Compact ~20WPC Tube Amp

it is simply using the first section of the 6SN7 as a grounded cathode volt-amp DC coupled to the second triode and then capacitor coupled to the 300B grid. This amp does not do the 300B justice (I’ve built it before).

As you know a circuit like this will result in a higher output impedance so wouldn't be the best application for driving a grid that has high capacitance and needs a lot of swing. The 300b has nearly the same mu as the power tubes we use in our OTLs. We also built an OTL using 300bs for a demonstration so I'm not speaking idly.

We don't run the tube with a lot of current either, despite that, direct coupled to the grid of a 300b will have no problem driving it to saturation. At that point the 6SN7 section will saturate too; when replaced by a much gutsier tube the difference was less than a 1% change so its very obvious that the 6SN7 can do this job quite well.

We use the 6SN7 driving the 6AS7G; if you compare specs to a 300b you see that the former has slightly higher grid capacitance and slightly lower mu; the 300b is easier to drive. Despite that we can get one 6SN7 section to drive as many a 14 6AS7G grids with no worries. One grid of a 300b would be like a walk in the park.

If you like the 6SN7 that much, if I were you I'd be looking a better topologies to make it happen.

Either of them (And other amplifier builders who have similar view) are far more qualified than me to debate this point with you. I can say that their comments make plenty of sense and are backed up by their SET amplifier performance results. I just listen and read in an attempt to learn.

@charles1dad 

What they are doing is a brute force thing by running lots of current through the tube so they can have lots of current drive on the grid.

Of course by simply paralleling elements in the 6SN7 they could do the same thing... but at any rate, there is a series impedance (the coupling cap) which limits how much current can be delivered to the power tube grid. So they are going for overkill.

But if you direct couple you solve this problem, although to do it effectively you have to run the driver tube with no voltage gain. That’s exactly how we’ve done it in our OTLs. The advantage of this approach is there is no overload recovery time- its instantaneous as there are no timing constants present between the driver and power tube grid. But it does make for what appears to be a more complex circuit, something that seems anathema to SET designers.

However, there is another advantage: the coupling cap that you do have to have to couple the signal to the driver tube can be a considerably lower value, usually by at least an order of magnitude, possibly more. This means that no matter what the coupling cap, the smaller cap will always sound better. So you have a double whammy.

The downside is the driver tube isn’t available for gain. But that’s OK- being a lower powered amp, an SET doesn’t need much gain (certainly less than we have in our OTLs).

The funny thing is you could execute this with the same number of tube sections as used in a traditional SET right now (three total, including the power tube). The reason you don’t see this is the fact that you have to build a negative power supply. But if you did, you could build an SET that was inherently lower distortion which means it would be more transparent.

The main limitation with some SET amplifiers is the choice of weak/inadequate driver tube. Just as you say, this driver needs to be robust. The DHT output tube demands this. He chose the 6EM7 rather than the 6SN7 for that reason

@charles1dad The 6SN7 has no worries driving a single DHT! That's not the problem- the coupling cap and the timing constants around it is. We ran into the same problem in our OTLs, but solved it by direct-coupling. We have a single 6SN7 thus driving 8 power triode grids with no distortion problems even when drawing substantial grid current. But to do this properly, you need a negative power supply and most SET designers aren't interested in the additional cost.

Which leads me to suspect that the 96 db sensitivity rating is exaggerated or (More possibly) the speaker’s impedance curve and load just isn’t suitable for an SET 300b.

@charles1dad That wouldn't surprise me at all. The impedance thing is why I wrote this article since so many speakers that are rated 8 Ohms actually have 4 Ohm woofer arrays and simply aren't meant to be driven by zero feedback tube amps! That was part of why I convinced Israel to try making speakers with higher impedances. It doesn't help that speaker manufacturers seem to often overstate their numbers as well.

If you can find a Dynaco ST35, its a lot more linear amp than its bigger brother. But it makes about 17 Watts/channel.

SETs tend to have a limited amount of what I call 'usable power'. By that I mean its usually not a good idea to push them past 20-25% of full power. Based on your comments your speakers are nowhere near efficient enough to work with your SET.

If that 96dB number is real, your speakers are similar to mine which are 98dB. I find that I need a bit of power; at least 30 Watts. I have an average room, so I suspect you'll need at least that much.

So the amps you are looking for (if you stay with tubes) will be PushPull. They tend to have a much higher amount of 'usable power'; usually close to 95%.

The ST70 is not a bad suggestion in this range; the RM10 is another good one. Either should be properly refurbished. Another amp that might suit is the Harmon Kardon Citation 2 if you can find one refurbished. They make about 60 Watt/ch.