The electrolytics in my preamp are about 20 years old and due for replacement. In replacing the ones in the power supply I thought I'd move up from Philips to Nichicon KZ but the 1000uF 50V cans are a bit large for this space. I could attach via some hookup wire or could try 1000uF25v. Is the 25v likely to be too low a voltage? The Philips they are replacing are 1000uF 63v, but just after that are the opamps which are rated at only 15v. I've attached the PSU schematic for reference.
View attachment Classe-Model 30-PSU.pdf
View attachment Classe-Model 30-PSU.pdf
The values used were pick either because they had a lot of them to use up or convenience. As the 78 and 79 series regulators will fry with more than 28Volts on their inputs, 35Volt types would be as low a voltage that I would be comfortable with.
Using flying leads introduces issues and I will not recommend that.
Why not look at Farnell, CPC or Mouser for replacements that are the correct size?
Using flying leads introduces issues and I will not recommend that.
Why not look at Farnell, CPC or Mouser for replacements that are the correct size?
First measure the voltage that is on the old caps and measure the AC secondary voltages please. Then we can give you sound advice. I think 63 V rating is over-dimensioned quite a bit but I could be wrong. Before the filter caps of the 7815 and 7915 there are series resistors probably to loose some voltage. Maybe the transformer they used had too high voltages ? Just measure and tell the results.
Do not use extended leads ! Belongs in the "failure category".
* I would replace the 4700 µF/16 V caps for 10 to 100 µF 25 V though. 4700 µF is way too large if they're situated right after the regs. Don't use low ESR caps there. It is silly that they used larger filter caps after the regs than before the regs. Wrong thinking, the first filter caps and the regs will filter out low frequency ripple, not the caps after the regs. Larger values are OK if series resistors are used (which they didn't). Have seen quite a few commercial audio devices that had oscillation of the regs because of too large caps after the regs.
Do not use extended leads ! Belongs in the "failure category".
* I would replace the 4700 µF/16 V caps for 10 to 100 µF 25 V though. 4700 µF is way too large if they're situated right after the regs. Don't use low ESR caps there. It is silly that they used larger filter caps after the regs than before the regs. Wrong thinking, the first filter caps and the regs will filter out low frequency ripple, not the caps after the regs. Larger values are OK if series resistors are used (which they didn't). Have seen quite a few commercial audio devices that had oscillation of the regs because of too large caps after the regs.
Last edited:
Thanks- OK, caps to the right of the regs need to be 25v or higher for the 15 volt regs and 16v or over for the 5v reg. Caps to the left, the input side, well I would measure the actual voltage but I suspect 35v is probably near the mark. If the voltage is around 22 then don't use 25v but go for 35v
There's a secondary reason for picking larger caps. Check a datasheet and you will notice that rated ripple current is correlated to can size which in turn is correlated to max voltage.
The proper way to replace them would be to find the datasheet of the originals and find a replacement with similar or better max ripple current.
The proper way to replace them would be to find the datasheet of the originals and find a replacement with similar or better max ripple current.
Hmm, interesting. That schematic is not even what is in there! Can't believe I didn't notice before. The earlier caps on the pos and negative rails is 1000u 63v but measures 15vdc in use. The second cap in the chain is not 4700u 16v but another 1000u 63v, and it measures 28vdc in use!
So the main power rail schematic cannot be right, as the operating voltage is 28v which would fail a 16v cap. But it seems I could either:
BTW: On the 5V rail the spec shows first 4700u 63v and then 4700u 16v, but both are 4700 16v! The first measures 5vdc and the second 10.8vdc in use.
The first cap on Neg input and Pos input measure at 15VDC. But the second measures at 27.8VDC!
Thank you for reminding me to do the actual in-use measurements!
So the main power rail schematic cannot be right, as the operating voltage is 28v which would fail a 16v cap. But it seems I could either:
- use a 1000u 25v on the earlier caps and 1000u 50v on the second caps on the power rails. OR
- use the 1000u 50v on the earlier caps and 470u 50v (or 100u 50v) on the second caps. The smaller secondary cap might free up enough physical room for the larger caps, though laid on their side.
