EL-95 Push Pull Project

[j-stylez]

Hello everyone!

After I had built a small SE EL-95 amplifier a while ago, I decided to take it apart and rebuild it into a push pull amplifier, based on two EL 95 tubes.

I was thinking to make it similar to a Princeton '65 Reverb up to the PI, but with a cathodyne PI since the El 95s dont need to be pushed as much as the 6V6 in the Princeton. My transformer gives me a B+ of 280, which is a little high for the EL 95 according to datasheet (250 recommended), but according to what I've read they should be able to handle it. I have a Hammond 125C for an OT.

I will also add grid stoppers to the EL 95s, maybe start with 1k?
Also, I will add screen stoppers on the EL 95s, I would go for 1k.
Then, i might also add a 100uF bypass cap on the cathode resistor of the output stage.

Could I delete the 1M5 bypassed by a 22pF before the third gain stage, or does this have some use without the insertion of reverb there...?

Do you think this could work? Any comments, questions, or tips are much appreciated.

Edit: I just realized that the PI in the 65 reverb is already a cathodyne PI so i'm not really changin the gain structure at all. So, I am thinking that I could just go with a 12AY7 for the first tube if i want it to be cleaner

[pdf64]

I think that the only thing that would stop the schematic working is the lack of a 0V dc reference for the 3rd stage grid; without a dc reference (eg to the cathode potential) electrons picked up by the grid will likely cause its dc potential to drift off and send the tube into cut off or saturation.
The PR has a path to 0V there via a 470k resistor and the reverb level pot.
That also acts acts (in combination with the 1M5) as a potential divider to bring the gain into sensible levels.

You could just replace them with a 100k to 1M trimmer potential divider and adjust to taste.

Yes, control grid stoppers are a very good idea, I think that values in the range 4k7 to 10k are normal for pentodes.
As the value increases, they provide improved mitigation for bias shift / blocking distortion under overdrive conditions.

[frus]

I think that the only thing that would stop the schematic working is the lack of a 0V dc reference for the 3rd stage grid

You also need the grid reference for 1st stage (e.g. 1M from IN to ground)

[j-stylez]

Thanks, I forgot to draw in the 1M resistor from Input to Ground. I am planning to solder that one right on the input jack.

Regarding the grid reference path for the third stage - the way you describe it, I would remove the 1M5 and 22pF, and replace instead with a, say, 1M pot, wired as if it were a volume pot, with one outside lug going to ground and the wiper to the grid?

If I dont want to use a trimmer- for a starting point, could I simply use a 470k series resistor followed by a 470k grid leak resistor? (alternatively, 220k / 220k if I want to lose some more gain?)

Also, I am wondering whether I need the 4th filter in my power supply. I see a lot of smaller amps with 2x12AX7 and a pair of pentodes just use a shared filter for v1 and v2.

[j-stylez]

I found this choke lying around. I remember it came from an old tube PA with a push-pull power amp with two tubes similar to 6L6.

However, I cant find any information on it. I am wondering whether I can somehow find out if I can use the choke for my




Is there any convention on specifying chokes by the numbers visible on this thing?

It sais 701YTM and GA0098 on it.

I think I'm just going to stick to my RC filtering for now since I cant find any information on it.

[The Dude]

Got any way to just measure the inductance?

[j-stylez]

No, I dont have the equipment for that. I'll just stick to my RC filtering in the power supply. Some parts I needed have arrived today so I'm going to start building. I decided to add a mosfet driven spring reverb and a mosfet tremolo. I'll put the tremolo voltage on one of the preamp tubes' cathode.

I'll keep the controls simple: volume, bass, treb, tremolo depth, tremolo speed, reverb.

Maybe I'll add a pentode/triode switch for the power amp. For the tremolo speed, I have a pot with a switch, so that tremolo can be turned off completely.

God I hate drilling the chassis...soldering is so much more fun

[j-stylez]

I have put it together, and updated the schematic. It works pretty well, however I have one Problem: There is some ugly distortion on decaying notes. It is especially noteable when damping the strings and strumming lightly, it makes very unharmonic distortion when the sound rings out, instead of the clean sound decaying as it normally would.

