el84 bias...How hot?

[Robert M. Martinelli]

Using a dummy load is the only way that makes sense to do a direct comparison and is the industry standard. The back EMF a speaker's coil produces doesn't necessarily reduce power in a tube amp either, in fact I would say theres usually an overall increase in power especially on the rise of the waveform which can be seen quite easily on a scope. It's possible to measure the difference but it would be very difficult and you would have to figure in the impedance curve of the speaker and a bunch of other factors in the frequency/time domain. It would be meaningful only if everybody used the same speaker which of course we all don't so we all use resistive loads when measuring power. What would probably be more meaningful is to say what an amps average db output was with a given speaker/cab. But that still wouldn't tell you how much louder it would be in certain rooms or certain places within the room etc. So you have to eliminate variables if you want to do a direct comparison of a certain part of the picture such as the power section of a tube amp. That's why resistive loads are used.

Hi again ETR,
I pretty much agree with what you said, that was exactly why I used the "apples to apples" phrase of yours, we obviously need to lower the variables' number to get better comparisons, OTOH, ( Thanks Bruce ) sounding good ( or loud ) is something that involves other factors..... the amp I love the most ( my 1964 AC30TB ) though a sheer 36 W ( fully cranked ) is on par with 50 Watters in terms of loudness, probably thanks to its higher harmonic content and better speakers' efficiency ( silver alnicos ).

I especially liked Bruce's phrase "what sounds good is good".....it exemplifies that, despite all our instruments, what makes an amp great is our brains, ear, taste, experience. If Bruce didn't have the ear, taste, brains and experience he has his amps would not be great amps, because, even if you try to take everything under control, there's always something that could be eluding you....that's exactly where brains, ear, taste, experience come to play. Instruments can ( and do ) help, but the heart of the matter lies elsewhere...

The only point I still disagree is about the supposedly higher output power you state a speaker should yield....apart from the back EMF, the losses in the coil gap and the coil wire, together with the speaker's partially inductive behavior always result in some power/efficiency loss, and, although you don't see this on the scope, the power being converted to sound is lesser. This is, of course, of little concern, since halving the power only lowers volume by a sheer -3dB.

I want to thank you all, I really enjoyed this thread!

Best regards

Bob

[ETR]

Hi again ETR,
I pretty much agree with what you said, that was exactly why I used the "apples to apples" phrase of yours, we obviously need to lower the variables' number to get better comparisons, OTOH, ( Thanks Bruce ) sounding good ( or loud ) is something that involves other factors..... the amp I love the most ( my 1964 AC30TB ) though a sheer 36 W ( fully cranked ) is on par with 50 Watters in terms of loudness, probably thanks to its higher harmonic content and better speakers' efficiency ( silver alnicos ).

I especially liked Bruce's phrase "what sounds good is good".....it exemplifies that, despite all our instruments, what makes an amp great is our brains, ear, taste, experience. If Bruce didn't have the ear, taste, brains and experience he has his amps would not be great amps, because, even if you try to take everything under control, there's always something that could be eluding you....that's exactly where brains, ear, taste, experience come to play. Instruments can ( and do ) help, but the heart of the matter lies elsewhere...

The only point I still disagree is about the supposedly higher output power you state a speaker should yield....apart from the back EMF, the losses in the coil gap and the coil wire, together with the speaker's partially inductive behavior always result in some power/efficiency loss, and, although you don't see this on the scope, the power being converted to sound is lesser. This is, of course, of little concern, since halving the power only lowers volume by a sheer -3dB.

I want to thank you all, I really enjoyed this thread!

Best regards

Bob

If you think about the ignition coil in a car (inductor) it may help you to imagine how an inductor can actually store and release energy in a burst. In that case enough to take the 12V car supply and make it thousands of volts to fire the spark plugs. The same type of thing happens in a speaker which is why you see the big voltage spike far above the voltage you get with a resisistive load attached. So if you include that spike on the edges of the waveform, which can be almost the voltage as with a resistive load and do the math for that instant you have many times the power.

The losses in the coil and gap are what I was talking about earlier with speaker efficiency and have nothing to do with the power being delivered to the load. Like I stated previously that should be figured in if you want to know about the db output in the room but it is out of the power equation as long as your impedance is matched.


