Improving the Coupling Capacitors for Better Tone Sounding

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
It is a fact that capacitors with a large varying voltage across them will produce distortion. Steve Bench used 70V rms in his test which is about +60dBu!!!

Steve Bench's tests were conducted with 0.1uF capacitors. When a properly sized capacitor is used as a coupling between circuits, the voltage across it is miniscule. A typical 47uF coupling capacitor in a 10K circuit will have a voltage across it easily more than 120dB lower than this.

On the other hand, in an EQ, there are significant voltages across capacitors so non-linearities are more likely to be detectable.

Cheers

Ian
 
What I enjoyed the most in this video, was this claim:

"Anybody who really know what they're doing in electronics will tell you that that that the coupling capacitors definitely are one of the more audible parts of any circuit"


I would love to know the opinion of people "that really know what they're doing in electronics" about this statement,
people that are very experienced and we can trust, even if just a few.

This is a complicated answer because it does not allow for a statement of magnitude, but just relativity, and in that sense the statement is correct simply because coupling caps are in the signal path, while almost all other caps are not (a few exceptions). Not only that, they are the biggest caps in the signal path (capacatively coupled outputs excepted, but those are not used in guitar amps that I have ever seen). But how much one actually hears caps, even when in the signal path is another matter.

Where he is wrong is that paper in oil caps are a good idea. We have already explained why they are not above. Like I've passed on guitars because the seller advertised that the tone caps were changed to vintage russian PiO (a right mess waiting to happen inside an expensive instrument that could damage other components or the finish, and since already old vs new prod, probably altready started that slow process; no thank you.

I do not believe (except with some limited exceptions re noise and other factors) that different caps types make any tonal difference. At the very best, this difference would be a point of diminishing returns. The only reason to change a cap is due to failure, needing/wanting a different value, or because the current cap type is suboptimal (old paper in oil, old electrolytic either of which may fail, and have likely already drifted significantly in value, or renmoving electrolytics from the signal path, and cheap ceramic discs, etc - tantalums can have a lifespam, just generaslly much longer than electros; noise issues w silver mica, but eg Fender nerds love them in the tone stack).
 
It is a fact that capacitors with a large varying voltage across them will produce distortion. Steve Bench used 70V rms in his test which is about +60dBu!!!

Steve Bench's tests were conducted with 0.1uF capacitors. When a properly sized capacitor is used as a coupling between circuits, the voltage across it is miniscule. A typical 47uF coupling capacitor in a 10K circuit will have a voltage across it easily more than 120dB lower than this.

On the other hand, in an EQ, there are significant voltages across capacitors so non-linearities are more likely to be detectable.

Cheers

Ian
Dear Ian,
Steve is talking about the DC biasing can change the performance curve D-E in caps too. Here we can linearize this curve of not! and this is a posible in ours units.

Best regards
opacheco
 

Attachments

  • IMG_7707.jpeg
    IMG_7707.jpeg
    169.8 KB
  • IMG_7708.jpeg
    IMG_7708.jpeg
    88.6 KB
CalavoBob,
I would like to do some measurements and see what I actually get between different capacitors and their sonic changes (if any!), but I have some doubts with your schematic:
1- What do you connect in the Rload/Rload terminals?
2- Do you have any instructions in order to use this DA tester?

Thanks in advence
opacheco
Hi Opacheco and others interested in comparing science with what we hear.
1. The Rload/Rload terminals are used to connect a resistor that the Radj2 pot uses to adjust for the cap's ESR (equivalent series resistance). I have been using a 1 kohm 1% metal film resistor there, and it seems to allow me to set the pot at a useful position in it's rotation. This pot doesn't really change things much when I adjust it though.

