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.
It is possible to add a filtering capacitor across the primary of your mains transformer. It works in combination with the impedance of the mains supply.
John Curl suggests ~10uF. It will have to be X rated.

It is possible to add a filtering capacitor across the secondary of your mains transformer. It works in combination with the output impedance of the transformer. We usually see 10nF to 220nF here.
 
Last edited:
Hi Wheeze,
I am not getting what you said about C1 position: in the usual PSU, we have a transformer, then a FWBR, then filtering (capacitative, inductive). In this last stage is possible, as you said, to use an inductor or a resistors to improve the filtering.
The voltage coming out from all above, supplies the Salas regulator in which is present C1: you said "IF C1 is after the FWBR", but I don't understand where else it could be....:confused:


Hey CLAUDIO,
Sorry for the confusion. What I MEANT to say, was if C1 IS in fact part of Salas and critical in value, then either leave it alone, OR, create a CRC/CLC filter BEFORE the Salas circuit, whereby C1 is no greater than 50uf >R1 or L1 >Salas C1.
Anything larger causes reactive loading on the PT's Secondary windings. I think you'll be happy with the results.
If you can use a choke, even betterk and if you have some Vdc to spare.
I'll be using a FWBR> 24uf >1H 5ohms >Salas
Salas mentions this somewhere, in one of his posts regarding passive filtering before the active filtering.
Regards,
W
 
In the original V1.0 version of the salas shunt, the BC550C was substituted for a mpsa18 by a few people. Would you change the buffer BC550C in V1.1 of the salas shunt, for the mpsa18 as well, or stick with the bc550c for the buffer posisition?

I ask, as I have a lot of mpsa18, and not so many BC550c!
 
It is possible to add a filtering capacitor across the primary of your mains transformer. It works in combination with the impedance of the mains supply.
John Curl suggests ~10uF. It will have to be X rated.

It is possible to add a filtering capacitor across the secondary of your mains transformer. It works in combination with the output impedance of the transformer. We usually see 10nF to 220nF here.

That would be differential mode filtering across on the primary. 10uf is too big and dangerous. 0.1uf X rated. If you wish to go from each leg to gnd, then use 0.1uf Y rated. BUT, it's already in kit form or you can DIY it and buy the board and parts.

Jon Risch AC filter/surge supp and white papers/schemo.
Kit or parts for it
DIYCable.com



Jon_Risch's Web Site for AC filter S/S
 
Hi Salas,
However HiFiNutNut reported oscillation when using a 470microF Rubycon ZL, that has a low ESR of 0.038 ohm.

Claudio,

First I would like to thank you wholeheartedly for your enormous efforts for publishing your extensive Speaker Workshop guide and I have known a very large number of DIYers have benefited from it. Your guide has helped me to obtain my speaker measurements. I am sure a lot of people are also thankful for your great contributions.

When I reported the oscillation it was with v1.1 (i.e. not v1) with the BC550c/BC560c buffer mod. Salas managed it by soldering a carbon resistor directly on the leg of the MOSFET while in my case the track length exceeded 1cm. v1.1 was meant to be a trial version at the time. The official version is v1, which is very stable. I lately removed the buffer (which then made it v1) and looked at the rails of v1 from my scope with a garden variety 100uF output cap and the screen showed a thin flat line.

With regards to pre-regulator, I have lately found the Fairchild Steath diodes to work very well as rectifiers. I also found a small value high frequency choke (390uH) works well. I understand that tube guys like to use LC or CLC with a large L. It is necessary because the PSSR is usually low with tube amps. With solid state, PSSR is usually very good at lower frequencies but poor at higher frequencies. So only a RF choke is needed. A small resistor (0.1R to 0.5R) in series is necessary to damp resonances due to the LCR circuit.

I have good results with my raw power supply. If you want more details, you can send me a private message.

Regards,
Bill
 
I smelled something brewing in the Salas kitchen, once more :D

I also found a small value high frequency choke (390uH) works well. I understand that tube guys like to use LC or CLC with a large L. It is necessary because the PSSR is usually low with tube amps. With solid state, PSSR is usually very good at lower frequencies but poor at higher frequencies. So only a RF choke is needed. A small resistor (0.1R to 0.5R) in series is necessary to damp resonances due to the LCR circuit.
The PSSR depends on the circuit topology, not that much on using tubes (especially penthodes) or semi conductors. Pi-filters are efficient ripple killers, more so if you combine several sections. Besides they filter noise introduced by (too) big caps after the bridge, creating ringing. Pi-filters are usable up to a certain current as coils are getting lumpy and stray fields are getting cumbersome. Also, the voltage drop and raise in impedance must not be counter productive.
The few cases I've studied benefitted from a distributed ripple rejection. The loads were steady, which is paramount to raising impedance in the PSU. For a dynamic load I wouldn't recommend it. Lowering a PSU output voltage can be done by inserting a small resistor between tranny and bridge, also between bridge and reservoir cap. That way you alter the conduction angle, changing the load currents from the reservoir cap.

could you explain why you think it is dangerous?
It wil stay charged for a while when you pull the power cord ;)
 
could you explain why you think it is dangerous?
That size cap, or any in x.00uf on AC mains is dangerous.
I've used them to break in caps and hold one helluva charge and take same when energized. If you energize it at the peak of the cycle it can explode. Lots of info on this. DC does not have a wave form...it's a flat line on a scope.
You also, playing with phasing.
I know what JC is doing and it was a product that Bob Crump sold. I suggest NOT doing so. I sent you links for doing AC filtering the proper and safe way. They work great.
 
