• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

30W sweep tube amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Howdy, Folks:

While not new to tubes, I am relatively new to the DIY hobby of designing and building tube amplifiers.
I have been in the downhole oil/gas industry for several decades, designing high temperature analog and power electronics, with the added requirement of very high reliability (which flies in the face of the high temperature, fun, huh!!).
Wanted to do something challenging and a little different. After reading numerous vintage articles (Crowhurst and the likes), I have decided to go the UL route. After looking at many candidate tubes, the sweep and RF transmitter tubes started to make a lot of sense. Finally have settled on the 6CB5 (after seriously considering the 1625, 12V version of the 807). Had to design a simulation tool to investigate its transfer and plate curves for UL operation, that was fun in itself.
Anyway, nothing concrete yet except for the overall topology of full differential from the input to the output transformer, as much DC coupling as possible and lots of local feedback loops, with no loop covering more than one RC constant.
Solid state CCS don't bother me a bit, they will be used instead of pentodes (no concerns about heater/cathode potentials plus lower minimum voltage) and modern SMPS for heater and likely +B.
I'll be concerned more about proper circuit design, things like operating points, phase shifts, levels, in short, the basics all around.
While quite capable of designing and building my own output transformer, not really looking forward to that. I am attracted to a toroidal design (all things equal, best coupling, lowest core losses and magnetizing currents) and would welcome any comments on providers out there. I know of Plitron and the Dutch equivalent but not of any others.
Anyway, nothing concrete for now, no details but the floor is certainly open to any pertinent suggestions and thoughts and ideas and, most of all, encouragement I can get from the group.
Oh, yes, what are you kind folks' ideas on schematic capture/sketching programs? While I have PCAD, it has no tube libraries and am loath to create any, so if there is something already out there, I would much prefer not to have to invent anything.

Thanks!

Rene
 
I'd worry about runing the 6CB5 in UL mode because of the limted voltage rating of the screen grid. If you ground the control grid and drove the tube via the screen, it would work just fine, and you could get a lot more than 30W from a pair. I have a nifty SE amp using the 6CD6GA in screen driven mode with Hammond 125ESE output iron. It is a surprisingly listenable amp for something so cheap.
 
Thanks, that actually sounds cool. One of the things under consideration is the low G2 max voltage, as you mention. There are a couple of ways around that, among them pushing the screens a bit while running the plate on the low side, say both at 250V, a lot of power can still be had. Another is the use of separate screen and plate windings, more money on the transformer but solves the issue. Another approach I've been considering is using a conventional single tapped winding, raising the plate supply to 350 or so, then using zeners to drop the mean voltage for the screens back down to the 175-200V range. have not seen that done but would not be surprised to see that it has. Planning on staying AB1 so driver power is not an issue while screen drive for sure requires power.
I'd love to see a sketch or at least a more detailed textual description of your setup, again it sounds cool. If one properly understands G2 gm it would appear to me that there is no reason why one could not make a reasonably linear amp out of it. You say SE and I am more on the PP camp but the principle you used should be adaptable.
Any suggestions on a sketch/schematic capture program?
Thanks!
 
Thanks, Tony. I am well aware of the screen G2 voltage limitations for sweep tubes, very similar to the issue with RF tubes. This is why I was thinking of the 3 options mentioned in the earlier post: one is to push them but not too much, run the plates at around 250 and still get a lot of power out of these guys: another is your suggestion, separate windings; another still might be to use zeners to get the G2 voltages down to withing spec. Lossy, yes, but should be effective.
I hope there will be a whole more posts from which we can all draw some good ideas.
And Sy, thanks for the tip, already looked it up, kind of neat. Like you, if I go that route, cathode followers will likely be replaced with MOSFETs. Can't tell what is the output tube type used?

Rene
 
The Berning amp gets 30W with a pair of 6JN6s - one of what seems to be almost an infinite series of duodecar sweep tubes with 17-18W anode dissipation. There's nothing particularly magic about the 6JN6 - other tubes of similar rating would do just as well. They'd also be cheaper, as the "audiophile recognition factor" is not in effect. One benefit of using the smaller tubes is that there are lots of them with no anode cap. If you're going to take the trouble to mess with anode caps and plate wires flopping in the breeze, you might as well get some extra watts for your trouble. The 6CB5 could get you 60W with no trouble (Tubelab George would say lots more).
 
Thanks, Wrenchone. Done the math a long time ago, well over 60W is indeed quite possible, just not sure I want to spend the money for the OPT but probably should go for 100W iron just for the margin, go for 50-60W out of the amp again just for the margin and personal satisfaction, even though in reality, I'll never use that much power. So, the engineer in me was being pragmatic and says 30W is plenty but it won't take much to tempt me to the dark side of a whole lot of power!
Plate caps and the T12 envelope (I'm actually after the 6CB5A) are just plain attractive to me and have the visuals I'd want. I don't mind dealing with their issues.
Any tips on a schematic drawing package?
Thanks again!
 
Thanks, Sy. The docs I found do say idle current is indeed 3.3mA! I looked up the plate current family of curves for Eg1 at 0 and various G2 voltages (from which we derive the gm for G2). the G2 voltage at idle will be around 5 volts or so! Only thing is that all this smells of class B operation and not AB, being that close to cutoff when idling. If the output transformer is not super tightly wound for super low leakage inductance, there is the high potential for notch distortion. But, since at least 2 people have endorsed the Bearning, I must either be missing something, or else the OPT is indeed super-duper.
This screen drive approach has suddenly intrigued me enough to make me want to dig deeper. Can't be sure that's good or bad! :)
Any more thoughts?
 
Yes, it's very close to Class B, but the tubes are surprisingly linear at that idle current. They're essentially zero bias triodes. One thing that helps a lot is the local feedback from the output tube plates.

I've built a couple of screen drive amps on that general plan. My last one used MOSFET screen drivers (IRF820, but this was in the mid 1980s when better choices weren't around), 6LF6 output tubes, 1200R plate to plate impedance, and something like 800V B+. No problem getting nearly 200W out of it as long as you didn't run that power for too long!
 
I have seen a pair of 6CB5's crank out 125 WPC without breaking a sweat. There is a rather simple solution if you don't want to design it all yourself. Pete Millett has a driver board that works well with sweep tubes.

http://www.diyaudio.com/forums/tubes-valves/203110-mono-push-pull-driver-pcb.html

I have tried it with a wide variety of sweep tubes including the 6CB5. Power output is dependent on the supply voltage and load impedance, but 125 watts takes 3300 ohms and about 650 volts, 150 volts on the screen grids.

Since I am also a stubborn engineer, and Petes boards won't physically fit in my latest amp, I am looking at a pair of driver boards I designed several years ago. I just got the 3300 ohm 100 watt Edcors, and haven't decided on the output tube yet. Physical constraints are the big factor since this will be a small rack mounted amp making 100 to 150 WPC.

His older board is designed for 12 pin compactrons, but I have wired all sorts of tubes into it for up to 250 WPC. There is a long thread that discusses a lot of the stumbling blocks that everyone climbs over when designing a sweep tube amp. Educational reading if you have time to waste.

http://www.diyaudio.com/forums/tube...-new-p-p-power-amp-design.html?highlight=6gv5
 
Again, thanks for the postings! One thing, though: we need to keep all this kind of quiet or the price of sweep tubes will go thru the roof! :)
Don't mind designing it all myself, like you kind of on the stubborn engineer side myself. This project will be as much for enjoyment as learning as actually using the critter. I am an experienced magnetics engineer and could even make the OPT but, likely, nah! too much work.
I really need to go to the basics of screen drive operation, run some simulations, etc, and decide whether G2 or G1 drive for me. One thing, using G2 drive makes lemonade out of the low G2 maximum voltage lemon. A comment made elsewhere is how well sweep tubes work in screen drive while "legitimate" audio tubes do not. In looking at the plate curves and the tabular data, it seems that the sweep tubes have a much higher G2 gm than others, that may well be the key.
I'll follow the postings per your suggestion, and hopefully learn a lot.
 
i am eager to take up SY re screen grid drive, i have a decade of 6S4 tiode tubes which i will use as LTP at 400volts plate for the big big swing required and mosfets as source follower...input tube can be a D3a or a Russian 6J9....

design and builid for the OPT's are simple and easy...

incidentally, the 6LU8's which used to sell for a $ now sells for 4 at Rogalski'es......oh well....
 
Any tips on a schematic drawing package?
I'm using LtSpice form linear tech's website, it is free and useful for drawing and simulating schematics. You can find a lot of tube's models here on this forum or using google.
IMHO I suggest to use an output transformer with separate windings for the second grids, that way you can apply a local negative feedback and also using a bigger anode voltage which leads to a bigger voltage swing more power will be delivered to the load, having a decent thd and tid factor thanks to the UL windings.
 
Last versions of my sweep tube line.
 

Attachments

  • PC240342.jpg
    PC240342.jpg
    689.1 KB · Views: 775
  • PC270356.jpg
    PC270356.jpg
    813.3 KB · Views: 751
  • side_view.jpg
    side_view.jpg
    524.5 KB · Views: 736
  • PC270358.jpg
    PC270358.jpg
    768.2 KB · Views: 699
  • Compactron PCB J2.jpg
    Compactron PCB J2.jpg
    632.4 KB · Views: 716
I really need to go to the basics of screen drive operation, run some simulations, etc, and decide whether G2 or G1 drive for me.

Most of the vacuum tube models that I have played with in LT Spice did not model screen drive even close to reality. Horizontal sweep tubes were intended to operate as a switch, with G1 used as the switch control, and G2 setting the saturated maximum current. This is why you can find screen curves for G1 = 0 volts for some tubes. Linear operation for G1 or G2 drive was not the design criteria.

About 5 years ago AES decided to sell off a large quantity os slow moving tubes. A feeding frenzy ensued, and several of us had large lots of tubes to play with. I was tinkering with screen drive then and I tested a lot of tubes and discussed the results in the thread. They sold off the 6CB5's for $2 each!

http://www.diyaudio.com/forums/tubes-valves/128533-tube-sale-aes.html?highlight=tube+sale+AES

The 6CB5 screen drive experiments start in post #94. Somewhere in that thread I discovered the fatal flaw in high powered screen driven amps. My amp kept blowing up rather violently without warning as I pushed the power up. I kept exploring the phenomenon, and figured out what was going on....

At the moment of peak conduction in a screen driven tube the screen grid is at its most positive potential, at this same instant the plate is near zero volts. This causes a rather large current flow from the cathode to the screen grid, instantly overheating it.

At the next half cycle the screen grid wires are still hot....hot enough to emit electrons, but now the screen grid is near zero, or negative while the nearby plate is hugely positive. In my case it was over 1KV, so the glowing screen grid becomes a virtual cathode and current flows from the screen grid to the plate. The current causes the tiny screen grid wires to melt...quickly, and an instant runaway occurs, ending in a tube arc and exploded mosfets that fed the screen.

Most music, even played at the clipping level, doesn't spend enough time in this region to provoke a tube arc, but testing at full power with a sine wave will!

Granted, my experiments were sucking big power out of small tubes, but they found a previously undocumented failure mode. A serious screen driven amp should incorporate some king of VA limiting on the screen grid.

Screen driven amps tend to have quite high output impedance, requiring some sort of feedback to lower it. All of the experiments I posted in that thread used no feedback at all, leading to rather high distortion numbers. More recent experiments have used local feedback from the output tube plate to the driver tube plate. I have also re-examined driving both G1 and G2 to reduce the peak voltages on G2, but those experiments are not quite ready for prime time.

I have recently started on a new amp that will see HiFi and guitar amp duty. It could see continued operation in clipping, so I doubt pure screen drive will be used, but I have just started testing stuff. I will start a new thread to chronicle the build as soon as I finish clearing off the workbench.
 
Hi George,

With the 6CB5, what kind of output could you expect from a 680V supply? Would you stick with your 150V on the screens? Would a quad in PPP work well into 1.5K A-A load?

I'm still looking for a good compliment using something "different" for my 200W Sowter outputs. It would be nice to only have four vs 6-8 tubes per channel while still getting ~150W.

How do these sweep tubes do from a distortion perspective vs. audio designated tubes?

Thanks,

Blair
 
Howdy, All:

Deicide67, I am near Georgetown, close to Jonah.

So, good folks, on the amplifier: thanks tubelab for the comment on how hor sweep tubes are used, i am aware. The curves for G1=0 vs variable G2 levels are quite commonly found for many pentode/beam tubes, even small signal ones. It is a way one can determine G2 gm when not explicitly given. have not done it yet but it is relatively easy to plot plate current for a continuously changing G2 voltage with G1-K voltage fixed at 0.

Thanks a lot for your analysis of the failure mechanism when pushing these critters so hard, makes perfect sense. Very happy to hear that at more sane levels of operation you do not expect that mechanism to be of concern. That is consistent with your analysis.

Back to a question I asked earlier in the thread for UL operation, it would appear to me that a zener system could be used to drop the G2 nominal voltage to safe levels with a connection to a traditional UL tap. Have not seen this done, can't think of a reason why not (other than to mind the dissipation in the zener system which could get significant) and have to assume that someone's done it as it is so obvious. Do any of y'all know of this having been done? It would be a lower cost solution to a transformer with separate plate and G2 windings, plus the coupling is potentially tighter.

I will continue to explore both my original concept of UL with differential all the way, and now the idea of G2 drive. I'm still thinking that something in the 30-50W range is going to be enough for my needs and will provide lots of headroom for peaks so truly high power operation is not needed, that should relieve some of the design challenges.

Keep the posts coming! Thanks again!

Rene
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.