Stereomour OG Power Transformer (and my mods)

[tumble2k]

I have heavily modified my OG Stereomour, and I would like to take some additional steps like adding a (solid state) shunt regulator to the driver stage. I am wondering how much current I can get out of the 160V winding of the PT. Right now each channel should draw about 55mA for a total of 110mA at 400V or 44VA. Would I be able to draw 20mA more or 52VA?

My Stereomour  has been driving my Omega RS3 Desktop full range near field monitors for 11 years now, and I have had a love/hate relationship with its sound: the upper mids had a lot of glare and the highs were missing. My first mod was to replace the paraded and input DC blocking caps with Mundorf Silver Gold. It made the Stereomour less glare in the miss but did not fix the highs. I tried removing the cathode cap on the 2A3, but that caused the Stereomour to lose its grunt.

I eventually replaced the cathode bypass caps the Elma Siimic II at 47 uF and the cathode resistor with a Mills 1.21K 12W. After a ridiculous 8 day break in period where the amplifier started sounding great, with no glare and good texture. The highs were there but still  lacking resolution.

The final mod was replacing the  power series resistor with a Triad C-7X 10 uH 250 ohm  choke. This finally got me some resolution and air in my highs. The mids were a bit hard and irritating so I replaced the power supply bypass caps with Kemet C4AQ 70 uF 500V film caps, and after a horrible but short 2 day break in period the Stereomour sounded gorgeous!

But I still want to see how far I can take it. For example, maybe I can fix bias the 2A3 with a string of LEDs so I can remove the cathode bypass caps. Add a delayed start for the B+ rail. Replace the 12AT7 with a 5670 (I have a 6.3 V transformer)… Sky is the limit!

[Paul Birkeland]

Right now each channel should draw about 55mA for a total of 110mA at 400V or 44VA. Would I be able to draw 20mA more or 52VA?

The Stereomour I power transformer ended up being quite a bit larger than it needed to be, and IIRC that had to do with commonly available bobbins from our winder.  PJ would know best how much current headroom there is on the HV winding, but I'd imagine you could load it down a bit more without any issues.

My first mod was to replace the paraded and input DC blocking caps with Mundorf Silver Gold. It made the Stereomour less glare in the miss but did not fix the highs. I tried removing the cathode cap on the 2A3, but that caused the Stereomour to lose its grunt.

Typically a zero feedback SET amp driving a loudspeaker with an indictive load in the treble will have slightly higher treble output than you'd otherwise expect, so a lack of treble may be more speaker or source related than amp related.

maybe I can fix bias the 2A3 with a string of LEDs so I can remove the cathode bypass caps.

I would definitely not recommend this.

Add a delayed start for the B+ rail.

Why?


You could try an LED in place of the 431 for driver biasing:
https://www.mouser.com/ProductDetail/Broadcom-Avago/HLMP-6500-F0011?qs=jT9z6tsiFNlBU7QuW0wo4Q%3D%3D

Another big factor that's totally absent here are the tubes you're using... 

[tumble2k]

Thank you so much for your insights PB!

I'll wait to hear from PJ about the transformer, but it sounds very promising!

The speakers are definitely a factor in the highs. They are full range speakers, so the resolution in the highs have some limitations. With a stock Stereomour, the Omegas were bright but not airy, which I translate to lacking energy above 10K or so. However in measurements (into an 9 ohm non inductive load) I only see a 1 dB drop in output power at 20 KHz.

Regarding the delayed start, I have been using Sovtek 2A3s as stock. I had one die on me because it was arcing on power up. I thought that maybe a delayed start might help with that. But if that's not useful I would so rather leave it alone. I haven't tube rolled because tubes are $$$ these days.

Regarding the LED biasing of the 2A3s, it appears that I need to get 60V at the cathode (I measured 58 V). As a beginner, I thought I could just get that voltage using LEDs. I would love to hear more why you don't recommend it. Noisy? Or just stupid?  If it's noisy, I might be able to add a PNP pass transistor? Maybe 50 mA is too high, although the HLMP 6000 has a max forward current of 90 mA. But with a forward voltage of 1.6V that would require 37 LEDs in series. Maybe that's just dumb. The Stereomour would look like one of those rice rockets from the 90's with under carriage lighting. 

