Capacitors in the cathode of preamp tubes

[M Gregg]

Assuming,

You don't want an LED in the cathode of a preamp tube what types sound the best..
ie Silmic, Muse, cerafine, Muse Gold etc..and thoughts on preferred ideas why might you use a cap resistor as opposed to an LED..wait for the Why do we not see LED's in audiophile amps? Thoughts and preferences and why? (Imaging)

What if any effect does the tolerance of capacitors have on stereo channels?

Regards
M. Gregg

[Struth]

Hi Guys

LEDs, diodes, current sources, etc, used to bias tubes for both pre and power apps are a form of fixed-biasing. For a PA it is a very inefficient form.

In a preamp, for hifi you want linearity and would presumably want to set the idle condition for maximum clean signal output. Whether this also demands or sets gain to maximum is a related concern. With Rk in place, its value can be tweaked for both/either maximum gain and cleanest output. If you wish to maximise gain adding Ck in parallel will do just this. Ck's value will determine if maximum gain extends to bass frequencies, and exactly how far that goes.

With the various fixed-biasing schemes there is no opportunity to focus the gain over a specific frequency range. The dynamic resistance of a diode or LED is very low, so stage gain is maximised for all frequencies.

There is a distinct disadvantage using the common active bias methods inasmuch as the noise of the biasing element is amplified by the tube. Mu between K and A is the same as for G to A. For this reason you often see bias diodes bypassed with a cap to shunt their noise to ground.

Have fun
Kevin O'Connor
londonpower.com

[MrCurwen]

Well, LEDs are solid state.

I'm guessing that the kind of person that has money to spend, reads some of the audio magazines but doesn't have any electronics skills to try things out for themselves, thinks that solid state is simply, well, evil.

I would agree that on the surface of it, this is not an unreasonable assumption from the person looking to spend money on hifi. The vast majority of SS stuff that most people, even hobbyists, are subjected to, is simply pretty bad.

So once the idea that SS is not good for sound has entered the persons head, he is not likely to buy stuff with SS parts, especially in the crucial position of cathode biasing. So, the people who make audiophile amps get this, and don't put LEDs there. Supply and demand, not necessarily based on any technical sense.

Another thing, from the point of view of manufacturers, is that a LED is a LED. There's really not much room for variation; sure there are 4 ohm impedance LEDs and 10 ohm impedance LEDs, but even that in most schemes wouldn't really make any discernible difference.

Where as with caps, you can have this and that brand cap, and the buyer is likely to have some preferences and opinions they have picked up from before mentioned magazines. This gives the manufacturer a brand based selling point to list.

Of course, simple tradition. RC is what was available when the classics of hifi were made, so RC is what people tend to use, or to look for in stuff they buy. There's a lot of people who use high output impedance PSUs and AC heating on DHTs as well, and prefer the sounds thereof.

[SY]

Originally Posted by Struth

For a PA it is a very inefficient form.

Can you expand on that? I can't seem to understand why that would be.

Originally Posted by Struth

There is a distinct disadvantage using the common active bias methods inasmuch as the noise of the biasing element is amplified by the tube. Mu between K and A is the same as for G to A. For this reason you often see bias diodes bypassed with a cap to shunt their noise to ground.

Have you measured the noise of a forward biased LED? Or the impedance? The cap bypasses I've seen were window dressing.

[andyjevans]

Well, my heretical answer to this is that I shun bypassed cathode resistors like the plague. Boutique parts or whatever - not interested. To me they dull the sound. Electrolytics are horrible and even polypropylenes are rather foggy. For the really clean sound use battery grid bias, or my favourite - filament bias with a DHT. I've now changed my whole hi-fi system so every tube is in filament bias, from start to finish. Very happy. For those unfamiliar with it, see:
VinylSavor: Filament Bias, Part 1: Concept

[Struth]

Hi Guys

Suppose you have a conventionally fixed-biased output stage, with say -40V on G1 that causes 50mA to flow per tube, and there are two tubes, for a total of 100mA for the output stage. The grid essentially draws zero current and thus zero power. The 40V of control appears at both ends of the grid leak, so there is essentially zero power to bias the tube.

If you now place a current source between the tube cathodes and ground, and tie the grid leaks to ground, it looks superficially like a cathode-biased amp except Rk is replaced with active devices. The bias circuit is adjusted to provide the same 40V difference between G1 and K that existed in the conventional circuit. With the same tubes, 100mA flows.

The active bias circuit passes 100mA while sustaining 40V, so there is a power dissipation of 4W. This is infinitely higher than zero watts required to bias the tubes in the usual way.

If the active bias circuit is voltage regulated instead of current regulated, then its dynamic impedance will be low and currents higher than 100mA will be able to flow during the signal cycle and the instantaneous power for the bias circuit will rise and fall with the signal. Overall, there will be 4W dissipation.

