18Watt.com

...incomparison to my AC30-ish build EF86 channel? I mean, I used stock AC30/4 values, Mercury Magnetic Woden clone trannies, Sozo caps, appropriate resistor values, etc...but I'm finding my EF86 channel to be more strong and tight and ballsy instead of the chimey, airy quality in the London's EF86 side.

Now, is it true that the London's EF86 side is more like a Selmer Zodiac "30" instead of an AC30/4? I find that the plate voltages on the London's EF86 side are about 185v on the plate and 65v on the grid, wheras my EF86 is (again, with the stock AC30/4 values...) landing at about 111v on the plate and about 108v on the grid. Would the voltages themselves lend to such a drastically different tone? Should I try for a larger separation between the plate and grid voltages?

I know I'm kinda rambling....but I'm sort of stream-of-consciousness typing here. My friend just brought his London over and that EF86 channel was all that I wanted my AC30/4 clone to be....(mostly... ). Incredible EF86 channel...


Thanks all...

The best thing would be to open the London and just copy the circuit. Of course, their PT may have different voltages and their OT could be voiced differently. Plate voltage does make a difference in headroom.

Another thought is to look at what kind of caps they are using. Maybe try Orange Drops in your EF86 channel.

I'm not sure their description at 65 Amps matches the Selmer circuit, but the drawing of the Zodiac is a little fuzzy. If the London is truly a cascode, that would probably be a good clue as to why it has the extra chime- but that's as much as I can guess.

I think rjgtr's idea is the best one, unless you can find the London schematic somewhere.

AFAIK it's a true pentode, and not cascoded triodes. With EF86s the voltages are very important for best tone and compression. I reckon that's what you should be experimenting with.

dehughes wrote:...incomparison to my AC30-ish build EF86 channel? I mean, I used stock AC30/4 values, Mercury Magnetic Woden clone trannies, Sozo caps, appropriate resistor values, etc...but I'm finding my EF86 channel to be more strong and tight and ballsy instead of the chimey, airy quality in the London's EF86 side.

That's to be expected when you're running such a high screen voltage, you'll be getting a mostly 'cold biased triode'-type sound that way. Lowering the screen voltage will bring you back towards pentode operation, and then warm biased triode sound if you go even lower. Temporarily replace the screen dropper with a pot and play with screen voltages, shoot for around 60V for a proper pentode sound.

Do you have any kind of a rough rule of thumb for small signal pentodes - eg what the ratio of screen grid to anode voltage should ideally be? I've not really played with pentodes so far, but have a few in my stash I'd like to build with...

Sort of, it's not too complicated. Once you've chosen your anode load resistance, work out the max current you can ever draw through it V/R. off the screen voltage at that current on the mutual characteristics graph in the data sheet- that'll put you immediately in the ball park. A little lower might be necessary sometimes.
Most people like pretty big anode resistors (e.g. 100k to 220K)so you can expect to aim for proportionately low screen voltages in the 50V to 60V region. If you're using lower anode resistors (like 56k-33k) you can expect to need more like 70V to 100V screen.

THANK YOU Merlinb! I believe you're right on the money... I don't want to copy the London EF86 circuit so much as understand why it does what it does, and then take those principles and apply them to my build. I've already come to realize (thanks to another guy on another board...) how much the cathode has to do with things (replacing a 25uf cap with a 68pf cap on the cathode cleaned up the EF86 tremendously...), so I'm just going to experiment with plate resistors, screen voltage, and plate caps and see what I can get. I had a feeling the fact that my screen was only a few volts under my plate wasn't an ideal setup...

What do you think of running a resistor to ground off the screen to drop some voltage off there??? They do this in the London EF86 circuit....a big 1w (470k?) carbon comp to ground in parallel with the .1uf screen cap... Would this be any different than just using a larger screen resistor in the first place?

THANKS!

zaphod_phil wrote:AFAIK it's a true pentode, and not cascoded triodes. With EF86s the voltages are very important for best tone and compression. I reckon that's what you should be experimenting with.

First of all, I agree with all the technical stuff you write (that I understand) but 65 Amps describes their circuit as cascode/pentode based. What's up with that?

http://www.65amps.com/London.htm

Thanks,
-Markg

What do you think of running a resistor to ground off the screen to drop some voltage off there??? They do this in the London EF86 circuit....a big 1w (470k?) carbon comp to ground in parallel with the .1uf screen cap... Would this be any different than just using a larger screen resistor in the first place?

