6S19P Amplifier

[EWBrown]

(title is a little reference to Animal House)

Have you considered using 6N1Ps for VA and PI stages (keep it an all-Russian tube design), I've run the numbers for these on TubeCAD, for general usage, and they look viable. They're fairly close to 6DJ8s, but more current for the filaments and differing gm and Rp, the amplification factor is 35. I like 6N1Ps as they're also dirt cheap, and if you get those which have the "diamond" mark and the letters OTK, those are the best ones, that have been military-qualitied.

The toroid power trannie as an OPT sounds like an interesting approach.
I remember once seeing a "cheap amp" SET design which used a 6.3 VAC filament trannie as the OPT, for roughly 3K:8 ohms. I never tried it out, but it has possibilities...

Tom, is that 95WPC amp one of the Velleman kits? Ramsey ELectronics in NY used to carry them a few years ago, but have dropped that line.

/ed B in NH

[TomMcNally]

Yep - that's the Velleman K-4000 ... I built it about 10 years ago. Aside from some early firecracker Tesla brand (now JJ) EL-34's ... it's been very reliable.

I got one of the very first ones in the US, I waited for a 115 V version. It was funny, they were selling for about $ 1000, and no one in the US was carrying them, so I called Jameco, who had their small kits, and asked them to see if they could get me the amp and matching solid state preamp. A woman called me back a few days later and said, sure - no problem .. the amp is $ 666 and the preamp is $ 190 (or something like that) and told me how to order it. A few minutes later she called back kind of stammering and hesitant so I said "You quoted me your cost didn't you" and she sheepishly said " ... yes" So I said "since the cat is out of the bag - I'll offer you 10% over that" and she called back a minute later and said "the boss said OK"

Anyway - I haven't modded it or anything ... it sounds good, and with the big Cerwin-Vega! speakers shakes the house.

[sorenj07]

Well, I won a pair of 10K a-a transformers out of a Hammond organ:
http://tinyurl.com/2jd8dq
I think this project is back on track. I'd planned for 8K a-a in my most recent designs but this should be all right.

I see myself as having two choices: Fixed-bias or cathode bias. Either way, The output tubes will be biased at 220Va, -85Vg, 40mA each. The PT will probably want to be able to put out 200mA comfortably, assuming minimal variation in current draw (class A, after all). A question: How close to the 11W dissipation should I go? At the moment, I'm clearly only at 8.8W, or 80% of the max. I've read somewhere that 80% is a sweet spot between tube life and power?

Fixed-bias:

Rectify 250V into one choke for the B+ supply, and have another set of diodes rectify into a cap-input Pi filter for the secondary B+ supplies at around 350V or so. This should give enough headroom for 150V of swing. However, will needing to smooth an entirely different B+ line not be worth the hassle? As an option, I have a schematic for a decent-looking MOSFET-based regulator that spits out a clean 300V from 340V input...

Cathode bias:

350V PT. Choke-input rectify to hopefully around 310V, minus 5V for the OPT, and 85V for the bias resistors, to 220V on the plates just like before. Will a 100-ohm 2W potentiometer have enough range to adjust between mismatched tubes? I'd assume half its value, 50 ohms, in series with another 50 ohms, in series with a 2K 10W resistor for 2100 ohms total, which according to the datasheets should put me right at 40mA. Could I get away with bigger-value pots, i.e. how much power should I assume these will dissipate? Max current at the peak of peaks should be 90mA, so if a 100-ohm pot is turned all the way, it should still only dissipate .81W right?

Now, for the rest of the amplifier. 150V of swing is no mean feat, but here's how I was planning it: the phase splitter consists of two 6N2P's in SRPP, joined at the bottom cathode and made into a LTP, direct-coupled (Aikido-style?) to a standard 5687 grounded-cathode driver stage, which then feeds into the 6S19P's either by RC or by the fixed-bias.. One question is where to try to inject some NFB, if I ever want to. A fully differential triode amp probably won't need too much, though.

One concern of mine is heater current: four 6S19P's is 6.3V 4A, then four 6N2P's are 1.36A, then two 5687's are 2A - total 7.36A. Maybe I can rectify the unused 5A line off of a standard Hammond transformer and use an LM-type regulator for the 5687's and 6N2Ps? I'd just wire the 6S19P's out of phase and hope that's enough.

Sorry to bug you all with questions and I'll try to come up with a comprehensive design, according to which general layout sounds better. What do you think so far?

[sorenj07]

This seems like a somewhat elegant solution. Buy an isolation transformer with dual primaries or secondaries, or both. Wire it so that you have 115V primary and two 115V secondaries.

1 bridge rectifier off the 240V (120V lines) goes to an almost-choke-input with a 3.3uF cheater cap to get 220V B+. Another BR goes to a full cap input supply which gets regulated to 300V or above for a good B+2 supply. For bias, I'd bridge-rectify off of just one of the 120V secondaries, then steal the layout from the Diytube Eiclone to get my voltages, adjusting a bit for the higher bias voltage.