BTW: On the 5V rail the spec shows first 4700u 63v and then 4700u 16v, but both are 4700 16v! The first measures 5vdc and the second 10.8vdc in use.
The first cap on Neg input and Pos input measure at 15VDC. But the second measures at 27.8VDC!
Thank you for reminding me to do the actual in-use measurements!
Last edited:
Very often, manufacturers choose "by convenience of having fewer parts to stock, with a bit of an over-rating that lets them serve many different purposes interchangeably". Or... "you can buy 100,000 caps at 1000 uF/63V cheaper than 10,000 caps in 5 different values".
(Maybe - sometimes there are "deals")
This is why you see that your system actually has a 1000 uF/63VDC cap where the 4700 uF is specified. "Because it is enough - whether 1000 uF or 4700 uF" (after all there ARE those regulators there, soaking up all the humpty-dumpty noise of rectification. The output-side caps are just electron reservoirs. And you're talking "preamp" here, where mostly balanced-current amplification stages are used.)
OK - so - like other recommendations, here's mine: 35VDC minimum. Choose the can using Mouser or one of the other distributers that has all dimensions ... to fit the same as the original caps. Don't over-worry about 1000, 2200, 4700 uF ... clearly the manufacturer wasn't worried. On the pre-regulation side, you can even go to 330 uF or 470 uF. It is a mistake to put too-large a cap on the input side, as that stresses out those poor little 1N4003 diodes during charge-up. As many of us learn "the hard way" in tube-amps, smaller is often better for the first stage of electron reservoirs. On the post-regulation side, anything from 470 uF to 10,000 uF is fine. The regulators won't allow themselves to "fry" if you use larger values. However ... again: larger values aren't necessarily your friends.
Good preamp circuitry (repeating myself) often uses balanced-current or "constant current" amplification stages. This puts a uniquely low "reservoir requirement" on the filter caps.
Have fun! That should be rule 1.
GoatGuy
(Maybe - sometimes there are "deals")
This is why you see that your system actually has a 1000 uF/63VDC cap where the 4700 uF is specified. "Because it is enough - whether 1000 uF or 4700 uF" (after all there ARE those regulators there, soaking up all the humpty-dumpty noise of rectification. The output-side caps are just electron reservoirs. And you're talking "preamp" here, where mostly balanced-current amplification stages are used.)
OK - so - like other recommendations, here's mine: 35VDC minimum. Choose the can using Mouser or one of the other distributers that has all dimensions ... to fit the same as the original caps. Don't over-worry about 1000, 2200, 4700 uF ... clearly the manufacturer wasn't worried. On the pre-regulation side, you can even go to 330 uF or 470 uF. It is a mistake to put too-large a cap on the input side, as that stresses out those poor little 1N4003 diodes during charge-up. As many of us learn "the hard way" in tube-amps, smaller is often better for the first stage of electron reservoirs. On the post-regulation side, anything from 470 uF to 10,000 uF is fine. The regulators won't allow themselves to "fry" if you use larger values. However ... again: larger values aren't necessarily your friends.
Good preamp circuitry (repeating myself) often uses balanced-current or "constant current" amplification stages. This puts a uniquely low "reservoir requirement" on the filter caps.
Have fun! That should be rule 1.
GoatGuy
There is something very wrong with that values. I think it is the other way around. You can not have higher voltage after the regs than before them ! The first filter caps are for filtering ripple and the voltage on them should be at least 3 V higher than the voltage number of the reg. So an 7815 will have at least 18 V on pin 1 and the 7915 will have at least -18 V at the input. If they work like they should you will see +15 V at the output of 7815 and -15 V at the output of 7915. Can't tell it much simpler....these are classic text book design rules.
Maybe you can draw a new schematic with the right values (as built) ?
And please take a picture of the board. If you want to change parts just do it right in one time. Saves time and the PCB.
Interpreting the given values and putting them in the (probable) right context I would:
- use the largest possible value (try if 2200 µF 35 V fits physically) for the primary filter caps of 7815/7915 and I guess you measured 28 V on those so I would choose at least 35 V rated low ESR caps like Panasonic FC/FM. Lots of choice there. For the primary cap before the 7805 I would use at least a 16 V rated cap. It is wise to choose caps that have the same dimensions as the old ones. Better still: measure AC voltages of the secondary windings to check of you are safe with 35 V rated caps for the 7815/7915. The series resistors are there for a reason.
