What are the low cost PP xformer options for 829B amp in pentode connection (separate g2 VCC) and NFB ? I want to bias it for class AB operation with class A extending to ~5W (normal listening). However I need as much headroom in class B as the tube allows since I'd like to use it with my digital piano (large dynamic range). I think I need one in the 30-40W range but I would like some comments how much can I expect from 829B in pentode mode class AB.
I've looked at edcor xformers - there is a price/size gap between 25W and 50/60W models. Also triode electronics has TF110 for cheap but I think it is way too small. Any other half-decent options below ~$50 apiece ? Obviously with 829B tube I can not use UL taps.
I've looked at edcor xformers - there is a price/size gap between 25W and 50/60W models. Also triode electronics has TF110 for cheap but I think it is way too small. Any other half-decent options below ~$50 apiece ? Obviously with 829B tube I can not use UL taps.
I want to do a 829B PP Pentode as well some day. When designing from scratch,it's good see what power transformers are also available. Then you either have to simulate it or plot composite load lines different combinations to see what works best.
Since there aren't many PP Pentode 829B designs floating around out there, it may be good to reference some 807 amplifier designs and application data. The 829B is pretty much equivalent to two 807's in 1 envelope.
Hammond or Edcor is about as budget as your iron is going to get in the 30+ W range. Running in AB with some NFB, you may be able to get by with slightly undersized iron, since the nominal power consumption for audio will be quite a bit lower than running it with sine or square waves. However it's always better to go on the larger side.
Since there aren't many PP Pentode 829B designs floating around out there, it may be good to reference some 807 amplifier designs and application data. The 829B is pretty much equivalent to two 807's in 1 envelope.
Hammond or Edcor is about as budget as your iron is going to get in the 30+ W range. Running in AB with some NFB, you may be able to get by with slightly undersized iron, since the nominal power consumption for audio will be quite a bit lower than running it with sine or square waves. However it's always better to go on the larger side.
This is my first tube amp (but I have EE background). I am pretty much set on the topology - pentode-cathodyne front end, pentode mode output, class AB1 with fixed bias, global NFB. I am inclined to go with edcor 25W CXPP xformer. Do you think it is enough or should I shell out for 50W/60W CXPP ? 829B is close to a pair of 6l6 AFAIK.
Pentodes and OPTs and non-DC couplings will make errors. Less phase shift between errors and corrections the better. In the real world may mean you need more than just one all-encompassing feedback loop to get the job done right.
I would recommend local feedback around the OPT, and rely less on global. I know GNF is what you are most comfortable and familiar with, but its hard to fix an error after the fact. Especially when dealing with less than perfect components.
Don't overlook that Triodes have intrinsic internal negative feedback that keeps them linear. The reduced "performance" relative to Pentodes may be well worth the reduced local feedback complexity to include them.
I recommend you dig up a PDF copy of OHShade's infernal diatribe on the 6L6, its in one of RCA's books dating 1938. Many ancient forbidden secrets revealed...
I would recommend local feedback around the OPT, and rely less on global. I know GNF is what you are most comfortable and familiar with, but its hard to fix an error after the fact. Especially when dealing with less than perfect components.
Don't overlook that Triodes have intrinsic internal negative feedback that keeps them linear. The reduced "performance" relative to Pentodes may be well worth the reduced local feedback complexity to include them.
I recommend you dig up a PDF copy of OHShade's infernal diatribe on the 6L6, its in one of RCA's books dating 1938. Many ancient forbidden secrets revealed...
Triode or UL wired 829B won't work if doing a single envelope push pull, since the screens are tied together internally. Paralleling the devices would kind of defeat the purpose of choosing the dual device envelope, IMO. Also, some pentodes do really well triode strapped, but doing so to an 829B, 807, 6l6 neuters the tube.
The point about keeping phase shifting to a minimal is a critical one when using GNFB. Try to keep it a 3 stage design as you are planning. You should be able to direct couple the input stage to the phase splitter if you play with it a little. That will remove 1 phase shifting object from the chain. As stated, finding clever ways to add local feedback, to reduce global is desirable, but sometimes it is not possible. Your probably already aware that compensation circuitry will require some tweaking.
You may want to look into using a high mu triode on the input rather than a pentode. If you can find one with enough gain to allow for the amount of feedback you plan to use. Either case, you will probably have to leave the input tubes cathode resistor partially bypassed. That way open loop gain will still be on the higher side, but you can still return feedback to that point.
