I've used the "SRPP" with choke between the top and bottom triodes with positive battery bias on the upper tube- this actually acts pretty much like a mu-follower. (operating point is determined by the choke R and any series resistance you add) See my old muscovite (sigh, sorry about the name) phono stage thread for details. This should
For my latest 45 PSE I am liking a lot 1/2 12AY7 per channel DC- coupled to the 49 triode. The 45's are working at 300V/33mA each and I using a rather uncommon load too, the 3.5K NP Acoustics latest version. I surely need more voltage drive than you need with the 2A3. Moreover, a 45 PSE has more or less the same input capacitance of a single 2A3. So it should work in your case too. The supply voltage for the input + driver is separate and I have settled on just over 460V but if you use the DC link caps is nothing different from a more common 250-300V supply. These are high voltage anyway, same procedures. The plate load of the 49 is 25K (3x75K in parallel) for 7 mA current. Anode voltage is 140V and self bias is 20V. The rest of the supply is for the 12AY7.
I have even gone a bit beyond with 490V for the driver and 320-330V/30 mA for each 45 (with self bias at around 70V). 330V for fairly used 45's (say 80% emission) and 320V for new ones. If the supply voltage is stable and use self bias, no problem at all and get full 5W without going into A2. With 300V plate voltage is close...
Place the resistor at the plate end of the choke winding. The resistor will add to the load and extend LF, it will also decouple the winding capacitance and increase HF. If any of that is useful, depends. But it is useful (and better IME) than specifying a 'winding DCR' for when you are wanting the Vdrop across the choke to equal the bias voltage of the next stage (stacked DC coupled stages).
If we're talking about low-mu triodes (2A3, 300B, 845, trioded 4E27) with mu about 4 or 5-ish, we're talking about a very different driver philosophy than we would about a mu of 8-ish (6V6 family, 6L6 family, 6550 family) or the different drivers for the larger mu of 12 valves (211, 304TL). (Any mu larger than that is a cult item unless first proven otherwise.)
Yes, multi-grid valves have a significant mu; it's between G1 and G2.
There's a difficult dividing line between the low mu pre-War output valves, which are difficult (at least) to drive from a driver supply B+ voltage derived from the output valve's B+, and the mid-late 1950's era mu of 8 valves, where this is not so difficult. The difficult low mu case was originally done, pre modern hi-fi demands, with inductors, even in commercial, cost sensitive, designs, but by the mid-1950's that had fallen away. Power supply costs were so steep in those days that a higher voltage B+ for the drivers was just too big a burden. So, things moved to "modern" mu of 8 output valves. Maybe not as intrinsically linear, but drivable from derived B+ and RC coupling. And, the horsepower races began.
Today, we live in a limbo where often ridiculous ancient designs are worshipped, and we're strangely unwilling to go to the bother of making a higher voltage B+ for our driver stages, just like it's 1937 again. I feel it in myself, and it bothers me. But that's the world we live in.
All good fortune,
Chris
Yes, multi-grid valves have a significant mu; it's between G1 and G2.
There's a difficult dividing line between the low mu pre-War output valves, which are difficult (at least) to drive from a driver supply B+ voltage derived from the output valve's B+, and the mid-late 1950's era mu of 8 valves, where this is not so difficult. The difficult low mu case was originally done, pre modern hi-fi demands, with inductors, even in commercial, cost sensitive, designs, but by the mid-1950's that had fallen away. Power supply costs were so steep in those days that a higher voltage B+ for the drivers was just too big a burden. So, things moved to "modern" mu of 8 output valves. Maybe not as intrinsically linear, but drivable from derived B+ and RC coupling. And, the horsepower races began.
Today, we live in a limbo where often ridiculous ancient designs are worshipped, and we're strangely unwilling to go to the bother of making a higher voltage B+ for our driver stages, just like it's 1937 again. I feel it in myself, and it bothers me. But that's the world we live in.
All good fortune,
Chris
My notes say this....
I have used a Hammond 156C (150H 3700ohms) with a 6SL7 and thought that it worked really well. Just had to play with location and orintataion to keep hum pick-up out of the circuit. This can be a nice variation of a SRPP. PeterP
Changing the "upper" Rk with anode choke, with Rw=1k (or Rw+resistor) you can get A~mu, lower output impedance and lower distortion then in "ordinary" SRPP. But it is desirable that "lower" Rk is bypassed (high rp of the tube). Choke has large AC impedance Z=2Pi*f*L and upper tube "multiplies" this with its mu - large "active" load for the lower tube. Damir
I've been using a cathode choke in the upper tube of a 6SL7 SRPP for a while now and it expanded the overall sound. Better highs, bass and just a bit more gain. Only down side is size and cost both of which are a small price to pay for the overall increase in performance. In my case with a 6SL7 SRPP I run it with about 3.3K (10K paralleled with a 10K pot. Note that with the pot at 5K the parallel combination is 3.3K) with a 220uF bypass cap on the lower tube. I used a Hammond 156C (150H, 8mA, 3700 ohms). I adjust the pot and the B+ to get about 150V at cathode of the upper tube section. Which means that it is running at about .5mA. I don't have a scope so all I can say is that is sounds good to me. As with all SRPP don't forget to bias the filaments. In this case to about 75V.
