el34 screen resistors wattage. why so big?

[J M Fahey]

"Typical" guitar amps use the tubes very conservatively: 50-60W from a pair of 6L6s or EL-34s is loafing for them.

Iīd love to see a pair 6L6 provide clean 60W (just before clipping).

An EL-34 pair can provide 100Wrms with Va=800V and 11k-aa.

Philips Eindhoven used to claim that in big bold title letters, but then in smaller print claimed 90W tops.
And we are talking a harsh, almost Class B amp, suitable only as PA in a busy train station and similar duty; no guitar player would use it.

Similarly, 6L6GCs can provide 100Wrms at 3k5-aa although 130W might be possible into just under 3k-aa

maybe in your dreams

, but we need to know the true cathode emission capability of this tube.

Thatīs the point, 6L6 is severely current limited and does not have what it takes to get such power.
Closest is Music Man which feeds them 650/700V on plates, biases them extremely cold, applies +24V to grids at any current they wish and slams them with cathode drive, none of that is available on conventional guitar amps, ... and yet it barely reaches 65W per pair.
With conventional AB biasing, no positive grid current available , just voltage drive?
Forget it.

If you use the same power supply and try different OTs there will be an optimum value _for that supply_. where maximum power from the supply is converted into audio power.

Yes, no doubt about it, you will always have an optimumvalue.
Now that it translates into 50/60 W per tube pair is something else.
Still talking about 6L6/EL34 of course.

It is best to look beyond what is done in guitar amps if you want to know what can be done with any given tube type.

Personally consider Guitar amps the ones extracting maximum power, if nothing else because they will *use* it, clipping is not only acceptable but desirable, non linearity and distortion is acceptable, etc.
None of that is acceptable in Hi Fi, not even in normal PA unless we are talking a noisy crowded train station as I mentioned, but if you have examples contradicting this, you are welcome to show them here.

Personally, *only* duty where I see such tubes providing higher power would be in Radio Amateur transmitters, Medical Diathermy apparatus (or electrical scalpels) or , say, plastic pouch sealing machines where specially shaped brass dies selectively "sew" thinfoil PVC sheets.
But none of that is Audio duty.

[KevinOConnor]

Hi Guys

Mike, in the two examples I gave the minimum voltage across the tube is taken into account since it is often the intent of a designer to get the most output from a given supply voltage. In each case it would be assumed that Vs is about equal to Va as in most guitar amps.In this context, one would set each supply voltage so that the expected peak plate current occurs at the lowest Vsat for the tube _at that current_, which makes the rest of what I said holds true

If you decide to have an uneven comparison then you can get whatever result you desire. Maybe. So, using the same screen voltage and taking no account of Vsat in the choice of Va, the 6k6 load will pull slightly closer to ground to deliver whatever the maximum plate current is with the unknown supply. This current will be lower than for the 3k3 load in the same conditions, so actually it works out just as in the controlled and specified example. The linear screen current is also in the same proportion as before - lower for the high-z and higher for the low-z. How hard the tube is driven into nonlinearity will determine how high Ig2 goes, but it will stay in proportion.

Vsat is the term I use in my books for the voltage across the tube when it conducts the most current, same as for semis. Obviously this voltage will be slightly lower for the 6k6 amp than for the 3k3 amp since the current is higher in the low-z case and the tube's Vsat is a bit higher.

Sort of related but maybe not, yet you might find it interesting: I've been working on some high-power tube amps recently. I found a surprising proportionality between Rg2 and Ig2 when the tube is near and above Vgk=0. There was also a happy resistance to grid current when a cathode winding was used and a lot more power is attained with this winding, too (this was always part of the design but it was tested with and without it). The CFB gives a low output impedance even before any other feedback is applied.

Another interesting discovery was a "natural compression" which occurs in UL connected outputs. The gain is not linear with level, which explains why this sounds so good in a guitar amp without a global feedback loop. All y MI amps use such an output, except for the Studio and KC-25 which have switchable output stage wiring and variable feedback.

