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.

BluegrassDan

Well-known member
GDIY Supporter
Joined
Mar 17, 2009
Messages
507
Location
Elizabethton, TN
I’m working on an external power supply that can power either a 2-channel or 4-channel version of my tube preamp and/or compressor. Would like it to be a universal plug and play for any of the designs I’m building.

Ian’s 12v and 48v designs are working great, providing there is ample heatsinking on the 12v. But I’m stumped on bein able to make a quiet regulated B+ psu.

I’ve tried the IRF820 MOSFET design, which just isn’t quite quiet enough. It’s “okay” but a traditional RC design is quieter.

Does anyone have any suggestions for a schematic to try? I am powering up to 4 channels, so 225-275vdc at around 100ma max.
 
BluegrassDan said:
I’ve tried the IRF820 MOSFET design, which just isn’t quite quiet enough. It’s “okay” but a traditional RC design is quieter.

Perhaps you have used it in a significant mis-configuration or perhaps you could elaborate what exactly you mean.

For the past decade or so variants of this circuit in the range of 100-300VDC have produced perfect noise floor in every single project: https://groupdiy.com/index.php?topic=51232.msg651854#msg651854

Preamps, vari-mus, mics, you name it. I no longer build tube things without it. What would be the point of reinventing something that works perfectly. It's hard to fail due to its simplicity and you can have a high voltage PSU with just a single electrolytic: one big one right after bridge rectifier.  Everything else can be polyprops.

I have also used it as an external PSU and a variant with switchable zener configuration of it is serving lab PSU duties. In this design any noise is a failed ground return path due to bad implementation.
 
Circuit like this will also give you very clean DC, works well enough for tubes although not regulator:
https://static.flickr.com/139/324693062_1bcc2f4ae2.jpg
It is cheaper, first cap can be 150u for 25mA, others ~47u, last one even smaller. Checked datasheets of modern lytic caps life, i don't see need to pay for polypropylenes under most circumstances.
I'm looking into transformer availability for something else. Where custom is expensive, back to back like in G9 should be ok if that is your need.
 
My3gger said:
Circuit like this will also give you very clean DC, works well enough for tubes although not regulator:
https://static.flickr.com/139/324693062_1bcc2f4ae2.jpg

I've used this with fatter MJE transistors in the past and it's only "ok". It's missing a crucial part where the noise of the zener reference is filtered away by an additional RC stage and hence performance is not good.
 
Kingston said:
I've used this with fatter MJE transistors in the past and it's only "ok". It's missing a crucial part where the noise of the zener reference is filtered away by an additional RC stage and hence performance is not good.

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
 
ruffrecords said:
Better if the zener is fed by a simple constant current source.

This could be upgraded into the MOSFET example PSU (also above) I have been using as well. In practice it outputs clean DC even without it.
 
Kingston said:
This could be upgraded into the MOSFET example PSU (also above) I have been using as well. In practice it outputs clean DC even without it.

This depends on your definition of clean. Most single ended tube stages have a very poor power supply rejection ration (PSRR) so any noise on the power supply appears at the output just a few dB lower.

Cheers

ian
 
ruffrecords said:
This depends on your definition of clean.

Clean as in there is no ripple to reject.

No multiples of 50hz, no unexplained crud from rectifier, no weird RF crap.

I haven't ever been able to achieve this with plain passive RC cascades.
 
Kingston said:
Clean as in there is no ripple to reject.

No multiples of 50hz, no unexplained crud from rectifier, no weird RF crap.

I haven't ever been able to achieve this with plain passive RC cascades.

No ripple is an impossibility since it implies infinite reduction.

Passive RC circuits can do much better than active regulators. Most regulators will give you 60dB ripple reduction, maybe 70dB on a goo day. The three stage RC filter can easily give you 90dB.

The best combination is perhaps a two stage RC filter followed by a regulator which should get you into the 100dB + region.

Cheers

Ian
 
Agreed.

I like to think of breaking it into two problems - passive RC for ripple r duction, and active elements for load regulation/voltage accuracy.  Then you get the best of both worlds.
 
This is an area I'd like to understand better ... 

I know that good regulated hv circuits can give very low 'hum' in even very sensitive applications.

And I know that serious passive filtering of the hv can also give outstanding results.

Do people use both together  ?  ie.  a big inductor with C-R-C-R following  + a common good quality regulator scheme?

..

I've used each seperately but not both ..  so not quite sure.

Thnx
 
In the past I have relied on several RC stages for HT ripple reduction for my mixers. One problem with this is that the output voltage depends on the load and the local mains voltage. This is not a problem for most line level  tube circuits as they are very tolerant of HT voltage. However, I have recently decided to change my HT voltage from 300V to 250V. This means my 6922 based SRPP stages can be run without heater elevation, Because the dc heater supply is then ground referenced it can also be used for semiconductor VU buffers which previously required a separate supply. But all this only works if the HT is no more than 250V. So now I need to regulate it as well as smooth it.

