|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
213.134.237.178
Is it wise to use a Pimm CCS in a PP el84 amp in pentode mode to get a stable screen supply ? Or are better methods available ? Zener string ?
Follow Ups:
Not that long ago I helped a guy working on a pp807 with vrtube regulated screens. He was having the same problems I had when I tried using gary's ccs feeding vr tubes for POWER pentode screens. It motorboats. Screens might be too capacitive or something. (I believe that's the concept of how screens isolate plate from grid, capacitors in series or the like, but I might remember wrong).
The ccs/vr tube combo works fine for signal pentodes, sounds great.
An idea I had later, and never tried was using the mu-follower output to drive the screens, essentially turning the regulator from a shunt to a series pass. Isolating the screens from the vr tubes at constant current should provide VERY stiff regulation.
The pp807 guy never tried it though, he went to zeners.
HELLO,Well, you would know from my G2 posts that I hate zeners or any kind of active device on G2. BUT, I DO have a suggestion - for your friend's application - that WILL sound better.
Use VoltSecond's zener trick and make NO zener above 6.8 VDC for a lower electronic noise floor. I series-connect them when I MUST use a zener string, and this sounds better than diodes rated above 6.8 VDC, VoltSecond is correct.
I liked it so much I bought a 5,000 piece roll of 6.8 VDC 1/2 watt zeners, after I A-Bed it.
Early ARC stuff is FULL of 47 VDC and 100 VDC zeners ( noise generators !! ) ie : SP3A-1, SP-6, SP-8, SP-10, D-75, D-76s, D-150s and these ALL will improve with series-strings of 6.8 VDC zeners in their place. Its easy to execute too !! Reversable also.
Have him try it, he will like it and never go back !!
I can't believe my eyes! Never thought you'd advocate zeners ever, haha. Yes I have seen your vehement G2 posts, and admit you're probably right, with quality chokes, something I have a shortage of (and a shortage of funds to purchase) haha. With what I've tried, to my ears direct connection to B+ with a 100 ohm resistor always sounds best, but like I said my recycled chokes are suspect and more playing is in order. I've been meaning to try voltseconds trick with my 46 tubes, but one of those many things I haven't gotten to yet, thanks for reminding me.
Don't let that stop you from trying your ideas. CCS or series regulators can still be an improvement over nothing.TRy the TL783 mos regulator as close to the G2 as possible no series resistor.
If you can't get what you want with a few $$ in parts then build the extra low DCR powersupply.
I use zeners to regulate the screens on the 6V6's driving my 45 homebrew. Nope, they're not transparent . . . they add a little high-frequency noise (not objectionable to my ears) . . . but I prefer the sound of the amp with 'em rather than without 'em. Total investment for the regulation was about $4.00. Clip 'em out if they don't do it for ya.
Jim,What is the voltage rating of zeners do you currently use?
Review this post:
http://www.audioasylum.com/forums/tubediy/messages/93330.html
Let me know by email how it works for you !!
Jeff, thanks for the tip! Interesting idea, and in my experience, Volt's never been wrong. Presently I'm regulating 125V on the screens using 22V zeners. I'll string together the <6.8V's and let you know. Thanks again, J.
Jim,Well, 22 VDC is better than 47 or 100 VDC for noise. Any voltage from 5.1 to 6.8 is OK he says, below 5.1 and the knee gets sloppy, above 6.8 and noise rises fast.
I tried it out and HEARD it in my ARC D-75's front end supply ( three 100 volt devices ! ) and liked it a lot, although that was just one of many things needing attention!
I think this is what Gary did on the 6AU6 drivers in the PP 47 amp - check his site for schematics.
Good thinking, but IMHO, any ACTIVE device to regulate pentode screens is HEARD. is not good enough, and causes a loss of transparency, and hence should be avoided. But, a STABLE screen supply is critically important. Do a "G2" Forum search under my moniker.
Jeff,
first excuse me for my bad English, I am from Germany.
I know the effect of low DCR screen supply very well. But I do not see a reason, why a well constructed fast regulation shouldn't do a good work. Could you send the regulation circuit, which you didn't like? Iam sure, it could be done in a way you would prefer it.
Frank
Hi,Thats OK, I don't speak German !!
