The simplistic Salas low voltage shunt regulator

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

iko

Ex-Moderator
Joined 2008
I would like to share a bit of good news. Just finished a high voltage version of v1.5, however, using n-channel mosfets. Although this thread is about low voltage, I think it is relevant in that this topology is the same as the one I recommended earlier to Tham. So that should give him some hope that it's not just simulated vapor ware. I used two trimmers, not set to the exact value shown as resistors in the figure (R1 and R10).

As I said, this circuit is currently running in reality, and it is stable. My intention is to use this (or some variation) for the dht headphone amp I'm developing in another thread.

:cool:
 

Attachments

  • reg-v15-hvn-real.png
    reg-v15-hvn-real.png
    24.3 KB · Views: 923

iko

Ex-Moderator
Joined 2008
Thank you guys!

My email button is off? I'll look into that. Well, you can usually find me as ikoflexer AT gmail.com ... who would've thunk? :)

Arjen, the topology should work with higher voltages but some parts will need to be changed. I'll direct you to another thread where salas already developed a HV regulator as well.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=134801&perpage=25&pagenumber=1

Arjen, if you're interested in such a regulator, have a look first at the salas version; I'm not sure if I should open a whole new thread for the HV I'm working on... ?

Andrew, have a look at this:

http://www.diyaudio.com/forums/showthread.php?postid=1886122#post1886122

That was the "Christmas tree" circuit, with what I thought would be protection for turn-on voltages for the jfet and for the ccs mosfet. I stripped that in the process when I tried it for real (with and without). Also, your points are not quibble for me, I always appreciate the time you take :) so do your "quibbling" as much as you want.

Stormsonic, I will show you a little later what happens with darlington in that position (a bit short on time right now). But you should know that in v1 I've been using the mpsa18 in that position :D I have tried a darlington there too, long time ago. I like the mpsa18; if you're not familiar with it, look it up. It's a nice part.

Andrew:
Can Iko refer us to the post where you rejected this Darlington idea to increase the voltage across the Q6 ccs.
I have no idea what you're talking about, sorry. Who rejected the Darlington idea?
 
20k R9 and D6 charge up the 470uF C6.
that turns on the pass FET.
Then we have to wait until the FET charges up C5.

What power will be dissipated in the 20k with 160 to 180V across it?

Without the 20k and Zener the Jfet will explode.

Once the caps are charged, what current will pass the 20k?
Will that be enough to keep the jFET turned on?

That Darlington proposal was raised months ago.
I thought you or someone else gave an explanation of why it was not possible.
 
Salas said:
Yes they are usable. Try them.

How did you like the subjective performance with those you built? For what circuits, and what you had there before? Thanks.

Thanks, i'll give them a try. I used them for the analog supply for a multi-channel volume control using multiple pga2320's and opamps. It replaced some teddyregs that were not working very well in this circuit. I think due to their high output impedance they are better suited to supplying less components and low currents. It now sounds much cleaner. There is more and better bass, and some nasty sibilance went away.
 

iko

Ex-Moderator
Joined 2008
stormsonic said:
Iko, MPSA is 100 Mhz bandwith, gain 500 to 1500.

BC550C is 300 Mhz bandwith with gain 420 to 800. Darlington connection to increase gain.

It's easy enough to try it out, I encourage you to do it. Try both, see which one you like.

I've extensively explored ideas like this, and arrived at v2 and then v1.5 after many, many things tried. But since you're curious, here's the darlington vs mpsa18.
 

Attachments

  • reg-v1-darlngtnvsmpsa18-zout.png
    reg-v1-darlngtnvsmpsa18-zout.png
    10 KB · Views: 796
diyAudio Chief Moderator
Joined 2002
Paid Member
stormsonic said:
Iko, MPSA is 100 Mhz bandwith, gain 500 to 1500.

BC550C is 300 Mhz bandwith with gain 420 to 800. Darlington connection to increase gain.

We are near instability in that circuit with too much hfe. Watch out.


col_s said:


Thanks, i'll give them a try. I used them for the analog supply for a multi-channel volume control using multiple pga2320's and opamps. It replaced some teddyregs that were not working very well in this circuit. I think due to their high output impedance they are better suited to supplying less components and low currents. It now sounds much cleaner. There is more and better bass, and some nasty sibilance went away.