- 10 to 100 µF 25 V standard electrolytic caps at the outputs of the regs and definitely not larger (if you like the regs to behave well that is). Check 78XX datasheets and you will read why. The regs don't fry or something like that but they will not perform any better and they possibly even oscillate at a low frequency with very large caps.
Don't worry about the 1N4003 diodes. They are not afraid of 2200 µF caps. While you are at it you can replace them for the better 11DQ10 diodes. Since it is your device you don't have to "get away with it" like manufacturers do. You now have the choice to make it a good PSU. The difference in price is marginal.
When you get the hang of it you might want to look how well decoupled the rails are locally at the active circuitry. You might win some points there !
Last edited:
Also think about auto-machining. Far cheaper to have one machine to fit lots of the same cap that lots of different ones.
The simplest way of knowing what voltage to use is to measure the input voltage to the regulators, then allow an extra 10% - just to be on the safe side.
You are right, of course. The lower voltages are the later caps in circuit. I was reading the board backwards to the schematic. BTW: the installed electrolytics (c.1992) were all axial and now most of my choices are radial and so 'fit' the space differently, being 'stood up'. So the NKZ-230 are just 4mm too tall. I'll work it out or go with Panny FC or Nichicon FW which can be stood up without hitting the top of the case, even in 2200uF 50v.
For after the regs, I'll use 100uF 50v, which are pretty small, freeing up a bit of space.
I'm with you on the diodes. Have several 11DQ10 Schottky's on my desk for both the preamp and amp.
Yes, once I've gotten the PSU sorted on both the preamp and amp, I'll begin thinking about the main signal circuitry, but that will be a separate project.
OK I was being conservative. Go for the next highest voltage after adding 10%. In the UK, mains voltage is usually within +/- 10%. Don't forget that the manufacturers also add a tolerance to their products.
If you read the texts of Nelson Pass you will see that Shotkeys are not necessarily the nirvana of diodes in low frequency supplies.
The old Schottky Snake Oil Rag...
Oh please... not more of this snake oil! There are exactly 3 reasons why a diode type may affect the "sound" of a circuit. [1] in an unregulated supply, if their lower forward voltage drop leaves more/higher voltage to drive the signal-path circuits. [2] if the switching transition has lower cross-over ringing, feeding lower hard-to-filter-out RF energy to the amplification stages, and [3] when reverse leakage increases ripple in delivered power, resulting in greater intermodulation distortion of well-balanced (yet still sensitive) amplification stage circuitry.
Note that [1], [2] and [3] are all negated by the use of series-fed high performance voltage regulators. The VR substantially drops the unregulated "reservoir" voltage to a near-constant level; its design specifically also attenuates ripple by at least 80 dB; design-choices that include snubber resistors in turn eliminate RF cross-over ringing or flyback noise. The use of larger caps both pre-regulator and post-regulator further drop feed-through RF, HF and LF noise by another 30 to 50+ dB.
So ... ANY "competent" diode will suffice. Specifically having Schottky-junction devices in THIS circuit can, and will do nothing whatsoever to the regulated, delivered DC voltage. Nothing. Further, there are snubbers in the circuit. This addresses flyback (and forward-charging) HF noise. Modifying the rectifiers might change it a tiny amount but snubbers+regulation sift that out quite effectively.
Don't change the diodes, is my recommendation. Actually - I'm not even sure I'd change the capacitors! Are they leaking? Is the preamp now plagued with audible 60 Hz noise (indicating capacitor failure), or perceptible intermodulation distortion? (indicating post-regulator cap failure) IF NOT ... then remember that "26 years" (as I recall the OP saying) is not a long time for modern capacitors to perform their duty. Yes, for pre-1970s equipment, and especially pre-1960s, 25 years would be entering the danger region. But not after the advent of competent electrolytic cap design in the early 1970s.
Sheesh. Snake oil.