I think the 25W CXPP xformer should do the job. I'd be surprised if the amp puts out much more than 30Wrms when it's done(unless your going AB2, but I don't think you are). Full power performance with sine waves may not be good, but nominal power for audio is much lower than sine waves in class AB.
The point about keeping phase shifting to a minimal is a critical one when using GNFB. Try to keep it a 3 stage design as you are planning. You should be able to direct couple the input stage to the phase splitter if you play with it a little. That will remove 1 phase shifting object from the chain. As stated, finding clever ways to add local feedback, to reduce global is desirable, but sometimes it is not possible. Your probably already aware that compensation circuitry will require some tweaking.
You may want to look into using a high mu triode on the input rather than a pentode. If you can find one with enough gain to allow for the amount of feedback you plan to use. Either case, you will probably have to leave the input tubes cathode resistor partially bypassed. That way open loop gain will still be on the higher side, but you can still return feedback to that point.
I think the 25W CXPP xformer should do the job. I'd be surprised if the amp puts out much more than 30Wrms when it's done(unless your going AB2, but I don't think you are). Full power performance with sine waves may not be good, but nominal power for audio is much lower than sine waves in class AB.
Thanks for the input. It is AB1 so there will be no additional drivers after cathodyne, it is essentially a 2 envelope per channel design with something like 6gh8 in the front end.
Pentode-cathodyne will be directly coupled. I am also playing with the idea of bootstrapping from cathodyne output to the input pentode plate circuit. Doing some spice simulation.
I understand the problems with global NFB. But I can not use UL with 829B and I want to use single tube per channel. PP xformer will be smaller/cheaper than SE one. Also the amount of NFB will be fairly limited as I want fairly large voltage gain, sort of an integrated amp.
The feedback will go to the pentode cathode circuit. For now I am thinking of taking NFB from OPT speaker output plus a small cap from one of the final plates.
Pentode-cathodyne will be directly coupled. I am also playing with the idea of bootstrapping from cathodyne output to the input pentode plate circuit. Doing some spice simulation.
I understand the problems with global NFB. But I can not use UL with 829B and I want to use single tube per channel. PP xformer will be smaller/cheaper than SE one. Also the amount of NFB will be fairly limited as I want fairly large voltage gain, sort of an integrated amp.
The feedback will go to the pentode cathode circuit. For now I am thinking of taking NFB from OPT speaker output plus a small cap from one of the final plates.
I briefly looked through the paper - lot's of useful info. However the proposed LNF has it's drawbacks. The cathodyne driver/splitter has asymmetric output impedance and does not fit well with feedback into output stage grids. Also it would be nice to use transformer output for feedback source but symmetric output winding is needed.
Somehow I knew it was coming to a suggestion of a fully symmetrical front end with lots of tubes and multiple feedback loops 
. BTW I am not sure output winding wire gauge is the same for all taps, also speaker loading is not symmetrical unless you want to use 8ohm only. On top of that 829B does not have separate cathodes.
I know that local feedback is both better and easier to debug. But I've done global in the past with solid state with reasonable success. Given that the amount of feedback is not extreme I think it can be done with reasonable stability.
. BTW I am not sure output winding wire gauge is the same for all taps, also speaker loading is not symmetrical unless you want to use 8ohm only. On top of that 829B does not have separate cathodes.I know that local feedback is both better and easier to debug. But I've done global in the past with solid state with reasonable success. Given that the amount of feedback is not extreme I think it can be done with reasonable stability.
Agreed,
For your first tube build, no need to get too complex and fancy. The topology you have been considering should do well (input stage, cathodyne, Push-pull Pentode with global feedback). Many amplifiers use that configuration. As with any design, it will just take some tweaking for it to come together. The edcor 25W CXPP xformer should do the job just fine.
For your first tube build, no need to get too complex and fancy. The topology you have been considering should do well (input stage, cathodyne, Push-pull Pentode with global feedback). Many amplifiers use that configuration. As with any design, it will just take some tweaking for it to come together. The edcor 25W CXPP xformer should do the job just fine.
Speaker symmetry around the 4 ohm tap is achieved at 16 ohms.vladn said:Somehow I knew it was coming to a suggestion of a fully symmetrical front end with lots of tubes and multiple feedback loops
Taps are taps, the actual impedance is a function of the amplifier.
All that matters is you got enough excess of choking Z to keep it
operating as a transformer rather than do-nothing length of wire.
Don't get me lying about what gauge(s) might be those wires???
