C'mon folks,
I know it's not homebrew cabinets or cloned amps, but surely someone out there can post their opinion on this? There seems to be a hell of alot of questions/threads going without a single reply here, what's up with everyone? It never used to be this way.
Seems folk can't be bothered to reply to little questions anymore, it used to be informative forum but now it's just a case of 'will anyone reply?'
Anyhow, thanks.
I know it's not homebrew cabinets or cloned amps, but surely someone out there can post their opinion on this? There seems to be a hell of alot of questions/threads going without a single reply here, what's up with everyone? It never used to be this way.
Seems folk can't be bothered to reply to little questions anymore, it used to be informative forum but now it's just a case of 'will anyone reply?'
Anyhow, thanks.
I have seen it posted several times that, at least before the regulators, the circuits prefer 'slower' caps. I suppose other parts of the circuit come into play, but there are technical explanations that are on here if you search hard enough.
I did not have great luck using Panasonic FMs before a set of regulators in my preamp. Sounded tinny to me.
One commercial example would be my 20 year old Harman/Kardon CD player. When I popped the top on it, the original caps were standard, 85C Matsushita (Panasonic) before the regs and finer, audio-grade caps were used in later parts of the circuitry.
Search threads for LM317, etc. to find different implementations and caps/sizes used.
I did not have great luck using Panasonic FMs before a set of regulators in my preamp. Sounded tinny to me.
One commercial example would be my 20 year old Harman/Kardon CD player. When I popped the top on it, the original caps were standard, 85C Matsushita (Panasonic) before the regs and finer, audio-grade caps were used in later parts of the circuitry.
Search threads for LM317, etc. to find different implementations and caps/sizes used.
So you have waited a whooping 4 1/2 hours? Good grief!
Right after the rectifiers is wise to use caps with "non so small" ESR. After the regulators, on the board is where low ESR counts and needs to be.
And more important, stay with the maximum voltage of the capacitors close to the voltage on the board, don't go higher just "to be safe" - ERS goes down with the maximum voltage.
Last edited:
I think folks have found the opposite to be true, i.e. that a somewhat higher value cap provides a better resulting power supply.
It all depends - my advice is forget boutique caps after the reg. Use medium esr types and draw 300ma + standing current from the reg (for the LM317 at least). I'll tell you why later if you really want to know. The other important stuff is to precede the reg with a really low cutoff (i.e. a few Hertz) RC filter using an ultra-low ESR cap and to use a high value ultra-low low ESR cap (200-1000uF) for Cadj. The other things you see like using external Vrefs to replace the reg's internal one amount to burnishing the turd for 3 pin regs. You may as well move to Jung types at that stage.
Personally I wouldn't use 3 pin regs for anything except digital circuits and CD servo duties. Critical audio amplification and DAC circuits need Jung type regs.
Personally I wouldn't use 3 pin regs for anything except digital circuits and CD servo duties. Critical audio amplification and DAC circuits need Jung type regs.
Last edited:
For the Jung-Didden Super Regulators a ceramic or film bypass across C5 may lead to oscillation. These gentlemen measured the regulator impedance down to micro-Ohms. You'll find that careful layout, star grounding, no daisy chaining of supply rails will work more to your advantage than boutique capacitors.
Its no secret - the output impedance graphs in the LM317 datasheet are shown for 500mA output current. Take a peek at the LM317L datasheet to see what happens at 40mA- not a pretty sight.
Impractical and unnecessarily expensive. To get to a few Hz with 300mA+ running through you'd end up with of the order of a 10,000uF cap at this point - such large values don't come at all cheap in ultra-low ESR. By all means use a nice big cap but just a standard one and parallel it with something that's still going to give a low impedance above 100kHz - like a high enough voltage high value SMT ceramic.
What's the point of an ultra low ESR cap here? I can't see how its going to give bang for the buck in the reference leg where the source impedance its seeing is at least 100R. Low ESR caps are best suited to low impedance circuits.
That stage I'd say comes before the time you start shelling out for ultra-low ESR caps of thousands of microfarads at high voltages
In the power supply for the Last PAS preamplifier (a "duded-up" Dynaco PAS) Walt Jung recommended a polypropylene capacitor for Cadj (if you can afford it!). I ran through a box of WIMA 4.7uF/400V caps when I was providing the boards as a GB about 5 years ago!
