suggestions for driver tube for 2A3

[Bigun]

Use of a line amp with gain may be desirable

that was my concern too, it was stated above that the ECC99 is a perfect driver for 2A3 but I'm not sure it has enough gain

PS - What's wrong with using a higher mu tube into a cathode follower, driving the 2A3?

nothing I'm aware of but many high-mu tubes also have high plate resistance

 

[rongon]

nothing I'm aware of but many high-mu tubes also have high plate resistance

That's why adding the cathode (or FET) follower might be a good idea. The high plate resistance, high mu tube drives a cathode follower, which has a very high input impedance and low Miller capacitance (it's an easy load). Then the cathode follower has the low output impedance and higher current to drive the 2A3 grid.

I tried this in LTspice. A 12AX7 to an ECC99 cathode follower to a 2A3. It looks like there's plenty of gain there. Maybe a single ECC99 can drive a 2A3 adequately. Let me try that next...

Tried again, this time with a better OPT model (not my model). I got 2W output at 2.5% THD with 0.7V input (peak).

The THD from the output of the cathode follower sims at 0.25%.

That's about what you'd expect from a single-ended 2A3 amp, no?

--

 

[rongon]

I played with it some more in LTspice.

I used a 300V B+ supply.

Using a single 5687 to drive the 2A3 grid.

CCS in the plate, set to draw 13mA (a decent spot for the 5687, well clear of the knee in the plate curves).

LED in the cathode in series with a 47 ohm resistor to get the cathode voltage up to 2.56V.

RC coupled using 0.22uF and a 470k grid leak on the 2A3.

2A3 is biased with -44V grid to cathode, using a 750 ohm cathode resistor and 330uF cathode bypass cap.

With 2.5V input (peak) there is 43.8V peak coming from the plate of the 5687.

It looks like using a single 5687 as the driver for the 2A3 means the output tube reaches full power just when you reach full output from your input/driver tube.

You can bias the 5687 cooler, with maybe 12nmA plate current. That way you can get a few more volts out of it. But you're not getting 6dB of driver stage headroom.

Anyway, that's what I found...

 

[methodarn]

that was my concern too, it was stated above that the ECC99 is a perfect driver for 2A3 but I'm not sure it has enough gain



nothing I'm aware of but many high-mu tubes also have high plate resistance

why ecc99? it's look linear just with B+ about 600V
try to 6s45p very low Ri - 1.5k, mu about 50! in real circuit it will be 48 and easely drive 2a3 at full power. but carefully, tubes with high transconductance such as 6s45p, 6zh9, e280, e810 etc. with not enough bias have current on grid! -2v would be guaranteed OK

regards, Dima Kohan

 

[Bigun]

That's why adding the cathode (or FET) follower might be a good idea.

I agree, it is quite appealing.

A 12AX7 to an ECC99 cathode follower to a 2A3.

If we were choosing a cathode follower we'd pick something with higher gm ?

why ecc99? it's look linear just with B+ about 600V

true if we want high current and high gain with a resistor load but it could be used with an active or choke load

try to 6s45p

I heard this tube once, it was very clean and powerful, perhaps too clean.

 

[rongon]

why ecc99? it's look linear just with B+ about 600V
try to 6s45p very low Ri - 1.5k, mu about 50! in real circuit it will be 48 and easely drive 2a3 at full power. but carefully, tubes with high transconductance such as 6s45p, 6zh9, e280, e810 etc. with not enough bias have current on grid! -2v would be guaranteed OK

regards, Dima Kohan

It's been discussed in this thread already. Tubes like 6S45P, or triode wired 6J9P or 6J52P, tend to create a higher proportion of odd harmonics to even harmonics than some people like to see from their driver tube.

Also, to get a grid bias of -2V on a 6J9P-triode requires a high plate voltage (180V or so), and you're only going to have a 300V B+ if you're using a 2A3 output tube. You can get around this by adding a second, higher voltage power supply for the driver tubes. Or as Bigun mentioned, use a CCS in the plate of the driver tube.

The Tubelab Power Drive looks like a good solution. CCS in the plate of your driver tube, RC-coupled to a source follower that's DC-coupled to the output tube.

Does require fixed bias on the output tube, though. But that's not so bad. Maybe overkill for a 2A3, maybe not.

Many ways to skin this cat these days.

--

 

[Ludwig Haus]

112A - 171/371 - 2A3

With use of IT coupling between driver and output the distortion products can either add or subtract (optional with configuration de secondary). Also, it is possible that the output tube grid can be positive at the same time the driver plate goes low so as to have the low Rp point of the driver at that moment…

Or DC couple (my favourite).

Either way, all DHT and save the D3a/ (and) E280F for low level stages (super quiet tubes when not oscillating!)

Also, just because its a 'three stage amp' could simply mean that you don't need a line stage amplifier (and you wouldn't) - call it a three stage amplifier from source to loudspeaker, integrated would be the word, but somehow sells less 'interconnects'.

Peas in a pod, fish in a bowl, meat at the farm… we're all the same.

LH/S

 

[rongon]

Last night, very late, I woke up thinking about the LTspice simulation I'd done earlier. It looks like there's distortion cancellation happening if I put a 12AX7 voltage amp into a cathode follower and that into a 2A3. I thought, 'What happens if I deliberately add 2nd harmonic distortion to the voltage amp by using a 12AT7?' So I tried it, with a 12AT7 with 50k plate load, unbypassed cathode resistor, 3mA plate current, 190V plate voltage. In the sim at least, it works very well. I was getting less than 1% THD at 2W output from the OPT (8 ohm load in the simulation). The dominant harmonic is the 2nd. Sensitivity of about 1V peak to full power out (a bit under 3W, just before high order harmonics jump up in the output indicating hard clipping). That's a good sensitivity, not too high, not too low. There's a good amount of headroom left too, since the 12AT7 bias is about -2.5V (g-k).

12AT7 (DC-coupled to) 6J9P-triode cathode follower (RC-coupled to) 2A3 -> OPT

Interesting, I think.

--

 

[Bigun]

Since there are two triodes in the 12AT7 envelope why not use one half as the cathode follower - since the 2A3 grid is a 'light load' there should be little distortion. The distortion may also be tuneable with changes in the cathode load of the follower.

One thing I always found interesting about this tube (never heard one though) was the GE datasheet has a while page on +ve grid current - it's way more linear in that region than in regular negative grid bias. I wonder what use the tube has with +ve grid current?

 

[rongon]

Since there are two triodes in the 12AT7 envelope why not use one half as the cathode follower - since the 2A3 grid is a 'light load' there should be little distortion. The distortion may also be tuneable with changes in the cathode load of the follower.

Is the 2A3 really that light of a load? I'm not so sure... But yes, I'll try to sim that this evening and see if it looks promising. It will probably be fine. Anybody know how I would look for slewing distortion in an LTspice sim?

A useful way to use a 12AT7 might be to do a Broskie-style CCDA; voltage amp draws the same plate current as the cathode follower, with the plate of voltage amp and grid of cathode follower at as close to 1/2 the B+ as possible. That is supposed to reduce power supply noise by quite a bit.

One thing I always found interesting about this tube (never heard one though) was the GE datasheet has a while page on +ve grid current - it's way more linear in that region than in regular negative grid bias. I wonder what use the tube has with +ve grid current?

Maybe even better on the cathode follower front. Perhaps the voltage amp can drive the cathode follower a little way into grid current without ill effects? Maybe, I don't know.

--