There was a SS Amp by Randy Slone, the Optimos amplifier, that did much the same "soft landing" technique, it killed VAS stage voltage gain as the output Mosfets approached the V rails. A modulated cascode dropped the VAS stage voltage down just before the Outputs hit the rails.
So I think this "soft landing" scheme could be real for the SP-10/20. Eliminates harsh saturation peaks in a high Fdbk design. That's what people complain about after all.
Which means we can't just bias the 6AU6 (or whatever tube there ) driver up into higher current without loosing this effect, unless carefully done.
Making the Amp much bigger, so it doesn't saturate, is the more modern approach.
So I think this "soft landing" scheme could be real for the SP-10/20. Eliminates harsh saturation peaks in a high Fdbk design. That's what people complain about after all.
Which means we can't just bias the 6AU6 (or whatever tube there ) driver up into higher current without loosing this effect, unless carefully done.
Making the Amp much bigger, so it doesn't saturate, is the more modern approach.
Last edited:
The SP-20 used PPP 6V6s, which essentially makes a different output tube, yet RCA retained the exact same driver (PP 6AU6 w/ Rp = 330k, Rg2 = 1M, Rk = 1.8k, Vg1 DC push-up R = 330k, Rnfb = 100k). That's the one Doug heard that impressed him. So it seems this driver stage is more flexible than it at first appears to be.
That's the thing that's driving me crazy. The sim is working fine.
With 6CB6 for the driver pentodes, THD is predicted to be 0.05% at 1W out with H2 dominating, and 1% at 9W out (just shy of clipping). Substitute 6JC6 for the output tubes and THD at 1W goes down to 0.02% and max power out goes up to 17W. That's with raw B+ of 330V, which is close to the original.
The "right OPT" was a garden variety 25W-rated Stancor with 10k p-p primary. That 25W rating was quoted with low end limit of -3dB ay 60Hz. So I'll bet it's comparable to the Hammond 1608 rated at 10W for 20Hz (-?dB), or maybe the Edcor GXPP15-10K which is rated at 15W down to 40Hz (-1dB). There is also an Edcor CXPP10-10K that's rated for 10W down to 20Hz at -1dB.
That's interesting. Wouldn't it be something if that's what's really happening in this circuit? If that is the case, would increasing the gm of the driver pentodes help matters? Simulation predicts it will. Substitute 6BC6 or 6EW6 for the 6AU6? That would be easy enough to do, since they have compatible pinouts as long as the g3 and k pins are shorted together.
6CF6 is a near-equivalent to 6CB6, except that it has a very sharp cutoff to the left of the knee. I take it that means it has an abrupt turn-on as plate current rises, or you could look at it as a very abrupt turn-off as plate current goes much below 1mA. That would mean 6CF6 would be a terrible choice for this circuit, as it really would go into cutoff early, being biased so cool (Ip about 450µA).
Interesting stuff here. Thanks for your thoughts on this, everyone. 🙂
If I were to build this pup as a single pair, 6V6 there is no question I would go to Heyboer and have them wind up an S265Q coil and stack it with either M6 or M4. This would buy me the ability to jump to PPP if I throw away the 16R capability and load the series connected secondary with 8 Ohms. Those OPT's are not to be despised... 🙂 I don't think I would rob my 6CB5A amps of their S258Q's for a PPP version( or perhaps a 6L6GC/KT66 in single pair ).
I freely admit the basic reason I have yet to build one of these is I have not gotten a thorough idea of how to improve it...LOL beyond replacing the VA/PI stages anyway...
Not that it really needed improving. There is another reason for disinterest; it is already damned brilliant, and building to recipe is not what gets me spending $$$ and effort.
Douglas
I freely admit the basic reason I have yet to build one of these is I have not gotten a thorough idea of how to improve it...LOL beyond replacing the VA/PI stages anyway...
Not that it really needed improving. There is another reason for disinterest; it is already damned brilliant, and building to recipe is not what gets me spending $$$ and effort.
Douglas
I don't know, simulation needed I think.
I did one calc. that makes me think it could be OK. When I reduce the current thru a 2x higher gm 6197 to 0.3 mA, its gm drops to around 750 uMhos, same as the 6AU6 at 0.3 mA. ( the 6197 datasheet uses a higher idle current for the tube, which gives it the higher nominal gm rating ) Now these calcs. are approximate, using Sqrt of datashheet current to 0.3 mA ratios to calc. the gm ratio. But being in the same ballpark should allow minimal bias re-work to get the same tube cutoff.
