• 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.

"Low" B+ Voltage Amplifiers

Status
Not open for further replies.
Jump to Latest
#61
rongon said:
5687 as a cathodyne, you think? The bias on an EL86 (or 6P43P-E, in my case) with 250 plate volts would be about -20V or so, right?

I should look into doing a straight up LTP from two 6AU6 into a PP pair of EL86 pentodes into 5k:VC.

I think I have a decent circuit proposal using a 6DJ8 cathodyne into PPP 6P15P-triodes into a Dyna A470 OPT (roughly 4.5k to 5k primary).
Click to expand...
Yep the bias is around -20V at 250V or so. However for a low voltage design I would go with the datasheet conditions of a= 170V, g1 = -12.5V, g2 = 170V, Ia=70mA in pentode mode.
The way I do it is to use an input transformer driving a LTP 6AU6 . This allows the full gain of the pentode to be preserved and it also allows me to hang the LTP off a negative -250V rail. I then direct couple the front end to the EL86. Of course this is a tricky arrangement to get right and not recommended for the novice. However what is worth considering is the Input transformer which allows for a 2V line level signal to generate enough gain in a 5687 to drive the EL86's directly. The 5687 can be used at 120V with g1=-2 and a gain of 18 with a Ia=36mA (a bit excessive maybe). OEP do some quite decent and cheap input transformers which fit the bill.
This amp is now my main amp and it does a damn fine job of just sounding great and transparent and detailed. Its the e3nd point of a very long evolutionary path of two stage Schaded amps.

Shoog
 
Last edited:
#64
I found I could get a 26LX6 to run at 5 mA with just a 5V knee drop (9.5V drop at 40 mA), but then it's gm is just 1650, same as a 3S4 (although the 3S4 wastes more V, like 10V). What's needed are some tubes that don't waste the LV at higher current (without mega heater bills like the 26LX6 or 26A7GT).

I have it. Fill the bottom of the amplifier chassis with a Jello like clear silicone caulk of some kind, that can be re-melted after probing/testing through it. Then one can run HV without any shock hazard.

http://frank.pocnet.net/sheets/137/3/3S4.pdf
http://frank.pocnet.net/sheets/049/2/26A7GT.pdf
http://frank.pocnet.net/sheets/123/2/26LX6.pdf
 
Last edited:
#65
I just tested a 13GB5 (Zenith) for knee voltages versus knee mA, and it is looking quite good too for LV operation.
(Vg1 = 0 all)
Vg2/Knee mA/Knee V
5/ 5/ 3
9/ 10/ 5
12/ 15/ 6.5
15/ 20/ 8
17/ 25/ 9
20/ 30/ 9.5
23/ 35/ 10
25/ 40/ 11
27/ 45/ 12
30/ 50/ 12.7
34/ 60/ 13.5
38/ 70/ 15.5
41/ 80/ 18

Compare the 13GB5 11V knee at 40 mA with
26DQ5: 10V knee for 40 mA
26LX6 and 35LR6: 9.5V knee for 40 mA
and 26A7GT: 5V knee for 40 mA
 
Last edited:
#66
you can run the tubes at 10v plate if you raise the plate resistor high enough that there is about 60% B+ accross the tube.
then feed the output to an op amp buffer with 2.2M ohm input impedance.
practically, 60v B+ with 220k plate resistor on 12AX7 sounds good.
the 6922 should work well with a 27k plate resistor 60v B+ as well
 
Last edited:
#67
OP Amp, 2.2 meg Ohm input??? Must be microAmp tube operation.

-------------

Just tested the 21HB5A tube for knee voltages, fairly similar to the 13GB5
(Vg1 = 0V)
Vg2/Knee mA/Knee V
3/ 5/ 3.5
7/ 10/ 5
12/ 20/ 8
22/ 40/ 11
31/ 60/ 14.5
38/ 80/ 17.5
 
#68
Ah-Ha! I just made some progress on lowering the plate knee V.

Using the 21HB5A tube, each time adjusted for 60 mA:

Vg1/ Vg2/ Knee V
0V/ +31V/ 14.5V
-2.625V/ +49V/ 10V
-5.25V/ +64V/ 10V
-10.47V/ +92V/ 12V

As you can see there is a broad minimum in the kneeV, using negative 2.6 to 5.2V on g1 and then higher Vg2 to compensate. The +Vg2 voltages look like they can remain in a reasonable range, and this gives a 31% reduction in knee V.

Tested again at 40 mA:
Vg1/ Vg2/ KneeV
0V / +22V/ 11V
-.534V/ +26V/ 9.5V
-1.06V/ +29V/ 9V
-2.62V/ +38V/ 8.7V
-5.25V/ +53V/ 9V

For a 21% reduction in knee V at 40 mA. I'll have to try higher current now to see if the % reduction is greater. That would be SUPER!!

I Assume this holds for other tubes. I'll go try another one too.
 
Last edited:
#70
Well, big progress now!!

1st, 21HB5A at 80 mA knee:
Vg1/Vg2/Knee V
0V / +38V/ 17.5V
-2.63V/ +67V/ 13V
-5.28V/ +73V/ 12.5V
-10.5V/ +101V/ 15V

Again a minimum knee for -5.28Vg1 and +73Vg2. A 29% reduction in the knee V from Vg1 = 0. So looks like the % reductions tops out for 21HB5A.

Then I tried a 26LX6 at 80 mA knee, and I got a 10V knee for any Vg1 and Vg2 combinations (that gave 80 mA) So there is NO optimisation here.

Then I finally noticed the OBVIOUS staring at me all along. Don't use the knee for the maximum current load curve, push the knee current way up above that (with high Vg2 and some neg. Vg1) and operate on the DIODE line instead (for the load line intersection).

