ok, i'm posting first off like that guy at the beginning of Monty Python and the Holy Grail - "not dead yet!" this project, that is. basically the only thing keeping me from building this sucker is my laziness at having to wake up early to go to this metal shop in Dorchester to have a suitable aluminum plate punched for me, 20x14x1/8". one other thing is these second thoughts I keep having.
i've been looking over the schematic and i notice a standard grounded-cathode gain stage between the 6SL7 phase splitter and the 6L6 finals. i was wondering if it would be OK to nix that gain section, seeing as most CD players etc. already have higher voltage output, and i like the idea of a separate linestage (i.e. i already have one). does this sound like an OK idea?
one more idea: the biasing of the 6L6 finals seems pretty basic, one resistor. if i wanted to really nerd it up, would it be worth it to substitute each tube's cathode resistor with a lower value, and a smallish little trimpot, to allow fine-tuning of the tubes' bias current? or, if i'm going to go to the hassle of modding the schematic, should I just figure out how to make a CCS? i figured out what current the schematic set to run through the resistor, i have it written down somewhere.
i've been looking over the schematic and i notice a standard grounded-cathode gain stage between the 6SL7 phase splitter and the 6L6 finals. i was wondering if it would be OK to nix that gain section, seeing as most CD players etc. already have higher voltage output, and i like the idea of a separate linestage (i.e. i already have one). does this sound like an OK idea?
one more idea: the biasing of the 6L6 finals seems pretty basic, one resistor. if i wanted to really nerd it up, would it be worth it to substitute each tube's cathode resistor with a lower value, and a smallish little trimpot, to allow fine-tuning of the tubes' bias current? or, if i'm going to go to the hassle of modding the schematic, should I just figure out how to make a CCS? i figured out what current the schematic set to run through the resistor, i have it written down somewhere.
idkTweeker said:
It's a posibility, seeing as I have a pair of those lying around. The problem is, I already have two pairs of matched-section Sovtek 6SN7GT's and 6SL7GT's and the sockets to go with them
. I don't suppose I could just use a 6SN7 to drive the 6L6's, if they have lower output impedance? my main question is: how do i figure out where my 6L6's are biased? i have the plate curves for the 6L6, i know that the cathode resistance is 700 ohms per tube, and the voltage at the centertap is 385 ideally, and that the load through the transformer is 3300 ohms per tube. when i apply this to a tube curve, i can't find a value along the curve that spits out 700 ohms for the cathode resistor. what am i doing wrong? does the grid stopper have something to do with it?
OK, reading now ( http://members.aol.com/sbench102/composit.html ). i'll also browse Valve Amplifiers to get more of an idea of wth is going on . thanks!
. thanks!sorenj07 said:
It's a posibility, seeing as I have a pair of those lying around. The problem is, I already have two pairs of matched-section Sovtek 6SN7GT's and 6SL7GT's and the sockets to go with them
Here's another crazy thought... with some extra 6SN7's you could just wire it up as a Williamson amplifier instead, using the Hafler/Keroes output stage -- this has been done many times, and sounds verrry good. This would give you much more reasonable (lower) gain, and likely lower distortion. There are many great examples of that circuit online (Acrosound catalog, for example). Your power supply / OPT would be perfect.
Save the 6SL7's for a phono preamp, perhaps?
sup jon. about the williamson, that's an idea and a half i'll definitely take a look, i think i have the article where Hafler and Keroes hijack the Williamson driver/splitter stage lying around somewhere. didn't make the williamson crowd too happy, iirc. Of course, it means picking up new biasing junk, but that's peanuts compared to a more convenient amp. besides, that quad of 6SN7's would be put to good use 
here's my power supply, for future reference
PT: Hevi-Duty E550, 125V -> 500V (4x), 1.14A
secondary winding resistance: ~5.1 ohms
Diodes: 2x6AY3 Fullwave
Choke: 193M
Cap: 400uF 500V
Output Voltage: ~390VDC, current, 260mA (this can increase to 300mA with minimal loss)
I have two 500uF 500V cans lying around, but PSUDII says that the current draw from using both paralleled would blow the 6AY3's. I'm considering installing both, and only wiring one for use until I pick up some sturdier 6DN3's, at which point I will have a beast PSU of 1000uF 500V
anyway, reading the Williamson circuit. also, checking out Jones' section on push pull
i'll definitely take a look, i think i have the article where Hafler and Keroes hijack the Williamson driver/splitter stage lying around somewhere. didn't make the williamson crowd too happy, iirc. Of course, it means picking up new biasing junk, but that's peanuts compared to a more convenient amp. besides, that quad of 6SN7's would be put to good use here's my power supply, for future reference
PT: Hevi-Duty E550, 125V -> 500V (4x), 1.14A
secondary winding resistance: ~5.1 ohms
Diodes: 2x6AY3 Fullwave
Choke: 193M
Cap: 400uF 500V
Output Voltage: ~390VDC, current, 260mA (this can increase to 300mA with minimal loss)
I have two 500uF 500V cans lying around, but PSUDII says that the current draw from using both paralleled would blow the 6AY3's. I'm considering installing both, and only wiring one for use until I pick up some sturdier 6DN3's, at which point I will have a beast PSU of 1000uF 500V
anyway, reading the Williamson circuit. also, checking out Jones' section on push pull
An externally hosted image should be here but it was not working when we last tested it.
