Single Ended EL84 tube amp plate voltage thoughts...

[kevtronic]

It's been a while since I've been here, but I find this is the place to go when your wondering about an amp
What brings me here today is an SE amp I'm working on that has a tube compliment of an EL84 and a 12ax7 and I'm trying to make a decision on plate voltages.
So, what I've done is tried the plate voltage on the EL84 at 250 and at about 310 and biased it both ways at about 95% (cathode biased).
The plate voltage on the 12ax7 was at about 170 (in general I kind of like my 12ax7's that way, but any who...)
With the EL84 plate at 250 it distorts very early and gets into metal territory in the upper end and on the o-scope the wave gets nasty distorted (using a 1k input at about 150ma).
It goes way beyond squared off by the time you hit 10 into sharp pointy skewed warpness - lol.
With the EL84 plate at about 310 it has more clean headroom, but at 10 the o-scope shows the same super distorted wave form.
So, I increased the plate voltage on the 12ax7 to about 220 and things are little better but still hairy at 10.
The amp only has a volume knob; no tone stack.
I'm trying to get as much clean headroom as I can.
The wave form looks very good at lower volumes, you push it a little and it does the norm flattening thing, but then as you hit the upper end the wave form is awful.
Any input, insight, advice, etc is appreciated.

[The Dude]

Not a ton of insight, but I'd go at it a stage at a time if you're looking for clean headroom. In other words, hang your scope on the 12AX7 plate first and get that looking the way you want it to with practical gain. If you feed crap to the EL84 grid, that's exactly what you'll get out.

[kevtronic]

Not a ton of insight, but I'd go at it a stage at a time if you're looking for clean headroom. In other words, hang your scope on the 12AX7 plate first and get that looking the way you want it to with practical gain. If you feed crap to the EL84 grid, that's exactly what you'll get out.

That is true and sounds like a good game plan. Thanks for the input.

Any thoughts about the plate voltage? This is the first SE EL84 amp I've tweaked out and I see a lot of people run the plates 300 to 350 volts. I was under the impression going past 250 for an SE amp took the amp out of the class A zone, not to mention the plate specs. Running the plates at a higher voltage seems to provide more clean head room and over all less distortion, but...

[The Dude]

Remember that if the amp is cathode biased, you need to subtract the cathode voltage from the plate voltage to get the actual voltage across the tube. The tube doesn't know if it's cathode is grounded or not. It only knows the voltage potential between the two.

[kevtronic]

Thanks for the heads up and reminder, always a good thing, however the 250 and 310 plate voltages are the actual plate voltages. Keep me thinking though - lol - never hurts, man, and it's appreciated, know what I mean

[mozz]

Subtract the cathode voltage from the plate voltage, wait, it seems you didn't measure dissipation did you? Maybe cathode resistor change might help. Also, i never use 150mv for input testing, more like 50mv.

[The Dude]

IMO, I wouldn't try to go over 300V plate to cathode, especially in a single ended amp. More important is bias, again, IMO.
This might help some:

Weber Bias Calculator

A single ended amp with a single EL84 isn't going to get real loud. You might also consider actually measuring output power just before clip. Maybe you've simply maxed out. But, again, you need to determine which stage is clipping.

[kevtronic]

Subtract the cathode voltage from the plate voltage, wait, it seems you didn't measure dissipation did you? Maybe cathode resistor change might help. Also, i never use 150mv for input testing, more like 50mv.

Thanks for the input. Always appreciated...

Dissipation was at about 95% for both plate voltages. I could be wrong, but I was under the impression that depending on whether a single coil fender type pickup (about 1/8 volt), a gibson type humbucker (about 1/4 volt) or p90 (about 1/2 volt) the pickups could produce from about 128mv'ish to say 500mV'ish in general. I kind of thought 150mv was a fair input. Am I missing something somewhere?