I have played thru an AC30TB/6 with blues and it was every bit as loud if not louder than any 85 watt Twin I have played thru. That's mostly due to speaker efficiency. The speakers in those amps are more like 5-6db more efficient than Oxfarts.

[bob p]

Paging Chuck...

chuck, I tried PM'ing you a couple of times about this but didn't get any response, so I guess I'll ask the questions here:

A. Cathode voltage questions.

With the circuit that you've described in Post # 19, what kind of cathode voltage change would you end up seeing if you didn't have any of the zeners in place? I'd really like to know how much of a cathode voltage change your circuit would see when driving the amp at full tilt without any zeners. We know that at a starting point your idle voltage on the 130R cathode resistor is 9.8V. How high does it climb if you drive it to max output?

B. Pi Drive Questions.

I'd also like to ask how much of a max voltage swing your phase inverter is putting on the grids of the output tubes.

C. Traces.

Have you ever captured the scope tracings that resulted from your zener mod experiments? If so I'd really like to see them.

thanks.

[Chuck H]

Well... I don't have that amp on the bench now. I'm working from memory. I seem to remember that the voltage on the 130R cathode would rise to about 20V!!! That's cold as ice! And, of course, crossover distortion was the result.

I didn't measure the actual swing from the PI only because WRT EL84's it was more than ample. And probably always would be. The input voltage to the PI was about 7VAC max IIRC.

I never captured any scope traces from the work. But I can say that in MOST amps that drive EL84's hard the crossover distortion is heinous! The hot bias and zener limiting on the voltage rise really does give a good representation of the classic cathode bias EL84 sound without the buzzy drawbacks evident with higher plate voltages and cooler bias settings when the amp is overdriven.

If I ever get the amp back on the bench I'll try to capture more details for you. But I can honestly say that if you ever put an EL84 amp with over 350Vp on your bench, you won't need any feedback from me. All will become clear in short order.

EDIT: I did have some PI voltage specs once upon a time but I Lost them in a hard drive failure. Going by memory, the PI node had about 310V at the rail and used an 820R bias resistor and a 47k tail plus a 1k feedback shunt. Feedback ratio was 22/1 off the 8R tap. Other PI values being typical (1M grid resistors, 100k plates). The OT is 8K for a pair of EL84's but also has a 10kR through a 1500pf cap as a shunt filter on the primary side. You can reverse engineer this if you like!?!

EDIT2: Also... It was recently brought to my attention that since diodes typically fail short rather than open a fuse might be a good thing to include in the circuit to avoid having the EL84 cathodes direct to ground with no grid bias in the event of a diode failure..

[bob p]

Well... I don't have that amp on the bench now.

yeah, i caught you off guard by resurrecting an old thread.

the answer to my question would be self-evident if i could put the amp on the bench, but all i have right now are a bunch of schematics and a plan for my next build... I'm normally not a fan of EL84 amps, but I thought that looking into the de-buzzing zener thing might change my mind.

hearing you say that the cathode voltage could rise up to 20V under load surprised me. i was reading Aiken's page on the AC30 where he referenced Pittman's book, saying that the voltages on an AC30 were "12.5V at 30W, 10V quiescent". Then I looked up the 6bq5 data sheet, looked up the anode+screen current swing from zero-signal to max-signal for a pair of cathode biased AB1, and calculated the anticipated voltage change on the cathode resistor. with a current swing from 80mA / 10V at idle to 114mA max for a pair of tubes on a 130R resistor, i was thinking that the voltage would rise would be a lot less than 20V. Aiken's page made it sound like 12.5V and my calculations were a bit higher, but they were still shy of 20V.

All these questions relate to trying to wrap my mind around the way that your cathode zener mod works and choosing component values.

