5E3 inspired amp with additional EF86 channel

[emi2345]

Hi there, I've drawn up a design for an amp, which will be my second build and the first for a friend of mine. It's a normal 5E3 but with an additional preamp channel consisting of EF86 pentode -> James tone stack and volume -> 12Ay7 triode gain -> 5E3 12AX7 cathodyne inverter.

The EF86 is a slippery beast, you change one thing and have to calculate everything again, but I think I've narrowed it down to the following circuit:

Vsupply 250V; Vg2 100V; Vg1 -2V; Va 130V
Ia 1.2mA; Ig2 0.2mA; Ik 1.4mA
Rk = 2/1.4 = 1.428k = 1.5k
Rg2= 150V/0.2mA = 750k
not quite sure what the gain is

I know that the high output impedance from EF86 won't load very well into James tone stack, but I was hoping the 12AY7 gain recovery would be sufficient for the 1/2 12AX7 cathodyne inverter. I can tweak the bright cap to try to balance out any frequency response changes, and I've designed in a tone bypass switch. It's rather like the Route 66 but using a 1/2 12AY7 into 1/2 12AX7 cathodyne inverter instead of a long tailed pair.

Any help or advice would be much appreciated!

[pdf64]

It's bad practice to leave grids floating (ie when the 'other' channel is selected);a Vdc reference is required, eg 1M to 0V, whenever plate current may flow.
The tube numbering is a bit wacky, there look to be 3 V1s! (12AX7, 12AY7 and a 6V6).
The signal from the pentode channel is being fed into the feedback loop (if engaged) inside the loop; that may lead to unexpected results, as it will be an error signal that loop will attempt to correct.
Gain wise, bear in mind that a 12AX7 LTP has a voltage gain of about 30 (= 30dB); so you're 30dB down on that, which is a big drop compared to a Route 66.

[Justin Thomas]

Gain wise, bear in mind that a 12AX7 LTP has a voltage gain of about 30 (= 30dB); so you're 30dB down on that, which is a big drop compared to a Route 66.

Just out of curiosity, would it be possible to just plug a 12AX7 into the AY7 slot and make up for anything lost? No harm in trying, right? I ask because when I did a build similar (EF86 - Volume - Tone -12AX7 - Cathodyne - 2xEL34) I had to put in an AY7 to tame the gain! Granted, it sounded awesome, but went from Zero to AC/DC by "3" on the Volume knob. I mean, <I> would love it, but I don't think the customer quite had that in mind...

Justin

[pdf64]

Yes, sure, that would pull things up ~10dB.
Sorry, scrub my comment about the feedback loop, I didn't notice it had 2 branches.

[Chuck H]

+1 on the floating grids. Dangerous for the tubes and pops like corn when switched. You really want to reference the grids to 0V for stability and eliminate any differential DC on either side of audio path switches to avoid noise.

Justin mentioned his similar el84 based amp. It should be noted that el84's typically start grinding at about half the PI swing as 6V6's, so that could be a significant consideration with a design like this where the power tubes will be doing most of the clipping. With the cathodyne type PI arranged with higher gain up front you might end up with the PI clipping as a prominent part of the tone. For better or worse I suppose. I expect it wouldn't any different than running a boost into a 5e3, which gets done all the time.

Another consideration is that an actual 5e3 has it's channels mixed in a way that adds additional voicing other than the tone control. You have omitted this characteristic and so the tone will be different. Again, for better or worse.

[Justin Thomas]

Justin mentioned his similar el84 based amp...

Hey Chuck, I know you're a huge fan of the 6BQ5, but check again... It weren't no typo. Methinks you have the Vox & Matchless on the brain too much...

Of course, I don't know if that changes any of your reply at all... I just kind of make something and go from there. It's worked so far!

Justin

[Enzo]

I would use a DPDT for the bias switch. One side to ground the cathodes, the other side selects either ground or bias for the grid resistors. As you have it you wind up grounding off the bias supply.

Just my personal opinion, but the power supply choke/resistor doesn't seem like much of a feature. I'd imagine it will spend its life in one position. Why would one wish to turn off the choke?

So you have a pentode input channel and a triode input channel, both feed into triode buffers, then the two mixing resistors are the same. Are your levels going to be even? Will the pentode and triode input stages put out similar signal levels?

You might consider a series resistor in the DC heater circuit. 6.3VAC rectifies and filters to almost 9vDC, take away a couple diode drops for the rectifiers, and you still have about 8v. You have about 0.8A tube current, a couple ohms ought to do it.

[Mick Bailey]

Once you lose the high-end early on you can't add what isn't there. Bright caps and peaking circuits can only deal with the spectrum you already have. I think you'll lose a lot of the advantages of the pentode input stage and also pick up a lot more noise along the way attempting to recover the lost sparkle. If this was me I'd probably experiment with pentode>CF>tonestack. No additional tubes, just different topology and a low impedance drive to your stack. Some though would still need to be given to signal levels.

[Chuck H]

Hey Chuck, I know you're a huge fan of the 6BQ5, but check again... It weren't no typo. Methinks you have the Vox & Matchless on the brain too much...

Of course, I don't know if that changes any of your reply at all... I just kind of make something and go from there. It's worked so far!

Justin

Ha! Nice. Well el34's can be pretty close in bias voltage to 6v6's under a lot of different circumstances. So that's a solid opinion on expected gain IME. Thanks for having my back.

