| gammelmalle |
When using a capacitor over the kathode-resistor, some people claiming that an electrolyctic will distort the sound. I've read the http://www.diyaudio.com/forums/show...or&pagenumber=1
but I only get more mixed up. Can it help to put another lower value non eletrolytic across the other. Lets say I have a 22uF Ck over a 1,5k ohm resistor. Will an 0,1 uF help ?
Thanks
björn |
|
|
| fdegrove |
Hi,
| quote: | | When using a capacitor over the kathode-resistor, some people claiming that an electrolyctic will distort the sound. |
They do and it ain't pretty.
With such a low bypass as 22uF you could envisage using a decent filmcap or composed values to make up the 22uF.
Space permitting of course.
| quote: | | Can it help to put another lower value non eletrolytic across the other. Lets say I have a 22uF Ck over a 1,5k ohm resistor. Will an 0,1 uF help ? |
I am no big fan of the following technique but the rule of thumb for bypassing electrolytes goes like 1/10 of the original value.
In your case that would be 22uF elco + 2.2uF film + 0.22uF + 0.022uF.
You may have to fiddle a bit to find the tonal balance that suit you and your system best though.
Cheers,;) |
|
|
| EC8010 |
Well, I realise this won't make me popular, but I don't have a problem with electrolytics. Modern ones are far, far better than old ones, and if you choose low ESR types, then I think it's hard to find fault with them.
More to the point, if you don't bypass the cathode, ra rises far faster than feedback acts to correct power supply noise. In other words, you throw power supply noise rejection away. Seems like a bad deal to me, but if we all agreed, we'd all listen to Quad 405s.
I'm just going to put my paws over my ears until the dust settles. |
|
|
| fdegrove |
Hi,
| quote: | | Well, I realise this won't make me popular, but I don't have a problem with electrolytics. Modern ones are far, far better than old ones, and if you choose low ESR types, then I think it's hard to find fault with them. |
LOL...I expected some resistance on this albeit not from you.:rolleyes:
Finding out what a cap sounds like isn't all that hard, use it as a coupling cap and tap off the signal at the output.
Listen to the sound with a good headphone set such as the Stax Lambda Pro and be your own judge.
Granted, modern low ESR electrolytes have improved a lot over the past 20 years, they still have a long way to go to reach the "perfect" cap status.
| quote: | | More to the point, if you don't bypass the cathode, ra rises far faster than feedback acts to correct power supply noise. In other words, you throw power supply noise rejection away. |
Sure, you have got to know what you're doing...one reason I so often emphasise the importance of a good PSU (read very high PSRR) and some circuits are more prone to that than others too.
Cheers,;) |
|
|
| EC8010 |
| Ah, well. I don't like to use bypass capacitors, not because they're electrolytic, but because they're capacitors. NiCds, diodes, short-circuit, etc, etc, anything to stop a cathode moving, rather than a capacitor. It's all about what happens when you momentarily overload that stage... |
|
|
| fdegrove |
Hi,
| quote: | | Ah, well. I don't like to use bypass capacitors, not because they're electrolytic, but because they're capacitors. |
Memory effects are the culprit plus the fact that most are just not linear at audio frequencies...
They always seem to slow down transient response, IMO and gobbling up detail, smearing etc.
I tried all kinds of caps for bypassing cathodes, not even the best filmcaps worked satisfactorily.
| quote: | | It's all about what happens when you momentarily overload that stage... |
Well, a none bypassed cathode resistor will have the valve overload quicker unless you rebias to prevent this.
One should design with this in mind right from the start IMHO.
Quite a lot more can be said about caps and stray capacitance, all of this deserves a thread all to itself, I feel.
Cheers,;) |
|
|
| ShiFtY |
Some electrolytics really do sound bad- I have some cheap speaker crossover ones that sound great, and some others that sound terrible. I have yet to try expensive film caps.
For a cathode bypass on a SE power tube, isn't the entire AC signal passing through the cap? |
|
|
| fdegrove |
Hi,
| quote: | | For a cathode bypass on a SE power tube, isn't the entire AC signal passing through the cap? |
Depending on the time contants, yes.
Not just for an SE amp though, it's the same story allover really.
| quote: | | Some electrolytics really do sound bad- I have some cheap speaker crossover ones that sound great, and some others that sound terrible. I have yet to try expensive film caps. |
In a multiway x-over, if you change one than best to change the lot.
