• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

What more do I need to know?

Hi,

What more do I need to know to compute the plate dissipation when I know the cathode voltage and the plate voltage? Is it simply P=VI?

As an example, here's the output section of an EICO HF87, notice the cathode to be at 35V and the plate at 430V. The bias current should be around 32/235=0.074A per EL34 right? Is there a way of computing the plate dissipation per tube?

How low a current can EL34, 6L6GC, 6550 go anyway?

Hf87s.gif
 

SY

diyAudio Moderator Emeritus
2002-10-24 10:19 pm
Chicagoland
www.SYclotron.com
If you measure at the cathode, the current is actually plate plus screen grid. Looking at some data sheets, you should be able to estimate the screen current and subtract it off, but the basic Ohm's Power Law still applies.

For the Eico (I have one sitting a few feet from me), the plate plus screen dissipation is (450-35)x0.074 = 30W per tube. The tubes run pretty warm, but I would typically get 5 years out of a set of EL34s.

For the intrepid rebuilder, there's an unused 50V winding on the power transformer which will allow conversion to fixed bias and a lower idle current.
 
so it should be:

Power dissipation = (B+ - cathode voltage) X idle current

rather than:

Power dissipation = plate voltage X idle current.

the datasheet for the EL34 says it's PDmax is 25W, isn't it too much for the EL34 to be running at 30W? or maybe it's just ok?

what would be the consequence of a low idle current in a cathode bias? I am not yet ready to move into fixed bias for my notes are still incomplete at the moment.

Thanks SY!
 
HOLD THE HORSES !!!!!

I am a bit confused.....

OK...we have 35V at the Cathodes with a 235 Ohm resistor....
So thats 35/235 = 148.9mA .........
Assuming 10% Screen current just for fun.....
That leaves 134 mA...for the two tubes...
This leaves 67mA Plate current per tube.....
Plate voltage is 450-35 = 415V

So Plate dissipation would be 27.8 Watts.......

Keep in mind that this is at IDLE....... Since when calculating tube behavior in an amp...it is usually done at OPERATING VOLTAGE durring full power output with signal passing through the amp....In that case the B+ usually is a bit lower..figure in the 415v to 425v range in this case.... Numbers may be in more safer agreement then...
The Plate Load also need to be looked at...to see how much of the AC swing exceed the plate disipation....
Ultamately you take an average of the Plate Dissipation over the full SINE wave for single cycyle..

Chris
 
Tim, Enzo,

I did not forget that, the 74mA is for "one" EL34 since 35/235=0.1489mA is for two tubes. Could the EL34 be run at less than 30mA in cathode bias?


Another thing that bothers me is about the grounding with metal body filter caps, I usually ground my chassis at the inputs, but if I would ever use those upright metal capacitors like in the ST70 which has their grounds at the caps body, would I be messing up my grounding since the body of the caps is going to be attached to the chassis therefore making another ground path to the case?
 
Here is a suggestion I would do...

Put in 6550 tubes...or KT88 tubes...don't sweat the heater current, you should be fine...Get rid of the cathode resistor...

Next, lift the ground point in the middle of the 120K grid resistors..and then put it to a negative voltage of about -48V...
You can tap the B+ with a resistor and diode and divider to get this....

Chris
 
cerrem said:
Here is a suggestion I would do...

Put in 6550 tubes...or KT88 tubes...don't sweat the heater current, you should be fine...Get rid of the cathode resistor...

Next, lift the ground point in the middle of the 120K grid resistors..and then put it to a negative voltage of about -48V...
You can tap the B+ with a resistor and diode and divider to get this....

Chris


Hi Chris,

Your suggestion leads into changing the bias topology from cathode bias to fixed bias, maybe later I will venture into studying fixed bias but for the moment will still study cathode bias ouputs. :D

Thanks!
 
Sch3mat1c said:
You cought me skimming again :D

Still, 27W is only 8% over ratings... ratings shmatings! Where's Geek with his sig when you need him. ;)

Tim

No problem, we all tend to skim from time to time.
:D

Anyway, so exceeding the 25W PDmax in the datasheet is just OK? I hope the 8% may only shorten the tube life but no apparent change in the tube character right?

Shortening it's useful life is not much of a problem as long as the tubes are still in production. :D
 

SY

diyAudio Moderator Emeritus
2002-10-24 10:19 pm
Chicagoland
www.SYclotron.com
OK, I took a minute to actually look at the data sheet. That's always a good first step, RTFM and all that. Eico's operation is almost exactly what is specified on the Mullard sheet for UL, 6600 ohm p-p OPT, 43% taps, cathode bias. The current divides out to about 62.5mA for the plate, 10mA for the screen.

Eico made a pile of these amps and they have an admittedly well-deserved good reputation. Dissipation in the output stage would not have escaped the designer nor the writer of the Mullard data sheet.
 
