didit wrote: ↑Tue Aug 20, 2019 3:11 pm
Hello Dave -
CF buffering into power tubes seems a sound idea. So far no one suggested LTP PI. Suggest that is also sound, coupled with CF -- better performance all round versus pushing a cathodyne; offering some gain, firm dynamics and ready potential to balance.
Also note that in UL, a pair of KT88 has relatively high primary impedance requirements. The A451 will work. However you'll likely find it works best modelled as 4.4K so each secondary tap winds up at 2x the loading spec - eg. 4 Ohm tap is best with roughly an 8 Ohm load. This may recommend having it wound with spec'ed 2-4-8 versus 4-8-16.
Best .. Ian
martin manning wrote: ↑Tue Aug 20, 2019 5:11 pm
I think a LTP with CF drivers would be fine, and very SSS-like. SSS had extreme +/- voltages with large cathode resistors. I proposed the bipolar supply (running off the common bias tap) to get away from the potential hazards (heh) of doing it that way. It's been proven out by several TAG members.
If I understand correctly, Colossal wants to keep it all-tube, but FETs are an option.
Ok, this is good counsel. LTPI it is then! Should we go with a 12AX7 and the Dumble values with a balance trimmer? Or another 12AU7 or even an 12AT7? I am using Martin's 160V bipolar supply running 120-0 and a FWB for the driver circuit. Driver tube will be 12AU7.
Ian, to address your concern about the OT primary impedance...I will be running a quad of 6550s at 450VDC under load and Ultralinear. This application is right out of the Genalex KT88 datasheet. The A451 is 2k2 on the primary so that would of course be 4k4 for a pair, so we should be compliant.
I am updating my schematic. Heyboer got the updated specs for the PT and choke.
Yes, I would like to stick with all tubes for this design. On to the preamp next!
I'm not sure a cathode follower makes as awesome a driver as it looks on a schematic. They'll drive the crap out of miller capacitance, so bandwidth won't be a problem, but it isn't a problem anyway. If you try to drive the OP grids much positive the CF falls on it's face. They run out of current, NFB collapses, output resistance shoots way up, and all of a sudden they're clipping. Using really big tubes for CF will get you a few more volts, but it's only a few percent addition swing compared to what you get without driving the grids positive.
And to design, build, and tune a direct coupled CF that is reasonably fail-safe... (They control the OP grids. What happens during power up? What if one of them fails? Do those output tubes then go into meltdown mode?)
Me personally, if I really wanted to drive a quad or sextet of big bottles big-boy style, I would probably consider an interstage transformer. Maybe put a 6V6 in front of it and have it wound with a mild stepdown like 2:1. That will outperform direct coupled CF's in just about every way - except perhaps bandwidth. But as I already said, bandwidth is not an issue.
I don't see any issues here. You can use a large grid stopper to limit CF driver current without fear of Miller effect, and anyway there is no risk of blocking distortion since there is no coupling cap. Transient on start-up is handled by the drivers coming on line with the main power (like a conventional bias supply), and until they start conducting (of if they were to fail) you have essentially the full negative voltage on the power tube grids, keeping them in cutoff.
Dave, you need 120-0-120 for the bipolar supply, not 120-0.
I'd make a few changes to that design. I don't think it's important to balance the +/- PT current, and removing the caps from the bias pot wipers may actually lower the ripple content in the outputs.
You do not have the required permissions to view the files attached to this post.
Colossal wrote: ↑Tue Aug 20, 2019 11:23 pm
I will be running a quad of 6550s at 450VDC under load and Ultralinear.
Of course you are. Simply lost track of the narrative somewhere.
martin manning wrote: ↑Thu Aug 22, 2019 1:57 pm
I don't see any issues here.
Agreed.
Abstractly reliability and weirdness in failure concerns would need thought. However CF buffer is 50+ years tried & true. Interstage transformer works OK, though with differing response.
Here's hopefully helpful thread from last fall on the topic - Post Phase Inverter Cathode Follower. Apologies for paying little attention this week, but day job needed it.
Best .. Ian
Last edited by didit on Sat Aug 24, 2019 12:43 pm, edited 1 time in total.
It's not the failure modes of direct coupled CF that bother me; that's just an add-on. What turns me off of the idea is that they just don't provide the real-world performance advantages that they seem to promise. A LOT of work for not much gain (no pun.)
Colossal wrote: ↑Thu Sep 26, 2019 7:52 pm
Bumping my own thread...
The transformers are almost done I would like to get some input on the preamp design next.
Can't go wrong with the ole bassman topology for preamp, the bigger cathode bypass caps give the bass the room it needs to thump. Not sure if other's know specific other topologies that work better. You could either go with the simple 5F6A style preamp, or for the more modern blackface AB165 I guess.
pompeiisneaks wrote: ↑Thu Sep 26, 2019 8:30 pm
Can't go wrong with the ole bassman topology for preamp, the bigger cathode bypass caps give the bass the room it needs to thump. Not sure if other's know specific other topologies that work better.
~Phil
xtian wrote: ↑Thu Sep 26, 2019 10:04 pm
The Alembic F-2B is a common choice for bass players:
Thanks guys, that's actually a great idea. I've been thinking about what preamp to mate to the amp and you can't go wrong with that classic topology. Any thoughts on a clipping stage after the recovery stage, before the PI? The grid leak would be a pot to adjust a cold clipper.
So the output transformer came today. It is an A-451 U/L clone. It is stupidly large and at least twenty pounds in hand. Power transformer and choke on the way.
geddyandalexworking.jpeg
You do not have the required permissions to view the files attached to this post.
Last edited by Colossal on Fri Nov 01, 2019 1:11 am, edited 1 time in total.
You can add a classic cold stage with a stereo lin pot acting as a voltage divider before and after it (reverse wire the second).
This way you can decide how much it will overdrive while keeping the output level almost constant (minor adjustments like adding resistors to ground are needed).
Just to resurrect this thread...I am back to working on the design. I've been speaking with Roberto and Martin, refining some ideas. I have the transformers on hand and working on the chassis drawing.
Here are where things stand, so far:
1) Power amp will be 4 x 6550 running in Ultralinear mode. I got some very nice Winged =C= 6550 pulls from Xtian (thanks dude!). Output transformer is a Dynaclone A-451 2k2 primary, 4/8/16 secondary, wound by Heyboer. It is massive!
2) Post PI cathode follower driver will be a 12AU7 running Martin Manning's CF driver design. A dedicated 120-0-120 winding was added to the PT for this purpose.
3) Phase inverter will be 6SL7 running as a long tailed pair with Aiken's design/values (82k plates, 680ohm cathode, 1M grid leaks, 47k tail).
4) The whole power supply will be C-L-C filtered before B+1. This will be 100uF (two 100-100uF F&T 500V in series for each node), 10H 57ohm 570mA choke, then 100uF.
As for the preamp, I am working on ideas. I have been discussing the idea of adding a clipping circuit as my friend likes the growly tones of Chris Squire and Geddy Lee. Roberto has suggested the idea of a Cascode -> Baxandall -> Cascode arrangement for the preamp, using 6SL7s. This is interesting! The clipping circuit may or may not be added after. A boost in front of the amp might be better and could easily push the second cascode to clipping.
I would like to get some input on the preamp design next.
