SET Asylum

Single Ended Triodes (SETs), the ultimate tube lovers dream.

Return to SET Asylum


Message Sort: Post Order or Asylum Reverse Threaded

facing the loss

108.213.92.252

Posted on June 26, 2012 at 10:18:40
GSH
Audiophile

Posts: 261
Location: S. Cal
Joined: January 19, 2011
Experimenting one day with my direct coupled high current low mu triode driver to 2A3 amp, I tried a paralleled by-pass cap across my final C to the output stage. I put an old Westcap PIO 0.0022/600v cap across the 50uf motor run that was already there. I had assumed that the Japanese made oil filled motor run cap was a "decent" cap. To my horror, I instantly heard a large difference. Now the metallic nature of symbols and all their reflections, phasing etc., were much more "present" as though this band of frequencies was now "boosted" by 1 or 2 db, or were now not -1 or -2 db....I liked it for a while but realized it was out of balance. I realized then that there was more loss or less quality (frequency wise) in a PIO motor run than I had thought. I also realized that to bring balance back I had to either remove the 0.0022 OR add numerous more small by-pass caps to the 50uf + 0.0022uf combination.
(I removed the experimental cap, for now but...)

I don't know the science or math behind what was verified experimentally. I do know that paralleling caps is NOT A NEW IDEA.
Whatever losses were present before the experiment, were now selectively less, selective to the point that with patience, one could experiment
with numerous other small value "quality" caps across any other PS or cathode by-pass cap and "re-achieve" balance, but with "less" overall losses, which "equate" to the perception of improved sonics.

This is what Dennis has done, it's no secret. The values he has experimentally chosen, are indeed his proprietary secrets. His version of "re-achieved" balance is what he deems his best for now, effort to compensate for the inevitable losses. These loses INCLUDE the less than ideal relationship between his driver of choice and the Cmiller of a 2A3.
To what degree has this been done? Very hard to answer, UNLESS somebody has the MATH to explain what is happening in a multiple paralleled cap situation, which nobody to date has provided. IMHO, this compensation for losses, IS THE PRIMARY "feature" of the Serious Stereo product (an item I've never heard and don't plan/need to BTW), which also includes attention to wire type and length. Is it imaginary? No. Is it perfect? No. Can we understand why this approach was attempted? Yes. Can the science only crowd poopoo this? of course. Is the poopooing backed up with math? Not yet, only partially in areas where known math applies, the rest of the picture remains unexplained, period.

So we have what I'm calling hard evidence that we have a significantly lossy system EVEN WITH the simplest of circuit design, DC SE.
Some will not try to compensate for it, pretending it's not there to begin with, or without the math to comfort aching minds, won't.
Others will, however they go about it. All they can do is report subjective findings, period, regardless of their "wish" to be scientific.

 

Hide full thread outline!
    ...
bypassing caps, posted on June 26, 2012 at 10:38:10
Ralph
Manufacturer

Posts: 2503
Location: Minnesota
Joined: April 24, 2002
what you just discovered has been known for a long time, although not by everyone in the audio community! If you bypass a coupling cap, you will get a smeared result. If you don't hear it then your system has resolution problems. With coupling caps, you just have to give it your best shot.

The problem has to do with the specs of the cap- series resistance and dissipation factor being two of the bigger players. Smaller caps tend to be better in these regards and can pass the signal in a 'faster' (that's how our ears hear it...) way. It is something that is measurable as well as audible- 'slow' caps will have less bandwidth although you might have to go quite high to see where the differences are.

Paper and oil caps are actually not bad in this regard and is why they are still around in audio. Teflon caps are good in this way too (but tend to be less prone to electrical leakage issues; they can be easily damaged with minor handling). The next best is polystyrene, which as some audio frequencies *can* be better than Teflon...

 

RE: bypassing caps, posted on June 26, 2012 at 11:05:48
GSH
Audiophile

Posts: 261
Location: S. Cal
Joined: January 19, 2011
As stated, I only tried it on a PS cap, there is no "coupling" caps in a direct coupled SE amplifier, unless a cathode by-pass qualifies.
And yes, I see that caps have different speeds or bandwidth preferences, but this is NOT good! It means only very very high quality (if that exists) can "replace" the need to compensate with somehow "chosen" multiple small value paralleled caps. I hate it! But, I accept the challenge of at least trying to compensate for true losses. It is seemingly a necessary voodoo.

Can anyone recommend a cap that when paralleled with quality small caps,
doesn't sound ANY different? for example a 50uf/600VDC ? This to me would indicate that the speed/bandwidth performance of the main cap was "good enough" to not be enhanced by whatever "superior, less lossy" effects the additional caps could provide, along with whatever "smear" they are also adding?

