You could simply substitute a FET bootstrapped follower for the 6SL7 in the Loftin-White design. It's direct coupled, plenty of drive for A2, cancels PS noise (no electrolytics anywhere, including the power supply), and no cathode caps. The bias is self adjusting for tube wear.
It burns some extra power in the output cathode resistor and takes a little iteration to get the operating parameters dialed in during the design phase, but from then on rock stable.
Sheldon
It burns some extra power in the output cathode resistor and takes a little iteration to get the operating parameters dialed in during the design phase, but from then on rock stable.
Sheldon
Last edited:
You are right, the L-W still amazes today, although there are obviously a few caps connecting the cathodes of both the output and input tubes.
I attached another design which uses only 1 single small valued cap in the whole amp circuit, which is the absolute minimum I came up with. The currents across this cap are tiny and it works into well defined loads and never sees any surges from overdriving, clipping or such.
I believe the cap could be replaced with a 1-1 interstage xfmr for a totally capless design, but I'm not sure if this would distort any less than a quality coupling cap.
Rgds
I attached another design which uses only 1 single small valued cap in the whole amp circuit, which is the absolute minimum I came up with. The currents across this cap are tiny and it works into well defined loads and never sees any surges from overdriving, clipping or such.
I believe the cap could be replaced with a 1-1 interstage xfmr for a totally capless design, but I'm not sure if this would distort any less than a quality coupling cap.
Rgds
Attachments
One small cap in series is certainly one cap too many, as is any interstage transformer. Nothing can beat a well done two stage with DHT finals and a small piece of 13 AWG Siltech ( silver with gold ) wire connecting the two stages.....
Jeff Medwin
You mean something like this?
Of course, it's not a power drive and no class A2 action.
Attachments
With the original Loftin-White take off, yes they are caps, but can be quality film caps. And the power supply needn't have very low ripple (that string between the cathodes completely cancel it. Few understand the design, but I can tell you it works. Here's a headphone amp I plan to cook up. Another option with even fewer caps is use that FET as a current source for the input tube, instead of as a bootstrapped follower. For plenty of A2 drive you can select an appropriate FET. In the current source case you can also use it as a u follower (probably a little better drive capability).
Attachments
One more bit of info. The bootstrapped follower is probably the better choice, because it acts more like a gyrator than a constant current source. It lets the tube set the voltage, whereas a current source doesn't respond to tube changes over time or between tubes. As a result, the voltage will vary more over time or tube change with a CCS than with the follower.
Sheldon
Sheldon
Hello !!
Thats more like it....simpler. KISS rules !!
NOW, WHY would one ever want "power drive" and class A2 action ( up to 5 Watts ) when the amp will be operating on efficient speakers, and will be using 50 mWatt to 500 mWatt of power 99 percent of the time ??
Why even care about "full power output" when you listen at 50-500 mWatt most all of the time?? In that case, design an amp for the BEST under 1 Watt performance !! Use as few a parts as possible !!
Instead of a CCS added, why not a high powered SHUNT resistor, on the driver's stage B+, to ground ?? A ONE PART shunt regulator, not a solid state device ??? Thats what I do, often called a "draw down resistor". It draws several times the current draw of the deriver tube. Mills MRA-12s for resistors are a nice sounding part there.
One other comment, a 2A3 drawing 250 VDC across the plate at 60 mA. will sound thermally stressed, and only last a few thousand hours before degrading substantially. Avoid that. Use a Golden Ratio plate dissipation, by reducing current, not voltage, till its about 62% of the manufacturer's 15 Watt rated dissipation.
At lower dissipations, the tube will operate cooler thermally, and it will sound far less STRESSED, ( because it is not thermally stressed ), and, it will last easily 10,000 hours, and up to 50,000 hours, sounding superior to ANY 2A3 operating at 15 Watts ( it's maximum ) plate dissipation.
This more conservative design approach ( cost versus performance wise ) allows you to buy a superb and expensive EML mesh plate, or a bargain priced JJ 2A3-40, and use it, in a superior mode, for 50,000 hours .
You haven't shown any power supply. One thing, with a 2A3 in a Direct Couple, is that you want the chokes to be at least 20 Ohms or less DCR, which I call "high fidelity". I personally prefer 10 Ohms or less, which I call ( and hear it ) as "ultra high fidelity". Two L/C sections in series to the 2A3 finals seems to be highest performing finals filtering in three decades of me testing this.
In a DC amp, use a tube rectifier and allow the voltages to all come up together. There is a tremendous bargain these days, the current production JJ 5U4G rectifier ( Antique Radio, Tempe, AZ ) sounds very very good on a good system. A no brainer, only $16.00. Have fun.
