Regulated B+ Solutions?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
That looks like it. There are other dual ones but they are low voltage. I have tagged the HV schematic at the end of this post. It is basically a discrete diff amp using a TL431 as a reference. The diff amp uses a pair of MJE13003 transistors. The TL431 is the set point so the minimum output noise is defined by the TL431 noise plus the ability of the loop to reject supply hum. The latter depends on the loop gain which I think is defined by the product of the MJE13003 beta and the bypass FET gain. Unfortunately the beta of the MJE13003 lies in the region of 8 to 40 so this might give us about 20dB lower ripple than the simple zener circuit.

I found one on eBay in the uk:

https://www.ebay.co.uk/itm/1pc-GZ34-5AR4-Kubota-Low-noise-Audio-adjust-HV-regulator-for-tube-amplifier-PCB/201343461287?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m2749.l2649

so I have ordered one to try out.

Cheers

Ian
 
I don't understand the point of a 5Z4P additional rectifier in that otherwise healthy looking circuit. The complication of the required heater makes no sense whatsoever. The circuit performs great without it.

Also what a bunch a silly greedy ebay weirdos protecting a bunch a trivially guessable parts values in that circuit. "we will mail you the BOM"

lol
 
Kingston said:
Also what a bunch a silly greedy ebay weirdos protecting a bunch a trivially guessable parts values in that circuit. "we will mail you the BOM"

lol

The values are quite easy to see on the silk screen.

Cheers

Ian
 
Yep - that's one of the Kubota pcbs around on eby ..  thanks for the correction Ian!

It has the flexibility for a tube rectifier, if that's what's desired  ..  sometimes tubes can  be fun to use, or so I hear  :)

I used the fred rectifiers myself, and utilised the spare pcb space to add a 'lay down' large electro filter cap in it's place - then using 'off board' dual JJ can caps in a 1ru rack. Additionally that allowed me to use higher voltage rated filter caps, thus allowing increased voltages.

In a 2ru context one can just use the taller onboard caps and no mod required. A pair of them could make one fine psu ..  with front panel adjustment and meters and all  :)

The main thing is it's cheap, works with different pass transistors and the pcb tracks are generously large with easy to access components, allowing larger resistors to be used  and the like. 

You can use whatever heatsink you fancy, for the pass transistor ..  it is located properly at the pcb edge to make addition of the heatsink easy. I used just the 1ru steel case 'top deck' for my two modules. I run with a low input-output voltage differential, so the heat dissipated is tiny.

I did use thick, large flat ceramic pass transistor insulating washer and proper sized mounting/bushing for safety.

The action of the output voltage adjust is very smooth and could easily use a potentiometer to allow easy and safe adjustment of the output voltage.

In my world, it beats out everything else in this category with ease.  If I were making pcbs, it would certainly be what I would integrate.

The noise floor one can achieve with this 11usd board, maybe another 30 in parts, with large caps,  ..  in the context of an external psu is just completely satisfying.

For example, in my single ended 'spud' stereo amp, using a 6L6gc as the final and 6SL7gt as driver  ...  this externally integrated kubota hv reg pcb made a massive difference to the hum perf, versus the stock onboard, unreg, big inductor psu.

Just the thing for a really high end, proper SE tube power amp with those expensive output transformers  :)
 
Hi Dan,

Here you can read my experience. Hopefully it can be helpful to you

https://groupdiy.com/index.php?topic=71117.msg906631#msg906631

Fabrizio
 
Breezio said:
Hi Dan,

Here you can read my experience. Hopefully it can be helpful to you

https://groupdiy.com/index.php?topic=71117.msg906631#msg906631

Fabrizio
Thanks for the link Fabrizio. If I read the scope traces correctly, at 260mA current draw the raw HT ripple is 1.4V peak to peak and after the regulator it becomes 60mV peak to peak. This is a reduction of 23 times or 27dB.

Cheers

Ian
 
Breezio said:
Actually even better, here is the very final design with measurements

So now the raw HT has 1.93V peak to peak ripple and the output has 18.5mV peak to peak which is a reduction of just over 40dB. However, R2 and C7 at 120Hz contribute nearly 16dB of this so the regulator is actually reducing hum by about 24dB.