 

RE: bypassing caps, posted on June 26, 2012 at 12:25:21
drlowmu
Manufacturer

Posts: 7989
Location: East of Kansas City
Joined: January 10, 2005
Hi GSH,

There is a difference between bypassing a power supply cap and a Rk cap.

Rk caps are somewhat more critical.

DynamiCaps allow for intelligent multiple film cap Rk bypassing, and according to Dennis Fraker, he finds they "add" very nicely.

A good solution / answer will be to use and trust your ears, to select bypass values and types. Use several differing values to get a linear music presentation.

There is no single Rk cap that will do all the frequencies at the speaker voice coil in a linear fashion. When you use a single cap on a Rk bypass, the music will unfortunately tune to the value of that one cap.

Cheers,

Jeff Medwin

 

RE: facing the over generalization, posted on June 27, 2012 at 12:47:15
You might have a very specific problem that should not be extrapolated to all caps. That last cap is, of course, in the signal path. Your problem may be more due to the specific motor run cap that you have used rather than oil caps in general. If your motor runs are old nos, been stored in temp extremes, etc. they will not perform well. Try a new ASC motor run; I have not found them to roll off the highs at all. I like Solens as well. Oh, and there is nothing wrong with putting in a nice 1 to 5 uf bypass cap on that last cap.

 

they're fairly fresh, posted on June 27, 2012 at 16:20:29
GSH
Audiophile

Posts: 261
Location: S. Cal
Joined: January 19, 2011
My final PS C is an ACS made in Japan, 50uf/370VAC, bought new 2 years ago, never tormented. I always thought they were good, and perhaps are better than a cheap electrolytic. I've used Solens too, reliable.

I say try putting any "good" small (0.05uf-0.001uf) cap across whatever final PS cap and/or cathode bypass cap you have, and tell me what you hear.
Then maybe somebody can talk about what happened: did some frequencies get "boosted" or are the same selected frequencies "faster" or ?

The smear issue has been mentioned numerous times, yet some recommend paralleling smaller better caps across larger caps. I have yet to hear smear. If smear is real then wouldn't multiple by passing (say 5 or 6 total) be disastrous? or possibly return to no-smear?

Anyway, something happens, and it "seems" to show that the ACS cap alone
is maybe 1 or 2 db down in a certain higher frequency zone. My guess is that the same thing I hear would be heard by most of you on whatever cap you have. And if you hear nothing, I want to know what cap you have please. Further, IF "returning to flat response" is what is happening by installing a selected group of small bypass caps, then it would seem then that most large (20uf-100uf) caps are not "flat". IF flatness is not the issue and speed is, then are we listening to different speeds across the audioband? Is that "smear"? Is that worse than non-flat cap response? Are these crazy questions?

 

RE: they're fairly fresh, posted on June 27, 2012 at 19:06:51
tube wrangler
Manufacturer

Posts: 1363
Location: USA
Joined: January 29, 2007

Nothing crazy about it. The problem is that the best capacitor in the world is really very bad as a linear, wideband, frequency neutral device.

No matter how good, in capacitors, all you get is junk.

Fortunately, some junk is much better than other junk. Is there any one capacitor that is, in fact, wide-band in an actual musical circuit where all parameters are constantly changing, where a single cap can work? NO.

That one cap-- regardless of quality, wants to form a resonant circuit--that is, it wants to tune-in to favor certain frequencies over others, based on the cap value and inductances and resistances in the associated circuit.

In order to get around this unwanted "tuning" effect (defect!), we have resorted to parallelling very high quality small caps with larger caps, coupled with very careful use of R.F. inductors.

In a simple SET amp, these capacitors will cost more than all the other parts in the amp combined-- you will end up using the world's best caps, and that is cheap compared to any usable alternatives.

I should slip-in here that push-pull amps do not have anywhere near these stringent requirements-- they're more wideband naturally, and are much more forgiving.

One or two small caps in parallel with a larger one in a power supply, a cathode bias system, or in a driver plate power supply will cause the amp to sound different, but not necessarily better. Usually, bandwidth will be gained, but incoherencies and hot spots in the frequency range as a whole will appear, rendering the amp pleasant, but not accurate.

Of course, the amp certainly isn't anywhere near accurate with only one cap in each of these places, either.

What we end up with is a requirement for as many as 11, 12, or 13 different capacitors of cascading values, installed at each of these places in the amp. If the right brands and values, and the right lead lengths are employed, there will be no trace of incoherency, and usable musical bandwidth will be widened, as will power supply and cathode bias response to signal. Dynamics will be improved so much that the only thing you can attribute it to is that finally, you have built a fast capacitor-- one that is fast at all the usable audio frequencies.