True, but the grid voltage limits between off and full on are the same with the tubes in parallel. Using a very different tube will require some checking of drive voltages. My guess would be that a KT88 will need some adjustments, especially if running UL. An EL34 or KT77 might just come out similar drive V range, with higher gm but also higher output current into a lower Z OT.
Last edited:
One of the previous options mentioned swapping the 1M common g2 resistor for the 6AU6 for a pair of bypassed 2M resistors. Un-bypassing them *WILL change the cut-off behaviour. Note the voltage amplifier curves for bypassed and un-bypassed dropping R to g2. Using the single 1M to a pair in push-pull will be nearly the same as a solid g2 voltage( either bypassed or reg'd voltage ).
The type 6AC7 datasheet shows this sort of behaviour. https://frank.pocnet.net/sheets/093/6/6AC7.pdf
Douglas
The type 6AC7 datasheet shows this sort of behaviour. https://frank.pocnet.net/sheets/093/6/6AC7.pdf
Douglas
I had cited this particular schematic in a thread that I posted years and years and years ago. There was contentious discussion...
I love it. Thanks for the pointers to the Heyboer iron. I looked up a vintage Peerless catalog. (Ain't the internet great? Back in the 1990s I would have had to go on a major search for a vintage catalog like that...) The S-265-Q looks very interesting indeed. Two primaries to choose from. Rated for 40W too. It's a heft sucker too. Is that 10 lbs each? Yowza.
I'm now tempted to build this as is, since it looks like it should work so well, and in such mysterious ways too. If I hate it then I can rework it to a different circuit.
smoking-amp suggested 6JC6 for the outputs, which looks like a good move to me.
In this thread, I'd tried out splitting the single Rg2 into one Rg2 for each 6AU6, but bypassing each of those with a capacitor to each 6AU6's cathode (not to ground). I tried it, and it made little difference. Unbypassed it would have made a lot of difference, because of screen grid degeneration. That wouldn't have worked out well.
We know from class A triode PP output stages that you can use a single cathode resistor shared between the two output tubes without introducing (much) cathode degeneration. But if you use a single cathode resistor for each output tube then you must bypass them, or you'll throw lots of gain away and increase rp of the output tubes by something like 35%. Bypassing the single shared resistor does increase gain a little, so it looks like there's some degeneration with the single shared Rk, but apparently not much. I expect the Rg2 for a differential pair of pentodes would work in similar fashion.
Ah, what's old is new again. There's nothing new under the sun. History doesn't necessarily repeat itself, but it does rhyme.
How do you get matched 6JC5s without buying a bunch and doing the deed? I planned a 6JB5/6JC5 P-P amp, but didn't consider the need for matching output tubes for P-P - egg on my face.... I would have to go back and look at the amp schematic to see if I made provision for equalizing the bias in the output tubes. It had a sand state front end - I have an embryonic bias balancing circuit that I have yet to try out - I had it planned for some 6LR8's with the little triodes implementing a folded cascode driver for the big output pentodes. I guess I could do something similar with the sand state front end...
Last edited:
How badly unmatched were they? Really far off from each other?
Was your 6JC5 design fixed bias? Wouldn't the two separate cathode bias resistors help balance the pair? It would make it so you could plug in whatever 6JC5s you have and measure the voltage at their cathodes. That way you could pick your own matched pairs.
Was your 6JC5 design fixed bias? Wouldn't the two separate cathode bias resistors help balance the pair? It would make it so you could plug in whatever 6JC5s you have and measure the voltage at their cathodes. That way you could pick your own matched pairs.
A while ago I found the Telefunken V69 and implemented the concept in two amplifiers. Just found out later that the plate to cathode feedback is used in many more amps, eg WE.
http://bilder.hifi-forum.de/max/857335/telefunken-v69_646189.png
I did not do simulations, but still very happy with the results, so yes, I would suggest moving on 🙂
http://bilder.hifi-forum.de/max/857335/telefunken-v69_646189.png
I did not do simulations, but still very happy with the results, so yes, I would suggest moving on 🙂
One must be careful when using the 6JB5 / 6JC5 / 6HE5 trio since each of those individual type numbers have a different plate dissipation rating. I have seen several different constructions with varying plate sizes. Some of these tubes wear one, two or all three type numbers regardless of which size plate is in the glass. Tubes branded with TV manufacturers logos are often worst case. I have a Magnavox tube that has the RCA stop sign logo on the top with the number 6JB5 inside it. The white Magnavox lettering on the side carries all three numbers. The 6HE5 was simply a repackaged 6EZ5 with its 12 watt plate. The two larger sizes came later as TV sets kept growing. I have seen 6JC5's with the small plate inside.