One thing to keep in mind....
WHat happens at idle is not a big a deal....
The B+ will DROOP durring operation...and THATS the voltage that really counts.....
Back in day, when men where men and Tubes were Tubes... The experienced designers new damn well that Mullards were conservatively ratted in the data sheets...and for a little excessive idle dissipation it was not a big deal... Because the B+ was already calculated to droop durring operation.... The amp worked fine, so end of story...These Engineers never figured that 30 years later Chinese tubes would be used in these amps that couldn't hold up to the NOS quality....

Chris
 

PRR

Member
Paid Member
2003-06-12 7:04 pm
Maine USA
www.diyaudio.com
> What more do I need to know to compute the plate dissipation when I know the cathode voltage and the plate voltage? Is it simply P=VI?

Should be, if you know where to divide by 2 and remember to deduct Screen current.

> HF87, notice the cathode to be at 35V and the plate at 430V. The bias current should be around 32/235=0.074A per EL34 right?

I think you wrote "32/235" where you mean "35/235".

430V-35V= 395V Plate-Cathode.

Cathode current per tube is indeed apparently (1/2)*(35/235)= 0.075A per tube.

From that, we deduct screen current. 10% is a good guess if you are too lazy to look-up EL34 data. 0.075A-0.007,5A= 0.067,5A.

> Is there a way of computing the plate dissipation per tube?

0.067,5A*395V= 26.7 Watts per plate.

> I took a minute to actually look at the data sheet. That's always a good first step

That's cheating. It also shows Objectivist tendencies.

If you do actually look at the data-sheet you find this condition:
[IMGHTTPDEAD]http://headfonz.rutgers.edu/EL34.gif[/IMGHTTPDEAD]

430V supply, just like EICO. 470Ω per valve is 235V shared between both valves.

Mullard wants screen resistors not shown on EICO, but otherwise the DC bias is identical. Mullard says you really get 2*(62.5mA+10mA) total cathode current, implying 34V at the cathodes. 430V-34V= 396V plate-cathode. 396V*0.062,5A is 24.75 Watts per Plate.

Screens seem to be eating (440V-34V)*0.010A= 4.1 Watts each.

"Design Centre" ratings are 25W plate, 8W Screen. We be OK.

Note that Design Center ratings include some allowance for utility-power variations. Later it became clear that tube companies should not try to account for utility tolerance, and that some designs and designers needed to crank to the safe limits. Design Maximum ratings are typically 10%-20% higher. These are "Design" ratings for an amp that does not yet exist. Since your amp DOES exist and you can check what it is really doing, you should check against the Design Maximum rating and not panic if the actual amp runs even a little hotter (Design Max has some allowance for production tolerances).

As far as I know, the 25W rating on EL34 is very conservative. Guitarists often use them at power outputs that suggest idle dissipation much higher than that.

> Keep in mind that this is at IDLE....... Since when calculating tube behavior in an amp...it is usually done at OPERATING VOLTAGE durring full power output with signal passing through the amp....

A Full Power test may be required for the sales-catalog data. And there were Full-Power 24/7 amplifiers (Ampex VTR servo motor amps, Bogens used to drive film projectors at 59.9Hz to sync color TV rates). But a speech/music system will NEVER run full power for more than milliSeconds. You design speech/music amps to run full-power a minute at a time, an hour in their life, and 1%-10% of full-power for many-many thousands of hours. If I had an amp that idled OK and glowed RED at full power, I would be careful in testing but unconcerned about speech/music use even into clipping.

> The Plate Load also need to be looked at...to see how much of the AC swing exceed the plate disipation....
Ultamately you take an average of the Plate Dissipation over the full SINE wave for single cycyle..


You can, though you can also look at the complete-amp data. At full power of 34 Watts, Mullard says 385V Plate-Cathode, 140mA total plate current, 385V*0.140A= 54 Watts input, 34 Watts output, 54W-34W= 20 Watts lost in the plates, 10 Watts each. This amp runs significantly cooler at full-power sine test tone than at idle. (An exact derivation is complicated by the UltraLinear connection and some output power sourced by the Screens; you need to check G2 dissipation, even though Plate dissipation is often low.)

> The B+ will DROOP durring operation...and THATS the voltage that really counts.....

Disagree.

Hi-Fi amps spend 99% of their life at 1% of full power.
B+ should not droop (much) in a hi-fi amp under speech/music signal.
Plate dissipation was usually less at full-power because much of the input power was sent to the load.

Home hi-fi tube amps get cooked by idling. Full-power cooking is so rare that it hardly matters, and is usually a slower tube-cook.

> How low a current can EL34, 6L6GC, 6550 go anyway?

You "can" go as low as you like. Note the fixed-bias condition: 30mA each plate. You could even take G1 voltage down to -50V with a 5K cathode resistor which will put your idle current around 10mA per tube

But gain will be VERY low until you wave the grid 10 or 20 volts. The amp will be "weak" for soft sounds, and blast-out when levels get higher. In sand-state they call it "crossover distortion". Tubes don't do it as bad as naked BJTs, but they sure can be under-biased so far that normal-level reproduction is badly warped.

Audio Power is expensive. You got a 34 Watt amp because you will use 34 Watts. At that power in this scheme, one tube will peak at ~200mA and the other will be near cutoff (and vice-versa for the other half-cycle). Gain is high because of the high current, but low because only one tube is active. A first approximation to a "good" idle current would make the gain of two idle tubes similar to the gain of one tube at 200mA. We need to know the change of gain with current; assuming that Gm varies with square-root of current is close-enough for most tubes for a Gm range of 2:1. Therefore each tube must run at half gain, which is one-quarter current, or 50mA each at idle.

This result is simplified. The "optimum" is broad. However, higher current usually sound better than lower current. So 62mA instead of 50mA is perfectly reasonable.

There is another aspect. If you try to set idle current low with a large cathode resistor, your efforts are foiled when Full-Power forces large currents to flow no matter what the idle condition is. In this topology, a perfect amplifier would pull a supply current of about 0.7 times peak output current, or about 140mA. But if you were "cool" and used a 5K cathode resistor for low-low idle current, attempting to reach Full-Power implies a drop of 140mA*10K= 1,400 Volts! Can't happen. The amplifier would drive itself deep into cut-off as fast as it can charge the cathode-cap. A large cathode capacitor will delay this, and for speech/music use you could surely run say 500 ohms and 500uFd, get idle current down to 45mA per plate, and not have much bias-shift in speech/music duty. (You would fail the 34W steady-state test-tone test specs.) Or to minimize bias-shift between idle and full-power, you would want to idle at 140mA/2= 70mA per tube. (However, at this supply and load, 70mA is past the nominal plate dissipation.)

So: lowest THD around 50mA, but more sounds better. Lowest bias-shift around 70mA, but anything over 63mA is past the nominal design rating. In that light, we have "proved" that about 60mA-62mA is a good operating point for these conditions.

In any case: EICOs (most of the Classic Age amps) are not big tube killers. They do run HOT. Some eat caps and resistors, but after a few proactive repairs they are set for the rest of your life.

> EL34, 6L6GC, 6550

6L6 has slightly lower Mu, should get a slightly larger cathode resistor. And of course original-recipe 6L6 would be far past its ratings here. When you get into 6L6GC, 6550, all recent EL34, etc.... I suspect these are all the "same" parts in slightly different shapes. And because they have been used sorta-interchangeably in over-worked guitar amps, for self-defense the better recent tube makers build their "branded" tubes to a fairly high rating. They don't usually publish new ratings; they just beef-up the tube until they don't get too many complaints about early deaths (or not considering the price: cheap mystery-house tubes will always be with us).

> How low a current can EL34, 6L6GC, 6550 go anyway?

You don't buy big tubes and starve them. Or if you do, yet ask for the big power they promise, the crossover distortion kills you.

Look at 7189, kind of a re-rated 6BQ7. Bottom of page 2 shows it running AB1, fixed-bias, ~150mA peak, 105mA DC at that power, but only 15mA idle. Fisher used it once, I have one here, and spent many years playing with it. The 15mA idle condition is badly distorted. Feedback hides it somewhat, but 'scoping the driver shows it has to slap the grids hard to get any small signal. Assuming Ipk/4 per tube is a "good" idle suggests 38mA, but that is slightly over the Design Maximum rating. By ear: 15mA total (7.5mA per tube) is too low, 50mA per tube sounds better but glows in the dark (and you can't get genuine 7189 any more); besides which the whole box runs so hot it stinks. I think I usually ran 25mA-30mA as a compromise between too-low and too-hot.
 

PRR

Member
Paid Member
2003-06-12 7:04 pm
Maine USA
www.diyaudio.com
me> if you were "cool" and used a 5K cathode resistor for low-low idle current, attempting to reach Full-Power implies a drop of 140mA*10K= 1,400 Volts! Can't happen.

Should say:
if you were "cool" and used a 5K cathode resistor for low-low idle current, attempting to reach Full-Power implies a drop of 140mA*5K= 700 Volts!

Still "Can't happen."
 

anatech

Administrator
Paid Member
2004-06-06 8:31 pm
Georgetown, On
Hi JojoD818,
I have one of these sitting in front of me as well (like SY), I rebuilt it to to original spec. I am toying with the idea of finding another one. I really like the sound of these. Curious about putting them in parallel mono each to hear what 70W sounds like. They do run warm, but no worse than most and better than some for sure.

I am using Electroharmonix and I like them the best so far.

If you are worried about the grounding scheme, Eico grounds at the filter cap lug. You can always use the phenolic mounts for these caps, that would isolate them from the chassis. This is done for negative grounds, or caps in series for high supply voltages. Needless to say the cans are completely covered by heatshrink (it comes in clear) with a top disc.

PRR - always enjoy reading your posts. I either learn something or get a laugh.

-Chris