• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Effect of DC Resistance in SET Power Supply

Member
Joined 2019
Paid Member
I am trying to get a better understanding of the consequences of high internal resistance in an unregulated single-ended triode power supply.

I understand the final capacitor in the supply is in the signal path and is responsible for supplying AC signal current. The capacitor is then "topped off" by the supply as it is drained.

Say for example, it is a 130uF capacitor feeding a 2A3 SET output stage. Will a high internal resistance in the supply prior to capacitor, say 2kohm, be detrimental its ability to recharge the cap efficiently? Or rather, are there real world audible effects of this high resistance on the dynamics of the amplifier?

I know voltage regulation is less of a concern in SET designs compared to AB/B PP since the power supply caps are typically not drained to the point that they are unable to maintain their voltage. I am wondering if the output stage PS cap could be sufficiently drained during real-world use (not pushed to clipping) to the point that the supply cannot refill the charge fast enough, so to speak, and affects dynamics.
 
Every aspect of a passive DC supply affects its regulation ability. Even the slightest loading
drops the DC voltage until the next 120Hz pulse can refresh it. Any transient signal waveform,
especially those changing much faster than (1/120) seconds, will alter the supply voltage during
(and for some time after) its presence, due to the current taken. A slower signal will cause a DC
voltage shift in the supply that we call sag. The DC bias current loading sets up a steady state
condition in the supply voltage that we call ripple.
 
Last edited:
Member
Joined 2019
Paid Member
Hmm I see. I had only read of sag being a concern in class AB PP amplifiers due to their widely varying load. This also applies to SET, or just to a lesser degree?

Some texts I have discuss adding series resistance to damp LC filter resonances, but wondered what the effect might be of increasing the internal resistance and making the supply less "stiff" in a SET design.
 
Last edited:
Unless the supply current drawn is constant DC, there will always be transient effects in any power supply,
in addition to the usual ripple. Resistance added to adjust the transient response of the passive supply will
make it slower to respond, all other things being equal, similar to tuning the woofer alignment in a speaker.
This will help.
PSUD2
 
Last edited:
Member
Joined 2019
Paid Member
Thank you both, I have used PSUD2, I'll revisit it and see if it helps my understanding.

Without getting too far into the gory details, the crux of my question is if I have ~2kohm of DC resistance in my supply before the output stage capacitor, with each output tube drawing 60mA at idle, will that pose a problem in class A. I think I am understanding it is a matter of maintaining my power tubes DC bias point as AC transients pull current from the caps and the supply refills them.

That is currently in my design, which is about a days work away from being completed. Since this resistance is also dialing in my output tube B+, it is somewhat vital at this point. If the "slowness" of my DC supply to respond to AC transients will be detrimental to the sound, some significant changes would need to be made.

I guess giving it a listen might be the only way to see if it is an issue! Gosh, I sure hope not, I thought I was at the end.
 
Last edited:

PRR

Member
Joined 2003
Paid Member
Happy class A is very nearly constant current.

Un-clipped speech/music will not have sustained stress.

So it should be inconsequential.

In SE guitar amps, which DO clip hard, we also have idle-buzz from small B+ caps. I often sketch added R-C filtering. People do not come back complaining about dynamics.

Time-constant of 5k load and 130uFd is 0.65 seconds. Time constant of 50Hz is 0.003 seconds.

I would worry more about the color of the capacitor, than any potential sag.

> about a days work away from being completed

So why rack your/our brains? Build it. Flog it. Put a meter on the B+ and see if it varies.
 
Member
Joined 2019
Paid Member
To learn and anticipate issues so that I might work on how to address them, of course PRR! This is a problem I haven't come across, I am new to the DIY tube hobby and this is my first big project, there is an element of anxiety given the time/money/effort that has gone into it, I do appreciate your input. The capacitors are blue and quite sharp :)

But will do, I will measure the B+ at load and see if it varies.

Thank you all for your time.
 
I think I am understanding it is a matter of maintaining my power tubes DC bias point as AC transients pull current from the caps and the supply refills them.
Since it's AC, peak current will be symmetrical until you clip the amp substantially. The AC transients do not drain the power supply caps.
If the "slowness" of my DC supply to respond to AC transients will be detrimental to the sound, some significant changes would need to be made.

