Output Transformer Selection

[JJH0906]

…I did do a load line calculator and the 3800 ohm impedance seem to push the tubes way out of the safe range, the 6600 ohm version fits much better but still unsure how that works out with the calculations that Merlin gives in his example.

How about taking and providing screenshots of the calculator with your data, then we can all see the context of your query (and also have a chance to spot any bad data)?

Here are three graphs from the www.vtadiy.com website that generates load lines based on input data.

[tschwarz]

OK, I'll bite. This is how I use the load-line calculator for pentodes. This is a step-by-step sequence:

1.) Select the tube type, select P/P or SE. P/P in this case...
2.) Set the HT. 360V seems to be the value for your design
3.) Set the screen voltage. To do this I need to estimate the voltage drop from B+1 to B+2 (e.g. choke or R/C filter). I also need to know if I'll use cathode bias or fixed bias. If it's cathode bias I'd subtract the value from the grid bias voltage field conservatively from B+2 and use that for screen voltage. This is not an exact value, more an estimate as the grid bias voltage will change later. I'm guessing 20V drop in the R/C filter and see a first grid bias of ca. 25V. So I'd use 360V - 20V - 25V = 315V to start with. For a fixed bias system I don't subtract the grid bias voltage as screen voltage is referenced to the cathode. So here in your example I set screen voltage to 340V.
4.) Now I'd set a more realistic grid bias voltage indirectly through setting the quiescent current. My formula for grid biased amps is Pmax / HT * 0.7. I didn't invent this formula, I'm using Merlin's. So I get roughly 58mA. I do this step even though the quiescent current doesn't have a direct correlation to OT impedance...

OTLoadLine.png

5.) Unless you wind your own OTs, you need to chose an available off-the shelf value. To find the range for that impedance the load line should run somewhere through the knee of the 0V grid voltage curve. We've already established, that to go steeper with the load line, you reduce the impedance. The tool suggested 7200 Ohms initially, which I change to 4000 Ohms. Voila, right through the knee. Hammond - as one OT supplier - makes several suitable 4k OTs in the 40..50W range. https://www.hammfg.com/electronics/tran ... assic/1750. The 1760J looks promising...

Cheers!

[JJH0906]

Oh... my... gosh!

I swear this is the best explanation I've come across for this, despite the number of videos I've watched and book references I've gone through! I'm sure the info is there in most of those sources but I haven't seen it put in a simple, step-by-step method such as this!

To fill in a few blanks, I am using a CLC filter but the unloaded B+ is huge, like 496V. I haven't constructed the voltage drop circuits for the other sections yet (Plate, screen, reverb, inverter, preamp), so I guess I can't answer what those voltages would be but this gives me a something solid to review. Also, I was planning on using a fixed bias to develop keep the signal as tight as I could. The backstory is this is intended to be a single-channel, "clean", pedal platform build so I'm try to keep any overdrive of the signal path as low as possible.

Thanks so much, this is amazing info!

[tschwarz]

To fill in a few blanks, I am using a CLC filter but the unloaded B+ is huge, like 496V. I haven't constructed the voltage drop circuits for the other sections yet (Plate, screen, reverb, inverter, preamp), so I guess I can't answer what those voltages would be but this gives me a something solid to review. Also, I was planning on using a fixed bias to develop keep the signal as tight as I could.

I use the nominal voltage (from the documentation) of the PT * 1.4 for my initial HT assumption. Unloaded voltage makes no sense here. You can measure the resistance of the secondary PT coil, though, and calculate the voltage drop across the transformer winding - assuming you have at least a general idea where your quiescent current will be. I largely ignore the other stages of an amp here and just add ~10..15% for the screen current and a wild estimate for PI and pre-amp. So with a real case I'm currently working on:

PT: 250V 195 Ohms. 6P1P-EV cathode biased P/P at ~350V HT. Pmax=12W.
Quiescent current: 12W / 350V * 0.85 ~= 30mA. That's per tube. So adjusting for screen current: 30mA * 1.15 * 2 ~= 70mA
Guesstimating 10 mA for PI and pre-amp. Thats ~80mA in total.
Voltage drop across the PT winding is then roughly 15V..20V.

Or, in your case, from unloaded voltage: my transformer is rated at 250V~ / 120mA. With the 195 Ohms resistance in the sec. winding I'd expect about (250V + (195Ohm * 0.12A)) * 1.41 ~= 385V unloaded DC after the reservoir cap, before the choke...

Cheers

[JJH0906]

Even more excellent information! You should write a book!

Also, just out of curiosity, I noticed that if a line was drawn between the Q point and the calculated point at the 0V knee, The line briefly passes through the maximum operating power line. Is this just a nuisance of load lines or is that something to consider?

To add to that, I plan on using the JJ 6L6GC power tubes, which lists its Ra-a as 5k ohms. Any thoughts and regards to that?

[tschwarz]

You should write a book!

Yeah - leaving the book writing to the truly enlightened. Actually I'm waiting for one of the real gurus on this forum to tell me I'm completely wrong.

I noticed that if a line was drawn between the Q point and the calculated point at the 0V knee, The line briefly passes through the maximum operating power line. Is this just a nuisance of load lines or is that something to consider?

This is not a problem for push-pull operation as each tube only operates roughly half the time (somewhat longer, the further you move from Class AB to Class A). So you can push the load line well over the Pmax curve. Obviously the Q-point would lie below Pmax.

It can have a very positive effect on the amp's sound though, to push further into Class A territory which you can do by either biasing "hotter" (i.e. a higher quiescent current) or by increasing the OT impedance. I have an amp where I can change cathode bias with a switch - in essence purposely run the tubes about 20% over Pmax instead of the normal safety margin of 15% under Pmax. This has a dramatic effect on the amp's behavior that is strangely difficult to describe: more sensitive during the attack - e.g. you hear much more of the plectrum action. Feels as if the amp were "anticipating" your next move on the fretboard. And a bit louder and a bit more distortion when run at full power (with an attenuator). Something really worth trying...

To add to that, I plan on using the JJ 6L6GC power tubes, which lists its Ra-a as 5k ohms. Any thoughts and regards to that?

Specifying Raa per se is not enough, it requires context: HT and screen grid voltage. The JJ datasheet must definitely be specifying those too. Then you can enter those parameters into the loadline calculator, adjust for your design's HT and screen voltage and derive the best OT impedance for your system...

Cheers