Electro cap replacement - Comment invited.

[Steve Jones]

I have been an advocate of replacing electrolytic capacitors in old gear for a long time. Because so much of the gear that comes into my service shop is in the 15-30 year old range, I believe that there is no harm and some real benefit in replacing these caps in general. (Tantalums of course are replaced without question). Many would disagree with me, for whatever reasons, but no-one has given me any really solid reason to leave ancient electros in old gear, usually it is a kind of "well, it might change the sound" kind of answer.

We know that these caps dry out and go down in value, and that they need power to keep the electrolyte intact and to stop it thinning. I would wager that just about every old piece of gear has had at least one or two times in it's life where due to failure or falling out of vogue it has been left under a bed or in a cupboard for years on end with no power. This is fatal to electrolytics. Then they get turned on again with full power, so they can't self-heal properly, and suffer even further. These caps weren't designed to have service lives in decades anyway. All of the old synths that I have re-capped have had positive sonic or envelope timing results from being re-capped, and power filter caps that have borne the brunt of surges and spikes for decades are definately candidates for failure.

So, after all the prattle, my question is this: If I replace the electrolytic capacitors in say for instance a Urei 1176 with good quality new caps, then calibrate the unit to factory spec, why would this have anything other than the effect of rendering the unit as close to how it was intended to sound new? Is there some "voodoo" or "X factor" in dried out old caps that will kill the sound of the unit? Some would say so, but my ears don't tell me that at all. So if anyone has any science that says otherwise, I would love to know!

 

[Mark Burnley]

Hi Steve,

Well, here's my ha'pennyworth...

I'd say in balance, 75% of the faults I find in equipment come from the power supply. The first thing I check when I know I've got a power supply fault is what's strapped across the rails...9 times out of 10 there is some sort of electrolytic that has dried out and "crowbarred" the supply, thus taking out voltage regulator/pass transistor etc etc.

A friend had an A80-MkII which was totally "dead". The machine was born in 1983, and the entire machine was riddled with electrolytic faults- the main power regulator board had 3 separate electrolytic failures which in turn had taken out most of the regulator transistor circuitry, and 12 of the 24 penthouse modules had PSU decoupling and signal-coupling faults. After scoping the channels in "Input" I found severe LF rolloff and even lobsided waveforms as the electronically-balanced output card had faulty output caps.

A full recapping was needed for a quiet life!

Okay, so there are several possible mechanisms for electrolytic failure-

1. Change of value- normally this value drops.
2. Electrical failure- Open circuit (if you're lucky)
Short circuit (if you're unlucky :sad: )

...and there are essentially only 3 uses for an electrolytic in a piece of equipment:

1. Power supply smoothing (large value after the rectifier- 470u and upwards!)

2. Power supply decoupling (smaller values anywhere from 1uF to 2200uF) These are usually combined with a small-value series resistor to act as a filter for the following circuit. The capacitor acts as a small local "tank" of charge that can supply current during a peak-demand moment when the PSU rail may "sag" momentarily.

3. Audio Signal Pass. These are your input, output and interstage DC-blocking capacitors. Don't forget that this also includes emitter-bypass capacitors- these allow the transistor to be biased by the emitter resistor for DC, but allow a low-impedance path for the AC signal, which increases the gain of the stage.

It's obvious for 1 and 2 that any faults in the capacitor are just plain bad and will cause headaches/problems/disaster. Shorts are obviously bad. Opens not as disasterous but- an open implies that the capacitor is no longer "seen" by the circuit- so any decoupling or smoothing is no longer being carried out, so ripple may be introduced on the supply lines and decoupling from stage to stage may be compromised. That's why I always replace old and seepy-looking capacitors in the PSU area/local decoupling.

Slightly more bizarre are the effects in 3. When a signal-coupling capacitor fails, it can do all sorts of bizarre and not always immediately noticeable effects. If the value of a coupling capacitor falls, then the circuit will have its LF response reduced. As this usually happens over years, most owners probably don't notice it. Also, depending on the output level of the equipment (the owner may find that to get a reasonable bass output level they have to wind up the levels) they may feel that the equipment has more "presence". A bass rolloff can be seen as a broad mid and upper boost....

...so you recap it, and they say "where's that great presence gone!?!"...and you explain that no, actually it's better because all of the bottom end is there now... :?

Short circuits are another strange thing to deal with because a leaky signal cap can throw the bias of the next stage off. Sometimes when this happens you notice it instantly- basically you're getting squarewaves out with sines in- the bias is so off that it's a Fuzz Face. Other times it's more subtle- a slight shift can give an asymmetric clipping (especially in discrete circuits) which can actually sound quite nice...the joys of Class A circuitry

I think when replacing a cap you can't just fire new ones in, but look at what is happening. Has this capacitor caused a session-killing fault? If so, does the same cap on other boards (in a desk etc with multiple boards) look of a similar age? When this cap is replaced with a new type (in the signal path), does it A/B well with a similar untouched channel? If you look at the signal flow within the equipment and see where the suspect cap is, could it possibly be doing anything good to the sound? And if yes- then why?

My personal system is to replace PSU caps with the same value or greater- larger caps are much cheaper now compared to when a lot of this gear was designed, and available in smaller packages. I also go for higher temp rating inside packed equipment. For signal stages I check the actual value on a bridge, and check for leakage etc. If it's all about "the sound" then I replace it with a measured value, if not I try and get a good lower-point by measuring and scoping. It depends on the equipment.

So I don't thing there's any "magic" or "mojo"- just certain things that can be explained, woked out and even replicated that makes everyone happy, and extends the life of the equipment!

:thumb:

Mark

 

[tardishead]

completely agree with you but all electros sound different. I replaced caps in old Helios and Neve channels (different brands in each unit) and they came out sounding different to the next one.
the problem is sometimes a good old cap sounds better than anything I have found to replace it - in particular Neumann mics. unless the cap fails there is no real point in replacing it. I am sure the designers were particular about their choice of caps

 

[Mark Burnley]

[quote author="tardishead"] unless the cap fails there is no real point in replacing it. [/quote]

Bingo!

That's exactly my feeling- people have asked me to recap stuff and I've just said "No!" because I knew that it couldn't be any better!

:grin:

Mark

 

[Steve Jones]

Great post Mark, that's the first time anyone has put a really convincing explaination to me as to why the "shotgun" technique may not be appropriate. Definately PSU filter caps take a pounding, I am sometimes wary of increasing the value though, as the turn on surge caused by charging a couple of big caps might cause the rectifiers to have a thrombo if they aren't rated high enough or are getting a bit tired.

I guess a good ESR meter and capacitance meter are essential in deciding wheter to "leave alone" or to replace, but on balance, with the litany of possible faults you have detailed, both obvious and subtle, and not knowing the history of the piece of equipment and possible protracted periods of storage time, it really becomes a difficult decision, particularly as my customers come back to me after re-capping their synthesiser filters and commenting on how they had forgotten how big their machine sounded new until now that it has it's "bottom end" back.

I will definately ponder what you have said though, what you say makes a lot of sense.

(Also, a possible "fourth use" of electro's, is in timing circuits such as are found in envelope generators and low frequency oscillators.)

 

[Steve Jones]

Hey Mark, well I am having trouble going to sleep now after pondering your post some more. I guess that it would seem that from what you say there could be little harm in replacing PSU filter caps and power decoupling caps as a matter of course, but audio interstage and emitter bypass caps should be treated on a case by case basis as these can clearly effect the sonic characteristics of the unit?

I guess where my head has always been is that I see electros as something that degrade over time and do have a limited service life. But because I am not a designer I have not logged a lot of listening time comparing the sound of different brands of caps, so I can see how loading a unit with different new caps with different ESR's may make a big difference in some circuit stages. I guess this is a trade off against the fact that an old cap is also effectively a different cap as well compared to the one the designer had on his test bench 25 years ago....

 

[Gus]

When I work on guitar amps I test all the caps with a LC102. Some of the old used caps test better than some of the new stuff. I noted the lc102 because it can check leakage up to 999VDC. I also checks ESR.

The older electros at 450VDC have under 1.5 ohms of ESR per 20uf. There are new ones like this :however I have tested some new electros rated at 500V and 22uf 4 to 9 ohms of ESR and note one measured 22uf. Atoms still seem to be well made.

ESR affects the "ghost notes" and "punch" in a guitar amp IMO.

 

[Mark Burnley]

Hi Steve,

Yes, I forgot about the 4th case- timing elements (which are often electrolytic or tantalum) are a case on their own- obviously they effect the sound in an indirect way- although no audio signal passes through, their capacity and charge-leakage will affect the time constants of compressor and limiter sidechains etc. Good point! These have to be assessed individually to work out how they interacted with the timing circuit.

It's an important point too about not just throwing so-many-thousand-uF of capacitor into a circuit- bigger is not always better as far as post-rectifier capacitors are concerned. And even increasing all the local-level decoupling caps will add to the total inrush current.

