First 300b Design, Please comment

[SGregory]

I have wanted a 300b SET for a long time. After considering many different driver ideas, I came up with the following KISS design. I am a newbie at tubes and I won't take claim for any creativity or being too original. I enjoyed the sound of the 12b4 preamp so much I figured if I could get a grounded cathode driver to swing close to 70V with a 2V input I should try it. I thought about using a CCS for the cathode resistor but held back for the first effort. Maybe try it later.

For this circuit B+ is 500V using a 3k primary with the bias set to 80mA on the 300b. The driver is a 5965 with the triode paralleled. This is decoupled from the output stage by a dropping resistor of 3k bypassed to ground with a 20uF GQ cap. The resulting voltage across the drive circuit is 450V. Ra = 14k resulting in a plate voltage of 222V at idle. With a 2V input it will swing 68V at an Zo of 3k. The bias of the driver is 16mA across the plates using a bypassed 270R resistor. Any thoughts or comments for an improvement in this design. Better yet will it work????

P.S. Filaments are regulated DC. Forgive the hand drawn schematic, I don't have any fancy software to do it yet.

[tomchr]

I'm slowly coming up to speed on some sort of high-ish power (10-15 W would be nice) SET amplifier myself. I'm currently looking at the 300B and have run a couple of quick sims with it.

The 300B is limited to 400 V on the anode. Granted, there are some types out there that claim to be able to handle up to 450 V, but from what I understand, they're highly unreliable when operated at that high voltage and tend to go into thermal run-away. I would limit B+ to 400 V. Also, I think with that high voltage and the 3 kOhm impedance of your OPT, you'll be quite a ways off the sweet spot of the 300B's. You'll maybe get a little more headroom, but you'll be trading off THD in a big way. At least that would be my concern. Also, if you want max power out of the 300B, your driver circuit should probably have lower output impedance as you'll be driving the 300B a bit into class A2. Many people add a cathode follower between the input stage and the 300B - or do what George (tubelab) does in the TubelabSE and add a source follower.

I suggest throwing the circuit into a SPICE (or the like) circuit simulator. I use PSpice 9.1 Student Edition which can be downloaded here. You can get models for a handful of tubes - including the 300B - on Norman Koren's site.

Good luck! Keep us posted.

~Tom

[SGregory]

I thought that my plate voltage on the 300B was pushing it. Per SE Cad it would be right at its limit of 400V. Easy enough to change the load and voltage. A 3.5k would be more useful if I don't like this design. Given a 3.5 k I could push up idle current to 90mA with B+450V which would put plate voltage to 350ish.

I was not intending to operate into class A2. However, I guess a good question would be what is an "acceptable" level of Zo for the driver stage? I actually thought that 3k was pretty good for A1. Unless I missed something (which I am sure I did) it should drive >>20kHz

I have downloaded pspice. Will take a while to figure that out. I am still pretty much at the stage of getting tubecad to work and drawing loadlines on paper.

I have a tubelab SE sitting in front of me. (minus xfmrs). However, in this circuit I didn't want to put in SS.

[tomchr]

The learning curve on spice can be a bit steep at first. But it's worth it.

Just remember that transient simulations are versus time, AC sims are versus frequency. So for distortion simulations, you'd want to run a transient simulation at the frequency of interest and figure the THD from the harmonics (it's set up as part of the transient sim).

Thankfully, spice is reasonably well documented and there are tons of tutorials available on the net.

~Tom

[SGregory]

Although I have not put this in PSpice, I thought it worthwhile, based on comments, to redesgn the driver with a CF using the 5965 tube. In this case I split the triode, one for gain and one for buffer. This drops the Zo of the driver to 165ohm. Total gain of the amplifier is 34.08 which is still within my goal. I am not opposed to other tubes or using two tubes for the circuit. I just have a few of the 5965 laying around.

I also dropped 300b current to 70mA at 350V plate. This reduced bias voltage to 70v. At full input now the driver has to get closer to 0 volts on the 300b grid. At that point I am not sure if there is enough current flowing through the CF to drive the 300b properly. At this time I still want to avoid A2 operation.

The other question I have with a CF do you or do you not want to bypass the cathode (Rkb) resistor?

