single-tube (6GN8) amplifier?

[geekysuavo]

hey all. i've hacked together an interesting idea for a basic class a amplifier using a 6gn8 triode/pentode...

how's it look? any comments or suggestions?
http://www.geekysuavo.org/throwbacks/images/amplifier-1.0.pdf (it's also attached)

some figures, which i'm assuming (??) to be relatively correct:
triode section
HT = 350V
Rl = 120k
Ia = 1.125 mA
Va = 215V
P(Rl) = 152mW
Vin = -1Vpp
Vout = 60Vpp
Av = 60
Ra = 44.29
mu = 90
gm = 2.032mA/V
Rk = 1.8k
Vg-k = -2.025V
P(Rk) = 2.3mW
rk = rk || Rk = 901
Ck = 220uF
f(-3dB/LF) = 0.8Hz

pentode section
Zl = 6.5k
Va = 145V
Vin = -2Vpp
Vout = 125Vpp
Av = -62.5
Ia = 31.54mA
Pout = 6.5W ... is this correct?
Ig2 = 15.5mA
Ik = 47mA
Rk = 53.6
Vg1-k = -2.5192V
P(Rk) = 118mW
Ck = 3000uF
f(-3dB/LF) = 1Hz

miller effect
for pentode, Cag = 0.1pF
Cmiller = (60 + 1) * 0.1pF + Cstray = 17pF
for pentode, Rgl = 1M
f(-3dB/HF) = 288 kHz

i know that's just a lot of numbers, but i figured they may be of use.

peace,
~ brad.

 

[Jeb-D.]

Your schematic doesn't match the data listed. Schematic has a B+ of 350V, but you list the pentodes anode voltage of 145V. Your transformers 6.5k impedance is AC, the DC resistance is probably less than 100 ohms. 350V B+ would exceed the plate and g2 voltage ratings, which is 330V.

The 6gn8's Pentodes plate is rated for 5W. That means the absolute most power you will be able to squeeze from it is 2.5Wrms (5Wpeak) in Pentode mode. But you will realistically probably end up with about 2W Pentode and maybe a little less in UL. That's with a properly designed circuit.

There are some similar tubes rated for more power. I can provide numbers if they are of use to you.

 

[Jeb-D.]

In case you are new to transformer coupled stages, check out:
http://members.aol.com/sbench102/po-dis.html

Keep in mind that your plate voltage will be the B+ (HT, whatever you want to call it) and subtract, bias voltage (for cathode bias) and subtract - (idle current*transformers DC resistance).

You will find the voltage from the plate to ground to be pretty close to B+ at the operating point. Now the tricky thing is (when compared to resistor plate load), the plate voltage swings above(way above) your supply voltage (B+) when an AC signal is applied.

 

[sorenj07]

Now the tricky thing is (when compared to resistor plate load), the plate voltage swings above(way above) your supply voltage (B+) when an AC signal is applied.

Yep. It took me the LONGEST time to figure this out before I could design my own amps Great site by the way.

 

[geekysuavo]

wow, thanks for the feedback! i wasn't getting any emails from the forums, so i'd assumed incorrectly that nobody had replied... oops.

ok, i do see that my operating voltage is much too high now. thanks for pointing that out. but since reactance (and thus impedance) of the output primary changes with signal frequency, how is there any chance that the frequency response of the power amplifier will be linear in the 20-20k range? i must be missing something here...

just to make sure, the triode looks o.k., right?

~ brad.

p.s. thanks about the site!

 

[geekysuavo]

as a side note, i'm probably not looking for too much power output, since this is kind of my first design anyways... i suppose i could always go with a 12ax7/el34?

BUT, would 2W per channel even be enough to push two small 8 ohm speakers into reasonable volume?? thanks again for the help...

~ brad.

 

[geekysuavo]

sorry for the barrage of posts... lol. i just figured out how loud you can make 2W depends on the sensitivity of the loudspeakers... good to know.

~ brad.

 

[geekysuavo]

well, now that i'm thinking about it, what are these mystical similar tubes to the 6gn8/6eb8?