2. Keep in mind that this test does not produce a numeric value for DA. It does however allow you to compare one cap to another so in generic tests, you need to have a really good reference cap. I went onto Mouser and collected a range of capacitor values from the more trusted manufacturers like Wima, F&T, Solen, etc. choosing what are recognized as the better dielectric types (teflon and metalized polypropylene) and looked for voltage ratings that will work in tube amplifiers (200V to 630V). You'll find very few manufacurers actually list the DA.
The oscillator source needs to be a square wave that allows you to alter the duty cycle. The article suggests a level of 1Vpp to 10Vpp. I've been using about 2Vpp to 4Vpp. The article also recommends setting the wave's pulse width (tw) between 1ms to 20ms, and the wave's period (T) to about 5x to 10x the pulse width. I've been using a pulse width of 10 ms and a period of 35 ms. This is a little shorter of a period than it suggests, but I got pretty good results. A longer period gives more time for the residue to settle.
I Y-branch the signal generator to feed the tester, and also channel 1 of the o'scope. The output of the tester is sent to channel 2 of the o'scope. Notice that the bandwidth limiting capacitor C1 causes a slight rolloff of the leading edge of the source pulse. This keeps the artifacts below 100 kHz.
In the tests that I have done, I have set the gain switch to 100. I use a +/- 15V bench power supply.
I put the high quality reference cap on the Cref terminals, and then the cap to compare to this reference is attached to the Ctest terminals. As seen in my posted pictures, I use banana posts for all the connections except the 2 signal jacks which are bnc connectors,

My oscilloscope is set with channel 1 at 500mV/div, and the time base at 5ms/div. Trigger to channel 1. Channel 2's vertical sensitivity will vary as you adjust the trim pots. Try starting with about 2V or 5V/div on channel 2, and adjust the coarse pot to get channel 2's signal to drop. Then use the mid and finally the fine adjust pot. Now you can try adjusting the Radj2 pot to see if it makes any diference. You'll need to increase the sensitiity of channel 2 as you get it trimmed to it's minimum ampitude.

So far, the best I've seen are the Solen polypropylene caps and the worst I've seen are very old large Sprague caps pulled out of old guitar amps.

Despite this lengthy description of how to use this box, the tests go very quickly, and you can go through a stack of caps in very little time. I've got quite a collection of old Fender tweeds that I want to do some listening tests with, to compare the very good DA caps with the very crappy ones to see if I can hear a difference. One thing I should add though, is that I'm an active blues guitar player and I have become fond of using preamp tubes that are just a little microphonic because it give the amp a little resonance that I find pretty musical. To me, tubes that have no microphonics at all make a guitar amp sound a little dull. Of course tubes that have rampant microphonics I find unuseable. There is a nice in between spot though that really sounds great.
 
Where he is wrong is that paper in oil caps are a good idea. We have already explained why they are not above. Like I've passed on guitars because the seller advertised that the tone caps were changed to vintage russian PiO (a right mess waiting to happen inside an expensive instrument that could damage other components or the finish, and since already old vs new prod, probably altready started that slow process; no thank you.

Yes. Just from a 'mechanical' point of view. It's OIL - sooner or later it's going to go somewhere it shouldn't.
 
When building valve circuits, like guitar amps and compressors, I generally used the Cornell Dubilier WMF Series axial polyester film/foil capacitors. Traditional I suppose, very well made, reliable, useful valve circuit values, and the tolerance is good.

In respect of PIO caps, I bought some years ago (along with teflon caps, polystyrene and others) for timing circuits and eqs, when they were dirt cheap. In 20yrs I have never seen one of those sealed PIOs leak. Values for those I purchased were generally pretty tight to spec. Size is a big issue with these caps.

I have heard of problems with wax melting in wax/paper caps.

Lately I have just been trying to use up whatever I have at hand because I want to save on costs.

A few years ago I was working on restoring an old B67 and had it open and witnessed one of the old tantalum caps go south. Quite spectacular (and PCB scorching).
 
In 20yrs I have never seen one of those sealed PIOs leak.
I have. NOS ones are sometimes even all greasy to the touch on the outside when you get them; I've also seen little puddles in the bins containing them at the local electronics surplus store. There may be some variance in this regard in terms of quality. To me it's not worth the risk, when better options exist.