Does anything seem intrinsically wrong from this approach? Allied Electronics has a 100VA transformer with just this configuration for a very reasonable price.

[EWBrown]

One characteristic of most of the step-down isolator trannies, is that the turns ratio isn't exactly 2:1 input to output windings, it's more like 1.8 to 1 as it is set up to favor the 120VAC secondary, which will typically deliver 130VAC unloaded (or a tad more) with 120 VAC or 240VAC input. The consequence of this is ,that if it is used as 120VAC in and 240VAC out, the B+ voltage will turn out to be quite a bit less than expected, if the current max is approached.

I had tried this a few years ago with a Triad 50VA trannie, and got about 260VDC unloaded out, where i expected around 305VDC. It's OK for lower current or lower B+ application, but didn't work out as I'd hoped.

(FWIW, it would be just about perfect for a 1626 "Darling" amp B+ supply).



OTOH, the An-Tek dual primary, dual secondary, 200VA (actually 250VA), 230VAC + 230VAC toroid has a perfect 1:2 windings ratio, and LOTs of current reserve, plus the usual 6.3 and 12.6 VAC filament windings. It won't "sag" under typical tube amp current loading.

This toriod is also quite reasonably priced, and even costs a bit less if you don't get it from his e-bay site.

The" trick" with these is to use both windings for the b+ supply, either in parallel with a full wave bridge, or as 460VAC CT with a hefty rect tube or two SS rectifiers. A dual mono setup could be run with each 230VAC winding having its own FWB and caps.

These toroids were designed with 50 Hz capability, and will have an extra approx 20% current capacity with 60 Hz input.

THey are kinda big and ugly, and best "hidden" under the chassis... :o

You could derive the grid bias voltage with a capacitive "tap" as was done with the EIclone "IKE" amp design. no need to tie up an extra winding...

/ed B in NH

[sorenj07]

I'd really rather have a somewhat presentable amp. I love the idea of using toroids and have used an Antek with success but I think I might as well just commission a custom job from Edcor if I'm going to be spending "retail" prices. I'll keep an eye out for a belled EI transformer that does what I want...

In the meantime, I'm trying to design the DC coupling between the 12AX7/6N2P SRPP LTP (quite a string of letters) and the 5687 driver stage, but I realize that I don't have a clue how to bias the 5687. From my very limited understanding, the 5687's cathode resistors must be of a high enough value to bias the tube "past" the 150V or so that will be present on its grids, correct? The 5687's should have a B+ of 320V or so.

Any help would really be appreciated :) I'd love to be able to finish this thing over winter break. I grabbed a ten-pack of those silvery K40Y-9 Russian PIO's to continue the theme, at .22uF 1KV. I'm considering using them on the input as well, to block DC if I decide to try to run some feedback. I probably won't.

[EWBrown]

Basically the 5687's cathodes have to be a little higher DC voltage than that present in its grids and the plates of the 12AX7 / 6N2P. So, a large value cathode resistor, suitably bypassed, will be needed..

(I'll atach the link here, shortly)

http://www.the-planet.org/6EM7.html


Gary Kaufman has a 6EM7 direct coupled SET amp on his site, the output section runs with the cathode at 140VDC above ground, at 50 mA, and the cathode resistors total 2800 ohms, the grid has 115VDC from the driver stage.




I "breadboarded" a single channel of this amp, just to try it out and it sounded darn good!

The reason that I mention this particular circuit , is that the same or a similar approach can be used with your DC 6N2P / 5687 design.

/ed B in NH

[sorenj07]

Ok, cool. I've since then explored the possibility of eliminating the 5687 stage. There should be just enough room to do it. Here are my schematics so far.

A note on the LTP supply: I want to drop A LOT of voltage in the tail resistor of the LTP (down to 400V, as seen in the Dissident Audio 6AS7 design) so that it acts very much like a constant-current source without actually being one. That's why I have a line of 650V or so. Or is this impractical? Should I put a big RC filter after the rectifier or something?

Finally, I should give credit to the DiyTube Eiclone for the bias adjustment scheme. How much current should I assume? All the resistor values and wattages kind of hinge on this, which is why I left them out.



What do you think?

[EWBrown]

From a quick look at your power supply diagram, it appears that some of the rectifiers will be "fighting" each other, especially in the bias circuit, vs the 230VDC B+ circuit. and the voltage doubler will be "floating" the lower end of the 120 + 120 VAC winding at 240VDC above ground. That will get directed through the lower right diode in the bias circuit, to ground, which means something is gonna get "toasted" in fairly short order.


I'll have to do some more detailed circuit analysis, but I see several
electrical "conflicts" here which will cause serious problems.




For convenience, I'll number the diodes in a left-to-right zig-zag descending order, so the uppermost left side diode will be D1 and the lower right hand diode in the bias circuit will be D10.

The four diodes in the 230VDC b+ circuit may have to be somewhat "heftier" than uF4007s, perhaps uF5408s (?) in order to survive any current spikes and peaks.