~ brad.

 

[pmillett]

You might want to check out the latest issue of audioXpress (out today?). I've got an article on a similar amp in it. There is some data (schematics, etc.) on the design here:

http://www.pmillett.com/midget.htm

The magazine article has the full construction details in it...

Pete

 

[Jeb-D.]

but since reactance (and thus impedance) of the output primary changes with signal frequency, how is there any chance that the frequency response of the power amplifier will be linear in the 20-20k range?

With a resisitive 8ohm load on the secondary- The reflected impedance on the primary is 6.5k. The inductance of the transformer is high enough to where its reactance isn't influential until the low frequencies.

A good way to look at it is like this- the plate is loaded with a 6.5k resistor in parallel with a large inductor. At DC, the inductor is pretty much shorts out the resistor. But as frequency increases the inductors reactance becomes large, to the point of where it's like an open circuit. All that the plate will see is the resistor (which is the reflected impedance). The transformers inductance will determine the low frequency cutoff. For a given impedance, a transformer with a larger inductance will be able to extend lower in frequency.

For a real world speaker load, where the impedance fluctuates, yes, the reflected primary impedance does vary. This will cause a non linear frequency response from a voltage standpoint. The amount of how much the output voltage fluctuates with impedance variation is how amplifiers output impedance is determined. An amplifier with a very low output impedance acts as a voltage source, where-as a amplifier with a high output impedance acts like a current source.

 

[Jeb-D.]

If you want to do a Triode-Pentode, you may want to consider building some one else’s design. The one pmillett linked to is a good one. Pentodes take quite a bit more skill to design with than triodes, which may be frustrating for a first time builder. You have to have a solid understanding of how the screen grid affects things. Also Pentodes have large output impedance, which doesn't really matter for a guitar amp since your operating in the flat area of a speaker’s frequency range. But for Hi-Fi, you will have to incorporate negative feedback to drop the output impedance. Doing this also takes some skill, trial and error ect.. to do properly.

If you are set on designing your own, you may want to check out some dissimilar dual triodes.
6EM7
6EA7
6DN7
Triodes do only put out half the power of an equivalent pentode, but those triodes are rated at 10W. So you should be able to get your 2W out. This is plenty, when paired with efficient speakers for a bedroom or office amp.

If you are set on designing your own triode-Pentode, I would use a 6LU8. It could put out 5-7W.

Here is the best web resource I have found on the fundamentals of Pentode design.
http://www.freewebs.com/valvewizard/pentode.html

It only shows Resistive loaded, but once you understand that + a transformer loaded triode. You shouldn't have a problem doing a transformer loaded pentode.

There is a lot to this stuff, and it seems confusing at first. But if you stick with it, you will be a master in no time

 

[geekysuavo]

wow, this is excellent information... thanks!

well, i guess i could either try triode/triode and use a 6EM7, or try triode/pentode and use an ECL86 (and probably have to use NFB)? perhaps a first hifi project should use the simpler SET and a cheap output transformer...

~ brad.

 

[Jeb-D.]

If I were you I'd probably go triode-triode, with some cheap but decent output transformers for your first project. www.edcorusa.com has some good open-bracket ones for cheap. You seem capable of designing your own with a little research + trial and error. It will be more rewarding in the end compared to building someone elses design.

Just remember, safety first!

 

[Tom Bavis]

6EA7 / 6EM7 is a good choice, since it has a hi-gain input triode, and octal sockets are easier to wire than miniature ones (The 9-pin 6FD7 and 6DE7 are similar, but rated a little lower). You can find a number of 6EM7 and 13EM7 circuits on the web too - single ended, push-pull, direct coupled...

 

[Jeb-D.]

For power supply design use this
http://www.duncanamps.com/psud2/index.html

 

[geekysuavo]

thanks for the link. i've been wanting to look at PSUD for a while... (no windows here ) i'm considering building an adjustable maida regulator for testing designs out, and probably an adjustable constant current filament regulator too, using LM317T's and RLC filters... (a la morgan jones) any recommendations there?

~ brad.