Douglas,
Your idea of another winding is attractive, but I think I'll stick with Eli's original concept.
Yes, I've got heat paranoia on this one! This thing has to run as close to class B as I can get it for reasons mentioned earlier in the thread.
Yes, I understand your concern, but VR tubes allow very little parallel capacitance and have a higher firing voltage, making them very susceptible to "winking". I'm inclined to try the zener and large parallel cap approach to start, and if there is a problem go for some kind of compound regulator. Probably an LED in each zener chain will be a good indicator.
Your idea of another winding is attractive, but I think I'll stick with Eli's original concept.
Yes, I've got heat paranoia on this one! This thing has to run as close to class B as I can get it for reasons mentioned earlier in the thread.
There is another sol'n available to the noise problem. Small ferrite inductors to de-couple the Zener from the capacitor. Good film caps will behave properly to nearly RF( way out of the audioband anyway. Even the metalized polypropylene do quite well( and offer more capacitance per volume ).
Any chance you'd make a run away from max power, and do a B+ at maximum g2? This way, you'd be able to do a McIntosh style arrangement with a tap on the anode coil equal to CFB turns. Probably only cost you 20% on power. No more shunt reg and associated issues then.
You could also try 8417's at 400V. At 'B' bias level, you should be quite cool.
Anyway, let me inquire with the winder on modifying the coil to get closer to your 6k/10% point. What sort of tolerance/leeway are you willing to accept on the exact ratio?
Also, I suspect you'll be able to avoid the loop NFB you have planned. I built a PP KT90 with 4k Dynaco A441 using the tertiary winding for CFB. It didnot need any other NFB. Wished I had a tap to run full pentode as I described above. 15% McIntosh instead of the full Unity-Coupled, and its Nutty voltage swing requirements.
cheers,
Douglas
Any chance you'd make a run away from max power, and do a B+ at maximum g2? This way, you'd be able to do a McIntosh style arrangement with a tap on the anode coil equal to CFB turns. Probably only cost you 20% on power. No more shunt reg and associated issues then.
You could also try 8417's at 400V. At 'B' bias level, you should be quite cool.
Anyway, let me inquire with the winder on modifying the coil to get closer to your 6k/10% point. What sort of tolerance/leeway are you willing to accept on the exact ratio?
Also, I suspect you'll be able to avoid the loop NFB you have planned. I built a PP KT90 with 4k Dynaco A441 using the tertiary winding for CFB. It didnot need any other NFB. Wished I had a tap to run full pentode as I described above. 15% McIntosh instead of the full Unity-Coupled, and its Nutty voltage swing requirements.
cheers,
Douglas
Douglas,
This is new teritory for me. Being a bit of a "purist" I have never before used pentodes as pentodes, nor run in any class than A (at least for proper "Hi-Fi). This is why I'm relying so heavily on Eli. Interesting though it is, it would be unwise for me to go off on a tangent because it's a workhorse I'm building here. Nor am I certain about the primary impedance since I've yet to find any "science" in it's choice with pentodes.
The loop NFB is only part of the NFB, as you know. The sole reason for limiting to 2 stages has to make the NFB as innocuous as possible. I am a worrier about blocking and am quite sensitive to unpleasant distortion spectra. I've "bought" this design on the assurance that it will be good on both counts. It doesn't have to sound as good as my 845SE, but it will live in the same open-plan ground-floor, though used continuously, and I've heard some reasonably good valve commercial amps sound unacceptibe here.
Maximum power may be an issue, as I may end up running a little cooler than planned, thus losing some.
I appreciate your interest and help, particularly WRT the transformer, but I must say that I had planned to buy European - not for patriotism but the fact that my last big purchase from USA cost me heavily in import duty. Anyway, it's not set in stone and if the price/performance is right I'm happy to buy and therefore share the winder's tooling costs with you. Time is also of essence, since I am in Petersburg (Russia) in June, and could buy them for 1/2 the cost there.... But maybe they'll go on another project..
The ferrite may be a partial solution for RF noise, but there would need to be many turns to get enough inductance to isolate the AF noise. Fortunately RF is an area in which I have rather more experience than pentodes I will in any case use a string of zeners to improve the noise figure and the temperature coefficient.