The 6S19P-V tubes should take very little grid bias current, in the order of microamps. The overall current drawn by the bias support circuitry is dependent on the resistor and pot values, and it should be only a few milliamps at most.

The "tap" cap (refer to the Ike diagrams) value can be determined by experimentation or a little application of Ohm's law.

Xc (in Ohms) = 1,000,000 /( (2X pi X F (in Hz) X C (in uF)).

Since we're dealing with 60 Hz AC, this can be reduced to

Xc = 1,000,000 / (377 X C (uf)

and further reduced to Xc = 2652.5 / C (uF))

Plan on some extra "leeway" for the bias voltage, maybe -125 VDC off the rectifier anode, then into resistors and pots, as needed.

Since you want about -85VDC for the grid bias, then plan on the ends of the bias pot being at -65 and -105 VDC, approcimately This allows for adjustment for mismatched and "aging" tubes. The scheme used in the "Ike" amp is perfectly good, the trimpots should be around 20-25K each, this keeps the current through each pot to about 1.6 to 2 mA, with a 40V drop, for a dissipation of approx 80 mW per pot.


Then it's a relatively simple matter of dropping 65V off the "ground" end and 20VDC off the "high" end of the trimpots.

I've never seen an SRPP phase splitter design before, theoretically it should work just fine.

HTH

/ed B in NH

[EWBrown]

There is a means to run both a full wave bridge and a voltage doubler off the same trannie winding (no CT required), so 2/3 of this circuit can be made to work (I posted a rederence to this one a while ago, I'll try to find it).

It's called the "4X8" doubler circuit and can be found here, over at "the bunker of doom", just scroll through all the "conspiracy theory" and "doomsday paranoia" stuff, and there is some good electronics to be found...


http://208.190.133.201/lit/4x8/index.html

(rant mode on)

I can't currently access this site since the corporate "net nazis" consider it to have "controversial content" Yellow_Light_Colorz_PDT_21

(rant mode off)

This "4X8" circuit is a true full wave doubler, and the normal (non-doubled) voltage is also available. It's advantage is that a non-CT HV secondary winding can be used.


The Bias circuit is the real "villain" here, it causes at least two virtual short circuits, and there are some diodes which end up being connected in parallel, and tying two different DC voltages together - not a good thing...

The doubler in your circuit would set the lower end of the secondary winding at 220VDC (just before the smoke starts pouring forth), but then one of the diodes in the "bias" circuit shunts that to ground...

the best approach would be to use something similar to the "Ike" method, or perhaps a backwards connected 6.3VAC filament trannie fed off one of the existing filament windings.

/ed B in NH

[sorenj07]

Yeah, I'll have to re-think the way I'm going to provide some juice to this amp. Anyway, here's a small update on the OPT's.

Without voltage measurements, I just assume that 10K a-a is what they are.

[EWBrown]

The DCR seems kinda low for a 10K A-A primary. I would expect around 100 ohms, on each side of the CT, unless this is a really hefty high-power OPT....

A good way to estimate the actual primary impedance, is to connect the primary across a low voltatge AC source (anything from 6VAC to 24 VAC, (DON'T use the raw AC powerline), and measure the AC voltage on the 8 ohm secondary, and you might as well measure the other winding's voltage, too (the K1 and K2 cathode feedback winding).
Anything like a filament transformer, AC output wall wart, train trannie, etc, is good enough for this purpose.

The primary impedance will be the (8 ohms ) times the square of the ratio of the primary to secondary voltages.

For a 10K winding, this should be about 35:1 ratio. (actually 35.35:1, but then these things are never that precise). The square of the voltage ratio would be around 1250, X 8 ohms for 10K A-A.

Cathode feedback was an interesting method to inject NFB directly into the power section of the amp. THis can work with triodes, tetrodes, pentodes, BPTs, etc... It wasn't a common practice.
Typically, the cathode winding impedance is around 50-100 ohms K-K.

HTH

/ed B in NH

[sorenj07]

Update! Measured the various impedances of one of the OPT's (I assume it holds true for both) and got some interesting numbers.



I'm not sure what to make of this. My first impulse is that this opens up a lot more tubes I could try and use, my second impulse is to go with 6S19P's all the more because it's quite close to the 4K impedance I calculated for...

[EWBrown]

I had earlier stated that I had never seen an SRPP input / phase splitter design. Well, here is one, usedin a 6V6 PP design:

http://www.bonavolta.ch/hobby/en/audio/6v6_1.htm

It should be applicable, with a few R value changes, to accomodate 6N2Ps with 300 to 325VDC B+, to drive your 6S19P-V PP amp.


/ed B in NH

[sorenj07]

sweet. i'll keep this one in the memory banks, because at the moment the only parts of my "amp" that i actually have is 8 6S19P's! Yellow_Light_Colorz_PDT_07 Unless, of course, you count the 550VA 1:2 or 1:4 isolation transformer I have! OTL anyone?