max power of EF86

[Leevi]

I'm using one EF86 as a preamp stage and would like to maximize its power.
The B+ is about 300V. Below the Matchless DC30 preamp using one EF86:

http://www.schematicheaven.com/newamps/matchless_dc30_old.pdf

Any ideas how to make it  even more powerful?

/Leevi

[tubesornothing]

power (i.e. current time voltage) or just voltage?  For voltage more B+ and a bigger plate resistor will do it.

[VMS]

Hi,

here is a pretty good data sheet that gives you lots of choices:

http://tubedata.milbert.com/sheets/010/e/EF86.pdf

[DummyLoad]

is more gain what you mean? or more voltage swing?

[RicharD]

I assume he is looking for an increase in gain.

Hey ISO,

Did we ever bother to document telemetry on this circuit?  I'm in the "have not" catagory.   

It'd be nice to know what the screen voltage is.  Mazda Belvu has 2 curves published for pentode mode, 1 with the screen at 180V, and the other with the screen at 100V.
http://www.mif.pg.gda.pl/homepages/frank/sheets/020/e/EF86.pdf
Given the parameters of the Matchless:
Ra = 330k, Rk = 2k2, B+ = 300V
With the screen set at 180V, the tube will not bias.  With the screen set at 100V, the tube will sorta bias but Vtube is only like <20V.  That can't be right.  I'm betting the screen is somewhere about 50 - 75V.  It'd be nice to know the following about the "classic" matchless circuit:
(DC) Vplate, Vscreen, Vcathode  (AC)  Vout/Vin

To increase gain, you want to make your load line more horizontal.  This is done by either increasing B+, increasing the value of the plate resistor, or better yet, a combo of the 2.  Then you usually need to jockey your cathode resistor to set the tube somewhere right in the middle between B+ and Vcutoff.  Of course with these pesky pentodes, your screen voltage comes into play.  Less screen voltage means you need less bias.

Personally, this is one of my favorite preamp circuits.  I think it's very optimal the way it is.

[DummyLoad]

yes, ckt has been documented. posting it for leevi - sent you email with all i have.

matchless pre-amp is nice. i'm running it @ ~~285V and it seems happy and sounds excellent.

 

[Leevi]

Thanks for the replies. Yes, I would like to get more gain from it, not
power as I wrongly stated. There are lot of old data sheets available, but
I wonder how reliable these 50 years old sheets are. E.g. you can exceed the max plate voltage.
That's why I'm looking for already implemented/tested circuits that are using today's standards.
/Leevi

[chocopower]

Thats the the way i did it... 

http://www.el34world.com/Forum/index.php?action=dlattach;topic=8349.0;attach=14016;image

2 diferent values in a swicht.

[RicharD]

Attached is a gustimated load line for for the Matchless EF86 stage using the measurement I took from a protoboard experiment.  It looks like a gain of about 150.  You're not gonna get much more out of it unless you drastically change the circuit.

[Leevi]

I'm using same resistor values as in the ISOTone's circuit but for instance
the plate voltage is only 80V instead of 141V (maybe because the B+ is 272V instead of 330V) On the screen there is 56V and on the
cathode 1.47V.
/Leevi

[RicharD]

In a perfect world, a perfectly biased tube will have a Va (plate voltage)  exactly 1/2 way between Vcutoff and B+ source.  Hard to achieve but not so hard to get close.  What's killing your circuit is the fact that your B+ is 60V (20%) low.  You're still biased for the same current so your Va is getting pulled south of center.  My plate curves are purely guesstimates but probably close enough for geetar amp work.  Working from your measurements, I am at least certain of the load line operating  and the -1.5V curve.  Really the only questionable result is the AV.

So.... can you increase your B+?  This really what you want to dew.  If you're stuck with this B+, you might try flowing a little less current by changing Rk to a 3k3.  That should get your Va up to the 130ish volt area.


Edit:
Typos in my graph.  point 2 = .83mA and point 4 = .67mA

[Leevi]

Thanks for the analysis Butterylicious. However I wonder the value number 2 (8,3ma).
The voltage drop over 330K resistor is 192V (272V-80V). According to Ohms' law the current
should be 192V/330K = 5,8 mA.

Could one solution be also that I decrease the plate resistor .g. from 330K to 100K?