There is not a one I would like, I say this conceptually, see my G2 posts on this Forum.
G2 is SUPER sensitive and regulators have their own time frames.
"G2 is SUPER sensitive and regulators have their own time frames"That's right, and therefore you have to push the time frame of your regulator below 10us. All regulators with a big storage cap couldn't do this. Have you ever tried a very fast regulator with a small output cap?
No I have not. I do just the oppposite to get them to sound OK, add about 120 uF.
Assume an output resistance of 5 Ohms of your complete tube regulator, this would means a time constant of 5x120us=600us. This means a corresponding frequency of about 300Hz! This means that the cap is working mostly in the domain above 300Hz and the regulator below 300 Hz. If I assume the cap to be a electrolytic one, not a very good idea! It is better to work with a small cap with a fast regulator with high internal feedback. The cap should work only in the domain above 40 kHz. A well done SS regulator should be fine. Noise isn't a problem, if the corner frequency is about 40kHz. Use a polypropylen and a ceramic cap in parallel.
This solution would be cheap and have to be considerd before using a lot of iron and huge caps.
I always like to use choke input supplies whenever possible. YOU can put that "fast regulator" on YOUR amp, but I won't put it on mine sir. Compare it to what I do and report back to us please.What is the "time constant" of four simple 25K 12 watt Mills MRA-12 non inductive R's placed across the B+ supply on C2 of a L1/C2/L2/C2 filter? It draws about 25 mills and THAT is the KISS Shunt Regulator I employ for G2.
Ok, Sir.
My English is good enough to recognize your unfriendly tone.
I simply gave a response to a technical problem. I have done so using a rough calculation as an argument. Everybody could follow it or not. That is the liberty of argument.
More than ten years of serious engagement with electronics lead me to very careful conclusions.
It is not easy to understand for beginners - as are some members of this forum - to use expensive eqipment without trying a more inexpensive solution which has not been considered deeply enough.
The only purpose of this forum is to share technical arguments as we cannot share perception through this forum. This does not exclude reporting on our perception, of course.According to my experience, a fast regulated screen supply sounds better than any other solution. I made this experience with a 829B push pull amp.
If your question on time constants is serious you should give me the values of C and L.
Hello,I knew from your post that you were simply trying to assist me, and I thank you for that. Also, I am very SORRY to have given my post the wrong tone. I apologise.
Allow me to explain please. I have been building amps for thirty years now, and I up until a couple of yeras ago, I'd actively series double-regulate everything, but technically and philisophically now a days, I am not so sure that that is the best way for ME to go about it.
An active regulator employs NFB, has its own time domain, and I feel that if I can get similar results, or maybe better results, without the complexity of single ( or double ) regulation, I would like to do that. I've not tried all regulators, but I personally do not think ANY of the regulators are really good enough for G2 audio.
Part of the modifications to the pentode amp I am now doing consists of REMOVING the Audio Research D-75's tube type Screen Regulator, and converting it to a passive Ultra Low DCR supply. It sounds considerably better to me without the stock active regulator on the screens!! I feel the stock regulated is just not good enough, in fact - pure poison to the purity, authority, etc. of all these D-75/76/150 amplifiers.
Now my friends, if I can effect a brute force Shunt Regulator for G2, with just four 12 watt MRA-12 Mills 24.9 K ohm resistors paralleled across the G2 B+ supply, I feel that is a preferred approach for ME, no time domain problems, no NFB employed, about the ONLY downside is that there is noise, from the resistor being across the supply, hence my use of Mills NI's.
When I was ten and twenty years into audio, I was looking for the best " magic active regulator ", but I don't do that much any more! (I designed and built an eight chassis 300+ pound amp in 1976 with 28 active regulators in it.) I trust you understand me now a bit better, and I thank you again for your sincerety and care.
Respectfully,
Jeff Medwin
Jeff,
excuse me please, I have to do a lot of work at the moment, I want to reply next week or weekend. I have some interesting ideas.
...with Jeff sticking to his guns and others chipping in with what works for them. GREAT food for thought! Anyone following this thread should have half a dozen ideas to try, from cheap to spendy, and will learn a lot along the way. As Jeff says, "try it"!
