+R signals are getting into each other). I think how the 22uF (for each side) is physically placed (connected) helps define the output loop area of that stage and you need the R so it's isolated (that is my very basic level understanding anyway--sorry if that's wrong!).

 

[Rusan]

Add 10pf caps between B+ and ground?

10pF will only be effective at RF frequencies. Ian's recipe of a local 220uF decoupling cap paralleled with a .22uF film cap is quite sound (pun intended), in that it will fill the requirement of a low impedance AC ground at all frequencies. You want to have a low Z ground path even down past the circuit's -3dB rolloff point, hence the large-ish 220uF value. The .22uF film cap in parallel will provide a very low impedance ground path far past the highest frequency of interest, by bypassing the big electrolytic's ESR (equivalent series resistance) and ESL (equivalent series inductance). Together, they synergistically provide wideband decoupling and prevent ringing, which is necessary for amplifier circuit stability at all frequencies.

Back to the inductive coupling thing, here's an example of just how easily things can go wrong. Someone once sent me a single ended tube audiophile amplifier, that was oscillating like mad. The guy was using high-zoot silver RCA interconnects, and had changed the amp's input signal wiring to silver as well. That was when it had started oscillating.

The amplifier's physical layout design was unbelievably bad, and this was a commercial product! The RCA jacks were situated adjacent to the IEC AC line jack, and just beneath and behind the power transformer. The output transformers were directly underneath and sideways-offset from the power trafo by only about an inch, and one of them was less than an inch from the RCA jacks.

When he'd installed the silver signal wires, he ran one of them touching against the output transformer bobbin. Besides being too close anyway, silver also has twice the magnetic susceptibility of copper. A nice healthy positive feedback signal was being induced in the signal wire, happily turning the little amplifier into an oscillator. I simply bent the signal wires down away from the transformer.

In the case of preamps, I thought separate power supplies were preferable to PSU housed on board in the same chassis. By separating things out, the big PSU magnetic fields get away from sensitive input circuitry. Is this not the case? Is it better to have the PSU and audio circuit in the same chassis?

Yes, and no. It's rather application specific, and not always for reasons of potential noise coupling. Consider that tube guitar amps and microphone preamps have on-board power supplies. Both are sensitive, high impedance, high gain circuits, yet do not suffer if properly designed and implemented.

It's of course good engineering practice to place the amplifier stages and power transformer as far apart on the chassis as practically possible, and even the direction the power trafo is oriented in can have an impact. Moreover, the importance of a proper grounding scheme cannot be overstated. The Valve Wizard and Aiken Amplification are some excellent online resources concerning this (and many other things tube-related) that you might find very useful and interesting.

 

[abbey road d enfer]

This is troubleshooting, but the real question is how to share a power supply between 2 SET amps, so I posted it here in the drawing board forum.

Th eproblem may not be power sharing.
You may have the defective channel radiating its oscillation via magnetic or capacitive coupling.Try physically separating both channels, particularly transformers.

 

[Rusan]

I thought you needed an R for each 22uF otherwise common impedance coupling can occur through the B+ and travel a long way back towards the power supply C (big loop area doesn't seem ideal and L+R signals are getting into each other). I think how the 22uF (for each side) is physically placed (connected) helps define the output loop area of that stage and you need the R so it's isolated (that is my very basic level understanding anyway--sorry if that's wrong!).

In my way of thinking (which may or may not be correct!), since Channel 2 is where the problem seems to be manifesting but is affecting Channel 1 as well, there could well be some common impedance coupling happening, but the root of the problem is possibly caused by other issues. Thus, addressing both the issues of inductive coupling and achieving a very low Z ground at all frequencies via a local decoupling cap is indicated.

The local decoupling cap arrangement of 220uF paralleled by .22uF, connected as closely as possible to the grid leak resistors, cathode resistors and output transformer satisfies all the requirements, mostly independent of what's going on back in the pi filter stages.

Th eproblem may not be power sharing.
You may have the defective channel radiating its oscillation via magnetic or capacitive coupling.Try physically separating both channels, particularly transformers.

Good call, Abbey. If the output transformers are too close to each other and/or the circuit wiring, tube sockets or anything, try separating them. Rotating one transformer 90° to the other can help if you're not able to physically separate them, but note that its magnetic flux radiation that can affect the circuitry rotates 90° with it. Duh, I even mentioned something akin to it in my last post, and didn't even think about how that might be happening local, in the audio circuitry, because I was too focused on what's happening between the power supply and preamp. I amaze myself sometimes...

 

[CJ]

Look at your grounding scheme also.

 

[ruffrecords]

Here's a more complete schemo. Add the highlighted blue cap and increase the size of the electrolytics in the preamp?

How does inductance cause noise? Like an antenna?

View attachment 137521

That schemo looks the same as the previous one. Can you post a complete one. Also some pics of your build would help.

Cheers

Ian