I can remember some Aikido something, maybe headphones, run by Stixx rather early in the thread that he was very much pro using the reg, but I can't remember if others have given feedback about Aikido preamps and the reg. Different strokes for different folks as they say. Unregulated it is best for you.
Yep, that was me...
For sure it has nothing to do with the Aikido circuit but of course can depend on a lot of other factors. I am running two headphone amplifiers with the Salas reg (Nick, you should start to protect that name...
), one being pure SE transformer coupled, and the other Aikido parafeed. Both made a big jump up in sonic performance with the shunt reg, and the LAST thing I would say that they are lifeless... both of them bring me very close to the music... and that's what it is all about.
Lack of life, flat, would mean the shunt regulator doesn't have enough top voltage before the CCS or the idle current is too low. If anything is should improve dynamics! I'm also running a tube headphone amp now, with shunt regulator on B+, not exactly the Salas design (I modified it to suit my ideas), but very similar topology. Nothing short in dynamics, I can tell you that.
Here is a little challenge. I am blowing the 2sk170 reference. My incoming voltage is 275Vdc filtered. My output voltage is 220Vdc. It is passing 62mA. The circuit is http://www.diyaudio.com/forums/powe...stic-mosfet-hv-shunt-regs-97.html#post2084348.
It will start-up and run fine. After about 30 minutes it will stop regulating and just apply full voltage. It has happened on both channels now. It goes quietly and the circuit looks to be running fine until you measure the ouput voltage.
It will start-up and run fine. After about 30 minutes it will stop regulating and just apply full voltage. It has happened on both channels now. It goes quietly and the circuit looks to be running fine until you measure the ouput voltage.
Don't! You need to pass a lot less current through Q4. So lower the current through Q4 and increase R4.
Here's what you can do. R4 = 220k; adjust R8 in advance so that you get about 1mA through Q4. Then you get a drop of about 235V across R4 (so a 1W resistor should do), which gives you a healthy 15V across Q4.
Edit: you can set the reference current out of the circuit, that's best. You don't want to play with any adjustments while in the circuit, bad things can happen. Hook up Q4+R8 to a 9V battery and check the current with the DMM. An R8 of about 330R should give you a current close to 1mA. Then try several values for R4, starting with 220k.
The shunt current has nothing to do with the current through your reference. People usually suggest 15-20% more current than the load draws at idle, but it's really dependent on the load and what it's max current might be. You want to have enough current through the shunt so that it can supply it for every current spike the load wants.
Here's what you can do. R4 = 220k; adjust R8 in advance so that you get about 1mA through Q4. Then you get a drop of about 235V across R4 (so a 1W resistor should do), which gives you a healthy 15V across Q4.
Edit: you can set the reference current out of the circuit, that's best. You don't want to play with any adjustments while in the circuit, bad things can happen. Hook up Q4+R8 to a 9V battery and check the current with the DMM. An R8 of about 330R should give you a current close to 1mA. Then try several values for R4, starting with 220k.
The shunt current has nothing to do with the current through your reference. People usually suggest 15-20% more current than the load draws at idle, but it's really dependent on the load and what it's max current might be. You want to have enough current through the shunt so that it can supply it for every current spike the load wants.
Thanks Ikoflexer
I understand that the reference current has nothing to do with the shunt current. How does the lower current effect the noise? Can I improve the situation with using a lower Idss 2sk170?
I have to orederr some 1W resistors in the range.
My circuit uses 20mA at idle so I can drop the shunt current as well and save on total dissapation.
I understand that the reference current has nothing to do with the shunt current. How does the lower current effect the noise? Can I improve the situation with using a lower Idss 2sk170?
I have to orederr some 1W resistors in the range.
My circuit uses 20mA at idle so I can drop the shunt current as well and save on total dissapation.
Oops, never mind, I made a mistake regarding the voltage drop across the jfet Q4. The voltage drop across Q4 is set by the Vbe of Q2 plus the voltage drop across R9. Perhaps the MJE350 gets hot!
In any case, the conclusion is still the same, that no heat sink on the 2sk170 will help you. You might need to have better heat sinks on the mosfets and on the MJE350.
In any case, the conclusion is still the same, that no heat sink on the 2sk170 will help you. You might need to have better heat sinks on the mosfets and on the MJE350.
The MJE350 gets pretty warm to the touch. I will look at how to get a small sink on that. The 2sk170 is passing 1.7 mA with the 120k R4. It is stone cold. So you are right about the heat sink not required. It may just help keep the voltage from drifting.
I dropped the shunt current to 45mA, That will help on total dissipation of the FETS. The main sinks are running comfortably warm.
I dropped the shunt current to 45mA, That will help on total dissipation of the FETS. The main sinks are running comfortably warm.
Oscillation might also explain the jfet death. It can happen, when things get hot the whole thing starts oscillating. It doesn't take much to fry a jfet. Maybe a 15-20V zener across it will save the jfet; that will get conductive only if the voltage gets too high. Under normal operation it will not affect anything.
It can be something upseting it due to feeding an active constant voltage load in your case. Look for some friendly decoupling scheme on the load or local compensation or manipulation of the output Zobel in the reg. The relationship with the load should be the key in this. Could be not that bad since it can stand 30 mins. Inherently it has very wide phase margin, so I would not be afraid of some possibly non optimum layout on first guess. Else you have the oscillation on a dummy too. Still all those are just scenarios to what you will discover the real way with the instruments. Good luck.
Dropping the current slightly and additionally improving the airflow over the heat sinks made a big difference temp wise. It has been running now for several hours with no issue.
The one sshvsr has been running for several weeks on the test bed with zero issues. I seemed to get into an issue with the new prototype preamp I assembled. With the second channel on the sink, it certainly ran warmer. Plus the airflow around the sink was more restrictive.
Time will tell. It was not going to act up with the scope hooked up to it.
The one sshvsr has been running for several weeks on the test bed with zero issues. I seemed to get into an issue with the new prototype preamp I assembled. With the second channel on the sink, it certainly ran warmer. Plus the airflow around the sink was more restrictive.
Time will tell. It was not going to act up with the scope hooked up to it.
negative SSHV ?!
Sorry to ask again, but does anyone have a schematic of a working negative SSHV? (like to exstata etc.)
i ve been giving juice to my exstata (The eXStatA Electrostatic Amplifier) with my prototype bipolar salas reg for a handful of hours now and so far i have not noticed any sign of oscillation or malfunction and it sounds great (how surprising?
Sorry to ask again, but does anyone have a schematic of a working negative SSHV? (like to exstata etc.)
That is the one that worked for pidesd.
http://www.diyaudio.com/forums/powe...tic-mosfet-hv-shunt-regs-112.html#post2229950
http://www.diyaudio.com/forums/powe...tic-mosfet-hv-shunt-regs-112.html#post2229950