The choice of such circuitry is wide but most builders pro and amateur alike repeatedly select the poorest options. Our 'Designer Notes volume-1' coming out before Christmas has a detailed note about this bias method.

Diode dynamic impedance is clearly illustrated by measuring the voltage across the diode while varying the current through the diode. The voltage difference is caused by the internal resistance of the device, and both are tiny.

Diode self noise varies with the device construction. Zeners are the worst, if you needed a high bias voltage where one might be an option.

Have fun
Kevin O'Connor
londonpower.com

[DF96]

Originally Posted by Struth

This is infinitely higher than zero watts required to bias the tubes in the usual way.

Exaggeration. All bias methods use some power (even grid current bias!). Fixed bias usually needs an extra negative supply rail and this is bound to consume some power. Not much, but some.

Originally Posted by Struth

The choice of such circuitry is wide but most builders pro and amateur alike repeatedly select the poorest options. Our 'Designer Notes volume-1' coming out before Christmas has a detailed note about this bias method.

We await with bated breath to see if you have anything new to add to the well-known issues often aired in the forum.

[Struth]

Hi Guys

DF96: no peanuts for me, thanks. Actual contributions are welcome.

SY: The active bias method becomes even more inefficient compared to conventional grid bias when you are effectively simulating cathode bias. The maximum current is set for the output tubes by the bias circuit. This will be based on observation of typical cathode biased circuits, the idle current, the voltage across Rk, and making sure the tubes idle within their ratings. In this situation, the simple approach regulates current and allows voltage to float to where the tubes keep it from rising. Typically this will be similar to values with Rk in place.

Current in this mode might be many hundreds of milliamps depending on the circuit, so the bias circuit dissipation is much higher than for the above example. The floating voltage that ends up on the cathodes can be problematic, and in threads elsewhere people have observed double voltage compared to self bias. It is therefore good to clamp the voltage across the current regulator to some value a bit higher than what the typical idle voltage would be.

There are many ways to do these things. About the worst method is to use a three-terminal regulator regardless of whether it is wired to regulate current or voltage. These devices have an extremely poor frequency response and their noise signature will telescope through to the output.

Have fun
Kevin O'Connor
londonpower.com

[tomchr]

Originally Posted by M Gregg

You don't want an LED in the cathode of a preamp tube what types sound the best..
ie Silmic, Muse, cerafine, Muse Gold etc..and thoughts on preferred ideas why might you use a cap resistor as opposed to an LED..wait for the Why do we not see LED's in audiophile amps? Thoughts and preferences and why? (Imaging)

What if any effect does the tolerance of capacitors have on stereo channels?

The purpose of the cathode cap is to be as close to a short circuit as you can get across the frequency band of interest. Hence, you want a type that has the lowest ESR across the audio frequency range.

A good electrolytic cap like the Nichicon KZ-series offers low ESR in the audio range. As do many of the electrolytic caps intended for use in switchmode supplies.

The advantage of an LED is that it has low AC (dynamic) impedance.

Personally, I would not buy a cap that doesn't come with a data sheet. As a minimum, the data sheet needs to list the ESR - or even better show an impedance plot. Many of the audiophool caps are not specified. Wonder why...

The tolerance of the capacitor has no impact on the system. The actual capacitance, however, does determine the LF rolloff of the amp. A large sample size of +/-1 % caps would, theoretically, produce a distribution of LF responses with pole frequencies that vary by +/-1 %. A +/-10 % cap would cause +/-10 % variation (again, assuming a large sample size so you see the full distribution). Do you really care if the LF pole is 2.00 Hz, 2.02 Hz, or 2.20 Hz? Probably not...

~Tom

[Bob Richards]

If I want to improve a low level tube stage, I will bypass the cathode R with a Polyprop cap, and stick a current source, or mu-follower circuit on the plate.

The current source gives you significantly better power supply rejection (hum and etc.), and better tube linearity and more gain.

There are those who say polyprop caps aren't perfect... I say they are excellent. WAY better than the weakest link in any system. Maybe the listeners head was cloudy. Maybe he got a bad Polyprop.

Caps that don't come with a data sheet... I've never heard of that being an issue. Get an ESR measuring device if you're actually worried. Data sheets are often wrong, or at least a bit optimistic.

I'd worry about noise generated by the LED, and the fact that it's not at all adjustable. If an LED is 10 ohms, and you try to bypass it with a cap to short out it's noise, the cap will either be huge because of the very low impedance of the LED, or a smaller size cap might create a shelf in the response; perhaps in the midrange freqs.

If you don't bypass the cathode R, it will inject some noise (maybe an issue in a front end preamp circuit), and the plate impedance will be significantly higher, so will be more easily distorted by the loading effect of the next stage.