that's the williamson. i'm good for everything except the 450V plate voltage, which i had worried about. i wonder what to do..
sorenj07 said:
That's the schematic I was picturing
. Your power supply voltage should not be a problem... certainly there were many Williamson amps that used lower than 450V -- Heathkit W4, for example. You can take some liberties with this design (as many others have done) if you don't go overboard on the feedback loop (or if you are able to tune it carefully). Definitely take a look at Jones' description of stability and the Wiliamson amp... I know he gives a clear description of the concerns. Much of this discussion is valid for a wide range of feedback amplifiers.sorenj07 said:
Hi Sorenj07, I would not exclude the 6SN7 stage, but rather stay with the original circuit. In the circuit of post #8, the input is 618mVrms. If this is too low for you, the original Williamsom required about 1,8Vrms. I also have a little problem with the 6SL7 paraphase inverter as in post #8. The splitting resistors of 470K and 560K do not give best balance. Rather the values should then be 470K and 494K. The original (concertina) phase splitter is always in balance with equal anode and cathode resistors. There can also be other problems with a hi-mu tube as an input stage, though not serious.
As for the 6L6GCbias, I find a resistor of 570 ohm per tube more realistic, but that wil depend on the plate voltage. (Your maximum dissipation/tube will be 30W, etc.)
Regards.
Johan Potgieter said:
I'll probably end up going with the Williamson, because I have the 6SN7's to do it, and I don't want to have to attenuate my CD player so much. I also notice that the schematic shares many values with the H/K 6L6 UL so I won't have to break the bank changing the circuit
I don't suppose I could use different values for the coupling caps? Johan Potgieter said:
Yeah, after reading about splitters a bit, I hit upon the figure 90% between top and bottom resistors, and 470K/560K certainly isn't that. If I were to wire the 6SL7 as a paralleled concertina, I suppose it would fix that problem at the cost of much reduced gain, right?
Anyway, just to avoid confusion, here's the latest and greatest schematic, except the .47uF and .047uF coupling caps should be reversed. I have 4 .56uF 630V Solens and 4 .047uF Orange Drops so I'm all right in that department.
An externally hosted image should be here but it was not working when we last tested it.
Johan Potgieter said:
Which schematic are you referring to? The Williamson looks like it has 1.2K for each tube, and the H/K, 700...?
jon_010101 said:
That's the schematic I was picturing
OK, sweet. As I mention before, minimal parts changes are needed. I just skimmed that section yesterday but I'll give it a close read today
sorenj07 said:
I wouldn't be entirely keen on a parallel concertina ... using the original schematic as-is, or a 6SN7 Willy-style would be a safer bet. A single 6SN7 gain stage wouldn't realistically allow for any feedback. Johan makes a great point that the concertina has better balance than the H/K's paraphase inverter (at least with good resistors). The concertina also has less high-frequency phase error.
Regarding caps, if you did go for a Williamson, I'd be inclined to use the .47/.56's for the first stage (after the cathodyne) and the .047's for the second (after the driver).
Sorenj07,
Yes, for reasons I explained in my post #75 (pardon not re-copying here) I would use the smaller coupling caps first.
The bias: For 700 ohms/6L6 the dissipation seems to be 18W/tube. That was probably intended for the old 6L6 at Pmax of 19W. Bias about-36V at 51 mA/tube. That is going pretty class AB, although it could be used like that. But the 6L6GC can dissipate 30W (beware, the 6L6WXT is somewhat lower.) Thus one could go to 72 mA/tube with 380V h.t. This will give a Vg1 of about -32V, thus a common cathode resistor of about 220 ohm. Again I cannot read the values on the original diagram clearly. If that cathode resistor was 500 ohm, that was high. Certainly 1K2 per tube is almost class B; don't know where they got that. OK, the listed h.t. there is 450V, giving about 415 Va-k. You could make it adjustable and trim to give Pa of some 25W.
Regards
Yes, for reasons I explained in my post #75 (pardon not re-copying here) I would use the smaller coupling caps first.
