Hi,,
I'm a tube newbee, but have built a couple of chip amps with discrete regulated supplies, buffers and other goodies. This looked like a nice project, by a respected designer, and I was able to get the tubes readily, some locally. Other stuff I had some of too.
http://tinpan.fortunecity.com/saints/668/primer/6ck4-pp.gif
The tube operating points seem ok. I calculate about 80v at the plates of the BZ7's. With a plate-cathode voltage of about 250 on the CK4's, I get about 105v across the cathode resistor, giving a bias voltage of about 25v (105-80), and a current of about 50ma/tube. That all seems to add up.
The power supply has me concerned though. It's simple enough, but my sims in PSUD don't give me confidence that it will work. The schematic seems to suggest that it's for a stereo amp, but there is no way I can get a B+ of 365V with a current draw of 200ma. In fact, with just a basic C filter (no neg supply, no driver section), I get about 350V at 100ma. So it seems marginal even for a monoblock. I'm thinking I need to up the secondary voltage and add another RC section (to reduce the input C) or use an input L. Am I on the right track?
Sheldon
I'm a tube newbee, but have built a couple of chip amps with discrete regulated supplies, buffers and other goodies. This looked like a nice project, by a respected designer, and I was able to get the tubes readily, some locally. Other stuff I had some of too.
http://tinpan.fortunecity.com/saints/668/primer/6ck4-pp.gif
The tube operating points seem ok. I calculate about 80v at the plates of the BZ7's. With a plate-cathode voltage of about 250 on the CK4's, I get about 105v across the cathode resistor, giving a bias voltage of about 25v (105-80), and a current of about 50ma/tube. That all seems to add up.
The power supply has me concerned though. It's simple enough, but my sims in PSUD don't give me confidence that it will work. The schematic seems to suggest that it's for a stereo amp, but there is no way I can get a B+ of 365V with a current draw of 200ma. In fact, with just a basic C filter (no neg supply, no driver section), I get about 350V at 100ma. So it seems marginal even for a monoblock. I'm thinking I need to up the secondary voltage and add another RC section (to reduce the input C) or use an input L. Am I on the right track?
Sheldon
Attachments
BZ has curves as linear as an AX, it's exactly twice in fact..(half Rp, double Gm and perveance).. I don't know what your distortion is coming from. I'll admit I haven't used one yet.
At any rate, first we have to fix your circuit...
No-no number one: wayy too much capacitance on the 5U4. Drop to 40uF *MAX*.
Instead of the two stage filtering, you can save some phase shift nightmares and use one 56k resistor for each channel. Which isn't necessary anyway since you have the output stages on the same rail (B+).
No-no number two: direct coupling. Oh the horrors! You MUST use NFB to maintain balance, otherwise you'll never get equal bias current in the output. That's a hassle, so coupling caps are just fine. In the same area, you've got positive grid bias, but the 2k cathode resistors might be large enough for that; I see no indication either way. (It would be helpful if voltages were marked on the schematic.)
No-no number three: your negative voltage supply isn't. It shows a coupling cap (from a filtered DC rail) to a filtered resistor divider. If you had included a diode, it *might* develop some voltage from the residual ripple, but nowhere near 30V. This needs to be a diode in series with a resistor to one 5U4 plate, to a filter cap and so forth, for instance as I used in Frankenhouse.
Finally; you noted regulation is half what it needs to be -- use solid state diodes instead. Tubes are just plain horrible at voltage drop. FWIW, at least using a smaller capacitor may give the 5U4 an easier job and increase output a little.
Tim
At any rate, first we have to fix your circuit...
No-no number one: wayy too much capacitance on the 5U4. Drop to 40uF *MAX*.
Instead of the two stage filtering, you can save some phase shift nightmares and use one 56k resistor for each channel. Which isn't necessary anyway since you have the output stages on the same rail (B+).