Who sais you have to have separate cathodes to implement local
feedback? Or a symmetrical front end? One extra triode (probably
in same envelope) after the cathodyne is suddenly a lot of tubes?
I'm sorry if I presented too much information that made it seem
far more complicated than necessary.
Can see now that I'll have to back my words with a schematic...
Gimme a few hours to whip something up. I figure you only
need the topology, and are more than qualified to fill in the
specific component values on your own.
kenpeter - I mentioned separate cathodes only because the schematics you referred to has cathode feedback in the final (plus two more NF loops).
On a different subject - for power supply I plan on using 225V-250V for output G2 and front end and 450V-500V for output B+. I'll probably use center tap on the secondary for 225V. Do the values seem reasonable for 829B or should I go with higher B+ ?
As for the power xformer I'll try to find reasonably flat toroid. I'd like to hide it in the amp base to make it more compact. Or maybe I'll hack a simple non-regulated fixed frequency switching inverter, I have few large kool-mu E-cores with bobbins and some HV mosfets laying around.
On a different subject - for power supply I plan on using 225V-250V for output G2 and front end and 450V-500V for output B+. I'll probably use center tap on the secondary for 225V. Do the values seem reasonable for 829B or should I go with higher B+ ?
As for the power xformer I'll try to find reasonably flat toroid. I'd like to hide it in the amp base to make it more compact. Or maybe I'll hack a simple non-regulated fixed frequency switching inverter, I have few large kool-mu E-cores with bobbins and some HV mosfets laying around.
Well, I fell asleep, and now I got LabView classes all day today...
I still intend to deliver on that drawing. But not today.
Now if you are willing to tolerate two pentodes and two triodes
in the front end (next spin, not this one) Why not a differential
pair of Triodes (with intrinsic local feedback), topped by the pair
of Pentodes instead?
Each Pentode rigged to serve doube duty as both a constant
current source for the Triode's plate, and a cathode follower
(of nearly unlimited current at the top of the sense resistor)
for driving the Load. This circuit we call a "Mu Follower".
I still intend to deliver on that drawing. But not today.
Now if you are willing to tolerate two pentodes and two triodes
in the front end (next spin, not this one) Why not a differential
pair of Triodes (with intrinsic local feedback), topped by the pair
of Pentodes instead?
Each Pentode rigged to serve doube duty as both a constant
current source for the Triode's plate, and a cathode follower
(of nearly unlimited current at the top of the sense resistor)
for driving the Load. This circuit we call a "Mu Follower".
I was actually experimenting in spice with SE mu-follower like configuration as an integrated gain/splitter stage using pentode at the bottom and triode at the top. Phase splitter is formed by adding triode plate resistor and another resistor from triode cathode to the ground - all DC coupled. Works kind of neat (except requiring pretty high supply voltage). The output impedance is again asymmetric but in a different way than plain cathodyne - the cathode phase impedance is higher because it is essentially bootstrapping a pentode. Neither cathodyne nor this circuit is suited for driving output into AB2 though.
Since there are precious few dual pentodes in a single envelope, it seems that you can't get away without 3 bulbs per channel. You could perhaps use something like a pair of ECF80, pentode LTP that doubles as a phase splitter, triodes as followers. It's not going to do lots of AB2 this way but it will give you a 'little' extension into positive grid voltages.
The nice thing about this approach is that you get both O. Shade style plate to grid feedback insertion points, as well as traditional GNFB.
The first you can get by returning OPT ends to the corresponding pentode LTP plates. Feedback goes down to DC in theory, bonus is that you only need one coupling cap per side, which is what sets the LF corner frequency, that's the one from pentode plate to triode follower grid. Fixed bias for the output tubes is set on the follower grids, of course.
Also, since the input is a LTP, the 'other' grid can be used for traditional GNFB insertion.
Downside: calculating the resistor network for the plate-to-plate feedback can be a bit tricky.
IRC, someone did something like this already... P. Millet?
The nice thing about this approach is that you get both O. Shade style plate to grid feedback insertion points, as well as traditional GNFB.
The first you can get by returning OPT ends to the corresponding pentode LTP plates. Feedback goes down to DC in theory, bonus is that you only need one coupling cap per side, which is what sets the LF corner frequency, that's the one from pentode plate to triode follower grid. Fixed bias for the output tubes is set on the follower grids, of course.
Also, since the input is a LTP, the 'other' grid can be used for traditional GNFB insertion.
Downside: calculating the resistor network for the plate-to-plate feedback can be a bit tricky.
IRC, someone did something like this already... P. Millet?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.