LIION DRIVE - 1AD4 Subminiature Pentode Overdrive

[blackcorvo]

So, I have a few 1AD4 subminiature pentodes, and I'd like to put them to use. So, searching the forums, I found some projects such as the Pentadrive (https://www.diystompboxes.com/smfforum/index.php?topic=70533.0) and the Cold War Drive (https://www.diystompboxes.com/smfforum/index.php?topic=71957.0), as well as the subminiature projects by tomasha (like https://www.instructables.com/id/High-Gain-Subminiature-Amplifier/ and https://www.instructables.com/id/Battery-Powered-Tube-Amplifier/), and I decided to use them as reference for my project.

My idea is to power it all off a Powerbank, with a DC-DC Step Up board and the filaments being powered from the 5v output. The Step Up board output is adjusted for an output of 45v.

I made the following preliminary schematic:

https://imgur.com/a/0VZydtp

I might mess around with the tone control once I have it prototyped. If anyone has suggestions of better values or improvements, let me know!

[PRR]

Input jack: Tip and Ring are swapped.

Filament power: you need 2.5V of heat. You have 3.6V battery. Why step-up to 5V and then waste-off 2.2V? (The LiON batt is probably more stable than the carbon-zinc batts these tubes lived on in 1950.)

The heaters have a + and a - for bias reference. The original plan was partly grid-leak in a 2Meg resistor referenced to the - end of the filament. Not easy to guess if your biasing gives good working points. As a guess: the output stage is cut-off, with Vg2 well under 45V and Vg1 well over 1.25V.

Two stages of Pentode, one CCS-loaded, smells like a LOT of gain.

100k Volume pot on 500k Tone pot may have a big dip mid-rotation.

Philosophical thought: WHY use a "miniature" tube and then so many side-parts? The side-parts like cathode caps are often bigger in total than the tubes. Cathode caps are not needed because the 12 Ohm filament is much less than the 500 ohm cathode impedance. The datasheet suggests running G2 at full 45V supply, eliminating an R and a C.

[thomasha]

If you move the 22r resistor to the first tube, between ground and cathode, you can get a colder bias. Second stage would get even ore negative too.
On the other hand, the grid leak resistors will drop some voltage too.

If you have a cheap oscilloscope it's easier to verify which configuration gives more headroom.

[amptramp]

I happen to have four of these tubes in my collection.

I do not like the idea of moving the filament resistor to the ground side because you have to keep the 100 µF capacitors on the filament line in such a way that they don't allow huge inrush currents but you have to prevent current from one tube getting onto the other or you would have a multivibrator with cathode coupling and R-C coupling from the first plate to the second grid.

[blackcorvo]

Input jack: Tip and Ring are swapped.

Yeah, I drew it wrong. Fixed!

Filament power: you need 2.5V of heat. You have 3.6V battery. Why step-up to 5V and then waste-off 2.2V? (The LiON batt is probably more stable than the carbon-zinc batts these tubes lived on in 1950.)

I originally thought of that, but that would mean the tube will drift in bias as the battery gets lower, no? The battery fully charge actually gets to 4.2v, fully discharged it's about 3v. But it's a good consideration, I'll implement that.

The heaters have a + and a - for bias reference. The original plan was partly grid-leak in a 2Meg resistor referenced to the - end of the filament. Not easy to guess if your biasing gives good working points. As a guess: the output stage is cut-off, with Vg2 well under 45V and Vg1 well over 1.25V.

Yes, it was all off when I tested it. 100k Rp and 470k Rg2 worked better, but still a bit fuzzy to my taste.

Two stages of Pentode, one CCS-loaded, smells like a LOT of gain.

A bit too much, in fact! First stage goes into fuzzy overdrive on it's own. I'm probably gonna ditch the CCS for now.

100k Volume pot on 500k Tone pot may have a big dip mid-rotation.

https://imgur.com/VYmVhlt Yeah, but I like the sweep much better than the mid-scoop it usually has with traditional values. I could change it to get a "mid boost" when it's in the middle, easily.

Philosophical thought: WHY use a "miniature" tube and then so many side-parts? The side-parts like cathode caps are often bigger in total than the tubes. Cathode caps are not needed because the 12 Ohm filament is much less than the 500 ohm cathode impedance. The datasheet suggests running G2 at full 45V supply, eliminating an R and a C.

I tried G2 directly on B+, gain went away. But I was using 100k Rp when I tried, might need to adjust that value for it.
About bias, do you think it's best to run both filaments to ground? And what about adding a capacitor between the gain pot and the second stage's G1? That'd make sure the bias doesn't shift as you turn the knob, and I could use a higher value for RG1 if needed.