[45]

If you want to stay within Class A1 why don't you use the recommended operating conditions by WE?
For example: 400V/60mA, bias -87V. If you choose a 5Kohm : 8ohm transformer you will get 8.3W (a little less in practice because of the transformer loss) with 3% 2nd harmonic, 0.5% 3rd harmonic and much less for higher harmonics. That's 3.05% THD approx.
If the load of your speaker drops to 5.6 ohm the primary load will be 3.5K and you will get for the same input (174 V peak-to-peak) 10.5 W with 5% 2nd harmonic, 1.2% 3rd harmonic etc. (approx. 5.2% THD).

Practically you could use any nominal 8 ohm speaker.

If the driver is a triode that can swing 61.5 Vrms with dominant 2nd harmonic you could get lower overall THD in a wide frequency band because of harmonic cancellation.

Cheers,
45

[zigzagflux]

Originally Posted by SGregory

The other question I have with a CF do you or do you not want to bypass the cathode (Rkb) resistor?

That resistor should not be bypassed. Your 20uF power supply cap presents a zero impedance to AC at the plate. If you installed the bypass capacitor, you would also have a zero impedance to AC at the cathode. Both plate and cathode are now shorted to circuit common. No more signal.

[dsavitsk]

My sense is that putting a 300B in this sort of topology is a little like cramming a Porsche engine in a Pinto. There is a lot you can do to improve things -- CCS plate and cathode loads, LED bias, Power Drive, Ultra-Path cap, Shunt regulators, DRD, etc. -- and without some of that you won't reap the benefit of an expensive DHT. Might still sound better than a KT88 or whatnot, but it won't sound as good as it can.

Further, the filament supply and the power supply are probably more critical here than things like whether you use a cathode follower or not, and they should not be after thoughts.

Finally, 4.7u is a little large for the coupling cap. 0.22u, or even 0.1u, would be fine with the 200K resistor.

[JPeitzman]

I can understand wanting to build an amplifier that is sand free. However if you are going to use a cathode follower you are basically back to the Tubelab SE topology. I would argue that the FET follower would do a better job than the tube. Additionally you could go back to your first schematic and use an active FET load and take the signal from the Mu output to get an equally low output impedance as the follower.

I am also wondering why you want to avoid A2 but don't use recommended points? At 350vp you should be around 50mA to 60mA @ -76 to -74 with that load. Of course this can vary per tube, but the recommended points are usually a good place to start.

Additionally you could try transformer coupled with fixed bias. It would cost a bit more, but it is an elegant solution to lose everything in the signal but a couple resistors, a couple transformers, and the tubes of course. You could also look to try a tube with lots of balls so that it can drive the output if/when it starts to pull grid current without the follower, something like the 6BQ5, D3a, or 12HG7. I posted an idea a-la Gary Pimm using the 6BQ5 CCS loaded, however you wouldn't need to CCS load it, simple resistors would also work. Take special note in Pimm's test of the 90V out @ .25% distortion figure. This type of circuit could, and would, drive the living snot out of the 300B at any of the 350vp operating points.

As mentioned above, DRD could be an excellent option for avoiding sand. With DRD you are back to two stage instead of three....two with a follower; and you would lose almost all of the junk in between.


James

[Miles Prower]

Originally Posted by SGregory

I was not intending to operate into class A2. However, I guess a good question would be what is an "acceptable" level of Zo for the driver stage? I actually thought that 3k was pretty good for A1. Unless I missed something (which I am sure I did) it should drive >>20kHz

Unless you're feeding it some simple waveform, operate into Class A2 you shall. Music programs have fast rising transients that can be way above the average value, and these will likely drive the grid positive. These DHTs also tend to pull grid current even before the grid actually swings positive. Finally, 300Bs need adequate grid drive. Most of the complaints about disappointing sonic performance are due to inadequate grid drive. For any DHT, the MOSFET source follower is just about the closest to ideal grid driver, even if you operate Class A1 most of the time. A decent grid driver will handle overdrive more gracefully, and not contribute to sonic degredation. There is no lack of current sourcing, and the Zo is much lower than anything hollow state can give (since MOSFETs have much higher gm's than any vacuum tube).

However, in this circuit I didn't want to put in SS.

I'd use the SS. If you are dead set against that, then consider using an interstage xfmr (and be prepared to spend some $$$$ for a good one, cheap IT's are a major sonic compromise) and connect it to a stiff driver like a 12B4, trioded 6V6, 6AQ5, 6BQ5, or the low-u section of those dissimilar dual triodes that were used as vertical deflection oscillator/finals. That's just about the best hollow state way to deliver grid drive to a DHT.