Jeff Medwin
Last edited:
Ideas for Goldenbeer
From my experience, in 2 stage DC 2A3 amps, regarding your schematic in Post #44 above :
(1) your idea of using a 405 VDC B+ is not good. You want a B+ in the range of 450 to 500 VDC for a 2A3 or Type 45 Finals tube, mainly to optimize the DRIVER tube in the circuit.
Most of what you hear in such an amp is the DRIVER tube, just as Thomas Mayer has also written.
(2) Avoid use of a 6SL7, if you seek the ultimate in clarity, great diction on spoken words, and proper enunciation. The 6SL7 is pleasant, but gets a "D+" grade in the areas I have mentioned, it always had me "wanting more".
(3) Use 1/2 of a 12AX7, a 9 pin miniature tube. It can not be beat in this topology. Never parallel a high mu Driver tube, such as Nobukazu Shishido did, in a L-W amplifier. The triode halves do NOT match, and it is heard as a SKEWING of the music, to be avoided at all costs !!!
One ( or many ) REASONS why you can't beat a 12AX7 is because it is a 9 pin miniature, and the tube construction pins come right out of the envelope of the tube, and directly into the tube socket. Where the 6SL7, or ANY high mu octal tube "messes up", is when they have to add an extra solder joint, on every pin of the octal socket, which causes its vagueness in sound - as a high mu driver.
Despite common practice, and ANY thoughts on Miller capacitance, the 2A3 and Type 45 are VERY EASY to drive in a direct couple, and so, half a 12AX7 can be the best possible choice for 99 percent of the builders.
Now, back to your idea of 405 VDC for B+, and my suggestion of 450 VDC to 500 VDC:
a) Since we hear the Driver, most of all, it is incumbent that we totally optimize the Driver tube's Op points. Most old timers who are really "pros", will tell you a 12AX7 sounds best with about 195 VDC on the plate...most brilliant and alive sounding. Most will agree, 150 VDC is the very LOWEST you can go, to get the 12AX7 vaugely in the ball park, but NOT optimized. " I " like 175 VDC to 195 VDC Ea, at under 1 mA., and shoot for that in every such DC amp I do.
b) TOTAL amp Voltages : Lets say we use 175 VDC on the 12AX7 plate, which is on the the grid of the 2A3. Self bias on the 2A3, running it conservatively ( 250 VDC P-K, 42 to 43 mA.) may be "about" -48 VDC grid to cathode. So, that puts the cathode at 175 plus 48 VDC, equals 223 VDC on the 2A3's cathode.
I like 250 VDC P-K on 2A3, so it now is 223 VDC ( cathode ) plus 250 VDC ( K-P) or 473 VDC on 2A3 plate. Add 5 VDC for I times R drop across output transformer primary, and you can see we need a MINIMUM B+ of 478 VDC if we want to OPTIMZE the Driver's operation.
Also, Goldenbeer, that is "optimization" at 175 VDC Ea on the driver. IF you bump the 12AX7 up to 195 VDC Ea, under 1 mA., and you need 498 VDC as your B+ to the Finals stage. ( 195+48+250+5 )
No way, 405 VDC B+ will be worth using, you are defeating yourself before you solder your first solder joint.
c) Further optimize the Driver's B+. SHUNT regulate it !!! Throw away any type of CCS, with all its unwanted parts and complexity and signature on the music !! Less is more. From the 475-500 VDC Finals B+, drop down through about a 13-14 K Ohm "drop down" resistor, to obtain maybe 360 VDC as the B+ for the Driver. Filter it with a 50 uF WIMA DC LINK film capacitor. Place ( about ) a 36.5 K Ohms Mills MRA-12 twelve watt resistor from Driver's B+ to ground.
You have just created a KISS one part, zero negative feedback SHUNT regulator, drawing about 9.5 to 10 mA. against he Driver tubes "less than one mA." draw.
What this does for bass response and stability, clarity of complex passages, for the Driver stage, is beyond what you would imagine. It GREATLY lowers the stage's Z. It eliminates a CCS and its unwanted signatures - thank goodness.
There are additional optimizations one can make to the all-important Driver stage, if one is thoughtful.
d) One other thing, despite all E.E. theory, use a film cap on the Final's Rk, and make it act as a partial Rk bypass, to YOUR ear !! I usually use 8 uF to 10 uF as a main Rk bypass cap, with smaller value ( < 1 uF ) film caps in parallel. Have fun.
What I have written herein, IMHO, will be better than 95% of the DC 2A3 implementations you can build.
