Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ilcaccillo]

Hi Klaus and everyone

I was asked to perform a recap in a U67 PSU.

Do you think for the 25uf 350V caps is better to replace them with 22uf or with 33uf?

I ask the same for the 500uf 35V, should I go for 470uf or the next value up 560uf?

Thank you so much

[Kai]

Electrolytic caps are usually supplied with high tolerances anyway.
-10 / +20 % would be a typical value.
At the time the PSU was built, -20 / +50 % was the norm.

Functionally it doesn’t make any difference.

Besides fit, choose the ones with better quality from what’s available.
105 (or even 125) °C, and the higher load life hours are the most important criteria.
Low ESR is welcome, but not necessary, these didn’t exist when the PSU was created.

Typical load life hours are 1000 - 3000 h at spec‘ed max. temperature.
This doesn‘t seem much, but this value doubles each time working temperature goes down by 10 °C.
Starting with 2000 h at 125 °C you end up at 1 million hours for 35 °C.
That 100 years.
Starting more typically with 1000 h at 105 °C you still have 15 year at 35 °C.


Practical experience shows that the bigger sized ones suffer less, so in this case the bigger really is the better.
This is why a lot of these old caps in the tube mic PSUs are still up to specs.

[soapfoot]

Old electrolytic caps were often given in tolerances like "-20%/+70%." In other words, extremely wide and often erring on the side of missing "high"

Over time a standard set of values emerged (your 22, 33, 47, 56, 68, etc) but in the middle of the last century and before, values were often in multiples of 5 and 10 (this was rather arbitrary).

In practice, there is not any difference between a "22µF" and a "25µF" capacitor (or between 47 and 50, etc)--not in use, and probably not even in manufacturing.

10% is considered a fairly tight tolerance in electrolytric caps (historically), and even there a nominal "50µF" cap could easily read as low as 45µF, and a nominal "47µF" cap could easily read as high as 51.7µF. I treat the old "50µF" parts as perfectly-interchangeable with modern 47µF parts.

That said, no harm is going to result if you move from 25 to 33 or from 50 to 56.

If you want to do some math, the corner frequency of the last filter pole (8k2 resistor, 25µF) is 0.8Hz. With 22µF it moves up slightly to 0.9Hz. With 33µF it'd be 0.6Hz

Considering it's part of a multi-pole filter array, that's a lot of ripple reduction any way you slice it (more than enough to get clean DC at output)

[soapfoot]

Excellent advice, and we can add that electrolytic capacitor technology has grown by leaps and bounds over the past 70 years, and particularly the last 20 years. In some cases it may even be possible to use newer organic polymer types instead of old wet electrolytic types, but always study the data sheet.

Whatever capacitor technology you choose, measure the old part and look at the data sheets for prospective new parts. Get the highest voltage rating, temperature rating, and service life that will physically fit.

When replacing half-century-old parts, it's often very possible to replace a 350v, 85˚C, 1k hour part with a 450V, 125˚C, 10k hour part in the same capacitance value (or better!)

"On paper" at least, that's a lifetime fix for most people.

I've lately been very, very satisfied with the KEMET PEG124 series. The measurable specs on these are extremely impressive--widely available from suppliers like Mouser, and the kind of performance that the old designers would've only dreamed of in an aluminum electrolytic. Data sheet here

For your C1, C3, and C4 for instance, here is a part that would likely fit (can't remember if the NU67 uses axial or radial leaded parts), is a 450v part with a 22k hour service life at 105˚C.

If you look at the derating curves, you could expect a very long life at normal ambient temps!

[klaus]

Funny, but in all the years I have serviced NU67 I have maybe once replaced a filter cap....

Recapping, quite common and necessary in mic pres and other line level amplifying devices is rarely if ever necessary in power supplies for microphones, unless, of course, there are symptoms like hum, noise, voltage drops, and measurable deterioration in capacitance.

[soapfoot]

Funny, but in all the years I have serviced NU67 I have maybe once replaced a filter cap....

Recapping, quite common and necessary in mic pres and other line level amplifying devices is rarely if ever necessary in power supplies for microphones, unless, of course, there are symptoms like hum, noise, voltage drops, and measurable deterioration in capacitance.

One reason for this may be that the capacitors in the original devices operate at ambient temperature and were fairly conservatively-rated!

