What's this JCM800 mod do?

[Alex R]

I found this mod in an old JCM800 4210. Someone had plugged an EF86 into it, seems clear it's for a double triode. Seems a bit complicated for a mere boost? I am not sure what those caps are meant to do. Sorry about the rubbish drawing, nodes are blobs; crossing lines aren't nodes, the mod is in red. Thanks for reading.

Here's a photo:



the thin green wire grounds pin 9

[J M Fahey]

As drawn, the mod is nonsense.
To begin with, the 2.2uF cap shorts grid to cathode absolutely muting the amp.
The rest is on the same vein.

NOW, try to redraw it with an EF86 plugged in that socket, It might connect added parts to proper EF86 elements.

That said, the Mod would replace 2 cascaded gain triodes with a single pentode, besides other problems he will *lose* a ton of availoable gain, even if done correctly.

As in trading original , say, 1600X to 2500X gain for, say, possible 100X or 200X .

To boot, original circuit allows to use 1 or 2 gain stages at will, making the amp very flexible ... that is lost with the Mod.

[Alex R]

Ah thanks Juan, I had drawn a false conclusion from the fact that pin 9 - grid on the EF86 - was grounded... but that's only with the clean channel switched in. Same goes for the connection from the stock stage 1 grid, it's only switched into contact with the internal screen. That part I don;t get but I'm probably getting confused by the switching thing - and my usual failure to check stuff properly. All the other EF86 connections look ok, it just parallels the first gain stage, injecting a bit of gain and a lot of noise and microphonics .

The EF86 internal screen - I have seen these grounded or connected to the cathode, but this one seems to be connected to the plate and the cathode via two caps - 0.22uF to the plate and 2.2uF to the cathode. Is that weird or have I f*d up again lol

[J M Fahey]

I'm too sleepy to analyze it, besides it's something non standard and you have it on your bench, not any of us, so I'd draw the EF86 pinout on a clean sheet of paper
http://www.r-type.org/articles/dixey51.jpg
and carefully, numbering all pins, would draw one by one what they have connected, where it goes, and finally adding jacks, grounds, power rails, whatever.

And then triple check that, resting a few minutes at least between checks.
It's very easy to "see" what we expect to see.

Then when integrated in the full preamp it may make sense.

[Alex R]

Interrogated the customer a bit! The mod has been in there a while, a previous owner had it done in London (UK) by someone called Les; it started to throw up problems and the amp has been through two further techs who did things to the mod. The second tech decided it was meant to be an EF86. Somehow or other it did make amplified sound that way but it was very poor and weak.

Pin 9 is not grounded to a switched but a permanent ground, I found on further investigation. Can't be a EF86 then, as pin 9 is the grid.

The existing first stage is disconnected at plate, grid and cathode, by the way. This is a complete replacement first stage. You might expect either a parallel pair or two stages, and it might have been either of these in the past, though the parallel wiring is determinedly done and looks original.

As a first step I took the 2.2 cap off the grid and grounded it, leaving it also connected to the cathodes to make an extra bypass cap. OK the amp now sounds pretty good, like a JCM800 but fatter, as you might expect with a 2.2uF cathode bypass cap on across the original 0.45. Does the paralleling of the traditionally wired triode, rh on my drawing - with that lh oddity with a 1Meg plate resistor and 0.22uF from grid to plate - make any sense to anyone? If not I might try rewiring it as a simple parallel pair.

[J M Fahey]

Nice verbal description but please redraw the schematic.

Do not draw tube elements because we are not sure what tube actually fits, 12A*7 family and EF86 pinouts and components do not exactly match, so just draw a mini layout, showing the socket , jacks, pots, whatever and we try to pull a valid schematic out of the mess.

One important detai you don't mention, for example, is that 12A*7 tubes usually have pins 4&5 joined, get one end of 6.3V , and pin 9 gets the other end; you tell us pin 9 is grounded but not a word about filaments.

What looks like a 1M plate load might be a 1M screen resistor and so on.

[jazbo8]

Probably a long shot... could it be a mis-drawn cascode circuit? Such as this one (not for EF86 obviously)?

[baddog]

Put it back to original specs and see what the customer thinks... Don't tell him anything and see if he likes it and if he does then tell him what you did. If he wants something else then go from there.

Keith