[Chuck H]

That input voltage seems ok if you're shooting for really clean with a guitar. But don't mistake gain with headroom. It doesn't take much to drive an el84. A pair of +100 gain triodes without much voltage division will actually clip regardless of what gets to the power tube grid. So if the amp is clipping at lower knob settings that doesn't mean that it's not producing it's full clean power. It just means the preamp has a lot more gain than it needs and CAN clip both itself and the power tubes when turned up higher. So you COULD just keep the volume setting low. If the actual number setting on the knob is bothering you then you'll need to reduce preamp gain. I'd add a voltage divider between the preamp triodes. Another gain reduction would be to remove any cathode bypass caps (you didn't provide a schematic) or reduce the plate load resistor value at the input and rebias that stage accordingly. You can use your scope at different points in the signal chain to see this analysis in operation. As far as a good plate voltage... More (but within max specs) is typically better. I'd shoot for 200+ preamp volts. That's actually a tall order since you'll want to keep the power tubes under 330Vp for class A. You may need to branch the power supply to do it. But since you like a preamp tone with about 170V on the preamp plates I would say that a voltage divider is the easiest solution. You may have a particular aesthetic in mind that dictates a very direct signal path with a minimum of circuitry. Ok, fine. Keep thegain knob setting down then. You could also experiment with a different preamp tube. I've always been partial to the clean tone of 12ay7's and they have about half the gain of a 12ax7.

[kevtronic]

IMO, I wouldn't try to go over 300V plate to cathode, especially in a single ended amp. More important is bias, again, IMO.
This might help some:

Weber Bias Calculator

A single ended amp with a single EL84 isn't going to get real loud. You might also consider actually measuring output power just before clip. Maybe you've simply maxed out. But, again, you need to determine which stage is clipping.

I hear ya on the 300 volts for the plate voltage, which is why originally why I dialed in and biased at 250 plate voltage. But, it does seem that some folks push the EL84's upwards of 350 on the plates even for an SE amp, which is what led me here (can/should I push past 250 on an EL84). It's the first time I've tried really tweaking an EL84 SE, so, figured I'd pull my head out of the amp, come here, and let you guys jump start my brain to thinking again. It'll be tomorrow before I can dive back in. I will be starting with the preamp tube and working toward the output tube. I've already checked for DC leakage, etc. I've also checked the tubes on my tube tester, but maybe one of them is just going into overdrive way early or something. I'm thinking it's a lack of experience with SE EL84's and I'm going thru a learning curve.

[kevtronic]

I'd add a voltage divider between the preamp triodes. Another gain reduction would be to remove any cathode bypass caps (you didn't provide a schematic) or reduce the plate load resistor value at the input and rebias that stage accordingly. You can use your scope at different points in the signal chain to see this analysis in operation. As far as a good plate voltage... More (but within max specs) is typically better. I'd shoot for 200+ preamp volts.

Thanks for the info. I'm familiar with everything your talking about except for "I'd add a voltage divider between the preamp triodes." Voltage dividers I'm familiar with, but not putting one between the preamp triodes. How, why, where do you go about that? That's new to me and I'm all for new tricks

I've got a lot of 9 pin triode types in the drawers, so I could switch to one with less gain, but at the moment I hoping to make sure I'm not just missing something in SE EL84 land. The amp sounded good with the 310 on the plates and the 220 on the 12ax7, but man what an ugly distorted wave form. The wave form on the output of the amp just seemed way over the top. It seems I should be able to clean it up at least some.

Oh, on each the cathodes of the 12ax7 I am using a 3k resistor and a 10uf bypass cap and on each the plates I am using 100K plate load resistors. The cathode on the EL84 has a 25uf cap strapped across it. The coupling caps are .01uf 600v orange drops. I am using 3k as a grid stopper to the EL84, subtle differences but I preferred 3k over 1.5k, 5.1k, and 6.8k values I tried. I tried different values as it seems marshall amp lovers play with the el84 grid stoppers as part of there amp tweaks.