This is new teritory for me. Being a bit of a "purist" I have never before used pentodes as pentodes, nor run in any class than A (at least for proper "Hi-Fi). This is why I'm relying so heavily on Eli. Interesting though it is, it would be unwise for me to go off on a tangent because it's a workhorse I'm building here. Nor am I certain about the primary impedance since I've yet to find any "science" in it's choice with pentodes.
The loop NFB is only part of the NFB, as you know. The sole reason for limiting to 2 stages has to make the NFB as innocuous as possible. I am a worrier about blocking and am quite sensitive to unpleasant distortion spectra. I've "bought" this design on the assurance that it will be good on both counts. It doesn't have to sound as good as my 845SE, but it will live in the same open-plan ground-floor, though used continuously, and I've heard some reasonably good valve commercial amps sound unacceptibe here.
Maximum power may be an issue, as I may end up running a little cooler than planned, thus losing some.
I appreciate your interest and help, particularly WRT the transformer, but I must say that I had planned to buy European - not for patriotism but the fact that my last big purchase from USA cost me heavily in import duty. Anyway, it's not set in stone and if the price/performance is right I'm happy to buy and therefore share the winder's tooling costs with you. Time is also of essence, since I am in Petersburg (Russia) in June, and could buy them for 1/2 the cost there.... But maybe they'll go on another project..
The ferrite may be a partial solution for RF noise, but there would need to be many turns to get enough inductance to isolate the AF noise. Fortunately RF is an area in which I have rather more experience than pentodes
I will in any case use a string of zeners to improve the noise figure and the temperature coefficient.Let's be clear. I'm a HACKER. Little, if anything, in the design is original. The ideas of real experts have been "borrowed" and assembled into something that seems coherent. I'd like to think the "I"s and "T"s have been dotted and crossed. If somebody spots a flaw or 2 in my thinking, PLEASE speak up. Devil take my ego; John's cash and ears are on the line.
Near Class "B" operation WORKS for HIFI. McIntosh, et al proved that case. "Skull Sweat" is clearly a necessary ingredient.
The low O/P impedance and high current of the FET buffers seem to offer some protection against blocking. Be sure to use film and foil coupling caps. between the buffers and the finals. Film and foil parts have a reputation for being able to handle current pulses well. An INTANTANEOUS transient of barely positive control grid voltage should be OK. With cap. coupling, significant B2 operation is an obvious non-starter.
Near Class "B" operation WORKS for HIFI. McIntosh, et al proved that case. "Skull Sweat" is clearly a necessary ingredient.
The low O/P impedance and high current of the FET buffers seem to offer some protection against blocking. Be sure to use film and foil coupling caps. between the buffers and the finals. Film and foil parts have a reputation for being able to handle current pulses well. An INTANTANEOUS transient of barely positive control grid voltage should be OK. With cap. coupling, significant B2 operation is an obvious non-starter.
It's the development of a concept. All ideas are welcome, but if we get drawn too far away, the concept is gone. There is not much that's new under the sun, so I'm not bothered what's original and what's not. It's the sum {insert sigma} that matters.
Class B is a gamble for me, I admit. But it is the only acceptable class (read power consumption) for this application. It's got to work.
True, class B1 won't happen with the current FET arrangement. It might have been possible with the FET to grid DC coupling envisaged earlier, but there are good reasons for the current arrangement.
I'm happy with the way this is going. As soon as I get all my quotes for OPT, I'll make a decision. If anyone want's to join me on a group buy of OPT to reduce cost, of course I would be delighted.
Perhaps I should point out that although I personally need (near) class B. There is nothing holding back using this in class AB or class A operation.
My thanks all contributors, and to Eli Duttman for his patience.
Class B is a gamble for me, I admit. But it is the only acceptable class (read power consumption) for this application. It's got to work.
True, class B1 won't happen with the current FET arrangement. It might have been possible with the FET to grid DC coupling envisaged earlier, but there are good reasons for the current arrangement.
I'm happy with the way this is going. As soon as I get all my quotes for OPT, I'll make a decision. If anyone want's to join me on a group buy of OPT to reduce cost, of course I would be delighted.
Perhaps I should point out that although I personally need (near) class B. There is nothing holding back using this in class AB or class A operation.
My thanks all contributors, and to Eli Duttman for his patience.
