Is it a stupid idea to build a single ended cathode bias 7591A amp?

[spacelabstudio]

I actually have a couple of 6GM5 (same as 7591A with 9 pin base) tubes laying around and i thought it might be cool to try to use one of them in the single ended amp I'm making.  Now I'm wondering, though, if that's a good idea.  For starters, I don't know of any others out there, and there's probably a reason for that.  Looking at the datasheet, the "Characteristics and Typical Operation" section doesn't even mention a cathode resistor for the Class A1 amp configuration.  When I draw load lines (and granted I'm a complete noob at this) I find I can't really draw any load lines that stay under the max dissipation curve and give me any grid lines to the right, which makes me wonder if class A just isn't the thing for these tubes.  Is this nagging doubt I'm feeling just a little common sense starting to creep it's way in?  Or should I just go for it?

Chris

[RicharD]

I don't see any reason why not.  The cut sheet I'm looking at   calls for a 3k load.  At 300V, it idles at 60mA on the plate and 8mA on the screen.  For cathode biased, ignoring the screen current puts you pretty close to 70%.  It calls for a -10V bias so 10V/60mA = 167 ohms.  Try a 180 ohm 2W cathode resistor.  This should put you below 70%.  Just watch your cathode current when you first light it up.  I'm all about SE amps these days.  I'm running a 6L6GC single ended and it's perfectly just loud enough for small clubs.

[HotBluePlates]

A 7591 is basically the available dissipation of a metal 6L6 (19w), in a smaller 6V6-like package, with a hella-lot more Gm. Higher Gm means it takes less voltage to bias the tube and to drive it to full output.

Seems like a great candidate for a SE amp.

Richard already did the math for you for the 300v class A condition.

... When I draw load lines (and granted I'm a complete noob at this) I find I can't really draw any load lines that stay under the max dissipation curve and give me any grid lines to the right ...

Just because they didn't draw line for grid voltage more than -14v doesn't mean they don't exist. Aslo, it's perfectly fine for a load line to stray slightly above the max dissipation line, if it's only for a small portion of the time. "Small portion" depends on the situation and your judgement. "Small portion" is longer when you're running class AB and the tube is shut off completely for part of the cycle.

If you look hard at the 300v 3k 60mA condition Richard cites, the idle point (as plotted on the G2=350v curves) sits almost exactly on the max dissipation line (any lower and you'd just be getting less power output). If you were plotting a 3k load line passing through (300v, 60mA) you might want to figure the point at 200v plate as a second point on the line. That's 300v-200v = 100v / 3000 ohms = 33mA higher than your idle point. 93mA is still a bit under the max dissipation curve; so this loadline is almost tangent to the dissipation curve.

For SE, the loadline will run through the idle point. In reality, your B+ will be 310v or somewhat more, to allow for drop across the cathode resistor and also for sag in the power supply at full-roar. It is normal for the idle B+ to be a little bit high. Your tube won't melt at 18.2w.

[DummyLoad]

want my advice, don't take my advice... here goes anyway.
take the AB1 P-P cathode bias condition - listed as AB1 but seems much closer to A1.

450Va - 400Vg2 10K load - 86mA at idle 94mA full power - values for 2 tubes. 450V*43mA = 19.35W/tube at idle - that's over the 19W spec. full power, we get sag and at full power current rises 47mA per tube even under load Pdiss of tube won't rise much more, it actually falls since the power is diverted to the load.

fender champ PT should yield ~390-400V B+ w/ 5Y3GT, ~420V-430V w/ GZ34 or ~440V w/ SSR with ~40mA idle. cathode biased P-P plan says 10K for 2 tubes - use 5K load, P-P plans say use 170R for cathode R, double that and use 340R for Rk @ 2W or >. closest available values are 350R and 330R - i'd use 350R, maybe even 390R to start with.

i'd graft SE output stage to a deluxe like pre-amp and turn the    PI into a CF. perhaps nirvana maybe close at hand. KISS.

peace, but hoping to get nuked by PRR. i love it when that happens.

--DL

[Shrapnel]

Okay, I *had* to take a look.   Both the 6GM5 and the 7868 have the 9-pin base....  Neonoval for the 6GM5, and Novar on the 7868. Other than that they have a very similar, but not the same, pinout.
The 7591A uses an octal with an obvious difference in pinout.  With the same specs as well (Well the 7591A mentions the triode  Amplification factor) for Class A.

We also know that the 7868 is considered a rebottled version of the 7591A by most people. I guess we can expand that to include the 6GM5 then.

Go for it. After all, it does provide SE data already. All other data looks exact to the 7591A and 7868. Could prove to be an interesting little SE amp.

[spacelabstudio]

A 7591 is basically the available dissipation of a metal 6L6 (19w), in a smaller 6V6-like package, with a hella-lot more Gm. Higher Gm means it takes less voltage to bias the tube and to drive it to full output.

Seems like a great candidate for a SE amp.

Richard already did the math for you for the 300v class A condition.

... When I draw load lines (and granted I'm a complete noob at this) I find I can't really draw any load lines that stay under the max dissipation curve and give me any grid lines to the right ...

Just because they didn't draw line for grid voltage more than -14v doesn't mean they don't exist. Aslo, it's perfectly fine for a load line to stray slightly above the max dissipation line, if it's only for a small portion of the time. "Small portion" depends on the situation and your judgement. "Small portion" is longer when you're running class AB and the tube is shut off completely for part of the cycle.

If you look hard at the 300v 3k 60mA condition Richard cites, the idle point (as plotted on the G2=350v curves) sits almost exactly on the max dissipation line (any lower and you'd just be getting less power output). If you were plotting a 3k load line passing through (300v, 60mA) you might want to figure the point at 200v plate as a second point on the line. That's 300v-200v = 100v / 3000 ohms = 33mA higher than your idle point. 93mA is still a bit under the max dissipation curve; so this loadline is almost tangent to the dissipation curve.

For SE, the loadline will run through the idle point. In reality, your B+ will be 310v or somewhat more, to allow for drop across the cathode resistor and also for sag in the power supply at full-roar. It is normal for the idle B+ to be a little bit high. Your tube won't melt at 18.2w.

It's good to see y'all's load lines look a lot like mine.  I'll go ahead and give it a shot.  I can always rewire for an EL84 if it doesn't sound good I guess.  Nearest output transformer I could find to 3.3K at 10watts is a 4K, so I was using that for my load line and came up with about 200R for the cathode resistor.  I'll buy a whole range, probably, from 160-240 and experiment.

The power transformer I'm looking at is a Hammond 269AX, which puts out 250VAC.  With a bridge rectifier, I have a theoretical max B+ of about 353V (or 367V taking into account they quote 115VAC for the primary and I have 120VAC coming out of my wall).  Between transformer losses and a sag resistor I'm going to try to get that down closer to 300V.  Merlin says 1.2xVrms is more realistic for output taking into account transformer losses, which only comes out to 313V after adjusting for wall voltage, so I probably won't need too big a sag resistor.  Or  I might can omit it altogether.

Chris

[spacelabstudio]

And actually, it looks like Edcor makes some 3.5K output transformers that would work.  The candidates so far are:

Mercury Magnetics Axiom Series, 4K:8 10W,
Allen TO11C, 4K:8
Edcor XSE15-8-3.5K or GXSE15-8-3.5K, both 3.5K:8 15W

No idea how to evaluate without hearing.  The MM transformer is far and away more expensive than anything else.  Worth it?  My DRRI has an MM in it and sounds pretty good, but better than...?

[jjasilli]

To summarize:  Power Amp design criteria -- 1.  which PT:  300 plate volts or 450?  2.  Which OT? 10W and under may saturate if that's what you want.  15W may be more robust. 

[spacelabstudio]

To summarize:  Power Amp design criteria -- 1.  which PT:  300 plate volts or 450?  2.  Which OT? 10W and under may saturate if that's what you want.  15W may be more robust. 

I think 300V for the plate.  That's what the datasheet says for class A and looking at the curves it's hard to find a usable load line for class A that starts at 450V.

Chris

[PRR]

> Is it a stupid idea to build a single ended cathode bias 7591A amp?

1) Yes. 7195 and Kin are warped: high peak current, high hain at high current, low Pdiss for cheapness. Good for push-pull, bad for SE. Note the huge THD in the suggested SE condition; this THD may not drop much at lower power,

A pair of 6AQ5 will make as much power, lower THD, less B+ filtering, and MUCH smaller output transformer. That's the un-stupid way to go.

2) It can work dandy as a DIY gee-tar amp. THD may be flavorful. Cost/weight of OT may fade in DIY euphoria.


> it's hard to find a usable load line for class A that starts at 450V.

Sure, that's possible. Although there's no reason to go high-voltage.

Start with Pdiss. Round to 20W for hasty math (but re-check with 19W at the end). Knowing Pdiss and B+ we know max idle current.

600V at 20W = 30mA
500V at 20W = 40mA
400V at 20W = 50mA
300V at 20W = 67mA
200V at 20W = 100mA

The SE peak current will be double the idle.

600V peak = 60mA
500V peak = 80mA
400V peak = 100mA
300V peak = 133mA
200V peak = 200mA

Maximum possible plate current is limited by G2 voltage. Page 3 of Sylvania 7195 data shows max current for various G2 voltages. At the knee we see about:

400V 250mA
300V 180mA
200V 100mA

For simple power supply we would very-much like to have G2 voltage near or a little low of plate voltage.

200V on plate can only work if we have a ~350V supply for G2. That IS stupid.

At 400V plate we need 100mA peak and 350V-400V on G2 allows more than twice that. This can work. We'll need a lot of G1 swing, though probably no problem. There is the risk of overload damage when available current is twice what we need. Or we could provide a 200V-250V supply for G2. That will work well, but holding a fairly steady 225V from 400V with varying G2 demand is awkward to the point of being stupid.

The 300V condition seems reasonable.

A trial value of load for SE pentode is always V/I. 300V at 67mA is 4.5K load.

Power output, for power pentode, should approach 0.4*Pdiss or 8 Watts.

Using 19W Pdiss we must hold down to 63mA.

The very soft low-current end of a 7591 swing will distort "bad". THD numbers will be "high" even before 8 Watts. As long as we are being "bad", can we cheat without getting caught? A low-impedance load will suck more current, and we "gain" more on the top than we lose on the bottom, THD gets slightly more-bad, 13%, but the "11 Watts" number is impressive. But this is a bench-numbers game. In a real world of loudspeakers it isn't real meaningful.

250V at 80mA 3K to 350V at 54mA 6.4K. Err 20%-30% low impedance rather than high.
 
Hammond 125DSE can be wired 2.5K (yes, it will stand 80mA when worked 2.5K for guitar) or 5K.

[spacelabstudio]

So, I did actually build this:

I think it sounds great.

http://ampnerds.spacelabstudio.com/projects/se1/index.html
http://ampnerds.spacelabstudio.com/projects/se1/rossi/index.html

Chris