Jeff Medwin
From my experience, in 2 stage DC 2A3 amps, regarding your schematic in Post #44 above :
(1) your idea of using a 405 VDC B+ is not good. You want a B+ in the range of 450 to 500 VDC for a 2A3 or Type 45 Finals tube, mainly to optimize the DRIVER tube in the circuit.
Most of what you hear in such an amp is the DRIVER tube, just as Thomas Mayer has also written.
(2) Avoid use of a 6SL7, if you seek the ultimate in clarity, great diction on spoken words, and proper enunciation. The 6SL7 is pleasant, but gets a "D+" grade in the areas I have mentioned, it always had me "wanting more".
(3) Use 1/2 of a 12AX7, a 9 pin miniature tube. It can not be beat in this topology. Never parallel a high mu Driver tube, such as Nobukazu Shishido did, in a L-W amplifier. The triode halves do NOT match, and it is heard as a SKEWING of the music, to be avoided at all costs !!!
One ( or many ) REASONS why you can't beat a 12AX7 is because it is a 9 pin miniature, and the tube construction pins come right out of the envelope of the tube, and directly into the tube socket. Where the 6SL7, or ANY high mu octal tube "messes up", is when they have to add an extra solder joint, on every pin of the octal socket, which causes its vagueness in sound - as a high mu driver.
Despite common practice, and ANY thoughts on Miller capacitance, the 2A3 and Type 45 are VERY EASY to drive in a direct couple, and so, half a 12AX7 can be the best possible choice for 99 percent of the builders.
Now, back to your idea of 405 VDC for B+, and my suggestion of 450 VDC to 500 VDC:
a) Since we hear the Driver, most of all, it is incumbent that we totally optimize the Driver tube's Op points. Most old timers who are really "pros", will tell you a 12AX7 sounds best with about 195 VDC on the plate...most brilliant and alive sounding. Most will agree, 150 VDC is the very LOWEST you can go, to get the 12AX7 vaugely in the ball park, but NOT optimized. " I " like 175 VDC to 195 VDC Ea, at under 1 mA., and shoot for that in every such DC amp I do.
b) TOTAL amp Voltages : Lets say we use 175 VDC on the 12AX7 plate, which is on the the grid of the 2A3. Self bias on the 2A3, running it conservatively ( 250 VDC P-K, 42 to 43 mA.) may be "about" -48 VDC grid to cathode. So, that puts the cathode at 175 plus 48 VDC, equals 223 VDC on the 2A3's cathode.
I like 250 VDC P-K on 2A3, so it now is 223 VDC ( cathode ) plus 250 VDC ( K-P) or 473 VDC on 2A3 plate. Add 5 VDC for I times R drop across output transformer primary, and you can see we need a MINIMUM B+ of 478 VDC if we want to OPTIMZE the Driver's operation.
Also, Goldenbeer, that is "optimization" at 175 VDC Ea on the driver. IF you bump the 12AX7 up to 195 VDC Ea, under 1 mA., and you need 498 VDC as your B+ to the Finals stage. ( 195+48+250+5 )
No way, 405 VDC B+ will be worth using, you are defeating yourself before you solder your first solder joint.
c) Further optimize the Driver's B+. SHUNT regulate it !!! Throw away any type of CCS, with all its unwanted parts and complexity and signature on the music !! Less is more. From the 475-500 VDC Finals B+, drop down through about a 13-14 K Ohm "drop down" resistor, to obtain maybe 360 VDC as the B+ for the Driver. Filter it with a 50 uF WIMA DC LINK film capacitor. Place ( about ) a 36.5 K Ohms Mills MRA-12 twelve watt resistor from Driver's B+ to ground.
You have just created a KISS one part, zero negative feedback SHUNT regulator, drawing about 9.5 to 10 mA. against he Driver tubes "less than one mA." draw.
What this does for bass response and stability, clarity of complex passages, for the Driver stage, is beyond what you would imagine. It GREATLY lowers the stage's Z. It eliminates a CCS and its unwanted signatures - thank goodness.
There are additional optimizations one can make to the all-important Driver stage, if one is thoughtful.
d) One other thing, despite all E.E. theory, use a film cap on the Final's Rk, and make it act as a partial Rk bypass, to YOUR ear !! I usually use 8 uF to 10 uF as a main Rk bypass cap, with smaller value ( < 1 uF ) film caps in parallel. Have fun.
What I have written herein, IMHO, will be better than 95% of the DC 2A3 implementations you can build.
Jeff Medwin
Last edited:
Member
Joined 2009
Paid Member
Jeff,
With a low DCR power supply the last thing you would appear to want in series is a tube rectifier. Yes, they are 'traditional' and all that, but in terms of technology they are the worse performing element from the tube era.