GoatGuy
Oh please... not more of this snake oil! There are exactly 3 reasons why a diode type may affect the "sound" of a circuit. [1] in an unregulated supply, if their lower forward voltage drop leaves more/higher voltage to drive the signal-path circuits. [2] if the switching transition has lower cross-over ringing, feeding lower hard-to-filter-out RF energy to the amplification stages, and [3] when reverse leakage increases ripple in delivered power, resulting in greater intermodulation distortion of well-balanced (yet still sensitive) amplification stage circuitry.
Note that [1], [2] and [3] are all negated by the use of series-fed high performance voltage regulators. The VR substantially drops the unregulated "reservoir" voltage to a near-constant level; its design specifically also attenuates ripple by at least 80 dB; design-choices that include snubber resistors in turn eliminate RF cross-over ringing or flyback noise. The use of larger caps both pre-regulator and post-regulator further drop feed-through RF, HF and LF noise by another 30 to 50+ dB.
So ... ANY "competent" diode will suffice. Specifically having Schottky-junction devices in THIS circuit can, and will do nothing whatsoever to the regulated, delivered DC voltage. Nothing. Further, there are snubbers in the circuit. This addresses flyback (and forward-charging) HF noise. Modifying the rectifiers might change it a tiny amount but snubbers+regulation sift that out quite effectively.
Don't change the diodes, is my recommendation. Actually - I'm not even sure I'd change the capacitors! Are they leaking? Is the preamp now plagued with audible 60 Hz noise (indicating capacitor failure), or perceptible intermodulation distortion? (indicating post-regulator cap failure) IF NOT ... then remember that "26 years" (as I recall the OP saying) is not a long time for modern capacitors to perform their duty. Yes, for pre-1970s equipment, and especially pre-1960s, 25 years would be entering the danger region. But not after the advent of competent electrolytic cap design in the early 1970s.
Sheesh. Snake oil.
GoatGuy
You're a funny guy. I am pretty much convinced that the caps need replacement after all these years. ESR measurement and a check of their value can confirm that. Of course one can wait till they fail but we would call that too late. Costs are low and life of the device will be prolonged so why not replace them ? Caps have become much better over the years but I would not use types with snake oil electrolyte. They sound awful. The ones with goat milk electrolyte sound much more organic.
After consideration I would not change the diodes as they all sound and measure the same. Any "competent" diode will suffice
After consideration I would not change the diodes as they all sound and measure the same. Any "competent" diode will suffice
Last edited:
OK jean-paul.
Change the caps. Maybe every 5 years, just to be sure. And get the yellow-stripe cryogenically treated matched cut-off Schottky diodes too, while you're at it. Don't forget to change out the interconnect wires from copper to solid silver, too! There's a lot of signal loss in untreated copper, Sir Snake says. Maybe replace the rubber feet on the exterior of the cabinet with the soft silicone acoustic isolators. And, perhaps consider replacing the transformer with a custom wound R-core job. Increases the accuracy of power delivery. Rhodium plated plugs, single-crystal power cords, and platinum-rhodium plated jacks. They're easy to replace, and look so much better.
Yes ... its funny. Snake oil usually is, when exposed.
GoatGuy
Change the caps. Maybe every 5 years, just to be sure. And get the yellow-stripe cryogenically treated matched cut-off Schottky diodes too, while you're at it. Don't forget to change out the interconnect wires from copper to solid silver, too! There's a lot of signal loss in untreated copper, Sir Snake says. Maybe replace the rubber feet on the exterior of the cabinet with the soft silicone acoustic isolators. And, perhaps consider replacing the transformer with a custom wound R-core job. Increases the accuracy of power delivery. Rhodium plated plugs, single-crystal power cords, and platinum-rhodium plated jacks. They're easy to replace, and look so much better.
Yes ... its funny. Snake oil usually is, when exposed.
GoatGuy
Ha ha, you are so funny. Come one, you must know even better ways to ridicule matters that you don't believe. Trolling is the way to go when things go against a belief.
BTW don't know why you put my name in italics but that IS my name, sorry for that. I don't need to choose an internet name or any other camouflage, Goatguy.
BTW don't know why you put my name in italics but that IS my name, sorry for that. I don't need to choose an internet name or any other camouflage, Goatguy.
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.