Thanks for all the information guys. I primarily drew up the schematic with what I had in my parts bin which is why it's kind of sparse. I'm also trying to get back to simply playing around with audio electronics again. I've spent the last couple of years designing RF products that have to work without fail and I've felt at times that I have lost the love of simply playing around and making things catch fire while not being under intense pressure to meet deadlines.
That being said, lets get to the discussions!
A small resistance (2 - 5 ohms) after the main diode bridge before the first filter cap will get rid of some diode bridge noise. Largish W or several 1/2W in // would help them not burn up on power on.
That's interesting. I've not done any resistance in series with the output of the diode bridge before on any design. I mainly do RF design and it's a big NO-NO to have series resistance anywhere as your PSU is intended to be as low impedance as possible. Of course I don't know how this applies to pre-regulation so I'll be happy to try it out.
Using several filter caps in // to get your 10,000uF will lower ESR.
This is one of those times where I knew better but I just did the layout with what I had laying around. Maybe I'll do a revised circuit with all of the goodies.
R1 and R2 supply about 5mA to the zeners which is good. They're quietest around there.
Another 1N4004 from R1/C1 junction to unreg'ed V will help with discharge of C1. Same with the R4/C2 junction.
Since I've not really explored this type of topology before(I'm usually on a tight schedule and use packaged lin-regs on most designs), is this more of a robustness addition to keep the base-collector/emitter voltage from becoming too large when power is cut? It was my thought that the zeners would pull the cap down fast enough but then again I haven't actually tested it either.
A resistor from the top of the zener strings to the bases with another cap from base to ground would get rid of more zener noise. Use a slowish time constant for voltage ramp up (they'll think it's tubes!) I did one once that used a really nice big poly cap there and had a 4+ minute time constant. No noise at all!
Elegant. I'm going to wire that up right away! Looks like I need to do a revised layout with all of the goodies for sure. I can find a happy resistor value easily enough but a quick simulation isn't telling me much.
An active current source (again 5mA) to the zeners instead of resistors R1 and R2 would be another refinement to isolate the reg'ed from unreg'ed.
In your opinion, how much do you gain by doing something like this for a circuit like this?
Use a differential filter on the secondary of your power transformer to get rid of some mains hash. Especially if your using a wider bandwidth torroidal.
I generally use a filter network on the AC input to the unit. Would you still suggest using a common mode filter on the secondaries too?
Edit: I also think HexFreds are a good investment over 1N4000 series. Schottky diodes are quieter too but some are voltage limited, good for tube heaters though.
Although I built mine using what I had laying around, I do also have a number of MUR860 that I usually use for designs I consider more touchy to power supply circuits. Bcarso also turned me on to the SiC diodes which are really good but pricey. Each of my Neve modules has R/C filtering too which is why I went a little sloppy on the main regulation.
- if you have a fast enough transistor and low-ESR capacitors on B and E, you may want to include a small base stopper resistor (or ferrite bead, if you're trying to get the very best audio-band noise performance) on Q1 and Q2
I have plenty of ferrites laying around!
- for short circuit (and on/off-transient) protection you may want to include a diode between B and E of Q1/Q2
This would be in addition to WOB's suggestion for one between B and C?
For good measure I'd include a diode between E and C as well.
Working with power MOSFETs and IGBTs has taught me to always have a flywheel diode. The darlingtons that I am using have internal reverse biased diodes for protection so I didn't add any extra. I guess anybody who built this would need to make sure that they are protected.
- the thing isn't very green, is it?
;D The good stuff never is!
A power resistor in the collector connection of Q2 sized for ~24V drop at maximum operating current will likely be cheaper than a large heat sink for Q2 (plus it serves as another overload protection measure
Honestly I don't know how much load I'll be applying once everything is finished. I took a guess but even with fairly large loads I only dissipate a couple of watts. Once I get everything working at full capacity I'll take a look at doing this.
Thanks again Abbey, WOB and JDB