Great. Congratulations.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
ikoflexer said:
I would like to share a bit of good news. Just finished a high voltage version of v1.5, however, using n-channel mosfets. Although this thread is about low voltage, I think it is relevant in that this topology is the same as the one I recommended earlier to Tham. So that should give him some hope that it's not just simulated vapor ware. I used two trimmers, not set to the exact value shown as resistors in the figure (R1 and R10).

As I said, this circuit is currently running in reality, and it is stable. My intention is to use this (or some variation) for the dht headphone amp I'm developing in another thread.

:cool:

That will guarantee special sound from your new headphone amp. I believe it will stay stable with that amp for load too.
 

iko

Ex-Moderator
Joined 2008
AndrewT said:
20k R9 and D6 charge up the 470uF C6.
that turns on the pass FET.
Then we have to wait until the FET charges up C5.
Andrew, spot on.



What power will be dissipated in the 20k with 160 to 180V across it?
Without the 20k and Zener the Jfet will explode.

Less than 2W. I have very little experience with high voltage and the tricks people do to protect parts from over voltage. When I don't find established methods, I just improvise. I agree, in that circuit, the jfet would not survive the turn on voltage without that protection. If you know of a better way to do it, please let me know.


Once the caps are charged, what current will pass the 20k?
Will that be enough to keep the jFET turned on?
Yes, the current would be a bit higher but the jfet's S-G resistor/trimmer I thought of setting for about 2mA. Then the rest of the CCS mosfet biasing mechanism should work fine.

That Darlington proposal was raised months ago.
I thought you or someone else gave an explanation of why it was not possible.

Yes, I tried darlingtons in a variety of variations on v1. If they didn't make it into v2 and v1.5, the reason must be because they didn't work well enough.

Here's a rough number of currently saved versions that I've tried; variations on this regulator: 397. And these are only the versions that I deemed worth saving, but there are others that I tried and never saved.
Code:
$ ls *salas*.asc *reg*.asc reg*.asc | wc -l
397

Yes, my wife wants to kick me in the shin knowing I don't get paid to do this and that I do it just for fun :D
 

iko

Ex-Moderator
Joined 2008
Salas said:


That will guarantee special sound from your new headphone amp. I believe it will stay stable with that amp for load too.

I hope so :) I would like to provide the dht with the cleanest and most dynamic supply. Also active load (gyrator) ... I hope to squeeze all that a dht has to offer :D Of course, it may be a royal flop in the end :rofl: but if I don't try it, I won't know, right?
:smash:
 
diyAudio Chief Moderator
Joined 2002
Paid Member
ikoflexer said:


I agree.

BTW, salas, I kept forgetting. But if you get a chance, I would recommend that you try in your shunts the mpsa18. You have good ears, and should be able to tell the difference.

Must be a better controlled for low noise part, with less GB product. Still high hfe. Possibly lends a perceptible sonic difference?
 
Administrator
Joined 2004
Paid Member
Hi Andrew,
the maximum current that a trimmer track can pass is determined by the maximum power the whole track can dissipate.
That's only at maximum resistance! You must de-rate the power by the fraction of the travel in degrees. Figure out how many mW per degree of rotation, and multiply that by the minimum rotation you see used in circuit. So, this is for the element itself...

However, the contact has not been considered at all. Multi-turn controls are not designed to pass any current through the wiper at all. Minimal at best. They are designed to operate as a potential divider, so you have current that passes through the element, but nothing to speak of through the moving contact. This is a critical error in the use of these parts. Consider a volume control that is familiar to us all. How much current is expected to pass through the element? Through the wiper contact? Consider that this is a huge control with large contact surfaces compared to trimmers. You can not use a trimmer as a rheostat!

All,
If you need to use an adjustable resistive element, they do make wire wound rheostats for this purpose. You may have seen these in old TV sets or industrial controls. A Lamda power supply comes to mind. If you only need to sample the voltage (ie: no current), then you can use a resistor as the pass element and place a higher value trimmer across this. The wiper then feeds the high impedance point that samples this voltage. jameshillj did suggest this (I think this was your intent?).

-Chris
 
Administrator
Joined 2004
Paid Member
Hi Salas,
Remember, darlington transistors are slow. Composite darlington transistor pairs need sufficient base current in the first of the part to be fast. Therefore, unless you are careful, then a composite darlington will also be slow compared t a single transistor.

How many RF darlington transistors has anyone seen?

-Chris :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.