Furthermore, a simple application of CV = it or C = it/V shows that 1.93V ripple (V) at 0.26 amps (i) at 120Hz implies a 1000uF smoothing capacitor.

Cheers

Ian
 
ruffrecords said:
Its ripple rejection is also limited because the zener current varies with the raw HT ripple. Better if the zener is fed by a simple constant current source. Other improvements include changing BJT for a MOSFET which allows you to filter zener noise more easily.

Better still, use an LM723.

Cheers

Ian


Ian!

the constant current source sounded like a interesting idea!
is this how it would look like?

 

Attachments

  • MOSFET PSU WITH CCS.pdf
    68.7 KB · Views: 15
Jumping in here:
I've also found the MOFET with zener reference to the gate to be a nice performer but found a 2 stage RC before the regulator was the best for ripple rejection. 
Regarding the regulator/multiplier, I fed the zener string with CCS of 5 mA or so, then took the zener output to the gate via an RC filter, making the time constant long for a slow ramp up of HT.  I used a high value R and a nice Solen 10uF at the gate. 
You could use the TL783 as a simple CCS for the zeners.

Sorry, no scheme on hand and the best I could do right now is a chicken-scratch drawing but, you get the idea. 

Edit:. Ian, are you putting a regulator on card per amplifier in your desk or just one big master reg?
 
5v333 said:
Ian!

the constant current source sounded like a interesting idea!
is this how it would look like?

I have not seen it done that way in a PSU before but I see no reason it would not work. At least it gives a defined current, A more common solution seems to be a JFET with source tied to drain but IIRC the actual value of current you get depends on the device parameters.

@Winston. I agree, the FET is best at regulating the HT; any ripple reduction you obtain is a bonus. A long RC time constant can give you a slowly rising HT supply if that is what you need but it should also reduce zener noise to very low levels. When used with typical single ended class A tube gain stages, where the PSRR is only a few dB, you really need your HT noise to be down in the few microvolts region.You also need to choose the zener current with care. It depends on the voltage used, but over 50V you only need 1mA to be well over the knee. 5mA would be too much for a 200V zener for example and the dissipation would be one watt.

Cheers

Ian
 
ruffrecords said:
You also need to choose the zener current with care. It depends on the voltage used, but over 50V you only need 1mA to be well over the knee. 5mA would be too much for a 200V zener for example and the dissipation would be one watt.

I forgot you're regulating to 200V, my bad.  I was referring to what I did for 300V output, a 5mA current to 3 X 100V in series seemed to be a good approach. 

Yes the RC from the zeners is primarily a filter but a long time constant with ramp up was a bonus to me and also allowed a polypropylene cap not too silly in size at the gate.
 
Another edit:. If you have a definable current source then you don't even need a zener,  a resistor is better.

A 10M45S  (available at Mouser Electronics) is another decent CCS for feeding the voltage defining component.
 
ruffrecords said:
I have not seen it done that way in a PSU before but I see no reason it would not work. At least it gives a defined current, A more common solution seems to be a JFET with source tied to drain but IIRC the actual value of current you get depends on the device parameters.

@Winston. I agree, the FET is best at regulating the HT; any ripple reduction you obtain is a bonus. A long RC time constant can give you a slowly rising HT supply if that is what you need but it should also reduce zener noise to very low levels. When used with typical single ended class A tube gain stages, where the PSRR is only a few dB, you really need your HT noise to be down in the few microvolts region.You also need to choose the zener current with care. It depends on the voltage used, but over 50V you only need 1mA to be well over the knee. 5mA would be too much for a 200V zener for example and the dissipation would be one watt.

Cheers

Ian

Do you mean gate tied to source?
 
Winston O'Boogie said:
Another edit:. If you have a definable current source then you don't even need a zener,  a resistor is better.

A 10M45S  (available at Mouser Electronics) is another decent CCS for feeding the voltage defining component.

Ccs and resistor you say.
Have you tried it?
Sounds even more minimalistic!
 
Back
Top