6SL7 RC loaded - Good gain match. More open with much better mid and highs. SRPP 6SL7 - Best of the bunch. All of the benefits of RC plus more dynamic.
I got the bright idea to replace the resistor in the upper section of the SRPP with a choke. I had a 3.3K resistor in that position and got to thinking that the 156C should almost drop in. Since I had already set up the bottom section with a 220uF cap, 10K resistor and 10K pot, I simply readjusted to get 150V on the cathode of the upper section. With the input to the amp shorted, I noticed an increase in hum. Not too much and not noticeable more than 1' away from the speaker. A very noticeable increase in gain and more detail to the overall sound as well (bass, mid and highs). This really sounds good. With the 6SN7 absolutely no hum, but gain was too low. Highs are definitely there. Still more compressed than the 6SL7 but definitely the best 6SN7 configuration that I've tried. I didn't move the coupling cap. It takes the output from the cathode of the upper section. Should move it to the plate of the lower section. This is really no longer SRPP but choke loaded almost C4S.
^This^
One of the justifications for dual power supply. (CCS loaded SN7) -cap- (46) -IT- (300B).
Have never found the SN7 to be happy with reactive loads, hence the ultra-low distortion solution of CCS. That stage will by nature require a high B+, I have 400V. If one choose to resistor load, same conclusion.
46 driver stage, on the other hand, prefers higher current and lower B+. I chose 300V, which is pushing it on a rating basis. But it effectively drives the 300B.
300B output stage is running around 480V to provide a good operating point (pk-pk swing) at the grid and heavy current for low rp.
Try deriving all these B+ from a single supply and you have legit problems with regulation and wasted power. Instead, provide driver and input from a dedicated supply.
Rather than trying to design around the power supply, allow it to fall where it needs to. If you try to constrain your input and driver to operate in nonideal bias conditions, you will endlessly be fighting strays, leakage, gain structure, and impedance mismatch. Same goes for two stage designs with 1V nominal input.
I've been listening to a pair of Hammond 156C in series for a few days now. 300H, 6.8K DCR, 8mA but seems happy enough with 9.5mA. It's very detailed with good high frequency extension. So yes, it does the job with a tube of Rp around 10K and an unbypassed cathode resistor of 180R. Sound is not as liquid as an amorphous choke but you're getting into money there, while a pair of 156C are cheap for chokes. I'd put the tonality above an active load to my ears, so as of now it would be my solution.
Yes, three stages and the first two direct coupled could easily be good. But I'm just so seriously fed up of optimising DHT filament supplies with choke input or filament bias or whatever. You could use an 01A into a 46 and it would be lovely but I've called a halt to all the weight and size of filament supplies for that kind of front end. I just want a nice sounding medium mu indirectly heated triode and a quiet life.
I spent a year going through output tubes with a higher mu that would simplify the front end. PSE 4P!L were good, and so were EL12 and EL38 in triode. Good tubes, but I still preferred a 2a3 or 300b. So I called a halt to that solution, which I called the "Inverted DHT Amp" because you could use a DHT in the front end instead, and that did indeed sound good with a 10Y driving an EL38 for example. It remains a viable option, just like a 3 stage amp.
Yes, three stages and the first two direct coupled could easily be good. But I'm just so seriously fed up of optimising DHT filament supplies with choke input or filament bias or whatever. You could use an 01A into a 46 and it would be lovely but I've called a halt to all the weight and size of filament supplies for that kind of front end. I just want a nice sounding medium mu indirectly heated triode and a quiet life.
I spent a year going through output tubes with a higher mu that would simplify the front end. PSE 4P!L were good, and so were EL12 and EL38 in triode. Good tubes, but I still preferred a 2a3 or 300b. So I called a halt to that solution, which I called the "Inverted DHT Amp" because you could use a DHT in the front end instead, and that did indeed sound good with a 10Y driving an EL38 for example. It remains a viable option, just like a 3 stage amp.
Andy, can you measure the HF response please?.
6k8 DCR is a lot of turns (capacitance)?.
-
Filament complications with type 46, yes.
Use 71A?. It has the same filament voltage and current as 01A.
-
Check out type 12A in lieu of 01A - same filament power, just as linear, similar mu, runs a bit more current and Rp is ~5k.
6k8 DCR is a lot of turns (capacitance)?.
-
Filament complications with type 46, yes.
Use 71A?. It has the same filament voltage and current as 01A.
