Cathode bias *cap* issue - weird whistle

[EFK]

Here's an odd one which I have not encountered previously. I'm working on an amp based off an old Supro, paraphase PI, cathode bias 6L6, @ 400V B+. The original amp I have did not have a bypass cap on the shared 6L6 cathode resistor. I connected one (30uF) to the cathodes, but just temporarily I did not connect the ground as I wanted to try playing it both ways using an clip lead to see how the difference sounded. I like it w/o the cap better, not very much of a difference but here are a couple of issues:

(1) I have always read that adding the cap adds more gain. Every cathode bias amp I've ever had or built used a cap, but when I clip in the cap, what I notice is almost a very slight "damping" of the sound. It might be making a hair more gain (it's really not that noticeable in this particular amp) but it seems to lose something, it seems to get a bit "flatter" if that makes sense. Is this normal?

(2) It sounds great w/o the cap as I mention UNTIL I get the volume up to about 9, at which point when I totally crank it, all of a sudden there is this weird squealing whistle riding over the notes. The sound it still there, still sounds the same, but there's this damn screechy whistle on top of everything. I feel that is an oscillation of some sort? But why - what does the cathode bypass cap have to do with this? Again, this is *without* the cap.

??? Any thoughts appreciated.

[Chuck H]

I have to assume that the AC typically passed to ground via the bypass cap does two things when you don't use a bypass cap. One thing is local NFB since AC is now present at the cathode. I don't imagine this is the problem. But the other thing is that AC is now present in that big resistor and it's leads. If modulation in the resistor and it's leads are creating any EM it could be picked up by other leads and components in the amp. Sooo... Try keeping sensitive leads away from that resistor or it's leads or perhaps even relocate some components. Since the offending EM is probably of a fairly high frequency (AF wise) the easiest solution may be to use a small bypass cap. Maybe a film cap with a value below 1uf would fix it. But I'm just guessing. Easy enough to try though.

[EFK]

Hmmmm. First of all, thanks (as usual) Chuck - makes perfect sense, even to me. Because of tight space, bias resistor leads are near the heater lines. I'll have to take a look at what else is nearby - your little plastic cap idea-r is a good one too, which I shall try directly...

[EFK]

.68uF/100V took care of it. I do want to maintain the 'unbypassed cathode' sound/feel though, so I wonder if I should try going lower. Obviously on preamp cathodes, a .68 bypass - while low in value - does very noticeably affect the gain. Does this hold true likewise for the power amp?

[Chuck H]

Well... I can't bring expertise to the table such as is available here at it's best (Steve Conner, Enzo, R.G., etc.) but here goes:

The lower impedance of the power tube cathode circuit requires a larger cap than a typical preamp tube circuit to pass the same low frequencies to ground. Just ballparking with an online calculator but .68uf cap across, say a 330 ohm cathode will bump frequencies above 1kHz. You probably would hear the effect in the tone there except the bump is only about 2.5dB. The other thing a bypass cap on the power tube cathode does is stabilize the voltage offset between the cathode and grid. A popular trick on the "18 watt" forums is to use a silly big bypass cap (like 2200uf) to simulate the effect of fixed bias. So the smaller the cap, the less effect it has on regulating the voltage offset. I would think a .68uf has very little affect. You could always put a switch on it to see if there is any tone change with the .68uf cap in or out, aside from getting rid of the oscillation of course. You could also experiment with smaller caps until you find the lowest value that stabilizes the amp (+ a little for good measure and unforseen tolerance issues). It's easy to do and will give you the satisfaction of knowing there is as little bypass going on as possible.

[Rhodesplyr]

In a push-pull output stage with a common cathode resisistor, a bypass capacitor doesn't really add gain. What it does is to keep the gain from decreasing when you move into the B region of Class AB bias--when one tube is cut off and the other is conducting more current for part of a cycle than both tubes combined at idle. As long as you're still in the region where both tubes are conducting, the AC voltage at the shared cathodes should ideally be zero due to phase cancellation, assuming good balance of the phase inverter and other components. So, installing a cathode bypass capacitor shouldn't do much at all. It's when you push the amp harder that an unbypassed cathode resistor starts to act as negative feedback. Greater current across the cathode resistor would mean that the cathode voltage would rise, following the tube's control grid. The bypass capacitor attempts (anthropomorphizing here) to hold the cathode voltage stable with respect to the tube's control grid. The effect is to increase the amp's clean--or relatively clean--output.

The size of the bypass capacitor is does have a significant effect on how the amp sounds at higher signal swings, especially when it finally starts to distort. With no bypass, you'll have more compression due to the NFB of the cathode resistor. With the bypass, you can get louder transients, to a point. If you add too much, you'll get a really farty, unpleasant-sounding distortion when the output stage finally begins to distort. More specifically larger bypass capacitors allows you larger output voltage swings in the bass, which is why many guitar amp cathode resistors are a bit under-bypassed compared to, say, Hi-Fi amps. You don't need or want response down to 20Hz.

In contrast, in a typical preamp tube stage, the cathode bypass capacitor is *always* working (and very much in the signal path) because there's no phase cancellation as there is in a push-pull output stage. The same is true of amps that use individual cathode resistors and bypass capacitors for each output tube. In both these cases, the bypass capacitor is there to overcome (partially) the NFB of the cathode resistor and is doing its job at all signal levels, not just when things get loud.

The smaller the value of the cathode resistor, the larger the bypass capacitor needs to be--and vice versa. I have heard it said that the bypass capacitor's capacitive reactance should be one tenth of the cathode resistor's value at the lowest frequency you want the stage to amplify; however, I've also heard that that's a rather crude recipe for picking the right value.

In general, with guitar amps, I think the cathode bypass cap value is best picked by playing the amp and listening.

[Ronsonic]

The cap will affect gain and frequency response. For your example the .68 across, just for example a 2.7K cathode R will result in more gain especially in the mid and upper frequencies, which is why you see that on the bright channel of classic Marshalls. A larger cap will have a lower corner frequency. Of course this forms a high pass filter based on the value of the resistor. A smaller resistor would have a lower corner frequency as well. F= 1/(2*pi*C*R)

[Chuck H]

The cap will affect gain and frequency response. For your example the .68 across, just for example a 2.7K cathode R will result in more gain especially in the mid and upper frequencies,

You did cover the formula and added the caveat that the resistor value will affect the knee frequency (+++) but by using the 2.7k R example and stating the frequencies affected feels a little misleading since a .68uf across a typical cathode value for a pair of 6L6's will only affect frequencies above 1K and then only 2.5dB instead of probably 7dB for a 2.7k R value on a class A biased preamp tube.

It's when you push the amp harder that an unbypassed cathode resistor starts to act as negative feedback. Greater current across the cathode resistor would mean that the cathode voltage would rise, following the tube's control grid. The bypass capacitor attempts (anthropomorphizing here) to hold the cathode voltage stable with respect to the tube's control grid. The effect is to increase the amp's clean--or relatively clean--output.

Yes... Like the difference between cathode and fixed bias?

The size of the bypass capacitor is does have a significant effect on how the amp sounds at higher signal swings, especially when it finally starts to distort. With no bypass, you'll have more compression due to the NFB of the cathode resistor. With the bypass, you can get louder transients, to a point. If you add too much, you'll get a really farty, unpleasant-sounding distortion when the output stage finally begins to distort. More specifically larger bypass capacitors allows you larger output voltage swings in the bass, which is why many guitar amp cathode resistors are a bit under-bypassed compared to, say, Hi-Fi amps. You don't need or want response down to 20Hz.

If there's no significant affect until the amp starts to clip, why would bigger cap values be more suitable for hi fi amps that arent intended to clip???

In contrast, in a typical preamp tube stage, the cathode bypass capacitor is *always* working (and very much in the signal path) because there's no phase cancellation as there is in a push-pull output stage.

Right. That's a big chunk of info missing from my post above. Thank you.

The same is true of amps that use individual cathode resistors and bypass capacitors for each output tube. In both these cases, the bypass capacitor is there to overcome (partially) the NFB of the cathode resistor and is doing its job at all signal levels, not just when things get loud.

But since an AB amp is only amplifying the half wave form that's not in cutoff, doesn't this become a moot effect? And since a shared bypass cap will be working anytime only one tube is conducting isn't that also the same?

In general, with guitar amps, I think the cathode bypass cap value is best picked by playing the amp and listening.

+++ Absolutely.

[Rhodesplyr]

If there's no significant affect until the amp starts to clip, why would bigger cap values be more suitable for hi fi amps that arent intended to clip???

I think "clipping" may be the wrong term here. Distortion, in the broadest sense, comes in many forms; for example, you could say that the onset of compression due to the cathode resistor becoming a source of negative feedback is a distortion of the input signal. I'd say that larger bypass capacitors allow the output stage to swing larger, cleaner transients--up to a point. With a large bypass capacitor, when the amp literally starts to clip, it will do so in a fairly nasty, un-musical sounding way, part of which is the longer charging time of the larger capacitor. You generally try not to run Hi-Fi amps into clipping at all, but you want the output stage to be able to follow its input signal accurately up to that point. Since musical instrument amps are amplifiers and sound effects in one, that principle doesn't really apply.

I've experimented with this in Leslie amps. Later Leslies like the 122 and 147 use a 150 Ohm common cathode resistor with a 200uF bypass. Older Leslie amps like the 32H had a common cathode resistor with no bypass at all. One transitional model, the 50C, introduced Leslie's use of bypasses. You can increase the power output before obvious distortion, especially in the bass, from the older amps by installing a cathode bypass capacitor, and you can tailor the sound you want when the amp is pushed hard by varying that capacitance.

The point I was trying to make about amps that use individual cathode resistors for each output tube and individual bypass capacitors for those resistors is that those bypass capacitors are functioning to increase gain at any signal level--not just when one tube of the pair stops conducting. Because the output tube cathodes are not connected, there's no phase cancellation at the cathodes. Here's a schematic for the power amp from a Hammond A-100 organ:



The same is true of the Heathkit W-5M Hi-Fi amp. The purpose, in both these cases, appears to be that it allows you to balance the output tube bias current, something you can't really do easily with cathode bias and a common cathode resistor.

[EFK]

This has been a very informative and clear thread! Great information - thank you!