sound of fixed bias versus cathode bias

[maulertone]

I have a Crate V33 that has four EL84s in the output stage that are cathode biased.I am considering converting the output stage to fixed bias operation.How would this modification affect the tone?

[stokes]

Cathode bias will give a more compressed sound with less headroom and a bit sooner break-up.Fixed bias should do the opposite,to some degree.Of course it depends on how "hot" you set the fixed bias.This isnt an easy mod,unless you have some experience working with PCB's in guitar amps,so be careful.

[Chuck H]

With el84s the tonal difference between cathode and fixed bias is really small. But if you want to eek every bit of volume and headroom you can then go ahead and do it. This will increase the plate voltage. Well, not really, it decreases the cathode voltage. But it's the same thing to the tube. So if the Crate runs a high plate voltage (close to or just over 400 volts) and more importantly, highish screens (anything over 350 volts) then you may not want to do this.

But you can still have fixed bias without placing the el84 cathodes at 0 volts. You can just measure the voltage on the top of the cathode R. Lets say you measure 11 volts. Get a pair of 5 watt 6 volt zener diodes and solder them in series across the cathode resistor with the cathodes toward the voltage (instead of ground that is). Your amp will be just like it is now at idle. But now there can be no voltage rise at the cathode when current increases. This voltage rise in cathode bias amps cools the bias and creates that cathode bias sag. Your amp would now be fixed bias.

The '11 volts, two 6 volt zeners' used above is an example. If you measured 10 volts you would use a pair of 5 volt zeners, for 13 volts use a pair of 7 volt zeners. You want the zener string voltage to be right at or just above the cathode voltage.

This is much easier and less expensive than building a bias supply and trying to mod on a PCB. Plus you can easily put it on a switch so you can decide which you like best. It's easily reversed and does no damage to the board.

Chuck

[Robert M. Martinelli]

AFAIK cathode bias tends to create more even order harmonics, so changing to fixed would probably give slightly more headroom but also a slightly harsher sound ( odd harmonics ).

As Chuck wisely noted, cathode bias, when achieved by means of a resistor creates a sort of sag, at least when the cathodes are un-bypassed, this resistor also acts as a sort of NFB, because the cathode voltage ( voltage drop on the resistor ) rises with the cathode current, so this design is self-regulating to some degree, this also tends to make the tubes' life easier ( more voltage at the cathode=less voltage drop across the tube=less dissipated power ).

I don't know about the voltages and currents inside your amp, but, if your B+ is, say, 320 VDC and your cathode voltage is around 10-12 VDC then your tubes are already biased over their S.O.A. ( similarly to an AC30 ) so, if you change to fixed bias, I would decrease the +B voltage some 15 to 20 VDC.

I would also like to re-enforce Stokes' suggestion about being careful....working on tube amps can be lethal!

Hope this helps

Best regards

Bob

[maulertone]

Thanks guys for the information and advise. The B+ voltage measures +323 VDC and the cathode voltage is 10.9 VDC,across an 82 ohm resistor.So total cathode current will be 133 ma;about 33 ma for each tube.I am running the stock Sovtek EL84s in the output.Should I lower the plate voltage to these tubes?

[Chuck H]

On the macro view that 33ma @323 volts looks like 10.65 watts at idle for a tube thats rated at 11 watts max. But fear not. el84s are often run over their max diss even fixed bias. At 323 volts you'll be fine without lowering the voltage. Not only that but we haven't figured the real math yet. Thats 323 volts on the plates with 10.9 volts on the cathode. So the working voltage is actually 312. Now figure that your using some of that current for the screens. About 6 or 7ma. And we have to understand that the typical 11 watt max for an el84 is the plate rating only. So... Now we have 312 volts @ 26ma. For an actual bias at idle of 8 watts per tube. So your biased now at 73% dissapation. Thats real conservative for an el84 IMHO. You could actually heat it up some for that creamy classic el84 sound. But, if your after more clean volume and headroom I would just leave it as is and strap two 5 watt 6 volt zener diodes in series across the cathode resistor. Banded ends toward the tube. Your amp will be more dynamic and defined and less squishy. As I said before, the difference between cathode and fixed bias for el84s is really small. But you can check for yourself for the cost of two diodes.