If you don't you may face a complete tonal imbalance + the fact that you gain in efficiency when swapping to filmcaps.
So caveat emptor,;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
I tried all kinds of caps for bypassing cathodes, not even the best filmcaps worked satisfactorily. |
Exactly. So perhaps it's not really the fact that it's an electrolytic, rather, that it's a capacitor?
In theory, we never overload a stage, and we design to avoid overload. In practice, we do overload. I've come to the conclusion that the important point is how quickly the stage recovers from overload. A cathode bypass capacitor extends the time needed for recovery into hundreds of milliseconds, and I suspect that is why you prefer to avoid using them. Balanced audio doesn't need cathode bypasses, perhaps that's why it sounds so clean? |
|
|
| fdegrove |
Hi,
| quote: | | Exactly. So perhaps it's not really the fact that it's an electrolytic, rather, that it's a capacitor? |
Oh,yes...absolutely.
I must have expressed that here on the forum at least half a dozen times.
| quote: | | In theory, we never overload a stage, and we design to avoid overload. In practice, we do overload. I've come to the conclusion that the important point is how quickly the stage recovers from overload. |
Yup...couldn't agree more.
| quote: | | A cathode bypass capacitor extends the time needed for recovery into hundreds of milliseconds, and I suspect that is why you prefer to avoid using them. |
Indeed, and most OPTs present a similar behaviour to my ears, although much less in the mids, they're wideband phase behaviour buggs me just the same.
A well designed system should have all the virtues of sand devices without their nastiness and conversley well designed valve gear should have all the virtues of sand devices without the valves' shortcomings...valves have a lot less shortcomings and the so called "valve" sound can be attributed mainly to these darned bypass caps IME.
| quote: | | Balanced audio doesn't need cathode bypasses, perhaps that's why it sounds so clean? |
That, plus other advantages, yes.
Oh, and as you said somewhere else Johnson noise won't get cancelled out by balancing.
Cheers,;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
Indeed, and most OPTs present a similar behaviour to my ears, although much less in the mids, their wideband phase behaviour bugs me just the same. |
Hmmm. It has to be admitted that OPTs are less than perfect at frequency extremes, what with leakage inductance and lack of primary inductance, but they are an extremely elegant solution to the impedance matching problem, and they enhance efficiency, which is not something that could be said of OTLs. I've always felt that it is the mid-range that is most important. Get that right, and then worry about the rest. It seems to be easier to get that right with valves than SS. Conversely, SS can manage good bass. |
|
|
| fdegrove |
Hi,
| quote: | | I've got to find something for us to disagree about! |
30 years of audio experience and what do you get?:devily:
| quote: | | which is not something that could be said of OTLs. |
Well, Class A designs are considered the "nec plus ultra" in traditional designs and are just as inefficient as OTLs give or take a Watt or two.
Given a choice I'd go OTL, all else being equal.
When it comes to ultimate music reproduction I don't mind footing the bill.
| quote: | | I've always felt that it is the mid-range that is most important. |
Yes, and after you have that spot on with an OPT coupled design, what can you do?
Nothing.
| quote: | | It seems to be easier to get that right with valves than SS. Conversely, SS can manage good bass. |
Yes.
Valves can deliver good bass too, you just need a lot of them in // to get low enough Zo and this is where a bi/tri amped setup comes in plus a good valved x-over.
Tight bass with valves running full range is much harder to do.
Matching a SS subwoofer to a valve based full range amp seamlessly may prove just as hard though.
Cheers,;) |
|
|
| Circlotron |
| quote: | Originally posted by ShiFtY
Some electrolytics really do sound bad-
For a cathode bypass on a SE power tube, isn't the entire AC signal passing through the cap? |
If caps on the cathode are such an issue, why not avoid it altogether and simply bias it directly via the grid? In some respects at least, that would have to be better? |
|
|
| ShiFtY |
Circlotron that sounds very interesting. How would you do that with, say, a DHT in SE? Would it kill the bass?
I am starting to understand why there is a lot of anti-capacitor feeling; I think they do add more to the mish-mash of a circuit... |
|
|
| dhaen |
| quote: | Originally posted by Circlotron
If caps on the cathode are such an issue, why not avoid it altogether and simply bias it directly via the grid? In some respects at least, that would have to be better? |
Well, many designs do, but it adds a tweak to the circuit.