Have fun

[KevinOConnor]

Hey Guys

Yes, Justin, a big typo - was supposed to be 470V, something typical.

You need a bad-ass tube for a 4kV supply. I used a pair of 4CX250s at 2kV to get 700W and that was plenty scary - lethal if you slipped up. So now I stick with <800V.

Have fun

[Mike Sulzer]

Hi Guys

Mike, in the two examples I gave the minimum voltage across the tube is taken into account since it is often the intent of a designer to get the most output from a given supply voltage. In each case it would be assumed that Vs is about equal to Va as in most guitar amps.In this context, one would set each supply voltage so that the expected peak plate current occurs at the lowest Vsat for the tube _at that current_, which makes the rest of what I said holds true

I think if you keep Vp and Vgs equal for both values of Vp, then your 10% rule is out the window. In the higher Vp case, you bias for lower Ip, but Igs will rise, not fall. If you keep Rs the same, the screen dissipation will be higher. The screen might even melt on the first power chord.

[KevinOConnor]

Hi Guys

Mike: That's funny.

Juan: You need to get out more. You can put a pair of 6L6GCs into a YBA-1A and get 80Wrms clean (visually clean on a scope). Hifi guys do this as well in other circuits for lower distortion. Traynor and Garnet routinely got 80-85W from a pair of 6L6s or 6CA7s.Both tubes can produce more power but the application here is for a portable amplifier that will be mechanically abused. Pete used a shaker table to check the integrity of the tubes at various voltages and power outputs and settled for a modest reduction from maximum.

You are correct that many guitar amps do push the limit of what a tube can do, as in the Fender PS-series and their 300W bass amps. The latter abuse the grids with too-high circuit resistance, risking failures by means of inadequate circuit values. Countering that, the screens are fused and grid drive is clamped to 700mV positive. ENGL's Savage amp is rated at 120W from a pair of 6550s, which can readily produce 150W.

A point often overlooked is that MI copied hifi and continued to generally live within hifi rules as far as how they designed things. Tube safety rules are broader and allow a lot more power for one thing, and allow many other options for tone creation for another.

OP: One issue not addressed is the physical location of the screen resistors and how that impacts their size. Most of these Rs are mounted on the tube socket. In an upside-down amp the heat from the tube is heating this resistor even before current passes through it, so there are two heating effects to accommodate: internal and external heating. You measure the temp of the outside of the R using a laser thermometer and then see how far away from its limits the part is. over time, some wire-wounds run so hot that the lettering is burned off but the part still measures and functions fine. In any case, when there is high ambient temperatures, many parts need to be rated a bit higher than just for what they need to do.

Another example is fuses. These should be used at 80% of their rating since the current through them "primes" them by heating them slightly. This will cause the fuse to blow prematurely.

Have fun

[Mike Sulzer]

Hi Guys

Mike: That's funny.

Not quite sure what you think is funny, but it appears that you have no idea how a pentode works. You need to look at and understand the curves (for example here: https://frank.pocnet.net/sheets/129/e/EL34.pdf). Over much of the useful range of the control grid, the screen current increases by several times, as much as a factor of five, when the screen voltage increases from 250V to 400V. See the first plot in the link. The dissipation rises by a corresponding amount.

Edit: I took a look at a schematic for a YBA-1. I did not find any voltages on it, but from the ratings on the power supply caps you can tell that the screen voltage runs at about half of the plate voltage as I have been saying it must if you do not want to burn the screens. (Yes, the plate supply uses two 450 volt rated caps in series, despite your disapproval.)

These high power amps are class B, or near to it. Everyone knows that you can do this. What is your point?