I quite like the LR8 plus external FET pass transistor design from The Art of Elelctronics.  I am looking to provide 250V at 300mA. Starting with about 340V brom a bridge recifier and a 470uF cap you have about 6Vpp of initial ripple.Like many regulator chips, the LR8 has a typical ripple reduction of 60dB which will drop this to about 6mV is too much for audio mixer work. So I thought o add two stage of RC ripple reduction before the regulator, each consisting of 100 ohms and 470uF. They will drop a total of 60V at 300mA leaving 280V for the regulator to drop to 250. Each RC stage provides about 29dB of ripple reduction so the total should be well over 100dB.

The big problem is the waste heat. Dropping 340V to 250V at 300mA generates 27 Watts. At least it is shared between the two dropper resistors and the pass transistor. An alternative is to use one of the high voltage LED SMPS we have looked at before and  just regulate its 300V output to 250V using the LR8 style circuit. At least then we start with just a couple of hundred mV of ripple.

Cheers

Ian
 
Matador said:
Agreed.

I like to think of breaking it into two problems - passive RC for ripple r duction, and active elements for load regulation/voltage accuracy.  Then you get the best of both worlds.

In my experience the additional RC is unnecessary. Bridge rectifier to a big elco, and then the MOSFET stage. It allows for very optimal and efficient high voltage transformer selection since it's not necessary to account for RC voltage drop due to load.
 
Kingston said:
In my experience the additional RC is unnecessary. Bridge rectifier to a big elco, and then the MOSFET stage. It allows for very optimal and efficient high voltage transformer selection since it's not necessary to account for RC voltage drop due to load.

I am sure that is true for small current draws where you can ensure the initial ripple is quite small. In such cases the 60dB or so ripple reduction provided by the MOSFET stage will be more than adequate. But I need 300mA or so for an entire mixer which means the initial ripple is a lot it will be for a single tube mic pre drawing 15mA or so.

Cheers

Ian
 
ruffrecords said:
I am sure that is true for small current draws where you can ensure the initial ripple is quite small. In such cases the 60dB or so ripple reduction provided by the MOSFET stage will be more than adequate.

This is probably true. Maximum I have ever needed is around 60-70mA in a two channel configuration. The channels had an additional 1k/10uF RC because the PSU was remote with two meter cable.
 
Kingston said:
Perhaps you have used it in a significant mis-configuration or perhaps you could elaborate what exactly you mean.

For the past decade or so variants of this circuit in the range of 100-300VDC have produced perfect noise floor in every single project: https://groupdiy.com/index.php?topic=51232.msg651854#msg651854

Preamps, vari-mus, mics, you name it. I no longer build tube things without it. What would be the point of reinventing something that works perfectly. It's hard to fail due to its simplicity and you can have a high voltage PSU with just a single electrolytic: one big one right after bridge rectifier.  Everything else can be polyprops.

I have also used it as an external PSU and a variant with switchable zener configuration of it is serving lab PSU duties. In this design any noise is a failed ground return path due to bad implementation.

The thread has several revisions. Is Rev 4 the final (and merged with the earlier schematics showing the bridge rectifier and caps)?
 
I have designed a little PCB for the Horowitz and Hill circuit (using the LR8 as a reference for a MOSFET pass transistor). The idea being it can be used either to clean up the output of a high voltage SMPSU or a regular rectifier and a couple of RC stages. If it works I am happy to give the spares away.

Cheers

Ian
 
ruffrecords said:
I have designed a little PCB for the Horowitz and Hill circuit (using the LR8 as a reference for a MOSFET pass transistor). The idea being it can be used either to clean up the output of a high voltage SMPSU or a regular rectifier and a couple of RC stages. If it works I am happy to give the spares away.

Cheers

Ian

Ian,

I read in earlier posts that your 48v circuit can, theoretically, be adjusted up to 240vdc? Would this be a reasonable B+ solution up to 100mA? Could a simple RC network be added in front of it?
 
BluegrassDan said:
Ian,

I read in earlier posts that your 48v circuit can, theoretically, be adjusted up to 240vdc? Would this be a reasonable B+ solution up to 100mA? Could a simple RC network be added in front of it?

I did try this a few years back and it does work. However, as it stands it is not short circuit proof. If you short the output there is a lot more than 125V across the TL783 and it dies instantly. One slip of the test probe and its gone. I never did work out how best to fix this. One thought is to place a 100V zener across the TL783 but you would still need a series resistor to limit the current through the zener and perhaps a fuse to remove the current before the zener died. I will check out the data sheet to see if there is a better solution.

Cheers

Ian
 
This should be worth a try.

TL783shortcircuitprotection.png


Cheers

Ian
 

Latest posts

Back
Top