[emi2345]

It's bad practice to leave grids floating (ie when the 'other' channel is selected);a Vdc reference is required, eg 1M to 0V, whenever plate current may flow.
The tube numbering is a bit wacky, there look to be 3 V1s! (12AX7, 12AY7 and a 6V6).
The signal from the pentode channel is being fed into the feedback loop (if engaged) inside the loop; that may lead to unexpected results, as it will be an error signal that loop will attempt to correct.
Gain wise, bear in mind that a 12AX7 LTP has a voltage gain of about 30 (= 30dB); so you're 30dB down on that, which is a big drop compared to a Route 66.

- floating grids - I've updated the schematic with a rotary channel switch with 4 positions, and S10-B corrects this with a 1M to ground for the grid of the un-selected channel

- fixed the tube numbering

- gain - LTP with a gain of 30 is similar gain to the 12AY7 triode stage, which the route 66 does not have, so it should be similar right?

Just out of curiosity, would it be possible to just plug a 12AX7 into the AY7 slot and make up for anything lost? No harm in trying, right? I ask because when I did a build similar (EF86 - Volume - Tone -12AX7 - Cathodyne - 2xEL34) I had to put in an AY7 to tame the gain! Granted, it sounded awesome, but went from Zero to AC/DC by "3" on the Volume knob. I mean, <I> would love it, but I don't think the customer quite had that in mind...

Justin

- how did that amp sound to you? route 66 has tone before volume, does it make much difference which way round they go? My schematic has the same topology as yours: EF86 - tone - volume - 12AY7 triode gain - 12AX7 cathodyne)

I would use a DPDT for the bias switch. One side to ground the cathodes, the other side selects either ground or bias for the grid resistors. As you have it you wind up grounding off the bias supply.

Just my personal opinion, but the power supply choke/resistor doesn't seem like much of a feature. I'd imagine it will spend its life in one position. Why would one wish to turn off the choke?

So you have a pentode input channel and a triode input channel, both feed into triode buffers, then the two mixing resistors are the same. Are your levels going to be even? Will the pentode and triode input stages put out similar signal levels?

You might consider a series resistor in the DC heater circuit. 6.3VAC rectifies and filters to almost 9vDC, take away a couple diode drops for the rectifiers, and you still have about 8v. You have about 0.8A tube current, a couple ohms ought to do it.

- the bias switch design came from the Ampeg Heritage B-15. I did notice though that it takes about 10 seconds for the bias voltage to develop when switched from ground, so it's possible to switch to fixed bias whilst this is happening causing over-dissipation. I've corrected this, will it be a problem that my switch is on-off-on? B+ is not connected in the middle standby 'off' position. Now I'm going to need a beastly Carling IM250-73 4PDT switch so I have an extra pole for my LED traffic light behind a clear fender jewel lamp: green for cathode bias, yellow for standby, red for fixed bias. I've included that circuit at the bottom of my schematic too this time.

- the choke switch is to preserve a possible 'pure' 5E3 option with no choke

- the mixing resistors might need to be adjusted, and I might need extra ones in the cascaded input circuits to control gain, I will have to adjust these when I've got it up and running

- DC heater circuit is the same as the Ampeg Heritage B15, and with that I'm getting 6.7VDC at the moment with an annoyingly high 244VAC mains. Could drop it a touch I guess.

Once you lose the high-end early on you can't add what isn't there. Bright caps and peaking circuits can only deal with the spectrum you already have. I think you'll lose a lot of the advantages of the pentode input stage and also pick up a lot more noise along the way attempting to recover the lost sparkle. If this was me I'd probably experiment with pentode>CF>tonestack. No additional tubes, just different topology and a low impedance drive to your stack. Some though would still need to be given to signal levels.

I considered this but I thought the DC cathode follower would change the sound of the EF86. I'm following a topology that's similar to the route 66, plus I have the tone bypass option. If it sounds bad I'll look at other options.

Let me know if my 4-way rotary input switch is a mad idea or a usable fun working one. The idea was 4 positions- Triode; Pentode; Triode+Pentode; Pentode+Triode. This design is turning into a sea of switches which I may remove after trying them out, fun experimenting and learning though.

New schematic attached, many thanks, Edmund

[emi2345]

That schematic contained errors, I've attached an updated one here.

[pdf64]

... gain - LTP with a gain of 30 is similar gain to the 12AY7 triode stage, which the route 66 does not have, so it should be similar right?

Yes, sorry I didn't think that through either, that should work out; however, the gain will drop some when the NFB loop is engaged.

Why such a big grid stopper on the EF86?

[emi2345]

I was following this tutorial: https://www.ampbooks.com/mobile/tuto...de-tutorial-3/ and reduced the suggested 150K to 100K to try and bring noise down a touch, if I have parasitic oscillation problems I'll push it up. The AC15 uses a 110K.

[pdf64]

Thanks, I'd never clocked that the EF86 Miller capacitance is so low.
I acknowledge that this isn't a high gain amp, so noise probably won't be an issue, but my understanding is that series resistance in the grid circuit of the input stage is a significant noise source, so it's best to keep the input grid stopper value as low as is feasible.
With ECC83 inputs, Merlin even suggests a 10k grid stopper and adding capacitance
Here's his take on EF86 The Valve Wizard -Small Signal Pentode

[emi2345]

I'd be interested to know his take on this, he doesn't mention Rgs in that webpage