[KevinOConnor]

Hi Guys

Mike, I find it funny that you think the tube will melt in one case and not the other. I don't know what you are assuming that would cause the melt-down in the first place - maybe that it is a Marshall amp? My example was to show that the assumption of "more power is possible with the 6k6 than a lower primary-z" was incorrect. For such a demonstration, things need to be kept simple so as not to confuse the issue. You seemed to want to confuse the issue by bringing in aspects that in fact do not change the outcome of what I stated, so I must assume that you simply like to argue?

The Ia/Ig2 ratio isn't as linear as I stated but for the purpose of the demonstration about output power, it does not matter. Neither does it matter if you want to discuss the screen resistors in each case of the example. In neither case will the tubes melt down nor will the grid-stops be burned up. The currents are lower all the way around for the high-z primary case.

Your assumptions about the YBA-1A are incorrect. Va and Vs are both about 560V in that amp. The tubes idle within a safe range towards the hot end (about 20W each) and the output power is over 80Wrms.Tube life in these amps is extremely long with original factory tubes still providing full output after fifty years of service. Note that all tubes can last this long if not mechanically upset.

Have you read RCA's Class-B Design procedure? It is all as one would expect until the last step, where the idle current is adjusted to be one-third of thetube's plate rating. That is hardly class-B ! Neither is the YBA.

I understand Costa Rica is pretty hot. maybe you should go into the shade for a while? Have a cool drink. Go for a swim.

Have fun

[nsubulysses]

Well I am not sure why i'm even posting this. Probably because you're all a little wrong which is worth noting. I won't get into the screen grid or primary impedance debate. Here are some finding that I've pulled from my invoices.

Juan, there are Musicmans that do 100W with two 6L6s. Musicman 2100-B does this. THey are in the grounded grid arrangement like you say and they are cathode driven with a JE1692. So yeah, it can happen, but not in a typical arrangement in this example.

Kevin, Mike, some YBAs do about 75-80W at clipping with high plate and screen voltage but there are others that only do 40 of 45W, depending on what year they are and what transformers they use. YBA-1As from my record of invoices do about 75W at clipping with about a 560V plate and screen voltage as you say (EL34 example). YBA-1s vary, I have seen some do 40W with 430V plate voltage (EL34) and some do 60-65W at clipping with plate voltage 515V (KT88). You two are arguing YBA-1As vs YBA-1s and they are not the same.

VHT Sig X runs two 6550s at 555V plate and screen for 90W at clipping with 35W 6550s. If you use a 42W 6550 or KT88 it will do about 100W at clipping.

Sound City 200 Plus does about 210W at clipping with four KT88, however plate voltage is about 735V and screen at 350-450V. Hiwatt DR405 has similar plate and screen voltage and clips at about 340W with six KT88.

I would love to see a schemeatic of the Engl Savage amp. I'm sure I probably never will. What does it mean it can readily produce 150WRMS but it's rated for 120W ?? You can clip the shit out of it and get more power, or what do you mean? Well, any 100W amp can produce more power, readily, if you push it way beyond clean. Is this what you mean?

[Mike Sulzer]

Your assumptions about the YBA-1A are incorrect. Va and Vs are both about 560V in that amp. The tubes idle within a safe range towards the hot end (about 20W each) and the output power is over 80Wrms.Tube life in these amps is extremely long with original factory tubes still providing full output after fifty years of service. Note that all tubes can last this long if not mechanically upset.

If va only about 560V, then it is not a really high power EL34 push pull circuit, which would use about 800V.

But in any case, the evidence I have (and we have none from you) is that the screen voltage is less than 450V, not 560V.

My example was to show that the assumption of "more power is possible with the 6k6 than a lower primary-z" was incorrect.

Yes, that was your point. In fact you wrote:

Similarly, 6L6GCs can provide 100Wrms at 3k5-aa although 130W might be possible into just under 3k-aa, but we need to know the true cathode emission capability of this tube.