No-no number two: direct coupling. Oh the horrors! You MUST use NFB to maintain balance, otherwise you'll never get equal bias current in the output. That's a hassle, so coupling caps are just fine. In the same area, you've got positive grid bias, but the 2k cathode resistors might be large enough for that; I see no indication either way. (It would be helpful if voltages were marked on the schematic.)
No-no number three: your negative voltage supply isn't. It shows a coupling cap (from a filtered DC rail) to a filtered resistor divider. If you had included a diode, it *might* develop some voltage from the residual ripple, but nowhere near 30V. This needs to be a diode in series with a resistor to one 5U4 plate, to a filter cap and so forth, for instance as I used in Frankenhouse.
An externally hosted image should be here but it was not working when we last tested it.
Finally; you noted regulation is half what it needs to be -- use solid state diodes instead. Tubes are just plain horrible at voltage drop. FWIW, at least using a smaller capacitor may give the 5U4 an easier job and increase output a little.
Tim
SY said:
No I didn't ask that, but there are at least two reasons that shouldn't stop you. First, I may not have known to ask. Second, and more important, once we post a question here, it's community property and someone else might benefit even if I choose not to pay attention.
That aside, thanks. But do I need to parallel an AX7? Would the current from a single triode be insufficient? I've got the parts so I may as well breadboard up the circuit and learn something.
Sch3mat1c said:
Thanks Tim. I didn't make that schematic, I just downloaded it and pasted, so I can't take any credit, blame, or ownership. I guess, for attributions sake I should have explicity noted that.
I'll try the direct coupling just to see what happens. But I'll make sure I have some good caps on hand too. Yeah, I wondered about that negative supply. I didn't give it too much thought, as might be apparent (course, the other possible explanation is that I'm way behind the curve here - well, the truth will be revealed as we go) I now realize, as you point out, that even with a diode there can't be enough ripple to get the voltage that high. If I model a simple C filter with just a 100uf C and a 100ma current source, I get about 7V p/p ripple at the point where the "negative supply" would be. Partly I didn't think about it much, cause I figured I can always just make a separate negative supply. Your approach is even easier. Mainly, what I'm trying to nail down first is what I'd need in the way of a power transformer, cause the rest of it is easier to change. Since I've got a 300-0-300 tranny on the way, I'll give it a try and see what happens
Sheldon
Sch3mat1c said:
Tim, I don't absorb this stuff all at once yet - sort of in little quantum packets at this point. So I hope you don't mind a few more questions that came to me after the previous reply.
Are you saying to add both caps and negative feedback? As I read the circuit, I would think that the differential input stage would have reasonably balanced voltages on the outputs, so I'll ignore that as an issue for now. But would that be an issue I need to worry about?
As regards feedback, the output tubes have separate, fairly large, cathode resistors. Doesn't this provide feedback at the cathode, or is it not enough? As I read the circuit (remember my level now), I see the first stage output (calculated as riding on 80V DC) as an element in the bias level. If I get around 50-55 ma per tube, then the cathode would be at about 100-110V (for a bias voltage around -25). I can't just throw in caps, as then we're talking a whole different deal, aren't we?
As far as the values on the cathode resistors being ok to get to the correct bias point, I'm not sure. But at least it all seems in the right ballpark. I'm basing that only on inference and sorta reconciling the currents and voltages by looking at the circuit values given and at the tube data sheets.
All theory at this point, and that's why I want to do this little project - it's simple, but not too simple. And I can try some different things with it.
thanks,
Sheldon
Sheldon said:
Getting the DC conditions right, and stable, is always a problem with DC coupled amps. The differential/long-tailed pair is very likely to have a slight DC offset at the outputs (i.e. between the anodes) since the valves are never perfectly matched - leading them to draw slightly different currents. Ohm's law dictates that different currents through the same value resistance (anode load resistors) will create a different voltage.
The PP output stage is in essence a differential amplifier of sorts, and without the coupling caps, the DC offset will cause the operating point to be different for each valve in the output stage, causing DC across the output transformer - which is BAD.
