I've built a self-split guitar amp. It's fine for that (though your plate supply has to be VERY clean), but I don't think I'd ever use it for hi-fi.
-
Feeling lucky? Check out the AK 24th Anniversary Giveaway thread and join in the fun!
Brainstorming a push-pull 6L6GC stereo amp
- Thread starter thorpej
- Start date
-
- Tags
- 6l6gc push pull ultra linear
I've built a self-split guitar amp. It's fine for that (though your plate supply has to be VERY clean), but I don't think I'd ever use it for hi-fi.
Hm, and I see the Eico HF-89 is essentially the same as the HF-87 but with fixed bias. Whatever I build, I'd like to do it with fixed bias, so perhaps the HF-89 will be my starting point. One thing I noted is that the HF-87 uses a 6.6K OT whereas the HF-89 uses a 4.3K OT.
Now, I won't have quite the same voltages as the Eico, so the design is going to need tweaking regardless (it's a little higher than I originally thought -- ~430V out of the voltage multiplier, before any resistive losses in the filtering network -- but still not the 450V that the HF-87 has. I'll draw up some load lines later today... I guess going with 4.3K, I could squeeze a little more power out of it. Transcendar makes an appropriate 4.3K OT.
Now, I won't have quite the same voltages as the Eico, so the design is going to need tweaking regardless (it's a little higher than I originally thought -- ~430V out of the voltage multiplier, before any resistive losses in the filtering network -- but still not the 450V that the HF-87 has. I'll draw up some load lines later today... I guess going with 4.3K, I could squeeze a little more power out of it. Transcendar makes an appropriate 4.3K OT.
crispycircuit
Super Member
Over the yrs. I've built 2 amps using the HF-89 circuit. They both turned out great. Using 10db of feedback and having plenty of gain. A simple circuit and easy to do point to point. I put volume controls on the input and ran CD direct. Highly recommended. It also would make a nice ST-70 ft. end. The Eico 89 xmers were quite large and the 87s were good size.
The Dynaco clone transformers made by the various manufacturers are indeed nice, but they have the wrong UL taps for 6L6GCs. A 50W 4.3K 40% transformer is a better fit for the application.
Yah, I'm starting to lean towards the HF-89.
Heh, of course he does. Thanks for the link.
One thing I am a little concerned about adapting the HF-89 circuit to use 6L6GC is the output stage's grid circuit. The 6L6GC specs 100K as the max grid leak resistor when using fixed bias, although I see this exceeded in 6L6GC designs on a regular basis (I have a 6L6GC fixed bias guitar amp that uses 330K, and the Fender Bassman 5F6 uses 220K). I could drop the grid leaks to 100K and halve all the values in the bias balancing network to get close to spec, but this is still above design center max (though at 400V, I won't be pushing the output tubes particularly hard, so maybe this is OK). Even if this is OK from a "don't blow up the valve" perspective, it's still going to require a fair bit of driving current from the LTP and larger coupling caps. @kward, I'd really love to know how your 807 amp does.
I note that the original Williamson amplifier uses cathode bias.
I drew up the beginnings of the power supply last night... My PT produces 78Vac across the entire winding. B+ is provided by a 4x multiplier. Bias is tapped off one leg of the secondary winding though a 1N4007 and a filter / divider network that ultimately feeds a 1N4757 51V Zener (I currently have it configured to supply -75V to the Zener). Can anyone tell me what the original HF-89's bias tap on the PT is?
jazbo8
Well-Known Member
The Williamson runs the KT-66's triode-connected in class A, not what you want I think... There is no tap on the PT, the bias supply is fed with a separate winding on the PT, at least, according to the schematic.
Yes, I mean what was the bias winding's voltage? I have been searching for this spec but haven't found it yet.
For my 807 type amp, the LTP is a 6FQ7, with 395V rail voltage and 47K plate resistors. Additionally I run a solid state CCS in the tail (Ixys 10M45). The output stage has 100K resistance between each grid and ground--that's 91K fixed resistance and the rest is accounted for in the grid stopper and bias/balance scheme. I use .33 uF coupling caps. The 6FQ7 drives the output stage easily, with headroom to spare.