/Leevi

[DummyLoad]

Ib = Vk/Rk = 1.47V/2200Ω  = 0.67mA or 670uA  - my Ti displays 0.000668182 - something tells me it's not 6.67mA 

respectfully,

--ISO 

[RicharD]

192/330k=0.58mA. You dropped a zero.  This made me realize I have a couple of typos.  I did the same thang twice.  2 should say .83mA and 4 should say .67mA  Point 2 is the 2nd coordinate when plotting a load line.  Your 1st coordinate is B+ @ zeroA.  Your 2nd coordinate is Ra/B+ marked at 0V.  Now you can graph your load line.  Technically Ra should be your plate resistor (330k) in parallel with your load pot (1M) in parallel with the input resistance of the following stage (probably >2M).  Ra should really be more like 248k.  This would put your 2nd coordinate closer to 1mA.  This also why my graph is slightly off.  We're counting on thumbs.  This is a guitar amp, it ain't gonna make that much difference.  The 3rd point is your plate voltage.  The 4th point is your cathode current.  Where these 2 points intersect with the load line is the operating point.  We know that measured to be roughly 1.5V.  From here is where the real guess work started since we don't have a curve sheet for Vg2=55V.  I looked at the 180V and the 100V G2 curves and guestimated where the +/- .5V curves should be.  Where your load line intersects these 2 points represents an input swing of 1V.  You graph straight down to find out your Vout.  Vout/Vin=AV  90/1 = 90 (ok... 3 typos).  

So if you change your plate resistor to a 100k, what happens?  272/100k = 2.72mA  1st, your load line got more vertical.  You're losing gain.  The more horizontal you make your load, the more gain you will get... up to a point.  Of course now your screen voltage has changed and so has your bias point.

Them Matchless guys fingered it out.  You need more B+.  What rectifier are you using?

[Leevi]

Exactly it is really 0.58mA. I didn't check the zeros since I trusted the result you had counted.
So there is not so much to do in my case. I can increase the B+ maybe 10V by changing
the resistors but it will probably not help that much. I'm using EZ81 rectifier and the amp is an SE amp with two EL84s.
/Leevi

[tubeswell]

Thanks for the analysis Butterylicious. However I wonder the value number 2 (8,3ma).
The voltage drop over 330K resistor is 192V (272V-80V). According to Ohms' law the current
should be 192V/330K = 5,8 mA.

Could one solution be also that I decrease the plate resistor .g. from 330K to 100K?

/Leevi

Get the screen voltage to 120-140 and try a 47k plate resistor. With HT = 280V this should get you about just under 3mA idle current in the tube of which about .6mA will be screen current. With a 680R cathode resistor the cathode voltage should be about -2V. This should get the load line hitting the Vg0 knee and the bias point roughly in the middle  = heaps of gain (if that's what you're after

[Merlin]

Do you want more output swing or more gain?
If you just want gain then there is no need to increase the supply voltage, just increase the anode resistor to whatever you feel like, and maybe increase the screen voltage by reducing the screen resistor. Even the classic Vox EF86 input has a gain of 200, so you could copy those values.

[PRR]

> I wonder how reliable these 50 years old sheets are.

Those guys knew more about their tubes than we EVER will.

> E.g. you can exceed the max plate voltage.

Sure. And get shorter life. For most old-time uses of tubes, frequent replacement was annoying, expensive, or dangerous. Grandma couldn't change her own tubes. A airplane pilot can't change tubes in mid-air.

Stage amps can afford to be a bit reckless. We want MORE, we can change our own tubes, and the cost is a fraction of the total cost (strings, picks, shirts) of being a musician. If 150% voltage cuts tube life to 20%, 1,000 hours on a $20 tube, that's 2 cents an hour.

Anyway: why worry about an EF86's plate voltage rating? You probably can't find 500V. The happy-spot for a resistor-loaded device is ~about~ half the supply voltage. So you probably won't have even 250V on this 300V tube. And raising plate voltage will not raise gain.

> I would like to get more gain

Start with the load impedance. In the Mattress plan you cited, 1Meg||470K is 320K. For a pentode, pick your plate resistor about equal; Matchless used 330K (they were not fools).

Find the pentode ratio of Plate current to G2 current. EF86 shows 3mA/0.6mA, the ratio is about 5.