Last edited:
Curious - did he say why? Seeing as ideally the Cadj needs to be at the very least 10uF (and preferably 22uF or more), that's going to be one fairly huge cap - polyprops don't get made in low voltages! Wiring such a physically large cap into the circuit will incur considerable series impedance in the connecting leads. Did he specify PP for low dielectric absorption by any chance?
One thing you don't want to do is a very low esr boutique cap after the reg. The cap directly after the reg has one purpose and one only: to keep the reg stable. They work best when they have some esr - after all, damping means removing energy and you can't do that without esr.
Before the reg you have more freedom. Boutique caps here are allowed, the only restriction being your budget. Not that it makes an audible difference, but it makes a nice conversation piece.
jan didden
Before the reg you have more freedom. Boutique caps here are allowed, the only restriction being your budget. Not that it makes an audible difference, but it makes a nice conversation piece
.jan didden
Last edited:
oops, my bad, the 4.7uF is on the out leg of the LM317 across a photo-flash capacitor, the 470nF is on the ADJ pin -- see the attached schematic -- the LM317 is not an LDO regulator (obvious) . The Super Regulators are LDO so, the output cap is part of the feedback network.
The "RS..." resistors were a suggestion of Mr. Yaniger and can be jumpered.
An externally hosted image should be here but it was not working when we last tested it.
Bigger is better -- Panasonics I have used --
An externally hosted image should be here but it was not working when we last tested it.
Just realized there's a more compelling reason than cost to avoid high value ultra-low ESR caps on the Cadj pin : leakage current. Chemi-con's ultra-low ESR polymer caps have a leakage current (after 2mins) of 0.2CV, giving over 1mA for a 16V 330uF. And that's after 2mins - what happens prior to 2mins is going to be worse. So output voltage stability (aka LF noise) is going to be seriously degraded by specifying a polymer cap in this position
That is unless one decides to run the whole 300mA+ standing current through the output voltage setting resistors..
Just realized there's a more compelling reason than cost to avoid high value ultra-low ESR caps on the Cadj pin : leakage current. Chemi-con's ultra-low ESR polymer caps have a leakage current (after 2mins) of 0.2CV, giving over 1mA for a 16V 330uF. And that's after 2mins - what happens prior to 2mins is going to be worse. So output voltage stability (aka LF noise) is going to be seriously degraded by specifying a polymer cap in this position
Did you see Jim William's methodology for measuring leakage? (Appendix B of ApNote 124). He used a real "boutique cap" in his noise machine -- a $400 wet slug tantalum.
I set this up to test the caps in an HP 465A Amplifier. The alligator clips have copper tape and copper foil connecting them to the battery, load and nanovoltmeter:
An externally hosted image should be here but it was not working when we last tested it.
One thing you don't want to do is a very low esr boutique cap after the reg. The cap directly after the reg has one purpose and one only: to keep the reg stable. They work best when they have some esr - after all, damping means removing energy and you can't do that without esr.
Before the reg you have more freedom. Boutique caps here are allowed, the only restriction being your budget. Not that it makes an audible difference, but it makes a nice conversation piece
jan didden
Hi Jan.
I was wondering what caps you'd recommend for standard ESR typs after a regulator please? I'm currently using the recommended 1000uf - LM7808CT - 1000uF arrangement for the simple PSU for a 'Shigaclone' CD transport. I'm using Rubycon ZLs which are apparently VERY low ESR. So I should be using something different after the LM7808 instead? I think the datasheet says to just use a 0.1uF ceramic cap? I'd like to get this working correctly, rather than JUSt using my ears to 'tune' it which others seem to be doing.
Many thanks,
- John
Just realized there's a more compelling reason than cost to avoid high value ultra-low ESR caps on the Cadj pin : leakage current. Chemi-con's ultra-low ESR polymer caps have a leakage current (after 2mins) of 0.2CV, giving over 1mA for a 16V 330uF. And that's after 2mins - what happens prior to 2mins is going to be worse. So output voltage stability (aka LF noise) is going to be seriously degraded by specifying a polymer cap in this position
I'm somewhat of a newbie but my research agrees with you about needing some ESR for regulator decoupling caps.
I read something in one of the voltage regulator articles on Just one thing about music - when it hits you feel no pain that you can get very good results from using an ultra-low ESR decoupling cap followed by a low value resistor to get the resistance that the regulator sees back up to a desirable (or optimal) level. Does anyone have a comment on this concept?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.