Regarding biasing of the driver tube, For the 5670, I was thinking about using a 270 ohm resistor in place of the LM431, which would bias the grid at -1V keeping the 3.6 mA current source. I didn't think about using an LED.

I'd like to ask one more question if I may. I saw on the Stereomour II forum that someone replaced the output transformer with a Kaiju one. I asked Eileen about it and she warned me that the Kaiju iron was not designed for the Stereomour. I measured the impedance ratio of the Stereomour to be around 560 so an 8 ohm load would appear as 4500 ohms at the primary (I think that is why the transformer is labelled 4K). Is the important design parameter the plate resistance (800 ohms for the 2A3 and 700 ohms for the 300b)??

There's the parafeed inductor, which I frankly don't understand how to choose. Maybe that won't work on the 2A3 at all. There's also the issue of whether the Kaiju iron would physically fit where the Stereomour iron is (I also have the Triad C-7X stacked atop the parafeed choke).

Thanks again for your insights!

[tumble2k]

Oh and thank you for this wonderful beautiful design! I have learned so much from it and gained a truly lovely sounding amplifier. I am driving it with a Schiit Modi+ and a Antique Sound Labs Mini Phono (15K output impedance). I try to use the Stereomour for background music, but I just get drawn in all the time!

[Paul Joppa]

The PT-6 is probably good for at least 150mA (75mA per channel) and possibly a bit more - I revised my design process between doing these two designs so my notes are not apples-to-apples.

The issue with biasing the 2A3 is that the bias drifts as the tube ages, and the bias voltage is very sensitive - the transconductance is about 5 mA per volt, so a 1% change in voltage (0.6v out of 60v) produces a 6% change in current (3mA out of 50mA).

The issue of Kaiju OT-5 transformers is they are 3K primary vs. 4K for the Stereomour OT-2. The difference is audible!

The chassis is drilled for either set, so there are two reasonable alternatives:

1) Keep the OT-2's but replace the 20-henry PC-3 plate choke with the PC-5 on the 60-henry tap. This is mechanically a little tricky, but I know PB has done it.

2) Replace both plate chokes and output transformers but increase the current from 50mA to 60mA. This increases the plate dissipation to 18 watts, 20% over the standard 2A3 spec but within the capability of most modern single-plate 2A3s such as the Sovtek, EH, JJ, etc.

In both cases, you'll nee to change the parafeed cap as well.

[tumble2k]

Really helpful information.

I'll definitely have enough current for a shunt regulator.

I guess I won't mess with the OT just yet. Those parafeed caps were $100 apiece on sale! I know the OT-5 PC-5 ain't cheap either!

Thank you so much for answering my questions!
Clif

[Paul Birkeland]

-1V of bias would not be a great choice.  The stock operating point is -2.5V of bias and the LED I posted should be a bit over 2V of bias.  As you decrease the bias voltage, dissipation will increase across the C4S, and down around 1V of bias, I'd be worried about grid current being an issue.  Down around there, you could also accidentally be limiting the power output of the amp since you can't shove enough signal voltage into the 12AT7.  If you use an unbypassed cathode resistor, that issue will get a lot worse.

I have put the new Stereomour iron on old Paramours, but I can't remember doing the PC-5/OT-5 on a Stereomour.  PC-5 and OT-2 work well together, but they are tricky to mount on a Stereomour chassis since the OT-2 mounting screws are going to prevent the OT-5 from mounting correctly to the chassis.  Stereomour I also didn't have the extra chassis holes.

[tumble2k]

-1V of bias would not be a great choice.  The stock operating point is -2.5V of bias and the LED I posted should be a bit over 2V of bias.  As you decrease the bias voltage, dissipation will increase across the C4S, and down around 1V of bias, I'd be worried about grid current being an issue.  Down around there, you could also accidentally be limiting the power output of the amp since you can't shove enough signal voltage into the 12AT7.  If you use an unbypassed cathode resistor, that issue will get a lot worse.