Jeff,We discussed pentode operation in a different thread. I think I understand your position and I realize that the ideal screen supply is a separate, low impedance source, so that it is isolated from the B+ sag that can be caused by the large swings in plate current. Regulation is out because it is a source of noise.
Problem is, this requires investment in a second power transformer (plus rectifier and smoothing network) for the screens, plus sufficient space to install it. What about using chokes with sufficiently large smoothing capacitors to provide a reasonably steady voltage for normal listening levels of, say, orchestral music, when the current rises only on transients?
What I have in mind is a PP EL34 stereo amp with an SS bridge rectifier followed by CLCLC network, consisting of:
100uF smoothing cap
5H choke
470uF reservoir, take-off to OPT centre-taps
10H choke
235uF, take-off to screens and earlier stages.Whilst I realize this is not quite the azpproach you recommend, would it be worthwhile to try or just a waste of time, in your opinion?
Ray,
You got it all wrong. No free lunch on G2.(1) Screens need their own totally-SEPARATE G2 supply
(2) Iron MUST ABSOLUTELY be under TEN Ohms - to get amp to boogie, and far LOWER than that is sonically MUCH more desireable. ( One should really read ALL my G2 posts.)
I currently use a Power Trannie that is 2.3 ohms in DCR, weighs about 18 pounds, and an L1 which is ONLY .16 ohms DCR !!
Low low low DCR RULES, must be given priority ...... on G2!!
( No ones ever heard this execution AFAIK. )
(3) Typically requires an external G2 supply to the amp, but its worth it to me, maybe not to others - 'till they were to hear it !
Jeff Medwin
Ray,Here is what I like to use now, iron-wise, with some minimum alternatives, done from memory so I could be off a digit or two :
Power Transformer....... " The Source "
Signal Transformer Corp's DU1/2, 500 VA, 2.3 ohms DCR, the best.Its got the STUFF man ......... provides " true old-time authority, musical freedom, and magic" to the audio's presentation!!
Signal Transformer Corp's A41-175-230, 175 VA, 7.44 ohms DCR, ( .76 A. ).No slouch, but sonically not in the same league as the DU 1/2, the MINIMUM one should use on G2. Easy to "fit in" versus a DU-1/2, but..... SO what, I 'wanna HEAR the amp at the best, so just 'gimme that DU 1/2 please.
L1
Signal Transformer Corp's CL-6-12, only .16 ohms DCR !!! at a 12 MILLI Henry rated inductance. This thing is dynamically cool to HEAR. Obviously Jeff ain't concerned ONE IOTA with "reaching critical inductance", NOR is there any concern over the much-trusted "P-Dud" computer program....this inductor just sounds good to ME and thats ALL I need to know and have.
Triad CX-40 (?) from the Allied Catalog, .32 Hy at 10 ohms DCR at .6 A. current capability, not too expensive, eleven dollars or so. But really Ray, the preferred Signal CL-6-12 .16 ohm inductor for the L1 spot just kills the Triad to my ears.
L2
Lundahl's 10 HY choke, I forgot the number, maybe LL-1338 /10 HY. But it consists of two 5 Hy / 18.1 ohm DCR windings on a common core, which I parallel to obtain 9.05 ohms DCR at 2.5 HY, for some HY. filtering with L2 !!I don't know of any alternative to this choke for G2's L2, at this time. Its about $80 from Kevin at K&K.
Now you "know" Ray, and maybe this is too much to do, but I LIKE what it does, a whole lot.
Using Pentodes?... with all else optimized, its *** ALL *** in the G2 supply implementation !!
Jeff,thsnks for the information. I'll file it away for future use. (I'm not in a position to acquire more parts at the present time).
One other thing occurs to me: the critical voltage to hold steady is the cathode-to-screen voltage and this requirement is handily met with small-signal pentode stages by use of a cap between screen and cathode. However, since that's probably not practical with a power pentode stage, it implies that the use of any resistance in the cathode, such as the unbypassed 10 ohm resistor that is often included as a handy means of checking bias, is going to degrade things. Is that correct?
Ray,Excuse me. How the hell would I know that?? All I know is what I build... and if it works ( sounds decent ) or not! I am a BIG technical dummy of sorts.