All of these advantages can and will be negated if any lead on any part of this (stack-up?) is too long, too short, or uses too small a conductor.

Sorry this got longish, and it's quite expensive-- this is what it takes.

---Dennis---

 

Yes, I use 0.22uf to 10uf caps for PS bypass, not 0.001uf, posted on June 27, 2012 at 19:52:51
What is the makeup of your power supply?

 

RE: bypassing caps, posted on June 27, 2012 at 22:53:47
hennfarm
Audiophile

Posts: 334
Location: Oregon
Joined: October 8, 2008
In your terms Drlowmu, what total Rk bypass value do you consider excessive?

 

RE: bypassing caps, posted on June 28, 2012 at 06:52:59
drlowmu
Manufacturer

Posts: 7989
Location: East of Kansas City
Joined: January 10, 2005
It is excessive when the total value slows down the music's presentation.

Jeff Medwin

 

RE: they're fairly fresh, posted on June 28, 2012 at 07:09:45
GSH
Audiophile

Posts: 261
Location: S. Cal
Joined: January 19, 2011
"Of course, the amp certainly isn't anywhere near accurate with only one cap in each of these places, either."

This is what I was horrified to discover experimentally, and wished I hadn't, but now would rather "face it".

"That one cap-- regardless of quality, wants to form a resonant circuit--that is, it wants to tune-in to favor certain frequencies over others, based on the cap value and inductances and resistances in the associated circuit."

If this is true, then this supports either finding and implementing
a fix/band-aid, as described or other (?), and/or choosing a circuit that minimizes C's. I like Gary Pimm's idea:

 

RE: they're fairly fresh, posted on June 28, 2012 at 07:14:09
GSH
Audiophile

Posts: 261
Location: S. Cal
Joined: January 19, 2011



Only one cap in the circuit, not very big either. The final C feeding the
300B's CCS is not shown, not sure how involved it is? It seems to be out of the "loop"

 

RE: they're fairly fresh, posted on June 28, 2012 at 13:59:27
tube wrangler
Manufacturer

Posts: 1363
Location: USA
Joined: January 29, 2007
I agree that you want to minimize caps as much as is practical-- but remember-- we can't yet get along without them entirely.

You're going to have to admit the truth, bite the bullet, and make some nice relationships with expensive cap people.

You can start with Mundorf, Dynamicap (Michael Percy Audio is a nice, co-operative source), Reliable Capacitor (wonderful helpful people-- S. Cal.-style), old Hovlands, Sidereals, and whatever else that is really great-- that you can scare-up.

Don't be bashful, and don't pinch pennies! GO RIGHT AFTER the best-- to "H" with the other stuff! This will save you time, frustration, and-- in the long run-- money.

You just can't beat actual performance. Don't try to. Embrace the best and the great people who build it.

---Dennis---




 

John Curl has publicly stated that the selective use of different caps is WHY his Vendetta Research phono was , posted on June 29, 2012 at 21:16:50
Cleantimestream
Audiophile

Posts: 5966
Location: Kentucky
Joined: June 30, 2005
Excellent. Ergo, anyone whom tries to upgrade that phono amp may face a startling change in sound.
The Mind has No Firewall~ U.S. Army War College.

 

Furthermore, a cap value ( .68 micro farad in front of a choke STILL means it is choke loaded.~nT, posted on July 6, 2012 at 10:38:29
Cleantimestream
Audiophile

Posts: 5966
Location: Kentucky
Joined: June 30, 2005
~!
The Mind has No Firewall~ U.S. Army War College.

 

Measurements, posted on July 8, 2012 at 23:34:07
Triode_Kingdom
Audiophile

Posts: 2980
Location: Texas
Joined: September 24, 2006
The effectiveness of bypass caps can be measured using an oscilloscope. If no measurable audio frequency voltage is developed across the existing cap, adding another cap will make no sonic difference. OTOH, if AC is present, its nature can be better defined with a spectrum analyzer. That may point to the specific requirements for additional bypassing.
--------------------------
Buy Chinese. Bury freedom.

 

RE: Measurements, posted on July 10, 2012 at 08:01:17
Tre'
Industry Professional

Posts: 9223
Location: So. Cal.
Joined: February 9, 2002
TK, I read your post as saying that there can be frequency dependent ESR in the main cap and once that frequency band is identified the ESR can be targeted and reduced by choosing a cap with low ESR in that frequency band instead of just adding a random value bypass cap.

That seems reasonable to me.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

Page processed in 0.024 seconds.