One more thing....since we're considering Class B, you should examine the curves with a quarter of the a-a load. In your 6k a-a example, you're applying a 1k5 Ohm load. Just guessing, but this is pretty thin. 6L6 data sheets come with 2k5 SE sugestions.
Draw the load line. a-a/4 as if it were SE. Now if you go Class A, you get to use a-a/2. Best to look at things at the simple end of the pool first.
cheers,
Douglas
Draw the load line. a-a/4 as if it were SE. Now if you go Class A, you get to use a-a/2. Best to look at things at the simple end of the pool first.
cheers,
Douglas
Patience, my foot. This is FUN!
John,
FWIW, I've looked over the quick start PSU and I have some thoughts.
4X 1200 PIV Schottky diodes would be dead quiet in the HT bridge. Snubber caps. are not needed.
Neither a ceramic cap. nor a MOV is present in the HT circuitry to protect the SS diodes from inductive kick back. Inductive kick back can destroy SS diodes.
Small "fudge factor" caps. might be needed from heater L/R to heater common to compensate for the 0.5 V. forward drop in D5 and D6, when the amp is playing music.
Douglas,
I've thought long and hard about the load and drawn a few lines, and done a few calculations. Ok I have found some science in it, but there are soo many compromises that I realise now why triodes are so nice! Anyway, forwards:
Power is not the main goal, but minimising 3H is (although to the detriment of 2H and 4H). I'm therefore thinking of keeping the Z as low as reasonable, though not rediculously low as I'd set it before; I had not realised the z/4 rule in class B Perhaps between 8K8 and 9K6 a-a? Corresponding to 2K2 to 2K4 SE?
I'll choose the exact operating point on test. Thing can only get better as the class moves "up"... and it might need to.
Eli,
Thanks for looking at the quick-start PSU schematic.
If I knew where to get 1200v Schottkys over here, I'd jump at them - maybe someone can give me a lead?
Point noted about diode protection. I'll implement some.
Fudging the "music-playing" heater volts can be done winding the heater TX.
Edit: added 2H / 4H comment.
I've thought long and hard about the load and drawn a few lines, and done a few calculations. Ok I have found some science in it
, but there are soo many compromises that I realise now why triodes are so nice! Anyway, forwards:Power is not the main goal, but minimising 3H is (although to the detriment of 2H and 4H). I'm therefore thinking of keeping the Z as low as reasonable, though not rediculously low as I'd set it before; I had not realised the z/4 rule in class B
Perhaps between 8K8 and 9K6 a-a? Corresponding to 2K2 to 2K4 SE?I'll choose the exact operating point on test. Thing can only get better as the class moves "up"... and it might need to.
Eli,
Thanks for looking at the quick-start PSU schematic.
If I knew where to get 1200v Schottkys over here, I'd jump at them - maybe someone can give me a lead?
Point noted about diode protection. I'll implement some.
Fudging the "music-playing" heater volts can be done winding the heater TX.
Edit: added 2H / 4H comment.
John,
On 2nd thought, 8X unsnubbed UF4007s should be fine in the B+ bridge. The chokes in the filter kill the small amount of switching noise the diodes generate; so, the B+ will be "clean". All that's left is to keep the SS "hash" out of the mains. Ferrite beads on the primary "flying" leads close to the trafo are a method for suppressing the noise. "Buddha" Camille's RRSF on the rectifier winding is another good noise suppressor. Hmmm, use a RRSF made from 2X 10 nF. HIGH WVDC ceramic caps and 2X 10 Ohm wirewound resistors. That's inexpensive enough and will kill most of the noise. Low cost ferrite beads on the primary leads kill the remainder. An additional pair of 10 nF. ceramics for a total of 30 nF. across the rectifier certainly would help without a significant cost increase.
On 2nd thought, 8X unsnubbed UF4007s should be fine in the B+ bridge. The chokes in the filter kill the small amount of switching noise the diodes generate; so, the B+ will be "clean". All that's left is to keep the SS "hash" out of the mains. Ferrite beads on the primary "flying" leads close to the trafo are a method for suppressing the noise. "Buddha" Camille's RRSF on the rectifier winding is another good noise suppressor. Hmmm, use a RRSF made from 2X 10 nF. HIGH WVDC ceramic caps and 2X 10 Ohm wirewound resistors. That's inexpensive enough and will kill most of the noise. Low cost ferrite beads on the primary leads kill the remainder. An additional pair of 10 nF. ceramics for a total of 30 nF. across the rectifier certainly would help without a significant cost increase.