For maximum voltage swing, try G2 resistor about 5 times plate resistor. 1.5Meg.

For maximum gain you want G2 as LOW as it will go. Sometimes under 40V. To get near here, double the G2 resistor: 3Meg. The plate/G2 current ratio is not constant at very low voltage; some experimentation is needed.

Since you have fairly strong signals, if you get HIGH gain you will have LARGE output signals. So you must compromise between gain and maximum output. Split the difference between 3Meg and 1.5Meg. We see Matchless picked 2Meg. They are ahead of us!

Finally, pick a cathode resistor to get the plate voltage roughly halfway up the supply voltage. If we have 300V supply, want 150V drop in 330K resistor, we want 150V/330K= 0.45mA plate current. Add the 1/6th G2 current, 0.53mA cathode current. The cathode voltage will be something near half of G2 voltage divided by Mu. EF86 Mu is 38. Assuming G2 winds up at 40V or 160V, then cathode voltage will be 0.5V to 2V. For first test, assume 1V, at 0.53mA, is 1.89K. Put in a handy 2.2K (as Mattitch did) and see where the plate voltage falls.

You are not going to improve significantly on Matchless's design; they knew this stuff AND were paid to experiment.

There are many "problems" with a single pentode, Volume, PI plan. But it is a cornerstone sound in e-guitar tradition. Love it, or do something with a pair of triodes.

[Leevi]

Thanks for the tips guys
The plate voltage I can reach is about 280V. I'll test the proposed ideas by changing the resistor values.

> I wonder how reliable these 50 years old sheets are.

Those guys knew more about their tubes than we EVER will.

> E.g. you can exceed the max plate voltage.

Sure. And get shorter life. For most old-time uses of tubes, frequent replacement was annoying, expensive, or dangerous. Grandma couldn't change her own tubes. A airplane pilot can't change tubes in mid-air.

Yes, I don't question that. I'm meaning rather today's manufacturing techniques
and better materials that make the tubes more stable and allow more characteristics for them.
Maybe I'm wrong since I don't know how much the manufacturing processes have changed since 60's.

/Leevi

[tubeswell]

I'm meaning rather today's manufacturing techniques and better materials that make the tubes more stable and allow more characteristics for them. Maybe I'm wrong since I don't know how much the manufacturing processes have changed since 60's.

I wonder whether its not so much manufacturing processes that may've changed as much as whether its the 'little' corners that may've been cut to trim costs e.g.; poorer h-k insulation and poorer quality iron in the plates etc etc

[VMS]

The plate voltage I can reach is about 280V.

Do you mean supply voltage?

Can you show us the whole circuit? Maybe there is something else that could be done to properly drive those EL84s.

[DummyLoad]

Exactly it is really 0.58mA. I didn't check the zeros since I trusted the result you had counted.
So there is not so much to do in my case. I can increase the B+ maybe 10V by changing
the resistors but it will probably not help that much. I'm using EZ81 rectifier and the amp is an SE amp with two EL84s.
/Leevi

leevi, would you consider posting your entire circuit? we (buttery & i) recently built a matchless DC30 clone for a friend, i can't imagine that you'd need more gain.

your Va seems amiss if you are using similar values of Ra, Rk, and Rg2.  i've attached the schematic of lafayette escadrille i recently built - it's in another thread, however, this one includes all the telemetry. since our B+ values to the first stage are nearly equivalent, it does seem odd that your Vp is only 80V. check for mis-wiring or a weak EF86.

[Ritchie200]

I'm meaning rather today's manufacturing techniques and better materials that make the tubes more stable and allow more characteristics for them. Maybe I'm wrong since I don't know how much the manufacturing processes have changed since 60's.

I wonder whether its not so much manufacturing processes that may've changed as much as whether its the 'little' corners that may've been cut to trim costs e.g.; poorer h-k insulation and poorer quality iron in the plates etc etc

Yeah, a good example are the spec sheets for the early KT88.  By the time they quit making the good ones max values were up 60%.  Still, the boys knew how to keep them happy and the old figures still apply in many ways.

Jim

[Leevi]

The plate voltage I can reach is about 280V.