It doesn't solve the problem completely though. The DC bias you apply has a source impedance, either through the grid-leak resistor, or the bias supply to the bottom end of the driver transformer secondary. If, at the moment of overload, grid current flows:
The high value grid leak resistor will be unable to hold the bias under control
Or,
The cap that decouples the bottom end of the driver TX secondary is called upon to supply stored charge - back to square one!
I have been looking into the second case problem, and believe I've found a solution:
No cap....
Supply the bias point with a low impedance source such as a cathode follower.
Of course that begs the question: Where does the cathode follower get it's supply, but let's not go there ;)
Shifty,
The bottom end response is better with grid bias, as is the power output. It's just more trouble. But in my opinion, worth it :)
Cheers, |
|
|
| gammelmalle |
Thanks for all the replys.
But if I tell you thats its a guitaramp. Not a quad 505 or something equal....
And expermenting with a Ck makes A LOT difference. (Bass boost)
And that is a thing I can understand. But I wanna know more about the distorsion (if) and why its "only" on electrolytic. (I asume that we must compare the same value i.e. 10uF or what ever.)
cheers, and keep your answers coming. I learn alot.
björn |
|
|
| dhaen |
Gammelmalle,
In basic terms the value of the cathode capacitor cannot "boost" the bass. However, too small a value can cause loss of lower frequencies, as there is more feedback, or degeneration at them.
For the giutar amp application, it's enough to have a sufficiently large value cap to cover the frequency range, but not so large as to cause overload recovery delay.
Cheers, |
|
|
| analog_sa |
| quote: | | The cap that decouples the bottom end of the driver TX secondary is called upon to supply stored charge - back to square one! |
Hi John
Not sure if i understand this. A transformer or grid choke will provide a very low time constant at DC so it's not quite back to square one :)
cheers
peter |
|
|
| dhaen |
Peter,
I was thinking about the overload recovery delay issue. There might be a few milliseconds extension due to that time constant.
You are right regarding general "capacitor distortion" though.
Cheers, |
|
|
| Circlotron |
| quote: | Originally posted by ShiFtY
Circlotron that [change cathode to grid bias] sounds very interesting. How would you do that with, say, a DHT in SE? |
Instead of running the bottom of the grid resistor to ground, run it to negative whatever volts you want for bias.
Also, this next method have been done a googleplex times before, but how would it go to use a zener diode instead of a cathode resistor. Maybe a bypass cap across the zener. The voltage across the zener would not really change with changes in cathode current. Just what you want! |
|
|
| dhaen |
| quote: | Originally posted by Circlotron
snip..
how would it go to use a zener diode instead of a cathode resistor. Maybe a bypass cap across the zener. The voltage across the zener would not really change with changes in cathode current. Just what you want! |
This is the equivalent to connecting the cathode to a low voltage positive power supply. I think it suffers from a number of drawbacks:
The valve will be in greater danger from runnaway than with other methods. It might be difficult to get the bias just right, and there are still some charge / discharge effects of the parallel cap.
Cheers,
edit: now if this was class D.... ;) |
|
|
| EC8010 |
| There's no problem with using a diode for cathode bias. Certainly, it's not as safe as resistor bias in terms of thermal runaway, but when was the last time you saw a fixed bias stage run away? If you're particularly paranoid, use a CCS anode load and benefit from the reduced distortion as well as reduced fragility. |
|
|
| gammelmalle |
Dhaen!
If I get it right it is "overload recovery delay" (what ever that is) that I must search for to get more info?
cheers,
björn |
|
|
| dhaen |
Björn,
I don't have any links to hand dealing with this issue. I'll look around.
It is a phenomenon which more than half of amplifier designers pretend does not exist. RC coupled amps are more prone than transformer or direct coupled types.
The problem is to do with the behaviour of the amplifier following an overload transient. It is possible that the DC conditions of the amplifier are changed by the overload, and it can take some milliseconds before they revert to their proper values. During that time the sound is not normal.
Cheers, |
|
|
| EC8010 |
| The proper term for the phenomenon is "blocking." As John says, it is mostly swept under the carpet, but you will sometimes find blocking oscillators mentioned in television books (which are actually a very good source of information in general). |
|
|
| fdegrove |
Hi,
| quote: | | The problem is to do with the behaviour of the amplifier following an overload transient. It is possible that the DC conditions of the amplifier are changed by the overload, and it can take some milliseconds before they revert to their proper values. During that time the sound is not normal. |
When a strong transient sends an amplifier into momentary clipping it will take RC coupled amps much longer to recover from this condition than xformer coupled amps.
