[L0rdGwyn]

Thanks @Xcalibur255 , I appreciate the compliment, we'll see if it beomes a reality! This one is quite a challenge for me...the amplifier schematic is relatively straightforward, in fact I think I have a decent draft in LTSpice. But the power supply is another story. Very likely I will need a custom-wound mains transformer to pull it off, the DIYer I mentioned who has built a similar amplifier used a custom mains with dual secondaries, one secondary having multiple taps for the bias / source follower V- supply. I figure if I take my time, I can work out the best approach.

 

[Tjj226 Angel]

Thanks @Xcalibur255 , I appreciate the compliment, we'll see if it beomes a reality! This one is quite a challenge for me...the amplifier schematic is relatively straightforward, in fact I think I have a decent draft in LTSpice. But the power supply is another story. Very likely I will need a custom-wound mains transformer to pull it off, the DIYer I mentioned who has built a similar amplifier used a custom mains with dual secondaries, one secondary having multiple taps for the bias / source follower V- supply. I figure if I take my time, I can work out the best approach.

..........?

Why would you need a custom mains transformer?

 

[L0rdGwyn]

B+ for the 841 is around 600V, B+ for the 801A in A2 is around 350V, then a negative bias supply for the FET source follower, 150-200V. The way I have seen this done was with a dual secondary mains, with both secondaries stacked to get the 841 B+, the bottom having multiple taps for the -V supply. Another approach might be to use multiple off-the-shelf transformers, but like I said, it's a work in progress, going to be looking into the options over the next week. But I am open to suggestions

 

[Tjj226 Angel]

B+ for the 841 is around 600V, B+ for the 801A in A2 is around 350V, then a negative bias supply for the FET source follower, 150-200V. The way I have seen this done was with a dual secondary mains, with both secondaries stacked to get the 841 B+, the bottom having multiple taps for the -V supply. Another approach might be to use multiple off-the-shelf transformers, but like I said, it's a work in progress, going to be looking into the options over the next week. But I am open to suggestions

When you say you want 600v on the 841, I assume you mean that you are going to be using a CCS right? Have you calculated what your driver requirements are for the 801a?

 

[L0rdGwyn]

When you say you want 600v on the 841, I assume you mean that you are going to be using a CCS right? Have you calculated what your driver requirements are for the 801a?

Yes, CCS loaded with mu-output. The driver will need to swing 150Vpp to drive the 801A to clipping. The 841 can do that well CCS loaded with a 450-500V bias point, 600V B+ is ballpark, could go to 625 or 650 for a higher plate voltage. At close the plate saturation on the positive grid wing (+40-50V on the grid), the 801A could draw something like 10mA of grid current. FET I am modeling right now is the IXYS IXTU01N100 for the buffer, source loaded with a CCS as well, right now I have to IXYS IXCP10M45S plugged in.

So yeah, figuring out a good way to configure the PS is going to be the challenge. Maybe there is a simple way to do it, but this is new territory for me, I have a lot of homework to do.

 

[Tjj226 Angel]

I might just be crazy right now, so please double check my math.

Lets assume you have a 2v rms line input. 2Vrms is about 2.83v peak or about 3v peak. You multiply this by the gain of the tube at 30mu and you end up with 90v peak or 180v peak to peak.

Since you are going to loose some gain due to various reasons, I think hitting 150v peak to peak with a 2v input is reasonable.

As such, I wouldn't bother making the grid bias on the 841 much more than 3v at 4-5ma. Two red LEDs at that current should give you something like 3.4-3.6v of bias. ; )

Now you are talking about a plate voltage of around 300 and you might be able to push your luck and get away with a B+ of something like 450v. This might not sound like a huge deal, but it might just be enough of a reduction where you can use basic 600v power supply capacitors.

Im not super familiar with hybrid mu followers, so maybe there is some extra math that needs to be done to get you to the 600v figure, but as far as I can tell, you ""SHOULD"" be able to get away with much lower B+

 

[L0rdGwyn]

Thanks @Tjj226 Angel , definitely something to consider. Looking ahead the other day, I thought of using 800V DC link caps or something similar if I do end up with a high plate voltage on the 841. If I thought I could get happy bias points with a more conventional power supply, I definitely would go for it. Thinking that over, an alternative bias point for the 801A could be 400V / 40mA / -20Vg with a higher primary load. Using fixed bias, would need maybe 425V at the top of the OPT. Would be cutting it very close trying to use that same rail for the CCS loaded 841...just don't think it will work, will either be compromising the 801A bias point or the 841. I think I am going to have to get creative with the power supply still, whether that be stacking, some sort of voltage-doubler arrangement, etc.