Sheldon said:
The cathode resistors are bypassed by fairly large capacitors so they're essentially invisible to AC. There shouldn't be any feedback happening there at typical audio frequencies. I'm not entirely sure what you mean by just throwing in caps - the cathode bypass caps are already there. If you mean interstage coupling caps: yes, you'll have to recalculate the cathode resistor values.
audiousername said:
Ok, I think I get that one. Doesn't seem any trivial way around it. I guess it's a matter of degree.
audiousername said:
Yes, I was referring to DC. I intrepreted Tim's advice on caps to mean coupling caps. By feedback (and I'm probably using the term improperly), I meant DC feedback, i.e., the tubes should find reasonably close current values, assuming the grids are at the same voltage. Or is this not true?
Sheldon
Yes there is stabilisation of the operating point when using cathode bias (which is why you're generally allowed to use a larger grid leak resistor). It works because the bias voltage is a function of mean cathode current, such that if the mean current increases, more voltage is developed across the cathode resistor and the valve responds by decreasing conduction. The two effects oppose each other, which is desirable for a stable equilibrium to be reached. A similar argument can be made for a decrease in mean current.
I'm not sure if this answers your question, though. Maybe Tim should answer it, but he might be regenerating in his alcove right now
I'm not sure if this answers your question, though. Maybe Tim should answer it, but he might be regenerating in his alcove right now
audiousername said:I'm not sure if this answers your question, though. Maybe Tim should answer it, but he might be regenerating in his alcove right now
*Steps out of dramatic sparking green glowy thing*
we are dyslexia of borg ~ your ***** will be laminated
(Can the moderators make an exception to that word? The pun doesn't work otherwise...)
Anyways... yes I meant DC feedback to stabilize it. You could have AC NFB as well, but with all triodes it isn't really necessary.
If the cathode-grid voltage is 100-80 = 20V, then you can figure the proper resistor for the same current with the lower voltage, thanks to Mr. Ohm. You'll need to add grid leak resistors to keep the output grids anchored at 0VDC. Maybe 220k.
The front end will work the same. The output will see much more dissipation due to removing the
-ized waste of power in the cathode resistors.Lemme dig out the data sheet for this thing... lesse... 12W dissipation. At 365V, that's under 35mA bias, which on the graph shows up in the cutoff region.
Hold that thought on the power transformer; something lower, say 200-0-200V would be better suited. That might give between 200 and 280V (depending if you use a tube rectifier) which will fit a bit better. Distortion still isn't too good, but you probably can't do it without red plate. Ehh. It's probably most linear around 150-250V with plate dissipation run as high as it/you can stand it (which is going to be 80 to 48mA per tube, respectively). Bias in the range of 10-30V. Around 3 or 4ka-a for the OPT. I wouldn't expect more than 7W from this.Tim
Sch3mat1c said:[B
Anyways... yes I meant DC feedback to stabilize it. You could have AC NFB as well, but with all triodes it isn't really necessary.
If the cathode-grid voltage is 100-80 = 20V, then you can figure the proper resistor for the same current with the lower voltage, thanks to Mr. Ohm. You'll need to add grid leak resistors to keep the output grids anchored at 0VDC. Maybe 220k.
The front end will work the same. The output will see much more dissipation due to removing the
Lemme dig out the data sheet for this thing... lesse... 12W dissipation. At 365V, that's under 35mA bias, which on the graph shows up in the cutoff region.
Tim [/B]
Ok, thanks. That's what I thought. It sounds like fun. I'm going to try it according to the original (more or less) schematic and then see where I need to go to get it to actually work. Plenty of useful options here. One question though; wouldn't the much smaller cathode resistors actually defeat make the output stage more likely to be unbalanced in the absense of overall feedback? I guess with cap coupling I could go to a DC servo - something I'd like to learn.