Chuck

[Robert M. Martinelli]

Hi,
if your B+ is 323VDC ( I expected it to be around that value ) and your cathode voltage is 10.9 VDC this means 312 VDC are dropping across the tubes, and if your cathode resistor is 82 Ohms you have 133 mA ( total ) ( this means the cathode resistor has to dissipate some 1,5 W ), so 33,25 mA are flowing through each cathode, and, even neglecting the screen grid current ( which will be in the 6-8 mA range ) you seem to be around 10, 5 W ( below the max static dissipation rating for EL84/6BQ5 which is 12W ) so it looks that there's no need to lower your +B. Anyway, keep in mind that, as already said, with fixed bias ( and with cathodes to GND ) all the + B voltage will be dropping across the tube(s), so you will have to use this value to state the plate dissipated power.

If you use a fixed "adjustable" bias network ( variable voltage divider ), be sure to add a resistor between the bias pot and GND, to prevent the bias voltage to go too close to zero ( your tubes will thank you ).

Hope this helps

Best regards

Bob

[Chuck H]

OOOOOO...Looks like we had a "simulpost" (Enzo coined that one). I wanted to mention that I was intrigued by your notice of the NFB caused by the cathode bias arrangement. It gave me some ideas that I can apply to projects. But as I think you mentioned, the use of a big enough bypass cap would virtually eliminate that factor. So with all else being equal (unless the Crate doesn't use a bypass cap) I think the series zeners are the easiest approach to this.

Chuck

[maulertone]

Thanks again for the information.The V33 has a -12v (7912) supply on board.What if I supply the grid bias from this source?Any special precautions to be aware of?

[Chuck H]

None that I know of. I do know that you'll want to eventually build an independant -V supply for the bias because you will want to adjust it sometime. But to try it out that should work fine. Before removing parts and changing things you could give it a run by putting a jumper across the cathode resistor, lift the ground end of the power tube load resistors and connect them to that -12 supply. That way if you find (like I did) that it makes no useful difference you can just put it right back as it was and forget about it. I took the trouble to equip my el84 amp with a cathode/fixed bias switch before any listening tests. The switch never moves. I could actually forget which position it's in and I would have to check to see. It's that close.

To be honest I don't notice any real improvement in sag elimination other than what you might expect from an extra 10 or 12 plate volts. el84's are pretty squishy no matter how you run them.

Chuck

[Robert M. Martinelli]

OOOOOO...Looks like we had a "simulpost" (Enzo coined that one). I wanted to mention that I was intrigued by your notice of the NFB caused by the cathode bias arrangement. It gave me some ideas that I can apply to projects. But as I think you mentioned, the use of a big enough bypass cap would virtually eliminate that factor. So with all else being equal (unless the Crate doesn't use a bypass cap) I think the series zeners are the easiest approach to this.

Chuck

Hi Chuck!
Simulpost....I like it! Hope Enzo doesn't want any royalties on that!

While I was pressing the "post" button I got the notification about your new post....funny!

Thumbs up for the "zeners" arrangement ( very simple and effective ), OTOH an adjustable bias network would allow him to "play" with the bias, even if, as already mentioned, the differences with respect to the "zeners" solution will be small.

As to the "7912", ( if he decides to go that way ), I would advise to look carefully what that IC supplies the voltage to, it could be a good idea to add another 7912 for the bias alone instead of using the existing one. This would eliminate the risk of unwanted coupling between stages and/or PFB.
( after all the thing is tied to the control grids..... ).
Best regards

Bob

[maulertone]

Looking at the schematic for the V33,the 7912 provides power to the relays and their control circuits (channel,boost and presence switching).Do you think the power switching of the relays might cause "snaps" and "pops" from the speakers,due to transients on the bias supply?

[Robert M. Martinelli]

Looking at the schematic for the V33,the 7912 provides power to the relays and their control circuits (channel,boost and presence switching).Do you think the power switching of the relays might cause "snaps" and "pops" from the speakers,due to transients on the bias supply?

yep, I would expect them to spoil the signal somehow, even if the relays are usually tied to a reverse-biased diode....you could overcome this with some complicated and expensive filtering, but....why bother?

1 - You could do as Chuck says, as the mod he suggests is simple, cost-effective and easy to remove should the results not please you.

2 - If you absolutely want to follow the "7912 path", here is some advice.

