Single-Ended 6S19P (Again..)

[sorenj07]

Edcor's new line of "guitar" OPT's that are good down to 40Hz, probably better at lower power, makes this project a bit more attractive.





I have everything I need except said OPT's and a couple PSU chokes. I'll use a 35VA isolation transformer into a doubler for HV, and I've got a 35VA 12V toroid for filaments. Not pretty but I can try to hide them in a box Yellow_Light_Colorz_PDT_07

[EWBrown]

Looks good! The 6DJ8 SRPP has sufficient drive voltage to achieve full power, though the gain is around 28.5, so up to +/- 3V signal swing will be necessary to max it out, with a capability of -89 / +88.5V swing.. Tubecad shows this to be a perfect combination!

FWIW, I ran a tube cad sim for 6N1P , with RA + RK = 1000 ohms, and current = 3 mA, the gain is nearly identical at 28.9. with a capability of a -88 / +87.7 swing, which is also sufficient.

go for it!, either driver tube is good to go. Yellow_Light_Colorz_PDT_03

/ed B in NH

[sorenj07]

Cool. I like the 400-ohm output impedance. Maybe this SRPP could even drive two 19's? I guess I didn't notice the bias point of -3V. Maybe yet another gainstage is necessary. I just don't seem to be able to get the gain I want out of a 12AT7, paradoxically.

Either that or I wire this as a straight up power amp and feed it with a linestage?

[EWBrown]

The higher mu tubes should give sufficient "swing" to the 6S19P grid , with a B+ of 325V, either 12AX7 or the similar Russian 6N2Ps, with at least 2X the gain of the 6DJ8 or 6N1P. The 12AT7 will also work with somewhat less overall gain than these other two, in SRPP.

Update:

With a 12AT7, 300V B+, I= 1.8ma, RK = RA = 1K, swing is -89 / +89.3, which is good, even a bit more than needed. Gain is 48.86, so a +/- 1.7V swing will max it out, which is better than +/- 3V.

and for 12AX7 with the same B+ and RA / RK of 1K, the current is 1.1 mA, and the swing is -75 / +75 which isn't quite enough, so 325V B+ would be needed which would then deliver a -82 / + 81.7 volt swing, with a current of 1.2 mA. Even that is cutting it kinda close... :o

So the higher mu tube gives less SRPP gain with 300V B+... One of those strange "mysteries" of the tube world Yellow_Light_Colorz_PDT_08


If you use it with a preamp, then a +/- 3V input "swing" isn't an issue, just it won't deliver full power with most CD players plugged straight into it.

Tube CAD, 1960 style:

FWIW, I ran your curves, according th the RCA tube manual method (extended down to vg = -5V as they didn't have a curve for 0V, and up to vg = -160V) and came up with:

Presuming that V P-K = 215VDC, @ 48 mA:


OPT primary impedance: 3295 ohms

Power out: 3.19W

THD = 4.55% (most of this is 2nd harmonic content which is good)

lower Z OPT will deliver more power and THD,
higher Z OPT will deliver lower power and THD.

You probably don't need the NFB loop with figures like those.


/ed B in NH

[sorenj07]

Wow. Thanks a ton for all the old-school legwork. Over 3W is pretty sweet! I have a friend who might be interested in a super low-cost SET amp that can put out that kind of juice. I'd thought it was an oxymoron, but maybe not...

Anyway, about NFB. It shouldn't be used to excess to make a terrible amp decent, but trying just a bit with this amp might still be worth trying. Am I right in guessing that it could improve bass response?

[EWBrown]

It may help bass response but what NFB does is to reduce the overall gain. and also lowers the effective output impedance (which should helpthe bass response). You can try it with and without, go for what sounds best...

I found earlier on that a 12AX7 or 6N2P SRPP wit at least 325V B+ will d eliver enough "juice" to fully drive the 6S19P, you might try tapping off before the main choke (off the rectifiers' cathode, then go thru a separate low current, high inductance choke for the SRPP B+. Typically the 12AX7 / 6N2P is only around 1.25 mA per channel with Rk and Ra = 1K.

A 12AT7 will also work, at 2 mA per channel, with the same 1K resistors, somewhat less overall gain but still enough total signal "swing" to do the job.

I've been trying to work something up with my 6C41Ps, but they are even more "demanding" for a high negative grid bias potential, I may have to go with a pentode driver stage with that :o

/ed B in NH

[sorenj07]

Sounds like an idea, but my original reason for designing this amp was to absolutely minimize cost, and I happen to have a couple 6DJ8's kicking around. If I decide to go out on a huuuge limb and drop the money for a discount box of 6N2P's, I will :)

This is on the back burner though. The 6S19P has gotten me intrigued for a linestage with gain of around 2.5. GC with 285V B+, 6.8K anode load, 2.2K cathode (cap bypassed), 20mA at -45V. whatever flavor or size of coupler, >= to .22uF or so, at 400V. Zout of ~350 :)

[EWBrown]

THat should work well for a linestage. THe biggest linestage I've seen once was one which used 211s :o :o :o Just shows that ANY triode will work in SRPP or as a simple grounded cathode stage.

and it works quite nicely . Decidedly massively overkill, however...

/ed B in NH

[sorenj07]

One quick question about the input stage. If I bump the 6DJ8 SRPP plate supply to 270V (resistors to 680 ohms), I get nearly 100V of swing out of it from 3.4V input. My plate curves only show +-60V for the area in red that I've sketched out, which looked to be as far as I could go in terms of linearity. Is your 3.2W figure for the amp based on this swing, or on driving the 6S19P to clipping? Just wondering how you calculated it. I'd like to learn how to do this too :)

[EWBrown]

The power rating I calculated was for the maximum power capability (before clipping) of the 6S19P. The VA stage will have to deliver a +/- 80V peak swing (160V pk to pk) to achieve that level . It's the overall peak swing that matters, the coupling cap remives the DC offsets, so something like -60 +100 is still OK...

Lesser swing will mean less power. It's another "square" function...

I did a quick write-up of the "paper cad" method, from the RCA tube manual, It's pretty easy, once you've done a couple paper trials it'll become second nature Yellow_Light_Colorz_PDT_02

It's located here:

http://diytube.com/phpBB2/viewtopic.php?p=12458#12458


About the only SRPP combo I found that had a reliably high enough gain was 12AX7 (or 6N2P) with 325V B+ and even that was fairly close...
I used the same combo for driving 300Bs, and it works pretty well. I generally like a bit more gain so I can get full power output, even with a portable CD player or MP3, which tend to run around 1V peak on the line output, where most "quality" players run 2V or even more output.

I'll run some more tube CAD SRPP sims and see what I Can come up with.

BTW if you need some 6N2Ps, let me know, I have LOTS of them from a bulk lot, , and can spare a few, gratis... Same pinout as 6DJ8 or 6N1P, and about 99% close to 12AX7 operating characteristics.

/ed B in NH

[sorenj07]

That would be very cool, I'd greatly appreciate it! However, running some sims in TubeCAD, I'm noticing that at 100K Zin, a 12AX7 SRPP hits -3dB at 14kHz... Is this something to worry about?

Unfortunately 300V is all we're going to get out of my PSU at the moment. It's a doubler off a 35VA isolation transformer that's already feeding 96mA.

Hmm. We're really between a rock and a hard place, eh? I'd be tempted to stick with the 6DJ8 though, because its output impedance is much lower, and at 300V for the 12AX7, I'm getting the same 100V output swing.

Would it be possible to just DC couple a simple GC gainstage to the SRPP input? It'd have to be a 12V tube if it's single. Maybe a 12BH7? I wouldn't need the gain from cap bypassing, so at least NFB would be simpler to hook up..

Problem is, I dunno how to DC couple to a SRPP, because I don't think you can hand-calculate it off plate curves the same way...

[EWBrown]

Looks like 12AT7 with 300V B+ and RK = RA =1K should work, (posted in another update), and the 12AX7 isn't quite so good, after all......

Update:

With a 12AT7, 300V B+, I= 1.8ma, RK = RA = 1K, swing is -89 / +89.3, which is good, even a bit more than needed. Gain is 48.86, so a +/- 1.7V swing will max it out, which is better than +/- 3V. Maybe experimenting with adjusting RK and RA can kick it up a little bit, but not significantly.

JRB usually considers an SRPP as having a max useful gain of approximately half the of the tube's mu, but that tends to be somewhat conservative.


I've been back-editing these posts in this topic, so some "newer" postings may clash with "older" postings, I'll get it all straightened out...

I wouldn't DC couple into an SRPP, things really get complicates and weird with that, best to just cap couple, a 12BH7, or 6CG7 set up for around 10 dB gain (3.X) should be sufficient, if you need the extra sensitivity. I'd go with the SRPP frist then add in the "up front" stage if it is really necessary. .

/ed B in NH

[sorenj07]

All right. And how seriously should one take the warnings TubeCAD spits out about "insufficient plate current"? Its max voltage rating of 130 for the 6DJ8 plate is a little wack by the way.

It looks like the 12AT7 wins out. Can't beat the gain of 47 at 2mA plate current. I wonder if the tube's "iffy" curves cancel out due to the SRPP topology?

[EWBrown]

I wouldn't worry about "insufficient plate current", tube CAD likes a minimum of 20% of the max rated plate current, but I run tubes at "insufficient plate current" all the time and they sound fine.
Iusually run 12AT7s between 1.5 mA and 3.5 mA Max, and they sound fine... Lower plate current (and higher plate load resistor) means more gain.

BTW, I re-ran 12AX7 and get an upper -3dB freq response of >1 MHz in nearly all SRPP cases.

/ed B in NH - almost time to dieappear from "the salt mine" for another day Yellow_Light_Colorz_PDT_06

[sorenj07]

i dunno about the 12AX7. looks like i'm shopping for a couple 12AT7's when i decide to get this project off the ground :o

EDIT: how does a CCS sound as a solution to this problem? might be able to get away with only one 12AT7 or so...

http://geek.scorpiorising.ca/GeeK_ZonE/ ... 9#msg42449

EDIT 2: quick question: do CCS' decrease output impedance? a 12AX7/6N2P type in GC with a CCS might be just the ticket then.