There's a rather trolly individual who barks on and on about stuff like this and uses PSUD models that are representative of what happens in the power supply of a class AB push-pull amp. I would take what he says with a grain of salt.
 
luckily, no one wants to listen to a clipping amp...:eek:

attachment.php
 

Attachments

  • clipping.png
    clipping.png
    388.8 KB · Views: 210
Happy class A is very nearly constant current.

It is, when averaged over a complete symmetrical cycle or longer. The instantaneous cathode current of a SE amp driven near, but not into clipping will vary from near zero, to near twice the idle current, PROVIDED that the load impedance remains constant.

take a cathode biased SET amp and drive it into really hard clipping, the voltage sags up

I have seen them go either way, depending on the tube, how it's loaded, and driven. The typical RC network coupling into the grid of the output tube can cause bias shift that usually pushes the tube closer toward cutoff, lowering the current, allowing the B+ to rise. Mosfet or transformer coupling can hold the grid voltage constant. Here the shift can go either way, depending on how the tube's parameters shift with applied voltage and current.

All of the statements discussed so far in this thread are valid when the amp is loaded with a resistor and driven with a nice symmetrical sine wave. Music is not often symmetrical, and speakers are never exactly 8 ohms resistive (or whatever their stated impedance is).

If there is a graph of impedance VS frequency for your chosen speaker, it's only valid with a single sine wave at a constant level. If the woofer cone is moving in one direction and a big drum hit or other transient tries to instantly reverse the cone's direction, the instantaneous impedance can drop pretty low, near zero.

As stated the time constant of the power supply is usually long enough that the B+ remains relatively constant even if the amp is cranked up pretty loud, but the typical electrolytic output cap in the power supply does have some ESR and ESL issues causing some voltage variation at the higher audio frequencies that won't be seen on a voltmeter.

I usually try to keep the series DCR in the power supply as low as practical (150 to 300 ohms) without buying expensive parts, and bypass that last electrolytic with a big polypropylene cap which has lower ESR and ESL than the electrolytic. Many SSE and TSE builders have found the same effect that I have, a 30 to 100 uF motor RUN cap across that electrolytic will have an audible positive effect on the amp's sound.

Note that many sellers, especially on Ebay will pass off cheap electrolytic motor START caps as RUN caps and they will NOT work, and could burst or leak. START caps are made for momentary connection to get the motor spinning. A RUN cap is made to operate directly on the AC line, pass AMPS of AC current and eat all the transients on the power lines for years. You need a 370 VAC cap for amps up to 500 VDC, and a 440 VAC cap for up to 600 VDC.
 
Member
Joined 2019
Paid Member
As stated the time constant of the power supply is usually long enough that the B+ remains relatively constant even if the amp is cranked up pretty loud, but the typical electrolytic output cap in the power supply does have some ESR and ESL issues causing some voltage variation at the higher audio frequencies that won't be seen on a voltmeter.

I usually try to keep the series DCR in the power supply as low as practical (150 to 300 ohms) without buying expensive parts, and bypass that last electrolytic with a big polypropylene cap which has lower ESR and ESL than the electrolytic. Many SSE and TSE builders have found the same effect that I have, a 30 to 100 uF motor RUN cap across that electrolytic will have an audible positive effect on the amp's sound.

Thanks for this insight, Tubelab. The PS capacitors in this amp are 130uF 600VDC polypropylene, one per channel, ESR from the manufacturer is 3.2mohm. DCR of the supply is ~1.8kohm, although I gather this should not pose a significant problem, I will confirm with measurements when the amplifier is completed. I am only waiting for my OPTs to arrive.

Here is a photo mid build - my first design, although I have had much help along the way. Assuming my concern brought to this thread does not cause any issue, I am optimistic it will be a very good sounding SET amplifier.

11188775.jpg
 
6A5G . . . should be nice.

Only 0.25A more filament current than a 6A3 or 6B4G.
And almost all other specs are identical.

6A3 max plate V 300V, 250V max for the others
6A5G more output power, 3.75W versus 3.2W for the others.

I can not understand the differences, they all use 250V and 60ma, same u, same transconductance, same rp, same RL, . . . and yet the 6A5G gets slightly more power out.

Should be interesting to tube roll, and listen for any differences in sound.
 
Member
Joined 2019
Paid Member
Thanks, I hope so. It is basically a 6B4G with cathode sleeves and the heaters center-tapped and referenced to ground via the cathode, so AC heating with no hum (in theory, we'll see). Pretty interesting tube, hoping for good sound from this amp, high DCR of the power supply notwithstanding :D
 
Last edited by a moderator: