Recapping amps: facts, myths, legends etc

[rontoolsie]

I had a CDS3 unit modded recently, and the four 47uF Promisic gold colored caps were upgraded with something better. But none of them measured more than 25uF!....so they were all about 50% out of spec. Performance was markedly improved after their replacement-but I'm not sure if that was because correct value caps were used, of if it was a sonic signature of the replacement caps.

 

[James]

Precis of things I've posted before here several times...What absolutely kills electrolytic capacitors (such a main PSU caps) is temperature. This is the caps' internal temperature, which depends on ripple-current far more than just the temperature inside the amplifier case. And with large/low-impedance transformers, the ripple current can be 5-6 times the idle DC current drawn by the amplifer: dissipating a watt or so inside reservoir caps is entirely possible, even in low-bias amplifiers. It all adds-up

A typical modern electrolytic cap will have it's lifetime specced by the manufacturer at something like 2000hrs at 105degreesC. On paper just 3months at full load and temperature!

However the actual useful lifespan - usually specced by capacitor manufacturers as 'to the point where rated Equivalent Series Resistance' doubles - essentially follows the Arrhenius equation in being related to running temperature: with lifespan doubling for each 10degreeC drop below rated temp.

So at 40degC that same cap should last 64*2000 or about 14years in continuous use. Try that at 45degC and perhaps one can see why Naim's rule of thumb of replacing main PSU caps at 7-10 years is about right - it really is about the useful lifespan assuming for full ripple current, 24hrs a day, at around 45degC internal temp. This is basic physics - not some Salisbury scam. 7-10 year replacements for a low-bias amp run 24/7 is just prudent.

Stick that same cap inside, say, a Class A amp exposed to heavy ripple (due to the constant high idle bias) and so running at 60degC and it will only last , on average, 3years or so - but you'd never leave such a thing on24/7, maybe 4hrs a day of use and suddenly the caps again might well last 10-12years or more. Equally, a low-bias amp used 4hrs a day might well be fine at 20-30years plus, if the cap seals have held and not lost electrolyte. Use a newer cap rated 3000hrs at 105egC instead and the goal moves back nearly 50% pro-rata; cheap or free when you are replacing the parts anyway.

(For my own amps I started with reservoir caps rated 60000hrs at 105degC - and film for everything else; so >40yrs left, really...even if left on 24/7.)

A couple of curious observations, if I may.

I bought a used CB NAIT2 from a PFMer (or was a Naim forum member, I can't remember) eons ago and asked for it to be sent to Salisbury for a recap before it was shipped to me. The worksheet/invoice stated the tants and a few minor bits were replaced, but the main reservoir caps remained original. I wonder why these weren't replaced when the bigger amps tend to get those caps refreshed.

Naim amplifiers, in closed boxes, arguably run more warmly than a ventilated one. Case in point, my DV HX-1.2mk2 idles stone cold, and barely emits any warmth from the vents or fins unless I'm really cranking up the volume, which I hardly do these days. Given my room temperature, on most days, don't rise much above 20 degrees Celsius, should their caps last 4x longer than Naim's do at 40 degrees? It's just that I have never heard about Dynavector amps needing a recap, and despite being manufactured in my neck of the woods, I have no idea who the service agent is.

Maybe I should turn the DV off when not listening.

 

[davidsrsb]

Bigger capacitors have longer rated lifetimes, the miniature electrolytics only a few mm diameter are the worst.
ESR always goes up before loss of capacitance. By the time a capacitors value has reduced out of spec, the ESR will have gone really high

 

[Snufkin]

A couple of curious observations, if I may.

I bought a used CB NAIT2 from a PFMer (or was a Naim forum member, I can't remember) eons ago and asked for it to be sent to Salisbury for a recap before it was shipped to me. The worksheet/invoice stated the tants and a few minor bits were replaced, but the main reservoir caps remained original. I wonder why these weren't replaced when the bigger amps tend to get those caps refreshed.

Naim amplifiers, in closed boxes, arguably run more warmly than a ventilated one. Case in point, my DV HX-1.2mk2 idles stone cold, and barely emits any warmth from the vents or fins unless I'm really cranking up the volume, which I hardly do these days. Given my room temperature, on most days, don't rise much above 20 degrees Celsius, should their caps last 4x longer than Naim's do at 40 degrees? It's just that I have never heard about Dynavector amps needing a recap, and despite being manufactured in my neck of the woods, I have no idea who the service agent is.

Maybe I should turn the DV off when not listening.

From memory, this came up in a conversation with Pear Audio re the HX1.2 mk2, they said that recapping was unlikely to be needed for a very long time, if at all. They thought the mk1 may need it at the 15+ year mark. I had an HX1.2 mk2 for a while and it ran slightly warm (tepid) but never hot. It was replaced by an HX100 which did run quite warm and had an audible hum until it had its voltage reset from 230 to 240. It is now quieter and cooler, thankfully.

 

[James]

Class A amplifiers often have a fairly large bridge conduction angle, so the rms ripple current is not as large as you might expect...

Bigger capacitors have longer rated lifetimes, the miniature electrolytics only a few mm diameter are the worst...

This quad of 33,000uF cans should last almost forever then.

 

[Mike P]

I've recapped fours amps myself:

Leak Stereo 70 (1960's) - A 'no brainer' really given its age. I did this for reliability as much as anything.

Cambridge Audio A1 mk3 - Bought of a car boot sale for £4! No improvement after the recap.

Linn Klout - Bought of ebay with unknown history. Noticeable improvement in SQ after recap.

Sony TA-F770ES - Purchased from a fellow PFM member. Noticeable improvement in SQ after recap.

 

[Robert]

Precis of things I've posted before here several times...What absolutely kills electrolytic capacitors (such a main PSU caps) is temperature. This is the caps' internal temperature, which depends on ripple-current far more than just the temperature inside the amplifier case. And with large/low-impedance transformers, the ripple current can be 5-6 times the idle DC current drawn by the amplifer: dissipating a watt or so inside reservoir caps is entirely possible, even in low-bias amplifiers. It all adds-up

A typical modern electrolytic cap will have it's lifetime specced by the manufacturer at something like 2000hrs at 105degreesC. On paper just 3months at full load and temperature!

However the actual useful lifespan - usually specced by capacitor manufacturers as 'to the point where rated Equivalent Series Resistance' doubles - essentially follows the Arrhenius equation in being related to running temperature: with lifespan doubling for each 10degreeC drop below rated temp.

So at 40degC that same cap should last 64*2000 or about 14years in continuous use. Try that at 45degC and perhaps one can see why Naim's rule of thumb of replacing main PSU caps at 7-10 years is about right - it really is about the useful lifespan assuming for full ripple current, 24hrs a day, at around 45degC internal temp. This is basic physics - not some Salisbury scam. 7-10 year replacements for a low-bias amp run 24/7 is just prudent.

Stick that same cap inside, say, a Class A amp exposed to heavy ripple (due to the constant high idle bias) and so running at 60degC and it will only last , on average, 3years or so - but you'd never leave such a thing on24/7, maybe 4hrs a day of use and suddenly the caps again might well last 10-12years or more. Equally, a low-bias amp used 4hrs a day might well be fine at 20-30years plus, if the cap seals have held and not lost electrolyte. Use a newer cap rated 3000hrs at 105egC instead and the goal moves back nearly 50% pro-rata; cheap or free when you are replacing the parts anyway.

(For my own amps I started with reservoir caps rated 60000hrs at 105degC - and film for everything else; so >40yrs left, really...even if left on 24/7.)

Agree with this.

The only point I’d add is on heat within the case.
Some manufacturers have a nasty habit of placing small psu caps next to hot running regulators, sinks and voltage dropper resistors, causing them to bake.
Sometimes can’t be helped as you often want the cap near the reg for stability, but it shortens life.

 

[Rallye_punk]

Good thread gents.

I have some questions. I have a pair of Nap 135's that I think could benefit from a recap. I know there are kits available including the 2 x big resi caps and some tant caps / resistors etc.

Is it worth just changing the big resi caps, as from reading through the thread these seem to be of prime importance?

Are the resi caps are more prone to 'wear and tear' than the tant caps?

Will the SQ be affected by poorly perfomrming resi caps rather than tants?

Thanks!

 

[peterm]

The caps that fail most often in NAP 135s & 250s are the small electrolytics on the power supply regulator boards, these can be replaced quite cheaply. The big reservoir caps may or may not need replacing. If I was doing it myself I'd try just the reg caps first (and maybe the reistors that get cooked - you can tell by the discolouration!).