Thank you for your patience in answering my questions. I actually meant using -1V for the bias of the 5670, but your comments apply to that too. With a current of 3.6mA, I think that a good bias point for the 5670 might be -3V (833 ohms). I think keeping the LM431 2.5V regulator might be okay too, but there will be more dissipation on the C4S. -edit I could run the LM431 at 3V too.

I have put the new Stereomour iron on old Paramours, but I can't remember doing the PC-5/OT-5 on a Stereomour.  PC-5 and OT-2 work well together, but they are tricky to mount on a Stereomour chassis since the OT-2 mounting screws are going to prevent the OT-5 from mounting correctly to the chassis.  Stereomour I also didn't have the extra chassis holes.

Good to know. Thanks!

[tumble2k]

Probably better to keep the plate voltage at its current setting by increasing the C4S to 6 mA and setting the grid bias to -4 V with a 680 ohm resistor. Q2 is now dissipating over 1 W, which is theoretically okay but I'll add a heat sink. It looks like one advantage of using the 5670 is that I lower the plate voltage, which gives more headroom for a shunt regulator.

[Paul Joppa]

Apologies, my earlier post was about the Stereomour II, not the OG version. The Kaiju iron won't fit on the older version

For the 2A3 bias, some people have replaced the electrolytic bypass cap with a film cap and been very happy with the improvement. We incorporated that upgrade in the MonAmour using a 47uF Solen. That could be a good upgrade for an OG Stereomour.

For the driver, the 5670 used in Kaiju runs at 4.3mA/175v/4v bias using a 431 regulator chip with a trimpot to adjust the plate voltage. Without the trim, any fixed-bias arrangement would have too much variance in plate voltage. Using resistor bias would reduce the variance, but it would be a 930 ohm resistor which must be bypassed.

But in the Stereomour II, we kept the 12AT7 but run it at 5.7mA/200v/2.1v bias. The cathode resistor is 365 ohms, which is small enough to not need a bypass. It may be possible to do something similar with a 5670, maybe 3v bias and 6mA = 500ohms(?)

[tumble2k]

Thank you for helping me with this. I will definitely try the film cap too.

One thing I don't understand is why you need to bypass the cathode resistor if it gets to be around 1K. For the operating point you specified I pulled RP=8.3K and mu=30 from the curves. This means that the plate resistance is increased by (mu + 1) * RK or from 8K to 23.5K.

This seems to have no effect on gain because the load resistance presented by the power stage is 249K.

A = mu * RL / (RP + RL + (mu + 1) * RK) = 27

If I keep everything the same and increase RK to 1K, the gain drops to 26.
 
Also the increased plate resistance should have no effect on the distortion because the C4S has an infinite resistance.

What am I missing?

Anyway, thanks again. I am excited to try your bias point!

[Paul Birkeland]

This seems to have no effect on gain because the load resistance presented by the power stage is 249K.

Signal current modulation through the cathode bias resistor induces degenerative feedback.  Rp goes up and gain goes down. 

A = mu * RL / (RP + RL + (mu + 1) * RK) = 27

If I keep everything the same and increase RK to 1K, the gain drops to 26.

PJ is commenting that the 5670 with a cathode bias resistor would need a bypass cap. 

Also the increased plate resistance should have no effect on the distortion because the C4S has an infinite resistance.

The increased plate resistance has to drive the 2A3 Miller capacitance.  The 249K grid leak resistor in parallel with the C4S is the load the driver tube must drive outside the Miller capacitance, and it's definitely not infinite.

[tumble2k]

Many thanks for pointing out Miller capacitance. The 2A3 would have a Miller capacitance of 16.5 pF * (4.2 +1) or 86 pF. The output impedance of the 5670 with an unbypassed 1K cathode resistor is 8300 ohms + 31 * 1K or 39K. The 3dB point is less than 47 KHz. There would be a 0.7dB drop at 20 KHz.

With a 500 ohm cathode resistor the corner frequency would be more than 78 KHz. This would result in less than a 0.3 dB drop at 20 kHz.

With a bypassed cathode resistor on the 5670 the corner frequency is more than 220 KHz.