Sy, thanks. I guess it's a choice between changing resistors or adding a socket. I'm lazy too. We'll see what happens. I'm going to get this going in a couple of weeks. I'll post measurements when I get there.
Sheldon
Just don't forget the first PSU cap and the negative voltage supply stuff; the original schematic *won't* work without these changes.
Yes, the cathode degeneration balances it a little - but not much. It would be perfectly balanced with infinite cathode resistors (i.e., CCS's), but 2k is far from infinite (though two times the plate resistance is pretty good in that regard).
Tim
Yes, the cathode degeneration balances it a little - but not much. It would be perfectly balanced with infinite cathode resistors (i.e., CCS's), but 2k is far from infinite (though two times the plate resistance is pretty good in that regard).
Tim
Sch3mat1c said:
Thanks. Sorry for not being clear. I meant I'd first try the amp more or less as shown, but I will use a smaller cap in the supply and rework the negative supply along the lines you suggested. I may play with some different supply configurations too, just for the exercise. If the amp's not too stable or is too noisy, I can use it as a test mule to try coupling caps, ax7's etc.. By then I should have a good idea of what I eventually need to make a reasonably stable stereo amp with a nice chassis and all. This is not intended to be the be all and end all, just a nice little starter project with a few twists to keep me interested I'm pretty much unconstitutionally unable to just follow a simple recipe, so I need those. And I'm already plotting the next one. It'll be based on an RC coupled 801 amp with a 6SN7 driver, that is shown on Steve Bench's site. It only puts out half a watt, but the speakers will be close to 110 db, and they will be tri-amped. I may be ignorant, but I'm an ambitious optimist.
By the way, I had wondered about CCS's on the output cathodes, but haven't really seen too many examples. Gary Pimm incorporates that on his 47 pp amp. http://home.pacifier.com/~gpimm/47.htm Each time I read that page, I pick up a little more. His stuff is pretty complicated, though. If I get there, it will be in small steps. One question; would there be any advantage in having separate CCS' s on each output tube, if they were made adjustable?
Sheldon
I've heard this amp bog standard using RCAs with a Coral Beta speaker. Very, very nice amplifier, around 6W as I recall.
As has been mentioned, anything with DC coupling causes problems, as a change in the 240V supply will cascade through to the output stage. If you use good caps, there really is nothing to worry about cap coupling, and this will release you from the tetchiness of adjusting operating points.
So, rather than calculate things ad insanem, build it, and see how it sounds. THEN fiddle with it!!
(Incidentally you can build this design with 6EM7 - dissimilar triode. The small triode is like a 6SL7 (mu of 68 at 1.6mA)and the large triode is like an indirectly heated 2A3 (mu of 5.4 with rp of 750R).
This cuts the bottle count and gives much the same performance.
Cheers,
Hugh
As has been mentioned, anything with DC coupling causes problems, as a change in the 240V supply will cascade through to the output stage. If you use good caps, there really is nothing to worry about cap coupling, and this will release you from the tetchiness of adjusting operating points.
So, rather than calculate things ad insanem, build it, and see how it sounds. THEN fiddle with it!!
(Incidentally you can build this design with 6EM7 - dissimilar triode. The small triode is like a 6SL7 (mu of 68 at 1.6mA)and the large triode is like an indirectly heated 2A3 (mu of 5.4 with rp of 750R).
This cuts the bottle count and gives much the same performance.
Cheers,
Hugh
AKSA said:
Thanks, that's my basic plan.
QUOTE]Originally posted by AKSA
(Incidentally you can build this design with 6EM7 - dissimilar triode. The small triode is like a 6SL7 (mu of 68 at 1.6mA)and the large triode is like an indirectly heated 2A3 (mu of 5.4 with rp of 750R).
This cuts the bottle count and gives much the same performance.
[/QUOTE]
Don't go tempting me with stuff like that. My resistance to cool tricks is low.
Sheldon
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.