Voltages in the preamp (plate, grid, cathode) stages:
192 1.6mV 0.54 (first)
176 1.8mV 0.75 (second)
209 0.2mV 0.25 (third)
220 4.8 8.6 (PI)

Any ideas where I could start looking? The cathodyne PI voltages seem off. I disconnected my NFB, that didnt change the Problem. In General, I am wondering if I have too many gain stages, for the small EL95 tubes. I could remove my third gain stage and use it for a tremolo instead. I changed power tubes and preamp tubes, that had no effect on the Problem.

Additions:
- added 1M grid stopper on PI grid as recommended by the valve wizard
- reverb and tremolo to be added later

[g1]

Yes, those PI voltages are off. The tube is barely conducting. That would be the best place to start.

[j-stylez]

I am trying to improve the design of my PI, so I have made the following calculations to find a proper bias point and values for my circuit.
Following the method described by the valve wizard, I have first of all drawn the load lines and made the calculations for my current circuit. I come up with the following for the circuit shown in my latest schematic (HT=220V, 56k load resistors on the cathodyne, 220k grid leaks on the EL95):

Drawing the load lines (yellow and orange)

220V/112k=1.96mA
220k*56k/(220k+56k)=44.6k
220V/89k=2.47mA

I chose a bias point of about 0.8mA:

0.8mA*56k=44.8V
44.8V-1V=43.8V
1V/0.8mA=1.25k

So to optimise the circuit with my components, I would pick a 1k2 cathode resistor instead of the 1k I currently have installed.

Do you think my PI is biased too hot/cold judging from my calculation and sketch? I used the valve-wizard bias point to have a comparison, I hope that's ok. All the credit for the graph goes to him, I just wanted to build on that to have a comparison.

Now I am thinking, I could change up my load-lines to be able to bias my PI better. I could either lower the grid leak on the EL 95's to 100k, or lower the 56k load resistors on the PI to 36k as the EL95 dont need much signal swing to be overdriven. I drew in that bias point if I were to lower the load resistors to 36k, also in the diagram (red and brown)

Calculating the bias resistor per above method, I would end up with 36V on the grid and 34.5V on the cathode, as well as 1.25V / 1mA = 1.25k cathode resistor value.

That would still be close to the rule-of-thumb of having the grid at about 1/5 to 1/6 of the HT voltage. What do you think of these bias points?

Edit: And any ideas why my PI voltages could be so off? I have checked my layout again and again, but I can't find any error in the wiring. I have included a picture of the layout in the cassis. I am not getting any hum or noise, that PCB on the right is a LND150 reverb driver and recovery circuit. I haven't connected that yet, I'll first figure out the amp itself before I add reverb and tremolo.

Edit 2: I am going to have to recalculate that...I was working with 220V HT, but that's measured after the 56k plate load resistor of the PI, on the pin of the tube

[g1]

You have a gross fault in the PI, not a little design issue.
The current appears to be around .1mA on that side of the tube. Are both side heaters lit for that tube? Resistors measure ok and all connections good?
Measure resistance from cathode pin at socket to ground.

[eschertron]

^^^ what G1 said. The difference between 1k and 1k2 for a cathode bias resistor is not going to be significant enough to make the amp work - unless you discover the cathode R is actually 1M when you thought it was 1k
Check component values, check continuity and resistance, check voltages on each pin, swap tube... somewhere you'll find what's been overlooked. Once the problem is fixed, then tweak your design for the mojo

[j-stylez]

I have a 6,3v heater supply, which i connected to Pins 3 and 4 instead of 3&4 tied together and pin 9. That sounds like I might have found the Problem. Thanks guys I appreciate your help, always!! Will have to try this on the weekend.

Edit: I mean 4,5 and 9 - sorry, it was late.

[g1]

I think you mean pins 4,5, & 9 but hopefully that will cure it.