Hey-Hey!!!,
The loading thing is *NOT* intuitive. I think is's Lundahl who makes this note in their pdf data sheets they publish for their OPTx's.
It can also be determined from first principles if you have a pen and paper.
On the CFB/Tertiary winding OPTx, I'd ship it to you directly as a sample. No biggie there. I have been playing about with this idea for a little while. The E-Linear circuit has been very nice, and I have put CFB through tertiary windings on the back burner.
I have looked at the winding schedule, and think about 8k would be possible. This is just the anode winding, and the whole load would have to account for the cathode portion. So 8k + about 11% cfb is going to give you about a 10k a-a load, or two 2k5 ones, one at a time.
cheers,
Douglas
The loading thing is *NOT* intuitive. I think is's Lundahl who makes this note in their pdf data sheets they publish for their OPTx's.
It can also be determined from first principles if you have a pen and paper.
On the CFB/Tertiary winding OPTx, I'd ship it to you directly as a sample. No biggie there. I have been playing about with this idea for a little while. The E-Linear circuit has been very nice, and I have put CFB through tertiary windings on the back burner.
I have looked at the winding schedule, and think about 8k would be possible. This is just the anode winding, and the whole load would have to account for the cathode portion. So 8k + about 11% cfb is going to give you about a 10k a-a load, or two 2k5 ones, one at a time.
cheers,
Douglas
John,
If NTC2 REALLY protects the bridge against kick back, NTC1 is superfluous. I'm worried that there will be no protection at turn off time, as hot NTC devices present but a few Ohms of resistance.
PERHAPS a MOV across the 1st inductor and eliminating NTC2 is a better method of protection.
If NTC2 REALLY protects the bridge against kick back, NTC1 is superfluous. I'm worried that there will be no protection at turn off time, as hot NTC devices present but a few Ohms of resistance.
PERHAPS a MOV across the 1st inductor and eliminating NTC2 is a better method of protection.
Douglas,
Points noted. Yes I did use pen and paper to determine odd and even distortion ratio's. Now you're going to tell me there's a spreadsheet?
Eli,
Yes, the MOV makes much more sense for protection. I'll change it.
I wanted an NTC in the primary side to reduce the surge current and occasional "bong" of the transformer (when switching coincides with peak). It doesn't have to work every time to be useful.
Points noted.
Yes I did use pen and paper to determine odd and even distortion ratio's. Now you're going to tell me there's a spreadsheet?Eli,
Yes, the MOV makes much more sense for protection. I'll change it.
I wanted an NTC in the primary side to reduce the surge current and occasional "bong" of the transformer (when switching coincides with peak). It doesn't have to work every time to be useful.
dhaen,
Just a minor point here; MOV's are consumable devices. Each time you "hit" em, they are somewhat consumed. They eventually explode in a messy way depending on the frequency and severity of the hits. Until recently this fact, though very real, has been obscurred by those who make them (go figure).
Alternatives are Silicon Oxide Varistors, gas tubes, neon bulbs etc... the best of which, SOV's are hard to find.
The point being, mount your MOV where it will do no harm, if and when it explodes.
Just a minor point here; MOV's are consumable devices. Each time you "hit" em, they are somewhat consumed. They eventually explode in a messy way depending on the frequency and severity of the hits. Until recently this fact, though very real, has been obscurred by those who make them (go figure).
Alternatives are Silicon Oxide Varistors, gas tubes, neon bulbs etc... the best of which, SOV's are hard to find.
The point being, mount your MOV where it will do no harm, if and when it explodes.
dhaun,
The cat is out of the bag... some manufacturer's are being forthcoming with proper data.
Here is a link about MOV's:
http://www.circuitprotection.com/04Databook/I_varistor_(351-388).pdf#search='Metal%20Oxide%20varistor'
The cat is out of the bag... some manufacturer's are being forthcoming with proper data.
Here is a link about MOV's:
http://www.circuitprotection.com/04Databook/I_varistor_(351-388).pdf#search='Metal%20Oxide%20varistor'
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.