In the NU67 C1-C3 are specified as 350v parts, and C1's 270VDC is as demanding as it gets. Since electrolytic caps in (more general) tube power supplies (e.g. hi-fi and instrument amplifiers) are often run quite close to rated maxima (and parts tended to be designed with this consideration in mind), this is fairly-conservative operation (especially considering that the chassis will remain close to ambient temperature--nothing like the brutal environment adjacent to a scorching 5U4 rectifier!)

Likewise for the filament supply--30V parts are specified, and 19.1V is the highest demand (down to 7.5V by C7). C5 will see the most ripple current of any capacitor in the supply (by far), so would typically be the first to show age-related deterioration.

The inductance of the power transformer secondaries will help dampen the inrush current somewhat, which helps at turn-on.

It's worth adding: even if old capacitors still "work" and measure close to nominal capacitance value, their equivalent series resistance may have increased with age (and almost certainly began higher than modern parts).

This is a reliability concern, as high ESR in a capacitor can cause it to run hotter as current is pulled through the resistance. The resulting extra heat can cause a destructive feedback loop of accelerated aging (increasing ESR further), building heat and pressure inside right up to the point of a messy catastrophic failure.

So if ultimate reliability is a concern, there can still be some merit to preventive replacement of aging electrolytic capacitors (particularly in high-voltage or high-current applications).

If ultimate originality is a concern, there can be some (limited) argument for keeping the old parts in (though removing to measure for ESR using proper equipment can be wise to get out in front of any potential issues). In situations where such provenance feels important, I generally prefer to keep the original parts... in a ziploc bag. In my experience this is enough to keep "vintage value" high enough while actually protecting the long-term health of the device.

If economy (or expediency) is the primary concern, many will leave the old parts in until failure (though this can obviously prove to be a false economy).

[klaus]

Thank you for an enlightening post.

But I am still puzzled by the discrepancy between your quite reasonable-sounding theory and my empirical history:

The resulting extra heat can cause a destructive feedback loop of accelerated aging (increasing ESR further), building heat and pressure inside right up to the point of a messy catastrophic failure.

.

Why have I not encountered such catastrophe in 35 years of servicing hundreds of U67 systems?

[soapfoot]

Probably this is largely due to their fairly light duty (ambient temperature, and operating at derated working voltage). High initial quality of original parts could also be a factor--I'd expect Neumann and AKG both to use the very best in that era, as they were designing for demanding professional applications.

I should clarify that my "high-ESR ultimately leading to failure" scenario is kind of a worst-case (but not by any means an edge case). It won't occur in every old electrolytic cap (or even the majority), but it's common enough to have seen it dozens of times in audio gear more-broadly over the years, and disruptive enough to be worth taking steps to avoid (though, like you, I've never encountered a catastrophic capacitor failure in an old Neumann supply). Bold added by the moderator.

Ripple current, working voltage, and ambient temperature are all interrelated variables when it comes to the service life of wet aluminum electrolytic caps. Sometimes borderline levels of demand with respect to one variable can be tolerated as long as the other demands are modest, and if the cap is conservatively-operated in all respects an unusually-long service life can be common (as you've seen).

One imperfect analogy might be brake pads on a car. If you always drive slowly, always gradually slow to a stop, and drive only a few miles per year then the brake pads can have a very long service life--but that doesn't necessarily make it a bad idea to replace them preventively if they're very old, as they're consumable items and the cost of the maintenance is much smaller than the cost of a failure.

It's a judgment call and qualified minds can reach different conclusions, but if I'm already in a device doing other work I tend to be dispositionally-inclined to spend the extra money and do the electrolytic capacitors while I'm at it. If done properly with the highest-spec parts, the device is then refreshed and has the highest-percentage chance of operating trouble-free for decades more.

If one of the capacitors has already failed in any respect (even drifting out of tolerance), I automatically do them all, no questions asked. Agreed. KH

[Kai]

Why have I not encountered such catastrophe in 35 years of servicing hundreds of U67 systems?

Because at the time these PSUs were built, parts were not spec’ed as precise as today.

This lead to making these parts as good as could be possibly done, often much better than today, with more reserve built in.
This is especially true for the bigger sized elements.

A big AL electrolytic cap that does not leak can almost last forever if not challenged.

Only the small sized ones, specially true for those extra small ones (by measure of that period) inside microphones, are not sealed equally good, and that’s often even visible from the outside. The reason is simple: a good seal needs some space that was not available if you want to build a small part.