There's that really interesting article by Eddie Ciletti that shows the square wave testing method, where typical output waveforms are analysed to see what's happening:

Eddie Ciletti Squarewave Testing

Check a few words by Peter Florance 1/2 way down this page too:

http://mixonline.com/education/articles/audio_geeks_speak/

...but I think in general with a PSU-area capacitor you just use sense and find one that is rated for the performance as far as capacitance/voltage/temp ratings are concerned.

Capacitors in the audio path need a bit more care and attention- and with foil and film capacitors in older EQ's I just leave them unless they have failed! They mostly seem to have very good lifespans because most EQ stages have very low voltage and current stresses on them.

Steve, don't lose too much sleep over capacitors- you'll have visions of giant 22000uF capacitors running after you in your dreams...not good!

:wink:

Mark

 

[Steve Jones]

great articles, Thanks Mark. (The part on tape machine alignment was good reading too). Just another point raised by the second article - Is ther really an ideal voltage rating for an electro? I repaired a Korg SDD 3000 delay last week that had 18 Volt electro's throughout the +/- 15 Volt power supply. I was very surprised that Korg would use caps that are so obviously underrrated for the application. Physical size limitations not withstanding, is there any difference or possible problems with putting a much higher voltage rated cap in place of a lower? For instance, I replaced the caps with 25V parts, but could just as easily used 63V caps or higher. Are ther any particular negative issues that anyone knows with high voltage caps in low voltage applications?

 

[owel]

I know higher voltages of the same microfarad rating is physically bigger, and likely also more expensive.

In your example, 18V electros at 15V supply, that's pretty close for comfort for me. I'd go with a 25V at least, or 35V, if not 50V.

 

[Gus]

I tend to like some caps close to voltages they will see if I want clean. I think this might be because the dielectric is thicker for higher voltages. This seems to be most apparent with electros and PET caps. I believe it affects the DA the more thickness of the dielectric the more materal for the electrons to get "lost" in. I sometime will use much higher voltage rated caps if I want to add DA "warmth"

This is what I think I hear. I know one can fool ones self.

I hope some people do there own experiments and give feedback on this.

 

[CJ]

It is hard to put a blanket statement on this subject like "always change lytics that are 15 years old" or something like that.
You have to analyze this on an individual basis.
Where is the cap located?
What would happen if it failed, would it take out an expensive and irreplaceable vintage transformer?
Will I do more harm than good if I replace a cap in the signal chain?
What does this cap look like?
Is their fluid coming out the vent cap?
Does it feel lighter than usual?
How long was this unit in service?
Did it run 24/7 or just once a year?
How hot does it get in there?
etc etc.
Most of the time, the lytics I change are in equipment than runs all day, handles a lot of ripple current, and is in a hot enviroment.
The rest of the time, I just leave it if it sounds ok and the unit is not overheating.

These NOS Siemans 100uf/25 volt that I just installed on this Valley People board must be at least forty or 50 years old. But when I checked them with the meter, one read 129 uf and the other 150 uf!
The leakage was high but was dropping, as expected when a cap is fired up for the first time in half a century.
cj

 

[soundguy]

As usual, I have little technical to add, but an observation on this subject...

I think the hip to recap thing came from guys that refurbished consoles and were vocal about it on the internet. Whenever I sell an old piece of junk, someone always asks "is it recapped" and I ask them what that means and they dont know. so, its a very chic thing.

If you need duplicate channels of aging junk to match, I think there is some value to recapping, only so you have a baseline that is common between all the modules in a console to work from. I liked the sound of my old console when I got it, it all sounded fine, but the channels didnt match. They didnt sound bad on their own, but they didnt match. I recapped the whole console, and yes, the sound changed, but at least I now dont have to remember that channels 3, 14, 9 and 27 are the good sounding ones with the voodoo... If I parted the console and racked all the EQs, I never would have bothered as they all sounded good and in 20 different homes, they didnt need to match. So I think a lot of the hoopla about recapping vintage gear in that way at least is really taken out of context. Ive recapped stuff that sounded broken and was very happy. Ive recapped stuff that had no problems and I didnt like the way the sound changed, so I put the old caps back in where they will happily stay until they start to sound broken at which point I'll change them. the only exception to this is when I did the console as now Im pretty happy that ALL the channels are doing the same thing, which is the big goal of a console, right? One or two racked pieces from that console, however, dont need to adhere to that standard, at least not for my dollar. My next thing is all the record and playback cards on my tape machine as I definitely have voodoo channels now and I dont like it. Im trying to put this off for as long as possible, but it really needs to be done. As much as I dont want to reinevent the sound of my deck, Im sick of trying to remember which three tracks work for vocals and which sound better for guitar... blech.

dave