One way to make this very obvious is the use of oversized coupling caps, the overload recovery will be increased in time and the condition will be added to the continuing sinewave as a superimposition on the continued signal but distorted and lagging behind in phase.
PSUs will take some time to recover from this condition as well and yes this is certainly all very audible.
A quick Google on " overload recovery delay" will yield a ton of information about this phenomenon.
How amplifier designers can deny its importance is beyond me, then again there are a lot of those "details" that seem to escape them.;)
Cheers,:cool: |
|
|
| dhaen |
| quote: | Originally posted by EC8010
The proper term for the phenomenon is "blocking." As John says, it is mostly swept under the carpet, but you will sometimes find blocking oscillators mentioned in television books (which are actually a very good source of information in general). |
Ha ha:D
A characteristic of the blocking oscillators I grew up with, was they always used a transformer:cannotbe:
Cheers, |
|
|
| EC8010 |
| quote: | Originally posted by dhaen
A characteristic of the blocking oscillators I grew up with, was they always used a transformer. |
That's right. Kobayashi's US Patent 1,913,449 uses a transformer to couple the anode back to the grid, but the amplitude stabilisation element is an RC network. (If you very slightly redraw his circuit, it looks like every audio power amplifier with global negative feedback and explains motorboating.)
What happens with coupling and bypass capacitors is this:
Imagine that you have one stage with its anode at +100V, and you capacitor couple this to the grid of a second stage which has +10V of cathode bias. The first stage's anode then swings to +120V, a change of +20V. The grid of the second valve tries to swing from 0V to +20V, but when it reaches +10V, there is 0V between grid and cathode, so the grid conducts heavily, and is clamped to +10V. The voltage across the coupling capacitor instantaneously changes from +100V to +120V - +10V = +110V.
The first anode now returns to its rest position of +100V, but the coupling capacitor is still charged to +110V, so the grid of the second valve must be at -10V. The cathode of the second valve is held at +10V by its bypass capacitor, so Vgk is -20V, and the second valve is switched off.
The grid-leak resistor has -10V across it, so a current flows, and the coupling capacitor discharges back to +100V across it, at which point, normal operation resumes.
The key point is that we deliberately make the RC time constant of the coupling capacitor and the grid-leak resistor quite large (100n and 1M gives 100ms), and it takes 5RC for normality to return. Thus, an amplifier can take half a second to fully recover from a momentary overload that might otherwise have been imperceptible. |
|
|
| dhaen |
Yes, that's roughly how I understood it. It's hard accept that a patent could cover such a fundamental circuit. I bet it had been used many times before (but not described). But that's like a lot of things, I think.
BTW Did you know that "Kobayashi" translated directly as "Littlewood"?
Cheers, |
|
|
| fdegrove |
Hi,
| quote: | | BTW Did you know that "Kobayashi" translated directly as "Littlewood"? |
So, he must be British after all...only just a little bonsai.:clown:
Cheers,;) |
|
|
| Circlotron |
| quote: | Originally posted by EC8010
The voltage across the coupling capacitor instantaneously changes from +100V to +120V |
I don't know about instantaneously. :scratch: I would think the cap voltage rises according the the time constant of the cap and the first stage anode resistor that is supplying current to it. |
|
|
| EC8010 |
| Oh, alright, not quite instantaneously, but a lot faster than it can discharge via the grid-leak! |
|
|
| gammelmalle |
| quote: | | A quick Google on " overload recovery delay" will yield a ton of information about this phenomenon. |
Well I only recieved three (3) and those wernt too good ....
But I'm still glacd that you tried.
thanks a lot
björn |
|
|
| dhaen |
| quote: | Originally posted by EC8010
Oh, alright, not quite instantaneously, but a lot faster than it can discharge via the grid-leak! |
Only someone who's worked for the BBC could coin the term "near instantaneous" (NICAM?)
It's either instantaneous or not;)
Cheers, |
|
|
| fdegrove |
Hi,
Maybe this article will provide a little more insight even though I feel EC8010 did a very good job explaining the phenomenon:
NON IDEAL BEHAVIOUR.
Cheers,;) |
|
|
|