That extra resistor provides "fixed screen voltage", screen voltage doesn't change at any time (more or less).
If you use only one dropping resistor you get "sliding screen voltage" -when the valve conducts more the screen voltage falls.
Sliding screen should provide more squish, that is more compression (a little bit like sag). Fixed screen voltage will stiffen things up;- less compression and a more distinct boundary between clean and overdriven. I'd probably choose sliding screen, but either way will sound great.

If you're feeling adventurous you can experiment with fixed screen but using an undersized caspacitor, so you'll get compression of low frequencies but not at higher frequencies.

sharkboy wrote:.... 65 Amps describes their circuit as cascode/pentode based. What's up with that? http://www.65amps.com/London.htm

Dan Boule who did most of the design of these amps used to be a regular contributer on 18watt.com and still posts now and again. I know for sure that these amps were designed to use a genuine EF86 pentode. However, it is possible to somewhat emulate a pentode using a pair of triodes connected together in what is known as "cascode" mode. You can even get a kind of fake EF86, which uses a dual triode with the extra components packed into the tube base, and which will plug into an EF86 socket - GT make these in the form of a convertor - http://www.groovetubes.com/product.cfm?Product_ID=2044 It will be less prone to microphonics than a true EF86, but I doubt that it will sound quite the same. However, I have found one old post from Randall Aiken saying he prefers the sound of a cascode over a real pentode...

The correspondence between a cascode and a tetrode/pentode is something that has always tickled me.

One (grossly simplisitc) way to think about it is this. The purpose of the screen grid is to shield the control grid from the anode. This gets us low Miller capacitance. What you might imagine is a nice solid sheet of metal. Trouble is you can't do this - so you make a conventional wire grid. But if you stop for a moment and think that at the location in the tube where the screen grid goes all you have is an equipotential plane that lets electrons through, you could notionally replace the screen grid with a solid sheet that receives the electrons, and is then connected to another electrode that is heated to emit electrons. Electrons that then move on to the real anode. The current through the pair of electrodes is identical to the current of moving electrons in the vacuum. Couple these two electrodes to a suitable bias voltage and you have a working pentode, albeit one that is a bit of an engineering curiosity. But we can now separate the two intermediate electrodes into two separate devices - after all the only current flow between them is done via a wire. Well look at that! Two separate tubes!

This is a terrible oversimplification of the dynamic nature of how a cascode works - it ignores all manner of operational and design issues, but does give one a flavour of why there is a correspondence. The description especially ignores the control grid in the upper tube. Since most of the gain of the cascode comes from the upper tube this is something of a serious omission.

MarkD once told me how much trouble he had getting his Vajra EF86 channel to sound right. He said he went back to some old radio books from the 40s and found a great discussion on how to leverage pentodes. His solution for getting more chime out of it was to give the tube its own filter cap with a huge dropping resistor to bring the voltage way down. Interesting stuff.

When I compare his schematic to other EF86 circuits like those from the Dr, it kind of dawns on me why a lot of them sound like Mississippi mud...

THANK YOU THANK YOU THANK YOU! This is a supremely helpful thread...

His solution for getting more chime out of it was to give the tube its own filter cap

Good advice- that goes without saying. Pentodes have zero PSRR.

Thanks for the Vajra background Phil. I also found it very informative. Maybe it's time to use that EF86 that bought.

sharkboy wrote:First of all, I agree with all the technical stuff you write (that I understand) but 65 Amps describes their circuit as cascode/pentode based. What's up with that?

I've looked at the site and concluded that they probably changed the London design to use the same cascode input as the Marquee. It looks like the London description was updated, but a reference to the ef-86 wasn't deleted properly.

Yeah, I thought that they went to cascode a while ago because of the scarcity of good EF86s.

rjgtr wrote:Thanks for the Vajra background Phil...

Dat wuz s2-Steve, not me....

zaphod_phil wrote:

rjgtr wrote:Thanks for the Vajra background Phil...

Dat wuz s2-Steve, not me....

Hey you're right - Well then forget you! THANKS S2!




P.S. I found your posts helpful anyway.