The 5U5G sucks 3A for a 5V heater - that's 15W burned right there - the biggest use of power anywhere in the amplifier. The plates will drop 44V off the B+ rail, which is hardly desirable, more wasted power.
The dynamic resistance of this rectifier is spec'd at 21R which pretty much negates any benefit you wanted from low impedance chokes.
If the issue is the B+ coming up too quickly - there are other solutions that stay out of the way once the rail is up and ready.
I assume your listening tests prove that the tube rectifier works best - but can we say why ?
With a low DCR power supply the last thing you would appear to want in series is a tube rectifier. Yes, they are 'traditional' and all that, but in terms of technology they are the worse performing element from the tube era.
The 5U5G sucks 3A for a 5V heater - that's 15W burned right there - the biggest use of power anywhere in the amplifier. The plates will drop 44V off the B+ rail, which is hardly desirable, more wasted power.
The dynamic resistance of this rectifier is spec'd at 21R which pretty much negates any benefit you wanted from low impedance chokes.
If the issue is the B+ coming up too quickly - there are other solutions that stay out of the way once the rail is up and ready.
I assume your listening tests prove that the tube rectifier works best - but can we say why ?
Last edited:
Member
Joined 2009
Paid Member
The sound/performance/cost and reliability of a SE amplifier depends a lot on the power supply and given that most(?) SE amplifiers are not ultra-low distortion but have a 'sound' all these things become a subject of personal preference. The low DCR/low energy supply has a few tricks - it isn't simply about low impedance but how it reacts to changes in load makes a contribution to the overall sound. This approach has been criticized for not being 'pure' but it has also garnered praise. A silicon regulator produces a different result which may not be 'better' for everyone.
Hello,
There is a HUGE difference between the effects of a Power Transformer, Rectifier, and a Choke. The first two are very active devices, whereas the choke is a passive and DRIVEN member of the trio. It NEEDS to be low in DCR and in mass , to be driven easily, AND, to not exhibit too much stored energy which gets released out-of-time to the music.
Where you can relax some, is with the first two parts of the power supply trio, the PT and Rectifier.
I do NOT like solid state rectifiers sonically, compared to a directly heated rectifier, And I certainly do not like what solid state rectifiers do on turn ons !! I can compute the DCR superiority of solid state over tube rectification, as well as anyone. But I like to build and listen to tube rectification, its an active device, nice device. Do as you please, by all means, I won't change.
Regards to you,
Jeff Medwin
There is a HUGE difference between the effects of a Power Transformer, Rectifier, and a Choke. The first two are very active devices, whereas the choke is a passive and DRIVEN member of the trio. It NEEDS to be low in DCR and in mass , to be driven easily, AND, to not exhibit too much stored energy which gets released out-of-time to the music.
Where you can relax some, is with the first two parts of the power supply trio, the PT and Rectifier.
I do NOT like solid state rectifiers sonically, compared to a directly heated rectifier, And I certainly do not like what solid state rectifiers do on turn ons !! I can compute the DCR superiority of solid state over tube rectification, as well as anyone. But I like to build and listen to tube rectification, its an active device, nice device. Do as you please, by all means, I won't change.
Regards to you,
Jeff Medwin
Member
Joined 2009
Paid Member
It's still got sharp turn on/turn off characteristics (but less storage). And if you're using capacitor input, then you've still got that problem.
Member
Joined 2009
Paid Member
Its mostly the charge storage that gives the problem with EMI from the switching with regular SS rectifiers, an issue which is mostly gone with the SiC rectifiers. The switching is still fast - but how much of this is a problem past the first choke ?
Assuming SiC is still not as good 'sounding' as a tube rectifier I'd like to know why. Perhaps a resistor in series with a SiC diode provides the same good sound (provides some slowing down of switching and a higher impedance) but is a more controllable and less power hungry option ?
Interesting article:http://www.6moons.com/audioreviews/spec3/2.html
"I felt that the sound of this diode performed much like a high-quality tube rectifier."
Assuming SiC is still not as good 'sounding' as a tube rectifier I'd like to know why. Perhaps a resistor in series with a SiC diode provides the same good sound (provides some slowing down of switching and a higher impedance) but is a more controllable and less power hungry option ?
Interesting article:http://www.6moons.com/audioreviews/spec3/2.html
"I felt that the sound of this diode performed much like a high-quality tube rectifier."
Last edited:
Member
Joined 2009
Paid Member
What about the 5AR4, it offers the lowest voltage drop and might be more compatible with a power supply designed originally for SS - then I could compare ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.