The bias: For 700 ohms/6L6 the dissipation seems to be 18W/tube. That was probably intended for the old 6L6 at Pmax of 19W. Bias about-36V at 51 mA/tube. That is going pretty class AB, although it could be used like that. But the 6L6GC can dissipate 30W (beware, the 6L6WXT is somewhat lower.) Thus one could go to 72 mA/tube with 380V h.t. This will give a Vg1 of about -32V, thus a common cathode resistor of about 220 ohm. Again I cannot read the values on the original diagram clearly. If that cathode resistor was 500 ohm, that was high. Certainly 1K2 per tube is almost class B; don't know where they got that. OK, the listed h.t. there is 450V, giving about 415 Va-k. You could make it adjustable and trim to give Pa of some 25W.
Regards
Johan Potgieter said:
Johan, Would you use small couplers first on the Williamson, or just the original UL circuit with paraphase inverter? Historically this has been done both ways (for the Williamson, at least) ... I'd still be inclined to do it the other way around, with big caps first, simply to cut all subsonics immediately before they hit the output tube grids and, ultimately, OPT. Thinking in terms of preserving the precarious LF stability, more-so than reducing driver distortion
.John_010101,
Your motive is right, but if one analyses the situation:
We are looking at such low frequencies where the (coupling) capacitors reactance etc. have already started to become significant, thus causing attenuation. Now the NFB tries to restore the situation until it runs out of steam. That means that it will try to keep output constant by diminishing feedback, thus the amplitude before each cap will rise, rather than fall after it, if that is clear.
The most likely place where something is going to run out of "head-room" as a result of this action is the anodes of the drivers feeding the power stage. (We are ignoring for the moment that the output transformer will also play a role here.) It is therefore logical that the last C be kept as big as possible, so that such amplitude raise could occur in earlier stages where there is more headroom - until NFB runs out, etc.
In that sense, whatever the circuit, and only considering the (mostly) two capacitive coupling stages, it should be clear why the early one should start attenuating first. There could however be other factors in the chain meriting consideration; this is only the relative simple argument regarding two RC coupling elements.
The main point here is therefore to avoid overloading. The value of the time constants, also relative to one another is another point, having to do with stability criteria.
Regards
Your motive is right, but if one analyses the situation:
We are looking at such low frequencies where the (coupling) capacitors reactance etc. have already started to become significant, thus causing attenuation. Now the NFB tries to restore the situation until it runs out of steam. That means that it will try to keep output constant by diminishing feedback, thus the amplitude before each cap will rise, rather than fall after it, if that is clear.
The most likely place where something is going to run out of "head-room" as a result of this action is the anodes of the drivers feeding the power stage. (We are ignoring for the moment that the output transformer will also play a role here.) It is therefore logical that the last C be kept as big as possible, so that such amplitude raise could occur in earlier stages where there is more headroom - until NFB runs out, etc.
In that sense, whatever the circuit, and only considering the (mostly) two capacitive coupling stages, it should be clear why the early one should start attenuating first. There could however be other factors in the chain meriting consideration; this is only the relative simple argument regarding two RC coupling elements.
The main point here is therefore to avoid overloading. The value of the time constants, also relative to one another is another point, having to do with stability criteria.
Regards
Further John_010101, and not to overstay my welcome,
The attenuation of subsonics, indeed the attenuation of any products falling outside the audible spectrum, is something that enjoys too little attention imho. One must remember that amplifers are not perfect. The design criteria might be met regarding the audio spectrum, but somewhere outside at both ends, any amplifier will exhibit reactive phase angles as the NFB goes to 0.
Any artifact coming along here could cause undesirable circuit response - more so if two should intermodulate. This is more prevalent in the super-sonic region. Some folks frown on filters, but I have made it a point to somewhere in a pre-amp install sub- and supersonic filters to attenuate such artifacts. In my case below 25 Hz and above 25 KHz. Everyone can have his own choice. I recall at least one published test a long time ago where this made an audible improvement. To me, it simply seems reasonable to do so.
Regards
The attenuation of subsonics, indeed the attenuation of any products falling outside the audible spectrum, is something that enjoys too little attention imho. One must remember that amplifers are not perfect. The design criteria might be met regarding the audio spectrum, but somewhere outside at both ends, any amplifier will exhibit reactive phase angles as the NFB goes to 0.
Any artifact coming along here could cause undesirable circuit response - more so if two should intermodulate. This is more prevalent in the super-sonic region. Some folks frown on filters, but I have made it a point to somewhere in a pre-amp install sub- and supersonic filters to attenuate such artifacts. In my case below 25 Hz and above 25 KHz. Everyone can have his own choice. I recall at least one published test a long time ago where this made an audible improvement. To me, it simply seems reasonable to do so.
Regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.