Simplistic MosFET HV Shunt Regs

diyAudio Chief Moderator
Joined 2002
Paid Member
3 questions. Is your Vin safely higher than Vout at about 50V? Are there any big capacitors on the audio circuit's B+ receiving side? Maybe the audio circuit idles at more current than the bench supply or the CCS setting can provide? Despite parts substitutions something must be basically difficult in the set up.
 

iko

Ex-Moderator
Joined 2008
Definitely it was enough Vout-Vin, around 50V. No caps on the B+ receiving end of the headphone amp. The idle current of the amp I measured to be around 110mA once it stabilizes. However, at turn on, it draws a lot more current. I don't have a value, I should measure it again and try to draw a curve. This might be bad enough for the regulator to deal with, since the shunt seems very sensitive to large changes in the load. I mentioned this before: two more sensitive points are the value of R1, and the value of the trim resistor. Changes in these two values can impose a huge change in the behaviour of the regulator. I can understand that you and a few others are running the regulator without a problem. Perhaps the loads are smaller so these sensitive points are not so sensitive for you and those that it runs ok for?

For instance, with a load of, say, 150mA, if R1 is too large, the output voltage will be very small. If R1 is too small, the current in M2 will be huge. And only a few ohms (in my case, 4 ohms) are needed to swing from one extreme to another.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Yes it seems like being difficult with power amps given your experience by now. I use it for phono or line stages. Never thought it would be powering more than the gain stages of a power amp. Why don't you try a Vbe control BJT on the CCS instead of LEDs? Actually if the current asked can be 150mA, there is no way it can be stable with its CCS set at 93mA circa 200Vout as with the 22R. Basically at such heavy use we need to develop a stable power version through testing since it looks like there is interest for such. 200mA at 200V means 40W constant...
 

iko

Ex-Moderator
Joined 2008
Salas said:
Yes it seems like being difficult with power amps given your experience by now. I use it for phono or line stages. Never thought it would be powering more than the gain stages of a power amp. Why don't you try a Vbe control BJT on the CCS instead of LEDs?

I did try, but it does not help with the instability.

Actually if the current asked can be 150mA, there is no way it can be stable with its CCS set at 93mA circa 200Vout as with the 22R. Basically at such heavy use we need to develop a stable power version through testing since it looks like there is interest for such. 200mA at 200V means 40W constant...

I did try to set R1 to 10 ohms, then 6 ohms. That's how I got it to be stable for a little more time, but then it did not last.

A stable power salas shunt and simplistic please, if possible :D
 
diyAudio Chief Moderator
Joined 2002
Paid Member
ikoflexer said:
A stable power salas shunt and simplistic please, if possible :D

Better watch what you wish for.:D

If your amp idles at around 100mA and this one works in practice, the shunt Mosfet is going to dissipate 20W and the CCS Mosfet 10W. The MJE350 1.7W regardless of main CCS setting.
 

Attachments

  • shunthvpfixed.gif
    shunthvpfixed.gif
    6.7 KB · Views: 1,331
Yes, after upping the gate stopper of the 9240 to 1K and changing my wiring I had it playing music for about 20 min. but couldn't get higher voltage than 170V.
I then accidentally shorted the output and replaced the 840 and the 9240 since, changed resistors R1 and R5 and added a base stopper on the MPSA (33R, could substitute for 2 x 100R in // ). MPSA is also new...
 
diyAudio Chief Moderator
Joined 2002
Paid Member
So, by eliminating those suspects, logically, you have to inspect if the IRFP and 840 are for GDS correctly wired, and if some BC560 is affected. Did you manage to understand any differences from the Maida sound in those 20 minutes? Also, do you know the idle current consumption of you audio channels (both)?
 
Did you manage to understand any differences from the Maida sound in those 20 minutes?

Haha, that would be asking too much I guess...
Music from an old Sony CDP being fed to 20€ headphones (the ones
for testing) doesn't really qualify for some serious evaluation.
Still, it sounded full, detailed and un-aggressive.

Will continue digging in my built...

Edit. As mentioned before my consumption is a measured 56mA.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Check all semiconductor orientations and only suspects left are those BCs. Don't throw away the other stuff after it works again. Just inspect them vs new ones with the DMM and keep the unaffected ones.

Stixx said:
Edit. As mentioned before my consumption is a measured 56mA.

And your CCS is set at 75mA?
 

iko

Ex-Moderator
Joined 2008
Stixx said:
I found one spare IRFP9240 in my parts bin and put it in.
Everything else I left unchanged ecxept that R1 is 27ohm
and R5 82ohm. Unfortunately I still have a little over 4V
output voltage...
Any clues what is still wrong, or am I about to pile up parts
like ikoflexer? :D

Every time something happens, please check all the parts, especially all transistors. I check them with the diode setting on the DMM.
 
Every time something happens, please check all the parts, especially all transistors. I check them with the diode setting on the DMM.

Thank you Ikoflexer for your input. I must admit that all this is very new territory to me...
5 years ago I knew absolutely nothing about tubes and SS, I used to build loudspeakers, and that was it.
Then the tube bug bit me, and it is itching still :D
Fast forward and 5 headphone amplifiers later I have a working Salas' HV shunt reg, YEEHAAH!!!!

Yesterday I obviously picked the correct BC560 to replace, switched the reg on and had a working unit. Apart from the usual crackling at warm up (~30s) there was no oscillation, and I just carefully checked output voltage and (very carefully, didn't dare to touch anything...!) dialed it to 210V. After seriously heating up my ballast resistor I shut it down, bleeded capacitors in my pre-CRC and connected it to the amplifier. Again no problems, just music!
Vin is ~248V, Vout a little over 200V which seems about perfect. Heat dissipation is nothing to be worried about, the heatsink on the 840 still remains rather cold (but it is a big one...), the sink on the 9240 climbs to about 45°C max.
Everything was still connected to the cheap testing gear but I couldn't resist to try my good set of cans (Senn HD600). Also a few words about my Aikido that uses the Broskie boards (one of them rewired to have a mirrored pair), 6CG7 (RCA clear top) and 6H30, parafeed output transformers (for the time being Hammond 119DA but custom made AE 10K:300 are sitting on the shelf) and my favourite ERO KP coupling caps.

Had it playing for 1,5 hours and thorougly enjoyed it... Having gone through quite a few iterations on my Aikido I can clearly say that this is a killer regulator. Putting the headphones on I had to check whether everything is still working because it is so silent... absolutely 100% dead silent!!
Sounds VERY open, awesome resolution, soft but extremely detailed highs also resulting in a very credible soundstage and well articulated bass, it is all there! :) :) :)

My applause to Salas for the good work and many thanks for guiding me so far through my built.
N., when you ever get near to Stuttgart reserve some time to have a beer with me...:D (or whatever you fancy...)

Will continue my testing tonight and probably start to build a second one (since parts are relatively inexpensive) with a tighter layout and set for 100mA CCS. Will also post a picture or two...
 
Hi Stixx

Cool to hear that you finished the Salas Regulator with success. Nice job, even though it was a bit troublesome along the road.

I saw that you have received your AR transformers. Please let us know if you detect any changes in sound when the AE tranny is mounted instead of your Hammond.

Applaus to you for the nice build :)

Kind regards

Karsten