Go ask an engineer such theory questions. Many an engineer will know ALL such answers but ( often and also ) doesn't have a "clue" as to what sounds good, or, how to get it !!
I am limited friend!!
Jeff (and others) thanks for your contribution. Last question (to make it complete) what capacitors (and what values) are you using in your CLCL approach ?
Since my G2 operates at anywheres from 300 VDC to 330 VDC, You need to use 400 VDC caps as a minimum. They do not see big turn-on high voltages, due to the shunt resistors in C2 section of the supply.
HI,Good question pipo, here is the rest of my circuit implementation:
For caps, I am EXPERIMENTING STILL.
Although not applicable to you, for the ARC D-75 / D-76 series amps I actually use TWO of these supplies !!!, one for the ARC's critical Cross Coupler B+, and another for G2's B+. Its all filtered L1/C1/L2/C2.
Right now, Jeff is using 1979 Photoflashes, series/parallel, about 1,000 uF for C1 and about 1,000 uF for C2, PLUS ( on the G2 supplies - at the C2 location ) a 220 uf @ 400 VDC Panasonic TSHB cap ( DigiKey ) with a MILLS MRA-12 24.9K ohm resistor ( only a MRA-12 ) across EACH of the four TSHB caps, which becomes four paralleled Shunt Regulator resistors, which equals 6.22K @ 48 watts for G2 SHUNT - B+ to ground. All four TSHB caps are located within 1 1/4 inch lead length of Octal Socket's Pin 4, which is G2.
Total capacitance for G2's C2 is thus - 1,000 uF plus 880 uF or 1,880 uF, and those Photoflahes measure LOW in ESR, act like little atom bombs.
If you don't have Photoflashes of that era, TRY the low cost Panasonic TSHB 400 VDCers. I find caps all have different sounds and some work needs to be done with the C's to get it acceptable. For wire, I use 14 gauge stranded copper, not plated in the power supply, and on the Audio Boards, Grounds are hefty Kimber Kable TCX and B+ runs are all paralleled pairs of Kimber Kable TCSS.
Due to low DCR of supply iron, combined with high capacitance / low ESR caps, one MUST always use a 47 ohm CL-80 ( from Mouser ) " Current Inrush Limiter" / Thermistor on the primary of the power transformer, and I just ordered four 8 A. at 1200 VDC Fairchild Stealth diodes for SS rectification from Mouser, as uF4007 full wave bridges blew-up on this supply's severe turn-on surges.
Signal Transformer iron is direct-order from Signal factory in NY USA, and their full line catalog downloadable on-line. Alternatively, for a power trannie, find any equal or greater than 500 VA isolation XFR on eBay with a 120 VAC to 240 VAC capability. ( Gives about 298 to 330 VDC final B+ for G2 ). These are big and heavy, but can't be beat by anything I know of. The Signal Transformer DU1/2 is a very nice part, may be 2.15 ohms as I see in notes, not 2.3 ohms in DCR, darn low, unflappable - stable!!
It is an interesting diversion to actually measure the noise on a competently designed regulator's output. See, for example, Errol Dietz's paper in EDN (this was, IIRC, in '89) on noise performance of LM317.Of course, a major part of competent amp design is, whenever possible, to use topologies and operating conditions which are inherently uncritical of their power supplies.
I infer that you mean that it's better to avoid pentode mode if that's going to place critical requirements on the PS design. That's what most people do, of course, especially commercial amp designers, mainly by using UL mode. However, UL is not universally accepted and pentode mode has much to offer. If it's only a question of taming the PS to give the screens what works best for them (i.e.a steady voltage from a low impedance source), then why not?
My point was that, with proper design, regulators are quite quiet, even the cheap 317 variety.Even more so in output stages where you've also got high common mode rejection. If things are approached without prejudice (you and I see eye to eye on that one!), a solid state regulator is a clearly good choice.
IMHO, a 317 type regulator is NOT good enough by a longshot for Pentode G2 screens, nor is any active regulator I've heard good enough.
In pentode mode, you want to stabile voltage, not current. Zener string can be used, or VR tube(s). Series drpping resistor selected for more than peak screen current.
This post is made possible by the generous support of people like you and our sponsors: