I've done some simulations for crazy drive and they all show gm doubling with an abrupt change transitioning from class A to class B, making it hard to find a useful operating point. I think in reality (smoking-amp has mentioned this) a low bias current of a few mA works well to avoid this, but the models used in simulations aren't accurate enough at low screen voltages.
I have built a lot of stuff with sweep tubes. I came up with what I called Dual Drive at about the same time Smoking Amp came up with Crazy Drive. Pure screen drive can blast the screen grid into meltdown if the amp is continuously driven into or near clipping. Since I make HiFi and guitar amps, I searched for an alternative, and tried Dual Drive. It didn't do mush to save the screen grid under severe conditions, and although it makes for some pretty curves on the tracer, I still have not made a good low distortion amp with it yet.
I have wired a pair of 36LW6's up to a UD board and cranked them up in plain old G1 drive......Somewhere north of 160 watts from two tubes I rediscovered the fact that resistors DO have voltage ratings, and will flash over when that rating is exceeded by a factor of 3.
The 6LW6 data sheet is the same as the M2057 sheet except for the tube base.
https://frank.pocnet.net/sheets/084/m/M2057.pdf
Simulations are not likely to work well for Crazy drive, since the operation depends on accurate modeling of grid 1 current when positive and also optimised resistors for the grids. It does look likely that idle current will have to be unusually low for minimal cross-over distortion.
I'm in the process of setting up a curve tracer that can do push-pull curves now, but it will probably be around Spring before I have it all working. I'll check Crazy drive then.
Until then, I'd say building the Crazy drive P-P case and measuring the real thing is the only practical way to find optimum biasing.
https://frank.pocnet.net/sheets/084/m/M2057.pdf
Simulations are not likely to work well for Crazy drive, since the operation depends on accurate modeling of grid 1 current when positive and also optimised resistors for the grids. It does look likely that idle current will have to be unusually low for minimal cross-over distortion.
I'm in the process of setting up a curve tracer that can do push-pull curves now, but it will probably be around Spring before I have it all working. I'll check Crazy drive then.
Until then, I'd say building the Crazy drive P-P case and measuring the real thing is the only practical way to find optimum biasing.
Last edited:
What exactly do you mean by this? Setting it up in pentode or triode is very simple and will yield great results. No specialised equipment at all.
I don't see any reason 6LW6 wouldn't work just fine for audio in pentode mode. ( it apparently suffers eventual runaway in triode mode)
There are certainly more -economical- Sweep tubes out there. I recently tried two 21HB5A tubes in parallel and was surprised to see that the pair gave near identical curves (both in pentode and a very nice triode mode too) to 26LX6.
Needless to say, I quit collecting 26LX6 tubes after that, since 21HB5A were on the $1 list once, and I have a big box of them.
Two 38HE7 tubes in parallel will give you an older style "doubled" 42KN6. (these 38HE7 are still on the $1 list) (on most 38HE7 you can run just the pentode with 21V 0.45 Amp between pins 10 and 12 for the heater)
Last edited:
One of these days I'll build a big parallel amp for absolute nonsense output (I have no need for output >100 watts) and I'm thinking the dollar menu is a great way to do it. Conservative idle dissipation, three or four pairs per channel, and a big fat SMPS for heater supply. You could easily get a few sets of replacement tubes and have them last a lifetime. I'm just allergic to odd sockets so haven't tried very many compactrons yet.
It'll be justified if I consider it a small space heater for winter use
It'll be justified if I consider it a small space heater for winter use
12 Pin Compactron sockets are readily available (Chinese, ceramic), and fit well. Same chassis punch (Greenlee) as for Octal sockets.
The 9 pin Novar Sockets (sometimes called "9 Pin Compactrons") are harder to find ones that fit well. (mostly Magnoval stuff out there, which don't fit Novar)
Not much on the Dollar Menu these days for power output except 25BQ6 and 38HE7 and a few 10 Watters like 6DB5 and some composite Compactrons. 6MF8, 6KY8, 6T10, 8AL9 etc
I found some nice 6197 pentodes on Ebay earlier this year for $0.33 ea. in quantity. Very close to a 12BY7. Good for drivers at least. Or "unreasonable" paralleled power designs. I'm working up some fiberglass boards with 16x 9 pin sockets for running 6197, 12HL7, 6HB6, 29GK6 tubes in parallel.
The 9 pin Novar Sockets (sometimes called "9 Pin Compactrons") are harder to find ones that fit well. (mostly Magnoval stuff out there, which don't fit Novar)
Not much on the Dollar Menu these days for power output except 25BQ6 and 38HE7 and a few 10 Watters like 6DB5 and some composite Compactrons. 6MF8, 6KY8, 6T10, 8AL9 etc
I found some nice 6197 pentodes on Ebay earlier this year for $0.33 ea. in quantity. Very close to a 12BY7. Good for drivers at least. Or "unreasonable" paralleled power designs. I'm working up some fiberglass boards with 16x 9 pin sockets for running 6197, 12HL7, 6HB6, 29GK6 tubes in parallel.
Last edited:
My exploding resistors had to do with my stupidity and not the tube choice. The 36LW6 does work quite well and probably the tube of choice when I do get around to building the big one. I believe that 200 to 250 watts per pair is possible. There are 2 US manufacturers of LW6's, GE and Sylvania, several different versions, and two different bulb diameters. Make sure that all tubes used in a given amp are the same.
There is another fat undiscovered sweep tube that I will not name yet. There is only one kind, they are all the same, and it's still fairly cheap and available. I haven't fully explored its potential yet either.
I'm looking at 1 KW, 500 WPC. I have seen my audio power meter hit 525 watts at 5% THD with two pair of 35LR6.....the 6LW6 is BIGGER!
I have attempted two different "massively parallel" output stages.
Years ago someone made the audiophool press with an amp that ran a bunch of 12Asomething7's in parallel for the output device. I have several hundred 6AQ5's, all well used and of different brands, so I got this Dumm Blonde idea to wire a bunch in parallel with separate cathode bias resistors for each tube. Despite running a large pile of different tubes through it, I could never keep the thing stable.....it wanted to oscillate if you looked at it wrong. One tube or another would go into red plate runaway, so I would pull that one and try another, only to see a previously cool running tube start glowing.....
I eventually lost my patience, cranked the power supply to max and smiled as it melted down slowly.....to bad cheap video didn't exist then it would have done well on YouTube. 6AQ5's have the plate so close to the glass that most tubes will crater over a red spot, then suck air. Sometimes you find one that gets hot evenly and the glass gets sucked right up to the plate and all warped out of shape.....you have to turn down the power supply as it begins to melt to make a melted candle out of one.
More recently I bought 100 13GB5's when they were on the dollar menu.....50 cents each in 100 QTY,......so I wired up a perf board with 12 of them. Each had its own fixed bias pot. Again current hogging and oscillation was a problem, but all was not lost. In both cases these experiments ran half the plates in parallel, each bank tied to one winding on a BIG 2 X 120 VAC to 24 VAC power toroid.
I got the idea to run a bunch of common 80 VA guitar amp OPT's, one per pair of tubes. Then I played around with different combinations of the secondaries until I could get over 600 watts out of the monster on 600 volts. No oscillation was seen and the only redness I saw was when I tried to push past 100 watts per pair of tubes. Too bad someone bought up ALL of the 13GB5's in captivity.
https://www.tiffe.de/roehren/ga399ma.pdf
An easier to build "shunt Schade" Sweep amplifier (could use _LW6 tubes):
https://www.diyaudio.com/forums/tubes-valves/109534-please-help-build-amp-father-3.html#post1329903
Attachments
They will also drop right into Pete Millet's Engineer Amplifier as well, just set the bias-
Engineer's amp 50W monoblock
PCBs that can use external sockets for the output tubes are available, so a build would be easy.
Engineer's amp 50W monoblock
PCBs that can use external sockets for the output tubes are available, so a build would be easy.
For a different OT impedance, it just means using a different B+ for the output stage. 5K is a little high for a _LW6 however, more like 1.9K to 2.4K or so.
A note on those "alternative local Fdbks" drawn in on the David Wolze "Snubber" schematic above, in post # 1154 , they won't work with a Concertina splitter, (as just found out recently in another thread!). Have to change the input stage to an LTP splitter (preferably with pentodes) to get equal output impedances to drive the low Z N Fdbks.
A note on those "alternative local Fdbks" drawn in on the David Wolze "Snubber" schematic above, in post # 1154 , they won't work with a Concertina splitter, (as just found out recently in another thread!). Have to change the input stage to an LTP splitter (preferably with pentodes) to get equal output impedances to drive the low Z N Fdbks.
Last edited:
The type of sweep tube does not matter much as long as it's big enough for the job. The impedance of the OPT and the B+ voltage are the major factors in power output. Assuming a mythical tube that can transfer all of the B+ to the OPT without loss, you get the following:
a 5K opt will make about 50 watts on 500 volts
it will make about 72 watts on 600 volts.
and about 98 watts on 700 volts of B+.
The peak current demands are low enough that nearly any decent sized sweep tube will work and make nearly these numbers.
Do your OPT's have several secondary taps so that you can get a 2500 ohm load by running the 8 ohm speakers on the 16 ohm tap? That's how I'm getting 3300 ohms from my 6600 ohm OPT's.
I don't normally spend much time with other peoples amps, but that one is an easy way to make big power. I have built 3 of them. The 125 WPC flavor mentioned in the original thread used 3300 ohm (miswired 6600 ohm) OPT's on 650 volts for 120 to 125 WPC.
I generally try to steer away from really high B+ voltages for safety reasons. Although high B+ will definitely give one better plate efficiency. (power output per power input or lower plate dissipation) That's one of the reasons transmitting tubes use HV, besides lower heater power too.
Looking at the datasheet for the 6L6GC as a starting point, one finds an optimum load Z for lowest distortion. (they only show the SE case as a graph)
The minimum distortion curve results from gm distortion decreasing with higher load Z, just like for triodes, but screen current distortion (causing curvature in the "flat" part of the plate curves) increasing with higher load Z.
At some point in between, a minimum distortion region occurs. One usually shoots for that OT primary Z or the datasheet will recommend that Zpri. But modest deviations from that only kick up lowish harmonics that are easy to fix with some N Fdbk. Large deviations do begin to pick up higher harmonics.
For P-P, the 6L6GC datasheet uses a 5600 Ohm OT at 450 V B+.
I use a heuristic rule comparing max DC current ratings (or could use peak current ratings) between tubes to arrive at a similar optimum load Z for other beam pentodes/tetrodes that don't have a suggested value on the datasheet, like most Sweep tubes. So a 6LW6 has a 400 mA DC rating and the 6L6GC has a 110 mA rating. So the 6LW6 is like 4 6L6GC tubes in parallel (as far as current handling). So one would expect about 5600 x 110/400 = 1540 Ohm Zpri as an optimum. If the tube Wattage is higher than the 6L6GC, then Zpri gets further pushed up, linearly higher, with use of a higher B+.
TV Sweep tubes however tend to have lower screen current to plate current ratios than the 6L6GC, so one would expect the optimum Zpri to be higher up than the simple calculation above. Many are at half the current ratio, and some even less (6LR6 and 6LG6 are nearer to 1/4 the ratio, these datasheets do comment on the low screen current).
Considering that changing the load Z upwards alters BOTH gm dist downward and screen distortion upward, I go for a "median" adjustment using the SQRT function of the screen current ratio versus the 6L6GC ratio (the inverse actually for Zpri adjustment). So the 6LW6 "optimum" would increase to around 2200 Ohms. Again, there is reasonable leeway around that, since dist. variation is mainly low harmonic near the optimum. Higher B+ for more power is another factor.
One could also make a good case that the adjustment should be linear with screen current ratio (instead of SQRT), in which case the 6LW6 optimum would be up around 3100 Ohms. Seeing as Pete M. found 4K for the 6HJ5 optimum (and 8K for 6JN6), this could well be the case.
So with a 5K OT I would suggest using a 35LR6 or 21LG6 rather than the 6LW6, since these tubes have lower screen current dist. and they have enough Pdiss rating (30 W and 28 W) to handle a 100 W OT.
Actual measurements will of course trump "heuristic" estimates. You might want to try several tube types and see.
Looking at the datasheet for the 6L6GC as a starting point, one finds an optimum load Z for lowest distortion. (they only show the SE case as a graph)
The minimum distortion curve results from gm distortion decreasing with higher load Z, just like for triodes, but screen current distortion (causing curvature in the "flat" part of the plate curves) increasing with higher load Z.
At some point in between, a minimum distortion region occurs. One usually shoots for that OT primary Z or the datasheet will recommend that Zpri. But modest deviations from that only kick up lowish harmonics that are easy to fix with some N Fdbk. Large deviations do begin to pick up higher harmonics.
For P-P, the 6L6GC datasheet uses a 5600 Ohm OT at 450 V B+.
I use a heuristic rule comparing max DC current ratings (or could use peak current ratings) between tubes to arrive at a similar optimum load Z for other beam pentodes/tetrodes that don't have a suggested value on the datasheet, like most Sweep tubes. So a 6LW6 has a 400 mA DC rating and the 6L6GC has a 110 mA rating. So the 6LW6 is like 4 6L6GC tubes in parallel (as far as current handling). So one would expect about 5600 x 110/400 = 1540 Ohm Zpri as an optimum. If the tube Wattage is higher than the 6L6GC, then Zpri gets further pushed up, linearly higher, with use of a higher B+.
TV Sweep tubes however tend to have lower screen current to plate current ratios than the 6L6GC, so one would expect the optimum Zpri to be higher up than the simple calculation above. Many are at half the current ratio, and some even less (6LR6 and 6LG6 are nearer to 1/4 the ratio, these datasheets do comment on the low screen current).
Considering that changing the load Z upwards alters BOTH gm dist downward and screen distortion upward, I go for a "median" adjustment using the SQRT function of the screen current ratio versus the 6L6GC ratio (the inverse actually for Zpri adjustment). So the 6LW6 "optimum" would increase to around 2200 Ohms. Again, there is reasonable leeway around that, since dist. variation is mainly low harmonic near the optimum. Higher B+ for more power is another factor.
One could also make a good case that the adjustment should be linear with screen current ratio (instead of SQRT), in which case the 6LW6 optimum would be up around 3100 Ohms. Seeing as Pete M. found 4K for the 6HJ5 optimum (and 8K for 6JN6), this could well be the case.
So with a 5K OT I would suggest using a 35LR6 or 21LG6 rather than the 6LW6, since these tubes have lower screen current dist. and they have enough Pdiss rating (30 W and 28 W) to handle a 100 W OT.
Actual measurements will of course trump "heuristic" estimates. You might want to try several tube types and see.
Last edited:
Remember that the peak voltage seen in the OPT is twice the B+ voltage when the amp is run below clipping into a resistive load. I have seen voltages over 4 X B+ in a guitar amp at full crank into a speaker near resonance.
High B+ voltage lakes life tougher for the OPT if you play it loud.
Pete told me that he favored a "conservative" design when he first published the 6JN6 version, because most builders were beginners.
I told him that I would be quite unhappy if I didn't get at least 50 WPC......that took about 5 minutes after the amp first made sound. He specced an 8K ohm OPT, but I have lots of 6600 ohm OPT's, so that's what I used. I powered up the board on a bench supply set for somewhere around Pete's recommended voltage. My old Fluke 407D goes to 550 volts, but the caps I put in the board were only rated for 500 volts, so I split the difference and turned it up to 525 volts. Somewhere around 60 WPC flowed forth. The most I saw was about 250 WPC with 35LR6's on 650 volts into a 2500 ohm load.
The best arrangement for high power with that board involved two power transformers. One ran the board at a safe voltage below 400 volts, and the other fed only the OPT at voltages around 650 volts.