I have half a mind to parallel the preamp tube, but, at this moment I'm trying to tame the beast with the two halves separated.

I'll have to make a schematic up for the amp as I don't have one at the moment. Maybe I can work that in to todays todo list.

Thanks again, man, it's all appreciated...

[Chuck H]

Ok... Since you want this to be a "clean" amp (again, not the same as headroom, but I digress) you should consider that a single triode with a 100 gain potential will have an actual in circuit gain of about 60 with a bypass cap. If your guitar has peak voltages of 500mV that would mean peaks of 30V hitting the following grid. That's enough to clip the next triode in the preamp OR even the el84 grid if you only used a single triode! Removing the preamp bypass cap will reduce the in circuit gain to only 19 and linearize the bottom end down to the knee of the coupling cap. If you use this triode alone to drive the el84 you no longer have enough gain to reach clipping (or possibly full power). And keep in mind that you may on occasion wish to run a lower signal into the amp and still achieve full power. In this case the amp absolutely SHOULD be able to reach clipping with a typical guitar plugged in and the volumes up. So...

3k cathode is a little cold for a preamp tube. This doesn't keep with getting the most linear response or the most clean headroom. Instead, you might rebias a single preamp stage with a 1k cathode and no bypass cap. This is much closer to center bias for the stage than 3k (more headroom). Considering that this would be the only stage feeding the power tube grid, lets assume a 220k grid load. That would give you a gain of about 30. Enough to clip the amp with a full guitar signal somewhere around 8.5 with a typical volume circuit. In this case your consideration of paralleling the triodes is a good one. The lower impedance drive to the power tube grid should reduce grid loading, and therefor bias shifting. That should mean slightly better performance on peaks for better dynamics. To go this route you would parallel the triodes and use a 470R cathode resistor un bypassed and a 47k or 68k plate load (I'd go with 68k for just a tad more gain). Use a 250k pot as a volume control in place of the 220k-470k power tube grid load. Use the voltage arrangement that gave you 310Vp on the power tube. Badda bing. That should be about as much headroom as you're ever going to get from a 4W amp. Any other considerations (OT impedance/speaker match, NFB circuits, fine tuning bias conditions, etc.) not withstanding.

EDIT: You could also series the triodes with the first being the typical 100k plate/1k cathode arrangement I mentioned and the next triode would be a cathode follower circuit. That would give an even lower preamp output impedance (better) but slightly less gain and a higher noise floor (worse). There's always a tradeoff to be considered.

[Mike Sulzer]

First: is the impedance of the output transformer correct for maximum power transfer to the load?
Second: Does the preamp stage put out enough voltage operating linearly to drive the power tube past clipping?

If yes, yes, then there is not a lot more you can do to get more clean power.

I do not think you need the second 12AX7 stage to get all the clean power, but if you want two stages, put a gain control after each to give the most flexibility from clean to very, very dirty.

[J M Fahey]

1) What is the actual power out you are measuring when amp reaches clipping?
2) momentarily forget the first triode. What signal level do you need to feed into the second triode grid so EL84 just starts clipping both peaks?
3) are you using NFB?

EDIT:

It goes way beyond squared off by the time you hit 10 into sharp pointy skewed warpness

Well, thatīs what tubes do when overdriven, specially if loaded with a speaker at the other end of a transformer.

The ones which distort plain square, without crazy peaks, skewing or getting into crossover distortion are SS amps.

[eschertron]

SE amps won't clip evenly anyway, no matter what you do. To minimize the odd non-square distortion that happens to an SE power stage, the tube bias is usually targeted at 100% in order to give the 'most' class A power available. I'm guessing at idle the bias voltage for your EL84 is about -11.. -14vdc. So clean power starts to run out by the time the grid is fed 20..25vpp signal. You can easily get that from the anode of a single triode stage, from hot pickups or a pedal output. If there's a lossy tone stack after the input stage (you haven't said) then a second stage will be needed to 'recover' the signal.