A 7912 is cheap, so I would suggest not to use the one already in the circuit , but to use another one for biasing the output tubes. The CMRR is so high you won't have to worry about pops coming from the relay switching circuit.

Only thing to remember is that the 78XX-79XX series ICs ( if memory serves me well ) need at least some 30 mAmps to function properly, so place a 390 Ohm 1W load resistor between the 7912 output and GND ( it will have to dissipate some 0,4W so use a 1W resistor for peace of mind ) also, it's advisable to bypass them properly if the filter caps aren't close, so be sure to add two .1 uF poly caps, one between the input and GND and the other between the output and GND. A rather small electrolytic ( 10 uF or so ) between the IC's output anf GND filters residual noise ( if any ) and contributes to lower output impedance.

Oh, one last thing - the 79XX and the 78XX series have different pinout : the center pad of a 7912 is NOT GND - it's tied to the input, so be sure to keep it insulated from GND.

Hope this helps

Best regards

Bob

[Steve A.]

OOOOOO...Looks like we had a "simulpost" (Enzo coined that one). I wanted to mention that I was intrigued by your notice of the NFB caused by the cathode bias arrangement. It gave me some ideas that I can apply to projects. But as I think you mentioned, the use of a big enough bypass cap would virtually eliminate that factor. So with all else being equal (unless the Crate doesn't use a bypass cap) I think the series zeners are the easiest approach to this.

Chuck

Chuck:

I had never heard of that zener trick- I need to try it out. Thanks! (I don't quite follow the explanation but I will take your word on it- I think that I have more faith than reasoning ability these days! )

FWIW Bruce Collins had suggested that the Vox AC-30 design had some natural NFB in the circuitry from the shared cathodes of the power tubes. Not that anyone would want to split the cathodes these days...

Steve Ahola

P.S. Crate uses a cathode resistor bypass cap in the V33- is there any reason not to? In the Torres book Dan thought that the cap helped balance output tubes which were not matched exactly... BTW perhaps the difference between cathode and fixed bias is more subtle with EL84's than other output tubes, but I can certainly hear the difference and feel the difference in the response: I always thought that cathode biasing EL84's was like throwing a nice warm blanket over your amp. And it would help cover up mistakes in phrasing, etc., that might be more noticeable when fixed bias...

Speaking of which, sloppy players often prefer cathode bias while the more precise players usually prefer fixed bias... Whatever- I love 'em both!

[Chuck H]

Whatever- I love 'em both!

Then both you shall have...

If you choose a zener voltage that is higher than your standing cathode voltage, but lower than it's maximum rise voltage, you can get cathode biased clean tones for that softer, less spikey attack, and fixed bias for overdrive so it doesn't become too inarticulate and squishy. IMHO EL84's are TOO spongy under heavey OD with cathode bias. And crossover distortion due to grid loading becomes very pronounced. I think fixed bias with EL84's is much better for overdrive.

My latest creation has about 9.5 volts at idle on top of the cathode R and a 100uf bypass cap. At full tilt the voltage on the cathode R rises to about 18. I bridged the cathode R and bypass cap with a 13 volt zener. I start to see clipping on the scope with about 12.25 volts on top of the cathode R. Since the cathode voltage can't rise above 13 volts now, the relationship between the grid and cathode voltage doesn't change (fixed bias) when the amp is clipping hard. So I get a cathode biased envalope when clean but fixed bias under hard clipping. The in between tones that are just starting to clip or clipping a little sound, well, in between. FWIW there is no pronounced audible shift between the two even though the 13 volt regulation is fairly sharp. Cathode voltage rises from 13 volts as it begins to clip to about 13.75 wide open. I am also using the "Paul Ruby" mod to reduce grid conduction. Now that the cathode voltage can't rise above 13 I can use 15 volt zeners in the "Paul Ruby" mod with no audible effect except for the elimination of crossover distortion. The tubes are well into cutoff before the zeners begin to conduct. The two together work very well indeed.

Enjoy

Chuck

P.S. I don't really see this as a silicootie infestation since the diodes are not "IN" the signal path but only manipulate the operating condition of the tubes. But there ARE 9 diodes in there. If you add spike protection to the power tubes you could get up to 15 diodes in the amp. Anyone looking inside would probably think they must be part of a clipping circuit.