[Chuck H]

Choosing the zener value is easy. Wherever the cathode voltage is, value the zener one volt over where the amp starts to clip. Any less and you risk actually hearing the diode clipping. All the diode is doing is locking the voltage rise (keeping the bias hotter and eliminating crossover distortion) and reducing cathode impedance (which prevents grid loading and eliminates crossover distortion). There are precious few EL84 amps that sound good overdriven. It's a fine balance of OT deficiency, speaker damping and overall frequency response. The cool "swirl" that happens with big bottles sounds more like fizz (or BWIZZZ) with small bottles. Even 6V6's seem to sound more hashy and hissy in the top end when overdriven, right? The blunt force of the cathode zener and the shunt filter seems to make the EL84's behave. The amp wasn't overly buzzy or hashy, but at least typically so. I wanted better. The Paul Ruby zener mod, the bias rise block zener and the shunt filter all work together to make the EL84's sound more like big bottles. But they still don't act like big bottles. A little more dynamic feel IMHE. Really nice. My amp only uses a pair wired like this. I've built two 4XEL84 amps for customers and they're no sweeter IMHO.

[barfoden]

hhmm great site..

I am a bit concerned about my new amp.. It has two EL84 in a unique series push pull configuration (no phase inverter tube) with a 91 ohm 5 watt bias resistor that measures ~9.3V. It has no bypass capacitor on the EL84 tubes and a single 12ax7 which are setup as two gain stages on series bypassed with 10µf caps for full range with a bass selecter pot between the gain stages and a single passive tone pot after the second gain stage before the EL84 tubes to roll of the top end if the amp is to harsh.

Here is my worries:::

9.3V / 91 ohm = 102mA combined current on the EL84 tubes. ~ 51mA for one tube. Assume the screen current is drawing 10% of plate current at idle, this will give a plate current at idle of ~ 46mA. Measured plate voltage is ~328V (plate volt - cathode voltage is 318V).

318V multiplied with 46mA = 14.63 watt at idle.. Wow that is hot for a 12W tube.

The power transformer (made in china) is not particular small but do get hot after 3-4 hours of operation. Besides the ~105mA total current drawn from the tubes, some power is lost in the 5W 91 ohm bias resistor, 5 watt 200 ohm sag resistor in the power supply and 3 watt 4.7K screen dropping resistor. I think i calculated around a total draw of 140mA from the voltage drop over the 200 ohm sag resistor. Seems a lot for an amp with 1x 12ax7 and 2x EL84 even though the amp is a hot cathode biased one. Would feel better if i could lower the plate dissipation so the EL84 tubes are around 12 watt plate at idle.

[Chuck H]

This should really be a separate post. Taking a new issue on the end of a finished thread doesn't usually get the same attention as starting a new thread.

Also, it would really help to have a schematic of your design for clarity. That is a general consensus here WRT unusual designs. A schematic is best, a layout is less good and a verbal description is least good.

Increasing the grids negativity with a cathode biased amp is usually done by increasing cathode resistance. You seem concerned about the additional resistance in the current path leading to additional sag. So you can always decrease the sag resistor proportional to the increase in cathode resistance. Obviously this will increase plate voltage but I think you have some room for that at 318V and your overall bias condition should still be more idealized. Your other option would be to add a bias supply to add -VDC to the grids. This wouldn't change the system resistance at all but does get into the combination bias thing that seems to have garnered some bad juju in the amp building community because of the old Fender lore. So that's entirely subjective to your beliefs.

[barfoden]

Sorry Chuck..

This is a kustom defender 15h I bought for cheap dough.. I have not received the schematic yet.. Hope to obtain it later on..

It is designed by James Brown who have designed some pretty famous amps in his life time.

Just a very interesting setup with the series push pull EL84 that eliminate the need for a phase inverter.. This unsual setup could be the reason the bias resistor is of such "low" value (91 ohm)..

I think I have seen an old Mullard design with this setup..
I guess I could minimize the sag resistor value and get some less power supply drag, but then I guess I really have to increase the bias resistor value..

I think I will buy a spare set of EL84 and request for a pair that would bias a little cooler in a cathode biased amp.. I had a pair of JJ EL84 and they

actually increased the bias voltage a little and was biased just slightly warmer than stock tubes.. I guess I will just play the stock tubes and see how long they last before showing a lot of wear..

[Chuck H]

Just a very interesting setup with the series push pull EL84 that eliminate the need for a phase inverter..

This is where it gets confusing. You can't wire power tubes in series, so what does this mean? The verbal description isn't doing it for me and I can't find a schematic on line. Perhaps you could hand draw what's going on? This question is essentially reduced to "I have an unusual el84 power amp. How do I bias it?" Which is very close to something like "I had a great soup at a restaurant last night. What is the recipe?"