If you don't need the extra dissipation, use the 6EZ5. It is pin compatible with the 6V6, 6W6 and the other octal audio tubes. This allows for some experimentation with other tubes. The 6EZ5 used to be on the dollar menu. I ran them in my SSE's where they worked fine. They are still relatively cheap ($3 to $5).
Rongon -
I would have to go back and look at the schematic for an old, old thread to tell - I never actually fired up the amp, so I have no idea regarding any mismatch between output tubes. These days I would just use an amp with matched 6V6 outputs and be done with it. Actually, I went back and looked at the schematic, and it ran on a combo of self and fixed bias, with a pot to adjust the balance of the output stage.
I would have to go back and look at the schematic for an old, old thread to tell - I never actually fired up the amp, so I have no idea regarding any mismatch between output tubes. These days I would just use an amp with matched 6V6 outputs and be done with it. Actually, I went back and looked at the schematic, and it ran on a combo of self and fixed bias, with a pot to adjust the balance of the output stage.
Last edited:
Sadly, the dollar menu is no more, as it passed away with Stan and ESRC... I wonder if somebody else acquired his inventory?
Last edited:
Thanks for the advice on 6JC5 and equivalents.
I see in the data sheets that the 6JB5 is rated at 15W plate dissipation, while 6JC5 is rated at 19W plate dissipation.
Back when they were a buck each, I got a bunch of Sylvania and a couple of Raytheon painted "6JC5/6JB5/6HE5" from ESRC.
The Sylvania and Raytheon tubes look identical inside and all are etched "FJCE" on the glass, like the two on the right, below. The one on the left has its paint rubbed off, but is etched "6JC5/6JB5" (no mention of 6HE5).
Which are these? 'Real' 6JC5? Relabeled 6JB5? Could any of these be relabeled 6HE5?
I was figuring on running them at no more than 15W plate+screen dissipation at idle.
The warehouse formerly occupied by ESRC was confirmed empty some time ago. I met a guy at the Dayton hamfest in 2022 who claimed to have bought the ESRC inventory. No idea if it's true. The dollar menu was not Stan's (ESRC) idea. It started at Radio Electric Supply to move excess inventory. The list lives on there, but has taken a somewhat different form. There are no $1 tubes. There is however a list, and a fixed price structure dependent on the quantity purchased. Buy 10 of the same type and they are 75 cents each. Buy 100 of the same number and they are 50 cents each. But 500 and they are 35 cents. When their inventory is below the threshold that number is removed. Back when I still lived in Florida I called up and said how many 25DN6's do you have, send them to me! There were about 150 which cost me 50 cents each. The 25DN6 was removed from the list, as were the 13GB5's that I bought. I also got 500 6KT6's which were counterfeit. The tubes are marked "6KT6 GT. BRITTAIN" They were bulk packed in containers which had oriental writing on them. A friend confirmed that the writing was Korean and stated that the box contained 100 electron devices, 6KT6. The tubes work fine in my circuits, but do not match a US made 6KT6.
I have seen two different plate structures in the 12 watt 6HE5. One looks like the oval plate from some 6V6GT's and the other is a box plate much like a 6W6. Many 6JB5's have the same plates. All 6JC5's and tubes with all three numbers should have the larger box plate with a ridge in the center, or a plain box plate with slightly rounded corners. I have a box full of these things somewhere, but I can't find it now.
Thanks again for the info. Good stuff!
OK, so I have two kinds of 6JC5s.
One I think is the 'box plate with ridge down the center':
Is this a 6HE5 "with the oval plate from some 6V6GT's", masquerading as a 6JC5/6JB5?
Just from looking at them, I'd think the one with the larger, squared-off plate would dissipate more heat than the smaller, oval-shaped plate. But I don't know much about this stuff.
OK, so I have two kinds of 6JC5s.
One I think is the 'box plate with ridge down the center':
Is this a 6HE5 "with the oval plate from some 6V6GT's", masquerading as a 6JC5/6JB5?
Just from looking at them, I'd think the one with the larger, squared-off plate would dissipate more heat than the smaller, oval-shaped plate. But I don't know much about this stuff.
Yes, the box plate version in the top picture is the 19 watt version. The etching shows two type umbers but the red Sylvania paint shows all three. The bottom tube might be able to eat 15 watts if it was properly assembled.