The PSSR depends on the circuit topology, not that much on using tubes (especially penthodes) or semi conductors. Pi-filters are efficient ripple killers, more so if you combine several sections. Besides they filter noise introduced by (too) big caps after the bridge, creating ringing. Pi-filters are usable up to a certain current as coils are getting lumpy and stray fields are getting cumbersome. Also, the voltage drop and raise in impedance must not be counter productive.
The few cases I've studied benefitted from a distributed ripple rejection. The loads were steady, which is paramount to raising impedance in the PSU. For a dynamic load I wouldn't recommend it. Lowering a PSU output voltage can be done by inserting a small resistor between tranny and bridge, also between bridge and reservoir cap. That way you alter the conduction angle, changing the load currents from the reservoir cap.

I was playing with the raw supply before a shunt regulator. The shunt regulator should theoretically have very high ripple rejection but it depends on an electrolytic capacitor for filtering. We all know that electrolytic capacitor does not work well at high frequencies due to its inductance. So in practice I found that inserting a 390uH in the raw supply did clean up a bit at high frequencies, resulting in cleaner sound. Not a night and day difference, but still audible.

This led me to investigate different configurations in my power amps - C, CRC, CLRC, LRC. Originally I had C = 3 x 10,000uF per rail per channel. These capacitors have inductance of 16nH and ESR of 0.015R at 20kHz. I tried CLRC 10,000uF - 220uH (0.05R) + 0.1R - 2 x 10,000uF. From LTSpice modelling, I must insert a resistor 0.1R so that it won't have a peak at around 80Hz. While I found the higher frequencies were definitely cleaner, the sound became a bit unnatural to me, perhaps due to increased impedance. I then tried more Spice modelling, and found that LRC 220uH (0.05R) + 0.1R - 3 x 10,000uF would give another a few dB ripple rejection. I tried it. The sound was subjectively the best comparing to C or CLRC. Then I wired up the 4 rails differently and measured them with my scope. The scope confirmed my subjective impression, that LRC is better than CLRC. Note that the total 0.15R resistance in series with the rail is higher than the impedance provided by the capacitors, in the case of CLRC, it was 0.015/2=0.075R, and in the case of LRC, it was 0.015/3=0.05R.

So I am going to redo my preamp level raw supply before the CCS-shunt regulator to change it from CLRC to LRC and see if I can find any subjective improvements.

The above may only work with soft recovery rectifying diodes with which no snubbing high Q film capacitors are used, or possible severe ringing can occur due to the interaction of the C and the L, unless you pack sufficient resistance for damping, but in which case, the increased impedance may do more harm than good.

My hypothesis is this: Filtering can only be done with either capacitor or inductor, (along with resistors), or both. Electrolytic capacitors are OK at lower frequencies but not higher frequencies. Bypassing electrolytic capacitors with ceramic or film caps (all of them are of high Q) can easily cause ringing. Inductors are naturally good at higher frequencies (but watch out for parasitic capacitance, as Salas reminded me, so the proper inductors need to be selected). The idea is to use electrolytic capacitors for lower frequency filtering, and inductors for higher frequency filtering. In this case, only a small RF choke is needed, with the significant advantage of much lower impedance (comparing to a large L or R), minimal voltage drop, minimal heat dissipation, minimized RF rediation, minimal space required, and lower costs. My hypothesis is in sharp contrast to conventional LRC or CLRC in which the large L is to filter ripples down to the audioband or even the mains frequencies.

The only thing I am not sure is this: there may be a significant amount of DC which may saturate the inductor core making the inductor ineffective. I don't know if this can be a case so am seeking experts' advice on this.
 
I'm building QuangHaos Salas for DAC End:
Sala-shunt-for-DAC.jpg


Any recommendation what leds & diodes to use? Thanks!
 
diyAudio Chief Moderator
Joined 2002
Paid Member
The only thing I am not sure is this: there may be a significant amount of DC which may saturate the inductor core making the inductor ineffective. I don't know if this can be a case so am seeking experts' advice on this.

It can be. If there is DC current you need a properly rated core in any inductor. In case of using small values to dump RF, small air core coils are feasible though.
 
Any recommendation what leds & diodes to use? Thanks!

I have not tried a lot of diodes but only a few.

The cheapy shottky diodes 1n5822 (and its series) actually sound pretty good, and sustantially better than the normal 600V 35A bridge, although I have heard comments that the 1n58XX series do not sound as good as other shottky diodes, such as D10XXXXX, etc. I measured with my scope and found the 600V 35A bridge (in both my preamp and power amps) produced horrific switching noises, if I show you the magnified spectrum, you would believe these diodes are dedicated noise generators.

Salas then recommend trying the Fairchild Steath. I tried them. They sounded a bit cleaner than the 1n5822. I measured the rectified wave forms, the result was much much better than the 600V 35A bridges.

Actually, I am spending this week replacing the 600V 35A bridges with the Steath on my 3 pairs of monoblocks (and other upgrades on them as well). I will possibly finish them tomorrow and will measure them again. It will be interesting to see if they make any improvements on sound, as these power amps are very well designed and should have a theoretical 115dB PSSR across the entire audio bandwidth. In theory, these diodes should make no difference to the sound of my amps.
 
It can be. If there is DC current you need a properly rated core in any inductor. In case of using small values to dump RF, small air core coils are feasible though.

Salas,

I thought of air core inductors for loudspeakers. But since they have quite a bit of windings I suspect that the parasitic inductance is on the high side, making the inductor less inductive at higher frequencies.

I am trying these:

http://www.bourns.com/data/global/pdfs/pm2120_series.pdf

The datasheet says that until the current reaches 5.21A the inductance of the 220uH will lose half. To me that is pretty good, as my class AB amps would never draw that amount of current.

The question is, with the inductor immediately after the bridge rectifier, what is regarded as DC? Would the rectified wave form be considered consisting of any DC component?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.