[soundguruman]

First... Merry Christmas, or Happy Holiday.

With 350 Vp and the screens at 335 using a cathode bias, how hot can I safely bias my el84s? I know the rules (75% yada, yada) and max diss and all. But how hot can I REALLY bias them?

I ask because I'm trying to kill crossover distortion. I am driving the piss out of the tubes and I am using the zener trick already. I don't want to run fixed bias if I can help it.

I don't need to squash the crossover distortion completely. In fact I think a really miniscule amount is OK. But I do want to try reducing it a little more.

Any ideas?

TIA

Chuck

you look in a dark room...
see if the plates are turning a little bit red.
If they are a little red, back it off a hair, until the red just goes away.
Now, that's hot enough.

[barfoden]

Hi Chuck.. Not sure that they will release the schematic.

But here is the words from the designer himself

"It's not a parallel single ended amp, it's a series push-pull design, which is fairly rare out there. Instead of using an inverter to put the out of phase signal on the 2nd output tube, you just ground its input, and then remove the cathode resistor capacitor. That way the signal coming out of the previous stage gets split between the two 'stacked' output tubes, and the one on the bottom sees an inverted signal. If you look at each input tube's input grid with a scope you'll see inverted input signals.
It's a trick that only really works with EL34s/EL84s, since they have a lot higher transconductance than a 6L6".

Have a nice weekend ..

[Chuck H]

Well... The way it's described still doesn't make much sense to me. It could be a cathode coupled type of thing going on. Similar to how dual triode tubes have been run in "self split" circuits. This still isn't putting the tubes in series with each other but I think the designer is strictly speaking about the drive signal being derived from one tube into the other. Ok, fine. If I'm right what this means to you is that the cathode resistance is indeed critical to accurate coupling and that makes the bias non adjustable for the most part. You might be able to increase the cathodes positive DC with a larger resistor and then check the AC with a bypass cap in series with a resistor. Let's start by assuming you'll need a 120R cathode resistor. You would then bypass that with 100uf cap in series with a 470R resistor. Why 470R??? Because 470 parallel to 120 would represent about 95R to the AC component and should keep your coupling about the same while still cooling the bias down. To avoid additional sag you can still chip some value off the sag resistor. At that point the Vp will be increasing to where you'll need to re juggle the ratios on those three resistors again!!! But some experimenting should get you there. The only other rub I can think of is that cooling the bias will mean the power tubes go into cutoff sooner. Since one tube is feeding signal to the other this might cause audible problems and could explain the silly hot bias.

Adding -V grid bias would be impossible if the driven tubes grid is to remain grounded. But the designer mentions looking at the grid signals with a scope, I'm not sure how you do that with a grounded grid so I'm guessing he doesn't actually have it grounded, but rather almost grounded. Still... Generating significant biasing voltage over a small resistance may require more current than you can safely source from the HV rail. And I suppose the grid of the other power tube will have a significantly different load requiring a second bias supply so the two grids may be balanced. I dunno. No schem to look at.

EDIT: One more niggle... Increasing Vp is going to change the coupled signal and therefor the ideal coupling resistance. The whole shebang get's into more associative math than I know I do think it can be fudged and guessed at pretty well and a happier balance can be met even without exact calculations though.

[Chuck H]

That's sort of where I ended up. The down side is that you go through those skinny little bottles pretty fast this way. Not a huge problem since watt for watt they're really quite reasonably priced I'm biased at 11.5W per plate at idle, but since I run the amp balls to the wall most of the time they're operating very hot AB1. Just the same as AC30's are biased class A but are usually operating in AB1. This seems to be a happy place for EL84's.

[g1]

Well... The way it's described still doesn't make much sense to me...
the designer mentions looking at the grid signals with a scope, I'm not sure how you do that with a grounded grid so I'm guessing he doesn't actually have it grounded, but rather almost grounded.

I was thinking the same Chuck, but then I saw he didn't say looking at the grid, but looking at the input. Maybe he meant the cathode? (edit: oops, he did say grid)
Here's a schematic that may be similar to what he means:

Edit: if you call it "self inverting push pull amp" google will come up with much more info.