Please provide a schematic and actual measurements. Otherwise, I do not believe it. Your claim is that you lower the transformer impedance to get more power. Everyone else says that you raise both the impedance and the plate voltage. The tube can only provide so much current (shown the VI characteristic curves); in order not to demand more than it can give, you have to raise the impedance as you raise the plate voltage for the purpose of getting more power.

I do not live in Costa Rico, and I am not vacationing there.

[nickb]

Hi Guys.
i'm working on a trainwreck express. it is a really nice amp, but i have a few questions:

1) the common screen resistor is 1k 5w, but given that the screen current is very small (like 10mA per tube at 400v) why we use such a big resistor?
i mean, a 2w should be plenty enough.

Using the a 6.6K OPT and HT=450V and driven flat out the peak screen current is about 105mA, or more usefully 47mA rms. So the screen grid resistor dissipation is 2.4W. Therefore a 5W should be used although keep in mind you would never drive that hard under normal operation so you have the option to chose a lower value to act as fuse. Your build - your choice.

2) the trainwreck uses a BIG 25w resistor instead of a choke. even there a 1k 5w resistor should be enough. i don't get why.

Taking the preamp current as 5mA I come up with an average of 55mA through that common resistor when the amp is driven continuously- I am assuming that node is bypassed by a 50uF cap. That is about 2.5W. So again using a 2X rule of thumb a 5W should be enough.

Other things matter - how well ventilated is it? Is it close to other hot components? If you have PVC wire close you might opt for bigger to get a lower surface temperature....

I took the data from simulation based on an EL34 tube model that accurately calculates the screen current.

PS: How does it sound?

[KevinOConnor]

Hi Guys

Mike: You don't seem to know anything about Traynor amps so here is the deal> The YBA-1A rated at 80W uses a 4k-aa OT. Vpk=400V at 400mApk, so easily done within the 560V supply, which does fall under load. At any full-load voltage of 529V or less, the tubes are within their rating, and this only represents about 5.5% regulation, which is better than the circuit needs, so actual drop is likely a bit more. It's been quite a few years since I've measured one, but it works well and makes its rated power.

The YBA-1 is onyl 40W rated, has much smaller iron and no fan. The tubes are loafing.

It was never my assetion that a lower-z OT yields higher power. It is just that for the limited info about the 6L6GC, it can do 100W at 3k5.

Your knowledge about the EL-34 is lacking. The Mullard app for 100W at Va=800 and 11k-aa is but one way to get maximum output from this tube.

nsub: You seem to be confusing "applications" with "limtis" - they are nopt equivalent unless a given app works to the limits.

The Savage 120 is designed to produce 120W and it does. The 150W was with respect to what a pair of 6550s can do with other circuit values. The Savage is like any other 50W PA schematic except for the voltages around the output tubes and the use of a separate Vs rail. There is no CF between the splitter and the outputs, so output power is modest.

Similarly, a pair of KT-88s easily provides 150W of output in the right circuit - more in a superb circuit. The Soundcity amp is only designed for the power you stated and is not indicative of the tube capability just the design limit.

The MusicMan cascode output is a good way to interface opamps to power tubes and to use semiconductors to control the tube's idle condition. The idle current is not zero but it is very low, so tube life is great except for in the combos. The tubes could in fact be set to zero idle and there would be no visible notch on a scope. In fact, the OT can be wired backwards (end-for-end) and there will be no problem until about 75% of full output, at which point there is a bit of oscillation - experience of the wrong kind... hehe - but this demonstrates the power of the feedback loop when using an opamp for the front end. The bigger MMs produced only 130W with 4-tubes, the same as the UL Fender amps from a few years earlier. Those Fenders did not need UL connections to achieve 130W, and could easily have made that output in tetrode/pentode mode.