Thanks tomasha and amptramp for the suggestions as well. I do happen to have one of those kit digital oscilloscopes in my stuff, just gotta find it.

This is the schematic after suggestions, gonna play with it more tomorrow.
https://imgur.com/VegzTjO

[PRR]

[blackcorvo]

Oh, I see. So obvious yet I'm so oblivious. Thank you!

[thomasha]

blackcorvo I usually use excel to save all voltages and calculate the grid bias obtained. It helps when you have more filaments in series. And you can always check if it's biased correctly and compare to the loadlines.

About bias, do you think it's best to run both filaments to ground? And what about adding a capacitor between the gain pot and the second stage's G1? That'd make sure the bias doesn't shift as you turn the knob

Both filaments to ground would be ideal, I'm not sure how many feedback there is between filaments, or if the anode current of one tube is forced through the filament of the other. I still don't know if it is wise to run these tubes like this or if it's a shoot in the foot.

Adding the cap helps in case there is some DC voltage at the grid.

[blackcorvo]

blackcorvo I usually use excel to save all voltages and calculate the grid bias obtained. It helps when you have more filaments in series. And you can always check if it's biased correctly and compare to the loadlines.

I never learned how to use Excel, never really needed it.

Anyways, I played with the circuit all night, and my conclusion is: it sounds like a fuzz/booster, could never get it to sound anything near an overdrive. Oh well.

Best result seems to be CCS on the second stage, so you can get a decently clean sound with gain all the way down, with a Si transistor (MPSA92 worked best for me), both G2 directly on 45v+, and 15k Rp for the first stage (as I preset on a trimpot). For full-on fuzz, diodes after the second stage's output capacitor work well.

Isolating the gain pot from the second stage's grid through a capacitor sounded better to me, and doesn't mess with the bias as you adjust the knob.

I like the tone control, goes from a rich bass response to one side, all the way to a fierce treble boost on the other. However, it indeed is a tad quiet in the middle. I'll try PRR's advice and use a 100k pot there.

[blackcorvo]

After many tries, I found out that the 1AD4 just doesn't sound good overdriven. Not even in triode mode!
However, it can still be used as a Booster, if properly paired with a transistor buffer with Bootstrapping (this is based off Merlin's "Bootstrapping for more Gain" section at: http://valvewizard.co.uk/dccf.html , which I've used before in other circuits with great results).
The CCS circuit put out around 3Vp-p @ 1KHz with an input of 200mVp-p. The Bootstrapped Buffer puts out 4Vp-p with the same input signal.

But before getting to that conclusion, I needed to check the best value for Rp. Apparently, I was measuring my trimpots on the wrong legs, because the loudest output (with G2 at 45v+) after some more tests ended up with Rp at around 200k! With that, I could use 2x 100K resistors in series for Rp and use the common point between them for the Bootstrap.

Also, the idea of using the Li-Ion battery and PowerBank circuit are too convoluted, so I'll revert for a 9v supply for the filaments.

Here's the circuit I ended up with:

https://imgur.com/vPeduFU

[blackcorvo]

I couldn't sleep this evening, so I got back to this project to see if I missed something.
Well, apparently, the two 1AD4's I was trying for this project were both damaged!  Oops!
Figured it out after I plugged one fresh from the box, and noticed how it sounded WAY off with the values from this project.

Now, with Rg1 = 4M7, Rp = 100k, and Rg2 = 680k, B+ 46.5v, it gives me great pentode tones with a hair of drive if you dig into it, and plenty of output!
I will now try a second stage to see how that does. But after a nappy. 

[blackcorvo]

Welp, I discovered why those tubes were fried.
For my tests I was using a switching step-up/down board to get exactly 1.25v on the filaments, and I think the trimpot on it might be bad, because I suddenly saw the tubes glowing visibly during my most recent test and one of them was completely fried, the other has lost basically half it's original gain.

Lesson learned: simple works best. Use a resistor!!!

Too bad! I did more trials and got a single stage to give me a gain of about 28.8dB (or a gani of 27.5 times). Rp = 100k, Rg2 = 330k.

I'll order some new tubes to keep testing further. They'll arrive sometime next week.

[blackcorvo]

Okay, tubes arrived.
Adjusting values, 100k/680k is indeed what sounds best to me. Lightly compressed when you dig into your picking, enough gain to push any amp input nicely.

I've decided to go back to the lithium battery I used originally. I now have an idea on how to implement it properly in the final built.

An addition I made was a "character" switch, which changes the screen grid capacitor value to give different frequency responses.