For 21HB5A at 80 mA on the diode line:
Vg2 = 88V, Vg1 = -5.3V diode line at 80 mA = 9V

and for the 26LX6 at 80 mA on the diode line:
Vg2 = 94V, Vg1 = -10.5V diode line at 80 mA = 5V

So the big Sweep tubes can operate at high current and equal the 5V loss of the 26A7GT. The smaller ones more like 9V loss. But this is at 80 mA!

No question one can make a LV amplifier now.
 
#71
Summary results now:

26LX6 can now make 200 mA max conduction (and 19000 gm) with just +10V on the plate, using +86V on g2 and -2V on g1. Or 150 mA max conduction ( and 15800 gm) with just +10V on the plate, using +52V on g2 and -2V on g1.

So g1 operation will require a g1 clamp at -2V max, rather than the usual 0V.
+50V B+ with 26LX6 P-P should get you 4 Watts now. (better than the 0.44 Watts with the 26A7GT)

Using Crazy Drive with the 26LX6 would require +22V on the plate to get 150 mA conduction, or +25V on the plate to get 200 mA conduction. Or +38V on the plate to get 450 mA conduction.

26DQ5 ($3) is probably similar to 26LX6 with this new approach, but I'll have to test it again.

And for 12HL7, using -6V on g1 for the max, and +93V on g2, gets the 40 mA conduction plate V down to 10V from 17.5V (with +75V on g2 and 0V on g1)
 
Last edited:
#72
I tested the 26DQ5 using the diode curve approach, and its getting down to 10V on the left side, but it has more kinks to the right of the diode line. Tested 6CB5A and 6EX6 and 6CL5 also and they are in the 10 V region too.

However, it soon occurred to me that placing the load line from the diode line (instead of the knee) runs through all those kinks for lower plate current curves. So large signal (well, relatively) would have compressive kinks on the load line. So now I know why the knee is always used for the max current on the load line.

So 10V or 15V min. drop on the tube is the best one can do (other than 26A7GT at 5V Knee/ 40 mA). That then puts 26LX6 using 50V B+ and 150 mA (12V knee) at only 2.85 Watts output for P-P. Only option would be to parallel more 26LX6 tubes.

And thats the best selected 26LX6, and --LX6 is the best tube type I have seen (even tested 36LW6, not as good knee V). So LV 50V B+ sucks for an amplifier!

Only options I see are to use the sealed impedance converter box for each tube scheme, or a floating hidden boost supply in the plate caps, or just plain cheat with the Volkswagoner scheme (LV - HV bottom panel hinge switch!).

You need +200V B+ at least, to get a decent tube amp.
Finito!
 
Last edited:
#74
Shoog said:
Yep the bias is around -20V at 250V or so. However for a low voltage design I would go with the datasheet conditions of a= 170V, g1 = -12.5V, g2 = 170V, Ia=70mA in pentode mode.
The way I do it is to use an input transformer driving a LTP 6AU6 . This allows the full gain of the pentode to be preserved and it also allows me to hang the LTP off a negative -250V rail. I then direct couple the front end to the EL86. Of course this is a tricky arrangement to get right and not recommended for the novice. However what is worth considering is the Input transformer which allows for a 2V line level signal to generate enough gain in a 5687 to drive the EL86's directly. The 5687 can be used at 120V with g1=-2 and a gain of 18 with a Ia=36mA (a bit excessive maybe). OEP do some quite decent and cheap input transformers which fit the bill.
This amp is now my main amp and it does a damn fine job of just sounding great and transparent and detailed. Its the e3nd point of a very long evolutionary path of two stage Schaded amps.

Shoog
Click to expand...

Hey, that sounds very interesting indeed! What kind of input transformers did you use? 1:1? Or did you use a step-up? I have a couple of pairs of input transformers that might be usable for this, but they're all close to 1:1.

One thing I don't get is how a 5687 triode with low ra and relatively low mu would be used for plate-grid feedback around the output stage. I thought for 'Schade feedback' you want a driver with high ra, like a pentode...
--
 
Last edited:
#75
A 1:1+1 is demanding enough on input duty so I wouldn't even attempt to do a step up.
The schade arrangement I have uses 6AU6 for the input, I was not suggesting using a 5687 as a schaded input. That would be for some simple two stage with some form of plate to cathode feedback or a loop feedback arrangement. You would have to work out the precise details yourself.

However a 6AU6 can work well at low voltages if you wanted to try the Schade arrangement. in my case however I give the 6AU6 a full 250V to work with - which is easy when you have a negative rail. I used a 220V shaver isolation transformer to get my -B voltage which is a cheap solution when you only need 20VA.

Shoog
 
Last edited:
#76
Thanks, I'm relieved to find out that I misunderstood you. A 2-stage 5687 to PP 6P43P with gNFB would be one way to do it, and yes, a 1:1 input tranny was what I was thinking of (no gain, just phase splitting). Or maybe use some kind of control preamp with balanced out and do away with the input tranny? Anyhow, lots to think about. Thanks again for your ideas.
--
 
#77
rongon said:
Thanks, I'm relieved to find out that I misunderstood you. A 2-stage 5687 to PP 6P43P with gNFB would be one way to do it, and yes, a 1:1 input tranny was what I was thinking of (no gain, just phase splitting). Or maybe use some kind of control preamp with balanced out and do away with the input tranny? Anyhow, lots to think about. Thanks again for your ideas.
--
Click to expand...
You might actually find that the 5687 is a bit marginal depending on how much feedback you want to apply. The ECC88 might be a better candidate for the job.

Shoog
 
Status
Not open for further replies.
Top Bottom