If I could have someone like Heyboer wind me a dual secondary traffo at a reasonable price, that might be an idea, I'm told they will do anything. I plan to regulate both supplies, so doesn't necessarily need to be fancy. The other appeal of doing something custom is I don't need all of the heater windings included on off-the-shelf transformers, just the B+. If I went for dual secondaries of 250-0-250, I could stack the two supplies for 350+350VDC and run the bias supply off the bottom. These numbers are all very rough, that might get me around a 650V B+ for the 841, 325V B+ for the 801A, and my bias supply for the IXTU01N100. These are just the ideas I'm throwing around and I'm probably overlooking something, but gotta start somewhere.

I had actually thought of using filament bias on the 841. At 1.25A on the filament, a 3-5ohm 20W resistor would get me 3.75-6.25Vg. Would have to be heat-sinked though. I'm probably going to experiment with LEDs on the 45 amp, if that goes well maybe I will use it on this amp rather than filament and avoid dissipating the ~10W.

 

[Tjj226 Angel]

Just use an antek transformer and have two voltage rails if it is really that big of an issue.

https://www.antekinc.com/as-4t430-400va-430v-transformer/

Just use a choke input filter on the winding for the 801a to get you down closer to 350v and do what ever you want for regulation, and then a CRCRC filter on the other rail for your 841.

If you don't need filament windings on the PT, then there is really no need to have one custom wound. There are plenty of options for cheap : D

------------------------------

Im also thinking this may not even work. If the 801a needs to have 10ma of current available to draw from, then the 841/mu follower would need to be biased at something higher than 10ma.

This is one of those situations where I feel very strongly that there is some stupid simple solution to make this whole darn thing super cheap and easy to put together. We just need to find it.

 

[L0rdGwyn]

@Tjj226 Angel I like your idea of using the dual secondary Antek with a choke input! Something tells me there will be grounding headaches with the stacked supply, using two completely separate rails might be a better approach. I was looking at Toroidy too, but their dedicated high-voltage transformers (no heater windings) are 230V input only...might just go with unused heater windings, but that kind of stuff drives me crazy.

So talking with a DIYer who knows these tubes well, even with a mu-output, the 841 cannot drive the capacitance of the 801A, so that is the reason for the FET source follower. I was planning to cap couple the 841 to the source follower, then the SF will be direct-coupled to the 801A grid. So the grid current will be supplied by the IXTU01N100 (or whatever other FET I end up going with, although I think this is a solid option). I had planned to use a CCS on the IXTU01N100 source, which to my understanding must be set higher than the max possible grid current. Increasing the drain current also increases the transconductance of the IXTU01N100, so the higher the better, within the realm of practicality. Dropping something like 275V across the IXTU01N100 at 30mA constant current, would need to dissipate around 8W. Still picking this stuff up, so I'll confirm in LTSpice.

If I went with something like the Antek, would need to use a capacitor coupled full-wave bias supply to get the negative voltage for the source follower. Will have to look up some schematics, would have to be below the negative grid swing of the 801A, around -100V plus headroom for the CCS. Like I said, I'm using the IXCP10M45S as the CCS at the moment. I have a week off of work starting tomorrow (!!!), I'll post a LTSpice schematic once I've messed around a bit more, should make more sense then I think.

What I am doing with the FET buffer is pretty similar to what Pete Millett has cooked up here with his "A2 buffer". Tubelab has his own version which he calls "Powerdrive".

http://www.pmillett.com/A2_buffer.html

This will be one cyborg of a hybrid amp. I'm probably committing some unspeakable DHT sins by going hybrid, but I have read overwhelmingly positive impressions about using a source follower rather than a cathode follower for A2 grid drive, and I'm not opposed to some sand assistance

 

[Xcalibur255]

Is the challenge in particular the thing you're going for? It seems like choosing a different driver would be easier, but if you're super interested in the 841 that is all the reason one needs I suppose.

Have you considered just 10 squaring the amp? That is driving the 801A with another 801A or 10Y. It's still an all DHT TTF design but should simplify your powersupply since both tubes would have the same needs, and you should still have enough gain.

 

[L0rdGwyn]

Yes @Xcalibur255 , using a different driver would make things considerably easier! If all else fails, that will be the fallback option, but I think this can be done. The 841 is a finicky tube, and I am certainly prone to tunnel vision, but the issue with using the 10Y as a driver is the gain just won't be enough for the 801A in A2. With a 2V RMS input, 5.6V peak-to-peak, the 10Y will only swing something like 40-50Vpp, where the 801A in A2 will need 150Vpp to reach max output. The 841 with its mu of 30 is just about enough gain for the 801A. To use the 10Y, would have to add an additional gain stage or use a step-up interstage transformer to couple it to the source follower, but even then the gain likely wouldn't be enough. I'll revisit the idea though, I have heard of people also using step-up line-input transformers to get adequate gain in all DHT amplifiers, but then the cost and added complexity starts to seem pretty comparable to what I am seeing with the 841 power supply.

 

[Tjj226 Angel]

@Tjj226 Angel I like your idea of using the dual secondary Antek with a choke input! Something tells me there will be grounding headaches with the stacked supply, using two completely separate rails might be a better approach. I was looking at Toroidy too, but their dedicated high-voltage transformers (no heater windings) are 230V input only...might just go with unused heater windings, but that kind of stuff drives me crazy.