 

[martin clark]

Spot-on peterm, those blue 10uF/50v axial caps prove critical to the regulators stability. Fortunately very cheap and easy to do, and suitable axial electrolytics are still available from Vishay.

 

[rontoolsie]

My first foray into the tube amp world was to replace a pair of NAP135 with an Audio Research D70mk2. Which had its large filter caps in very close proximity to the very, VERY hot 6550 tubes. Needless to say less than 18 months later, one of them failed, leaking caustic slurry over the main PCB which had to be replaced at a cost of almost $600-this was 1986 dollars too. Even caps rated at 80C in this environment are on death row.

One reason why Naim seems to be unusually 'sensitive' to requiring periodic recapping is that their products are never tossed in a dumpster when upgraded/replaced-they find their way back into active use, and after 40 odd years many of the older models are way out of spec-I don't think even the most optimistic projections in the 1970s would have foreseen how many of their units in working order are scattered around the globe and still in daily use in the following century.

How many fine Armstrong 621 sold in the 70s are still out there in the field today-compared to 12/120 pairings? I don't think I've ever seen ONE.

 

[Tony L]

By saying that the world is full of working Quad 303s that predate anything Naim! Most will be on their second or third set of PSU and coupling caps though. A wonderfully reliable and serviceable amp.

 

[Rallye_punk]

I've had 30 plus year old Naim amps that have never been touched and sounded incredible. Then others with issues. I guess a lot of it is pot luck too.

 

[hifienthusiast]

I had an Onix OA21S identical to Tony's about 10 years ago. It sounded excellent and I found no sign of cap leakage. I wish I never sold it. I have kept the Onix BWD1 tuner.

Now I have OA22 which is less powerful but again there is no sign of cap leakage.

 

[Rallye_punk]

The caps that fail most often in NAP 135s & 250s are the small electrolytics on the power supply regulator boards, these can be replaced quite cheaply. The big reservoir caps may or may not need replacing. If I was doing it myself I'd try just the reg caps first (and maybe the reistors that get cooked - you can tell by the discolouration!).

Ok thanks. Where can you get the precise resistors? Ideally i'd want the same type but I guess this may not be possible nowadays!

Spot-on peterm, those blue 10uF/50v axial caps prove critical to the regulators stability. Fortunately very cheap and easy to do, and suitable axial electrolytics are still available from Vishay.

As above, thanks!

 

[peterm]

I got a complete service kit containing all that was needed (plus more!), with new rubber feet, fastening screws, heat sink paste etc from LesW. He could no doubt advise on suitable replacements, in many cases superior to those originally used by Naim, and his cd of instructions makes everything clear, including how to set the bias current which was strangely below spec on a NAP180 I got recently (see my post on the audio forum)!

 

[RustyB]

I think the Audioab stuff was underrated which will cause overheating and leaking.


True. The 8000P has 63V smoothing caps, and the rail voltage is...63!

 

[sq225917]

I assume that's new Audioflat stuff, not the none iag original product.

 

[RustyB]

No, original CST built 8000P.

 

[peterm]

Whenever I recap, I normally fit higher capacity or quality items than what was in there originally, as modern caps are much smaller than the ones available at time of manufacture. Tony - you should check the value and size of those BHCs - you will probably find higher quality and higher capacity modern equivalents in the same form factor nowadays - a lot of manufacturers were governed by the physical size of the cap to dictate the value.

The only thing I have like for like recapped was a recent Cambridge Audio CD3 - for the first time I replaced every electrolytic with same values, except for slightly increasing those used to decouple the 4x TDA1541As. I sounded great afterwards...truth be told it sounded great before

One other item - if you have sent your amp back to the factory to be recapped, note that SQ differences might not only be related the caps - I would fully expect a manufacturer to reset bias, and this can also have a major impact on SQ.

Richard

Interesting point about bias.
I posted a while ago on the audio forum about 2 NAP 180s I had, an old one from the first year of production and another from the final year. The older one sounded much the better of the two.
When I checked the bias the old one was on the high side of the normal range (36 - 38ma) but the newer amp was low at 30.3 and 30.8 ma.
I increased bias on both boards to 37.5 to match the older amp.
They now sound the same.
Don't know how the bias came to be so low!