A few years ago I was working on restoring an old B67 and had it open and witnessed one of the old tantalum caps go south. Quite spectacular (and PCB scorching).
I've never seen a tant fail. Heard about it. They are supposed to last much longer than electrolytics (and I have seen quality electrolytics from the 60s still in service and fine). I've used them in builds but not recently as I have other options in stock for those values. What happens; does it damage things? Electros are mostly just messy.

What was fun was testing out one of my vintage Mac SE/30s (took them home years ago when a former employer was getting rid of them - loaded to the tits with all the RAM they could take and running OS 7.1 which was insane) to see if any of them were dead to cannabalise it for the case to make a fish tank for my daughter. The first one powered on but then the screen went all weird so that's the one I picked. Taking the guts out, one electrolytic on the mobo doubled in size and burst (yellowish and foamy) and the clock battery also leaked. The battery was far worse - corroded component leads so the battery holder and some adjacent parts fell off the board. I'd post a pic but I'm not at home. RAM is still good though.

Incidentally the other SE/30s are fine and worked when I powered them on; time to sell them before they go too.
 
Last edited:
Almost anything can be destroyed if it faces enough strain

The hermetically sealed metal PIO caps with the glass seals tend to be quite robust and I’ve never seen one leak its oil

However I’m sure you could make one leak if subjected to enough force

With surplus or used parts it’s hard to know what stocks may have been subjected to before you encountered them
 
'Twould not surprise me at all if tests showed that pretty much all the mojo of a C37A is in the capsule.

The tube circuit, being a cathode follower, ain't mojoin', and the output transformer is also 'idling', and of very high quality, so the sound of that mic is it's unique capsule, modified a bit by it's rather less-than-transparent head basket.

Though collectors would have your head for it, I've always been curious how different it would sound with a more open mesh basket (like Royer did with the MA-37).
 
Last edited:
I've never seen a tant fail. Heard about it. They are supposed to last much longer than electrolytics (and I have seen quality electrolytics from the 60s still in service and fine). I've used them in builds but not recently as I have other options in stock for those values. What happens; does it damage things?
I've definitely seen tantalum caps fail. Probably because their voltage was too close to the power supply rails. They got very hot, and some even caught fire. Several times I've seen pc boards burned up around tants that blew. I'm not very fond of them, but they're used all over the place in the beloved Neve 1073 preamps. I haven't seen them fail there though.
 
I've definitely seen tantalum caps fail. Probably because their voltage was too close to the power supply rails. They got very hot, and some even caught fire. Several times I've seen pc boards burned up around tants that blew. I'm not very fond of them, but they're used all over the place in the beloved Neve 1073 preamps. I haven't seen them fail there though.
Worst thing about them is they usually fail as a short; if on supply rails, damage to the PS is almost certain.
 
replaced a ton of shorted tants, usually because they got stuffed backwards,

just replaced a Mica that read 546 ohms, went from the plate to cathode of a reverb tube on a vox ac 15.

got some Vitamin Q's, slight oil film on them but no PCB's.

Russian oil caps - have yet to see a leak,

do power supply caps influence sound? some folks think so,
 
The tantalum that I saw fail turned into a little fiery ball of coal and let out a lot of magic smoke. Not quite as dramatic as the time I saw (and accidentally made) a heater tab on a EL34 socket vaporise (emission of blue light as the metal of the tab was consumed), but unpleasant enough.

The PCBs were in slots at the front of the B67, and the glowing tantaulm scorched the underside of the adjacent PCB as well as the surface of the PCB it was on. I went about replacing all of the tantalum capacitors after that.
 
I worked as a production failure analysis engineer at a company that had gotten reports of strange performance in the field. Customer had sent quite a few back and sometimes I could replicate the odd video stuttering, but wasn't sure why. Two of the boards had an area about 3 inches in diameter in the power supply that was badly scorched such that I had no way to tell what had caused it, until one morning one of the returned boards blew up right before my eyes, and I could see that it was a tantalum that got really hot and smoked. Now that I knew the culprit, I examined a pristine board to discover that where there was supposed to be a 12V tantalum before a 5V regulator and a 6V tantalum right after it, they had been installed in reverse! The 6V cap was on the high voltage side of the regulator. Some further investigation around the company revealed that the board had been redesigned about a month before such that the positions of the 2 caps were interchanged intentionally, but the engineering change notice (ECN) had been sent to everyone except the folks that were stuffing the boards. Whoops! I guess the lesson is make sure everyone in the company gets the ECNs. Of course that was an incorrect application of those poor tantalums, but instead of just opening up they shorted with disastrous results. We had to recall them all.
 
I have a set of old MXR units on the bench right now with at least one blown tant. Find them dead all the time in 70's and 80's synths and digital devices.

Keep em in guitar pedals, timing circuits (maybe) and maybe U87's if you must, but outside that I've got no use for em.
 
do power supply caps influence sound? some folks think so,
There are mechanism of action for it. Regulators frequently used for audio don't like HF peaks in the impedance curve of the decoupling arrangement, op amps can become (marginally) unstable. Even just adding decoupling can lower ESR and increase stiffness to the point of instability: E.g., a single X7R cap has sufficient ESR to prevent resonances, but many of them in parallel on every op amp's PSU legs can result in significanty lower ESR than is good for the arrangement.

What I have learned is to use (for audio circuitry PSU decoupling) a) the worst (high ESR) electrolytics, b) no film caps or NP0, instead X7R ceramics, c) as little decoupling as possible for the usual suspects (5532/5534 etc.), d) add a little resistance if you need many decoupling caps in parallel.
 
Last edited:
Tants are common in Studer tape machines, and ATRs use them all over the place! Seen several failures the worst is when they explode and leave black burnt tant on the pcb. Several fail short and take out a regulator, easy to find, albeit more destructive. I like them for their longevity so they stay in for signal. Get replaced when taking on more strenuous duties. I'd never put them in a PSU for any reason.
 
I've only seen larger PSU sized oil caps leak with age, but still a small percentage looking at lots of 70-80 year old gear with them. I've never seen an installed hermetically sealed oil signal cap with leakage unless it was truly mechanically abused, also lots of them here in 70 year old gear. Any that are surplus tossed around in bins have certainly been dropped and mishandled, probably have dents and dings weakening the structure. They definitely have a euphonic sound. What they do exhibit depending on length of usage is higher to extremely high ESR, certainly part of the sound. I've seen some that were in 24/7 usage for 50 years in radio broadcast that had ESR so high there was treble loss, though they still measured dead on for C and showed no DC leakage across them. Sometimes some of that is the euphonic sound you might like in a piece of old gear - it's not about accuracy. Sometimes the piece is already so colored from the transformers that a new cap is desirable.

I had a Collins 212A once that had all the oil caps (they are all oil except cathode bypasses, PSU B+ also (LCLC), ripped out in a hazmat panic. Restored it (new coupling caps) and several other 212A's that had not had the oil caps removed. Never had to replace an oil cap that was still there, and the SE preamps sounded obviously different, console with missing oil caps versus several other consoles with orig oil (which all sounded pretty much the same) in isolated form on a Heathkit bench supply. The new caps made for far less 'color'. No one is buying and using these consoles for a lack of color!

Plenty of other ancient tube and early SS gear I've restored lots of sounds better with new caps. Age and hours used plays a part in all cases, but the oil cap gear has something different going on that can be desirable, whereas everything else has been a pretty clear improvement with new caps. The color in everything else tends to be a 'cloudy/grainy/spitty' sort of distortion/harmonic overlay that's best removed in favor of clarity.

So there's probably some hi-fi words to enrage......don't have any better'uns.
 
Last edited:
Back
Top