The typical tiny SMD EL caps of today leak like hell, e.g.

[ilcaccillo]

Thank you all for your replies.

I feel sorry that I didn't keep the original electrolytic caps I replaced in the U67 PSU, because I could provide some measurements of those old caps when compared to the new Panasonic 105º caps I used to replace them. Unfortunately I threw them in the bin after replacing them.

I measured a couple of them and ESR was quite high, much and much higher than a new Electrolytic cap. With those higher ESR reading I really don't think they were up to spec after 60 or 50 years, and I don't see any reason for not performing a full electrolytic re-cap when servicing a U67 PSU or mic.

Present-day electrolytic capacitors are much better than anything ever done in the past

[klaus]

(...) I don't see any reason for not performing a full electrolytic re-cap when servicing a U67 PSU or mic.

With that last aside (in bold) the ice you are treading on gets decidedly thinner.
You should not replace a condenser mic's coupling cap or its cathode bypass cap before doing a deep dive into the sonic consequences.

[Kai]

I measured a couple of them and ESR was quite high, much and much higher than a new Electrolytic cap. With those higher ESR reading I really don't think they were up to spec after 60 or 50 years, and I don't see any reason for not performing a full electrolytic re-cap when servicing a U67 PSU or mic.

Measuring ESR on a Linear PSU’s cap has it’s difficulties:

If the PSU wasn‘t used for a long time the cap has to be “reformatted”, by putting power on it for several days, before ESR can be judged.

ESR has to be measured at the frequencies the cap has to filter, somewhere in the range below 500 Hz, with 60 and 120 Hz most important.
Most ESR testers default to higher frequencies for the measurement.


ESR of a coupling cap, or a cap somewhere else in the audio path, in a condenser mic, doesn’t mean much (if at all) to the audio.
There are other things that are important, like DC loss, and audio quality.

[ilcaccillo]

With that last aside (in bold) the ice you are treading on gets decidedly thinner.
You should not replace a condenser mic's coupling cap or its cathode bypass cap before doing a deep dive into the sonic consequences.

I see I was not clear in the way I phrased it, my post was only concerning a full Electrolytic Re-Cap of the PSU only.
This is what I wanted to say "I don't see any reason for not performing a full electrolytic re-cap of the PSU when servicing just a U67 PSU or a Mic+PSU"
I only meant the 6 PSU filter capacitors.

But anyway I already replaced the Electrolytic capacitor (C9) in a few U67 microphones, I only used high quality Panasonic or Nichicon,
and there was definitely not degradation in audio quality or sonics, actually I felt the opposite the mics sounded better in my opinion.

I personally don't trust any 50 or 60 years old Electrolytic capacitors in any piece of gear, had too many problems, failures and repairs to do over the years due failed old Electrolytic capacitors, and had zero problems or disadvantages when replacing very old Lytic caps by modern high quality ones.

This is my personal experience, I totally respect if you guys have a different experience than mine.

Thank you Klaus and Kai

[Kai]

To say this clear:

I don’t usually fix stuff that is not broken.
Broken for me means there’s a real problem with the mic, not just a component has “out of specs” measurements by today’s measures.
Even then I try to avoid replacing more parts than necessary.

Therefore I have few experience with sound changes when replacing electrolytic caps.
There’s just no possibility to A/B a new with a broken part.
So no comment from me on this.

Astonishingly I’ll have faced more problems with capsules (usually contamination), tubes (wear), transistors (noisy), and contacts (dust and dirt), than electrolytic caps in historic microphones.

The same is not true for other older and not so old electronics, electrolytic caps are number one source of problems in those.

[ilcaccillo]

To say this clear:
I don’t usually fix stuff that is not broken.

I understand and I do the same.
Although a PSU Electrolytic capacitor from 1960 might work and not seem to present problems it will be out of specs after 60 years,
and will not perform better than a new, modern and high quality Electrolytic capacitor, so even do it's not completly broken I don't personally see any reason for not Re-Capping the PSU and replacing the 60/50 years old Electrolytic capacitors.
It's cheap enough and then you can be sure that the Lytic Caps are performing 100% and have much lower ESR than the very old caps.
This is my personal view, it will never make the PSU worse, it will be the same or better.

Broken for me means there’s a real problem with the mic, not just a component has “out of specs” measurements by today’s measures.

In the PSU “out of specs measurements by today’s measures" is a good enough reason to replace them

Thank you for your input Kai