Now, one thing I would consider doing differently if I were building this amp today is using a 12AT7 instead of a 6FQ7, only because I think the 12AT7 might be an excellent match for 6L6s, but I haven't done any homework to know where I would bias it or what kind of voltage swings I could get out of it.
Now, one thing I would consider doing differently if I were building this amp today is using a 12AT7 instead of a 6FQ7, only because I think the 12AT7 might be an excellent match for 6L6s, but I haven't done any homework to know where I would bias it or what kind of voltage swings I could get out of it.
Ok, that puts my concerns about the 6SN7 being up to the task to bed... it's electrically equivalent to the 6FQ7.
I've been thinking about using a CCS, but I don't really have any experience with them at all. I'm looking at the 10M45 data sheet, and it seems pretty straight forward... As a hypothetical example, it looks like it's: connect the A-pin to the cathode of both triodes, connect the G-pin to ground, and connect the K-pin to ground via a 90 ohm resistor for 12ma of constant current (6ma per triode). Is it really that simple?
Kevin Carter of K&K Audio sells some nice CCS kits for reasonable prices that might make it easier to implement. His forum is on AudioAsylum with links to his web page. Otherwise I've always built them with perfboards from the Shack. Not sure if they have them anymore as I haven't purchased any for a while. If you build one, use a couple of 9V batteries in series or a bench power supply to test them with your ammeter in series or a 1 or 10 ohm resistor in series so you can measure the voltage across it.. You can preset the current this way before you fire up the amp.
Have fun,
John
crispycircuit
Super Member
Eico HF-87&89 use EL34s for outputs not 6L6s......
Yes, I know. The point of this exercise is to build a 6L6GC-based amp, and the basic Eico HF-87/89 topology is one I'm considering, adapted, of course, to use the parts at my disposal.
Thanks for the reference. Luckily, I have a bunch of small perfboards in my box-of-random-parts, so I think I'll go with the bare 10M45S. I may just pick up a handful of them on my next Mouser order and build a couple of just to tinker with them.
Seems like the CCS kits (available from various places) have series resistors on the A pin that seem to be intended when used as a plate load ("to prevent oscillation"). I'm assuming that these aren't needed when the A pin is connected to a tube's cathode.
Ok, now I want to make sure I understand the HF-89 bias circuit.
The PT has a 75V bias winding. The diode and the first cap act as a negative peak detector, yielding -105V.
The 2 10K pots are in parallel, so combined with the 10K resistor form a 15K resistor that then neatly forms a 50% voltage divider with the 15K resistor that goes to ground. Each pot gives a range of -52.5 to -42V -- with the pot turned all the one way, the divider is 15K / 15K, and all the way the other way it's 15K / 10K.
NOW, on to the balancing network... Again, a voltage divider. For this example, let's assume we're being fed -47V from our tap on the bias supply. But the balance network is a little tricker... With the balance pot right in the center, seems like we have a divider formed by th 25K on each wide of the wiper and the corresponding 100K resistor, which would yield -37.6 for each tube. Pot rotated all the way towards one extreme, one tube has no divider at all, really, and gets the full -47V, and the other gets -31V or so. At 25% rotation it would instead be one tube would have -41.7V and the other -34V. I feel like I have must part of this wrong, because it's only linear for a very tight range in the middle of the pot's rotation.
Anyway, for dead center, it seems correct... I don't know what the typical bias voltage is for the HF-89, but looking at the EL34 data sheet, -37.6 seems completely plausible.
So, it seems like you could adjust the Eico balance network a bit to get the total grid circuit resistance in line with spec, so long as the ratios between the pot and the resistors were kept the same?
jazbo8
Well-Known Member
Basically, but it is likely to work as is without alteration. You use the 10k pots for the coarse adjustment, and the 50k pots for the fine adjustment. If you need more or less "raw" bias voltage, you can change either the 10k or the 15k resistor in the PS section.