Sorry, I mean supply (or B+) voltage as I have stated in my previous post
/Leevi

[PRR]

Plate voltage is a total red-herring. Higher plate voltage will hardly raise gain. The 300V rating could be used with a supply voltage far over 500V. If you could raise B+ to infinity (say over 1,000V) and increase the plate resistor, you could get slightly more gain; certainly not worth the cost.

Post YOUR schematic. Gain should be adequate unless you want utter-overdrive. And for that, a pedal is probably the way to go. If you are not getting pretty good gain, maybe the problem is somewhere else?


> today's ....better materials

Hmmmm?

In 1953, the US Military *depended* on good vacuum tubes. US Nickel would prepare sample billets, RCA GE KenRad would test them, and approve the ones they liked (we didn't know enough about cathode metallurgy to specify the impurities; trial and test). This was a huge pain for US Nickel: it was only about a ton a year and the tube makers could reject all the billets and ask USN to make more samples. And they did it so the godless commies couldn't sneak up on us.

Further: TWA, Pan Am, Santa Fe, and the military kept good records of Millions of hours of tube operation. Most tubes were good but a few weak ones could be very annoying to an airliner, fast freight, or arctic radar station. "Five Star" reliable tube programs examined all the materials and every tiny step of the factory, reducing potential troubles.

There were similar workers in the USSR and in China, though perhaps not as many.

Tube production collapsed. The USSR collapsed. China has gone through several cycles of future and past. Much of that knowledge was lost.

Yes, *other* materials improved a LOT. Silicon is a shining example. The testing and methods used around 1970 to get large "perfect" Silicon might have improved vacuum tube materials.... but by that time nobody cared. Owensboro was rote-assembling stock made years before and knew it was doomed. Nobody was in a position to put research money into a dying technology.

I'm mildly astonished that we get new tubes as good as they are. I bet the old men were bought-off their state pensions to consult in the tube factories. And generally improved metallurgy and global markets mean there probably IS a good material available, if you know what you really need. (For oxide cathodes, you don't want "pure" Nickel, some impurity is needed. We know to dope Silicon with Phosphorus or Boron; we may not know what magic salt is needed in cathode Nickel.)

[Leevi]

PRR:
You'll find the schematic in the following location:
http://www.rikstone.com/
Select Rikstone C10->circuit and Rikstone C10->Power Supply
/Leevi

[PRR]

You have a mixer between pentode preamp and power tubes. This is loss; also somewhat low impedance.

Input sensitivity is around 50mV, which is indeed high (low gain).

The Matchless, also ISOtone's amp, have a triode gain-stage between pentode preamp and power tubes.

Early Fender Champ has pentode to power tube, but no mixer, and wasn't intended to give high gain.

Also: look at your two-triode path. The gain of the first stage is somwhat low because of the low-Z tome control values. Still it has gain like 20*50= 1,000. The pentode preamp gain is 100 or 200, much-much less than the triode path.

My feeling is that you need another gain stage in the pentode path. The simplest way to do this is to move the mixer from V3V4 grid to V1B grid. V1B can serve both V1A and V2 preamps.

[Leevi]

Yes, the mixer is probably a problem. I have earlier used the EF86 in the following
project where the gain was clearly higher than in this amp:

http://www.kolumbus.fi/risto.kivioja/RikstoneAmps/CircuitH5_1.pdf

I don't want to add a triode since I have tried it before and IMO it will change sound i.e. it's not any more a pure EF86 sound.

One solution could be that I separate the preamps by using a switch like a-channel, b-channel, both?

Leevi

[VMS]

One solution could be that I separate the preamps by using a switch like a-channel, b-channel, both?

The 'both' option could be a problem because the channels aren't in phase.

I'm not an expert in resistive mixers, but what if the triode side mixing resistor was something like 470k-1M. Wouldn't there then be less loss on the ef86 side?

[Leevi]

The 'both' option could be a problem because the channels aren't in phase.

Yes, that's is true but not so unusual in tube amps. I think even in some Fender amps channels
have been mixed in different phases.
/Leevi

[Leevi]

The mixer was really a problem. After I had removed it the gain increased remarkable.
Then I set the plate resistor to 220K (like it is in VOX AC15) and cathode resistor to ~800 Ohm.
The measured voltages are now:

Supply = 267V
Plate = 122V
Screen = 36V
Cathode = 0.680V

IMO it sounds better now. Thanks for the help!
/Leevi