As far as the topic of the thread goes: Most guitar amps use the tubes conservatively with respect to the capability of the tube power wise. The envrionment the tube lives in is mostly unknown and is sometimes compromised by design - upside-down, in a combo - so screen resistor values have to be higher than is ideal from an engineering standpoint. Ohm's Law predicts that the higher Rs will have to be a higher power rating as it will necessarily dissipate more power than if it were a lower value. The harsh environment and heating drives the power rating for the part higher. if you want to clip the amp all day then the rating needs to be still higher. Fortunately 5W WWs are only 45-cets, so it is inexpensive peace of mind.

Have fun

[Mike Sulzer]

Hi Guys

Mike: You don't seem to know anything about Traynor amps so here is the deal>

True enough, but I think it is more important to understand how tubes work. Your original point was that if you have two designs that have the same power output but use different impedance transformers, and thus run different currents, then if you use the same value screen resistor, the screen resistor in the higher voltage case dissipates less power. Yeah, so what? The screen dissipates more power in the high voltage case because you would have to scale up the screen resistor in order to keep it the same. Carry this too far without scaling it up and you burn the screen.

Also, I think we all tend to be a casual with the term screen voltage, meaning either the actual voltage screen to cathode, or the voltage of the screen supply, that is, the voltage on the other side of the screen resistor. Sv. says never let the actual screen voltage of its EL34s be greater than 510 volts, period. And some calculations from the curves show that it if it is much greater than 400V in the high current part of the cycle, that is, when the control grid voltage is near zero, then you are in real danger of going well over the 8.5W average power limit.

It is obvious that if you operate up near the quiescent plate voltage limit of 825V, you need to use a much lower screen voltage supply.

Your knowledge about the EL-34 is lacking. The Mullard app for 100W at Va=800 and 11k-aa is but one way to get maximum output from this tube.

Good, let's se this quite different operating set up that gives 100W.

[nickb]

Hey Guys

Yes, Justin, a big typo - was supposed to be 470V, something typical.

You need a bad-ass tube for a 4kV supply. I used a pair of 4CX250s at 2kV to get 700W and that was plenty scary - lethal if you slipped up. So now I stick with <800V.

Have fun

If the OP will indulge me for a moment.

You must start a new thread and tell us all about the 2KV project, warts 'n' all. I think it would be a very unusual and and interesting read. When fun meets scary...

[KevinOConnor]

Hi Guys

Mike: You mentioned Cost Rica in another post so I thought you lived there. In any event, some of your responses sounded like you were having a heat stroke.

Your said about the YBA-1A:
"But in any case, the evidence I have (and we have none from you) is that the screen voltage is less than 450V, not 560V."

You must have a very bad copy of the YBA-1A schematic if you think Vs at the tube will be 450V. It REALLY is at about the same voltage as the plate.

You write about 6L6GC output:
"Please provide a schematic and actual measurements. Otherwise, I do not believe it."

See "The 400" chapter in TUT6), where maximum output from various tubes is explored (pgs.5-14 and 5-15 for 6L6GC. Are you one who believes that the plate power rating of a tube is what its audio output power can be? If so, you are mistaken. That rating is simply how much waste heat the tube can dissipate, not how much power it can CONTROL.

nick: When you do your sim are you using a high-z or low-z signal source for the tube grid? Most amps are driven by a high-z source and the extreme of grid-1 &2 current in real and typical guitar amps are not so high despite their nonlinearity. In any case, it is better to assume a worst-case and spec a part with conservative use.

[nickb]

nick: When you do your sim are you using a high-z or low-z signal source for the tube grid? Most amps are driven by a high-z source and the extreme of grid-1 &2 current in real and typical guitar amps are not so high despite their nonlinearity. In any case, it is better to assume a worst-case and spec a part with conservative use.

I used a 40K ohm source impedance.

Using the numbers you provided (thx for those) on the YBA-1A, I see that at 80W out the EL34s are running a little outside of their datasheet limits. Hardly a new phenomenon in this application area I'm sure that practically speaking it is of no consequence since continuous sine waves into pure resistive loads are not terribly enjoyable, musically speaking.