So talking with a DIYer who knows these tubes well, even with a mu-output, the 841 cannot drive the capacitance of the 801A, so that is the reason for the FET source follower. I was planning to cap couple the 841 to the source follower, then the SF will be direct-coupled to the 801A grid. So the grid current will be supplied by the IXTU01N100 (or whatever other FET I end up going with, although I think this is a solid option). I had planned to use a CCS on the IXTU01N100 source, which to my understanding must be set higher than the max possible grid current. Increasing the drain current also increases the transconductance of the IXTU01N100, so the higher the better, within the realm of practicality. Dropping something like 275V across the IXTU01N100 at 30mA constant current, would need to dissipate around 8W. Still picking this stuff up, so I'll confirm in LTSpice.

If I went with something like the Antek, would need to use a capacitor coupled full-wave bias supply to get the negative voltage for the source follower. Will have to look up some schematics, would have to be below the negative grid swing of the 801A, around -100V plus headroom for the CCS. Like I said, I'm using the IXCP10M45S as the CCS at the moment. I have a week off of work starting tomorrow (!!!), I'll post a LTSpice schematic once I've messed around a bit more, should make more sense then I think.

What I am doing with the FET buffer is pretty similar to what Pete Millett has cooked up here with his "A2 buffer". Tubelab has his own version which he calls "Powerdrive".

http://www.pmillett.com/A2_buffer.html

This will be one cyborg of a hybrid amp. I'm probably committing some unspeakable DHT sins by going hybrid, but I have read overwhelmingly positive impressions about using a source follower rather than a cathode follower for A2 grid drive, and I'm not opposed to some sand assistance

Oh ok, so you are doing a full class A2 fet follower. That will definitely help.

With the Antek transformer, you will notice that the input side of the transformer has a second 120v winding for 240v. You can use that winding for your negative bias supply so long as you aren't using too much current from the transformer.

You wouldn't really have stacked power supplies. You would just have multiple power supply rails. Grounding wouldn't be too bad in either case.

 

[L0rdGwyn]

Well I have my week off starting today, so I'm going to dive back in to the 801A amp design looking at some of the discussion that we've had here. I really like the idea of using a toroid - I am planning a separate PS chassis with an umbilical, so using an internal toroid vs. a potted EI core that I have to mount on top would allow me to stack the two chassis. My girlfriend is already wondering where the heck I'm going to put such a big amp...minor details, there is always more room

Very exciting news today though - Sowter had originally told me 12 weeks on the 45 parafeed transformers. Well I received a shipment notification today, it has only been a little over 3 weeks! Not sure what changed, but the transformers will be here Friday, perfect timing. Looks like the 45 amp is going to be built way ahead of schedule, going to be a busy weekend.

 

[Tjj226 Angel]

Well I have my week off starting today, so I'm going to dive back in to the 801A amp design looking at some of the discussion that we've had here. I really like the idea of using a toroid - I am planning a separate PS chassis with an umbilical, so using an internal toroid vs. a potted EI core that I have to mount on top would allow me to stack the two chassis. My girlfriend is already wondering where the heck I'm going to put such a big amp...minor details, there is always more room

Very exciting news today though - Sowter had originally told me 12 weeks on the 45 parafeed transformers. Well I received a shipment notification today, it has only been a little over 3 weeks! Not sure what changed, but the transformers will be here Friday, perfect timing. Looks like the 45 amp is going to be built way ahead of schedule, going to be a busy weekend.

Ugh. Your'e making me look bad. Now I need to drag myself out of bed and find that darn drill bit I am missing to finish up my 45 build. : P

 

[Xcalibur255]

Yes @Xcalibur255 , using a different driver would make things considerably easier! If all else fails, that will be the fallback option, but I think this can be done. The 841 is a finicky tube, and I am certainly prone to tunnel vision, but the issue with using the 10Y as a driver is the gain just won't be enough for the 801A in A2. With a 2V RMS input, 5.6V peak-to-peak, the 10Y will only swing something like 40-50Vpp, where the 801A in A2 will need 150Vpp to reach max output. The 841 with its mu of 30 is just about enough gain for the 801A. To use the 10Y, would have to add an additional gain stage or use a step-up interstage transformer to couple it to the source follower, but even then the gain likely wouldn't be enough. I'll revisit the idea though, I have heard of people also using step-up line-input transformers to get adequate gain in all DHT amplifiers, but then the cost and added complexity starts to seem pretty comparable to what I am seeing with the 841 power supply.

You clearly know more than I do so I should probably stop offering my 2 cents knowing I'll have change coming. I'm curious about this though, because people definitely do a 10Y driving another 10Y or an 801A driving another 801A and those designs don't always have step-up transformers involved. Maybe it's because they are staying in A1 and not pushing for as much power as you're going for?