-
Check out type 12A in lieu of 01A - same filament power, just as linear, similar mu, runs a bit more current and Rp is ~5k.
Last edited:
I have all these DHTs but I just want a simple 2 stage amp. If I'm at a loose end I may one day make myself a 3 stage amp, but as I get older I want light amps I can easily move around from bench to listening room. I'm done with heavy lifting. I try not to go over 12kg per chassis if I can help it, so I separate the PSU and sometimes the filament supply if it has chokes.
I don't have measuring equipment set up right now but hopefully getting some. But the high frequencies like violins, cymbals, percussion and the top of the piano are all there.
I understand, if the compromise is okay then thats all that matters.
However, consider that you can series heat both 01A/12A and 71A with the one filament supply.. 45V, 250mA.. you need a shunt to trim for correct voltage.. I did exactly that in a previous build and could dig up the schematic if you like?.
However, consider that you can series heat both 01A/12A and 71A with the one filament supply.. 45V, 250mA.. you need a shunt to trim for correct voltage.. I did exactly that in a previous build and could dig up the schematic if you like?.
Last edited:
Honestly, you want too many things in a simple package. A 2-stage all triode thing with 2A3 and 300B has always been a power amplifier in my experience.
Simple + lightweight + good performing you only have 3 solutions for the driver:
1) pentode
2) cascode
3) triode with gyrator (like D3a, C3g...). The gyrator is "simple" if you can use existing boards.
For the cascode you could even try the solid state solution with the Rod Coleman's folded cascode. This would make your amp even lighter.
Ale might have the boards for both folded cascode and gyrator...
If it were for me, I would go with the pentode. Start from these below, for example.
Attachments
Last edited:
The plate of the 6SN7 will be nowhere near the 1200V supply. Spec sheet sez 450V maximum for continuous operation.
Further limit is set by the max plate dissipation. So looking at a practical plate current we soon come to a limit on plate volts.
I'm thinking a 630V cap to the next grid is going to be OK. No KV caps are needed.
A resister in series with a choke forms a frequency selective circuit. Simply adding another problem.
It spite of that Patrick Turner often used it in his driver circuits.
If the CCS's in series are high enough impedance the Q point becomes undefinable.
So now we have drift.
90% of the weight is in the power stage and its share of the power supply. Coming from hi-fi line level, using a power tube duller than a 6V6, you really want two stages before the power stage. This can really be a 12AX7, even if your 'friends' make fun of it. Before "everything is mixed through a hundred TL072s", it was "through a dozen 12AX7s".
Yeah, perhaps safe under normal circumstances. There should not be a problem on power up as the time constant of the plate resistor/coupling cap is perhaps longer than the time it takes for the heaters to allow full conduction. But what about power down using an over ambitious PS having a couple mF? It is known to happen. Or powering up with no tube in the socket?
I like your direct coupled 2 stage input solution and may come round to it at some point, though a single stage would be my preference if it's good enough.
I think you can add a few things to your list. I already have working in my amp a simple, lightweight solution which is a true triode with enough mu and a choke load consisting of 2x 156C. It sounds good and has lifelike tonality to acoustic instruments. For me a solution doesn't have to just "perform" although it needs to do that, it has to produce a level of tonality which can satisfy a listener to subtle acoustic orchestral music and opera. It's the tonality that counts, and for this I prefer a choke to an active load.
I've tried a pentode stage as well and I couldn't get the same tonality as a true triode, though I concede that I might not have optimised it. I don't use pentodes as a rule. A cascode is an idea except for the output impedance, and I haven't tried that or SRPP. SRPP lowers the gain of the tube so that's a problem unless it's a SRPP-choke which is an interesting idea I haven't tried yet. We are adding a tube in both cases. I find choke loading pretty "simple" and always liked it in terms of tonality, and there are a few chokes with inductances over 150H like LL1668 or Hammond 193A for starters. Not big or heavy.
Then there are other solutions like a 1:2 step-up transformer on the input (SUT). The Lundahl one's I've tried compromised the sound so I rejected that one, together with an op-amp or FET on the input for step-up. Both "performed" but didn't satisfy the level of tonality achieved with a choke load. There are several possible solutions, in fact. It's just a question of finding the one that sounds right. When you have a 2a3 or 300b output tube the bar is a high one for driving it with something that sounds as good as these tubes do. So I'll pass on the 12AX7 mentioned and indeed most of the common 9 pin tubes, though there are a small number with much better tone. More interesting to look at some more uncommon tubes with bases like side contact, loctal, B5, B7, rimlok. There are some good but rare ones there if you know where to look for them, which is generally in Europe.
Interesting that Patrick Turner used a choke and resistor in series. He was a good man and much respected. Thanks for that! There are indeed more solutions than meets the casual eye here if you think outside the box a little. There have been several interesting posts in this thread already.
Last edited: