Soft start kit alternate tubes

[johnsonad]

I wasn't sure where to post this and technical topics seems a good fit.

I'm open to the idea of an alternate driver tube in my 300B amps which use the soft start board. Their operating point is; 350v, -75v, 50mA, 5k.  I looked back though some old posts and saw the 6sn7 mentioned though I'm assuming it would need to be a 6sn7gt or gta due to the plate voltage. In my current set up (100dB sensitive speakers) this would work but I'm planning on using a pair of 95dB sensitive speakers and will need a little higher amplification factor I'm assuming (BeePre as the preamp)? Ideally I wouldn't have to change the board or if so, minimally and I understand that the socket would need changed or in the least rewired. 

Given this, can anyone suggest an alternate tube to try? Would a 6072 (12ay7) be an option (amplification factor of 40)?

Thanks,
Aaron

 

[Paul Birkeland]

I'd suggest rolling back to the 12AT7.  Since you aren't directly coupled, you can remove the 300K plate load, 33K/10,000uF paralleled combination, and run the v1.0 driver circuit. 

 

[johnsonad]

Thanks Paul. I have a stash of 12at7's from the v1 and that's an easy conversion.  I don't need extra gain but don't want to loose what I have at the moment.  Did you guys have luck with any others tubes when deciding on the 5670?

 

[Paul Birkeland]

It'd be easy enough to leave it as-is until you got your new speakers.  (Do you have 6dB of extra clicks on your BeeQuiet?)

 

[johnsonad]

I do. I'm about 18dB down on CD and phono and around 9dB down on tape with the 100dB sensitive speakers.  Ideally I wouldn't mind trying a 6sn7 or a 6cg7 but I'm not in a hurry.  Gain isn't the primary focus, but trying the tone of a different driver tube at some point. 

 

[Paul Birkeland]

You'll lose 6 dB with your speaker change.  If you go from a 5670 to a 6CG7, you'll lose another 6 dB of gain, and you'll need to bias the cathode of the driver stage up pretty high so you can drive it with significant voltage from your preamp.

If you're willing to depart from 9 pin dual triodes and drill some extra holes in your Paramounts, there are some other options.

 

[johnsonad]

I'm on a 12x18" chassis plate and have plenty of room for alternate tubes. I would even be open to two tubes, one for the shunt and the other as a driver. The base hole is already large enough for an octal. What did you have in mind?

 

[Paul Birkeland]

You could try a 6688 triode wired, 6C45 (careful with plate voltage), 6GK5, 12GN7 triode wired, etc.

For my "utility" Paramounts, I used the 7 pin 6J6 as my regulator tube.  These are inexpensive, draw pretty light heater current, and it's easy enough to use one or both triodes depending on how much regulator current you can afford. 

Do note that I think a pentode wired 3S4 (feed the screen from the 300B cathode) would drive the 300B pretty nicely.  This would have the added complication of rectifying the 6V winding and regulating it to 1.5V or 3V, but you would have the reward of an all DHT amp.  (I did this recently with a loctal pentode recently)

 

[johnsonad]

The 6J6 sounds easy enough and I have access to hundreds of them.

Tell me more about the use of the 3S4 please. Have you tried it in this configuration yet? Would it be direct coupled or capacitor coupled? This is a fairly microphonic tube as a triode, would that improve at all wired as a pentode?  Would the filament need the same attention as the 300B gets or could I drive it from a battery pack as is done in the Quickie?

 

[Paul Birkeland]

You could use a battery.

You get max transconductance with 1.5V on the filament.  You'd have 75V on the screen (more than max, but not by too much), and I'd run the tube at 100V.  With a 25K plate load, this should leave you with about the same gain as the 5670.  I'd run 6mA through the tube, and use a 25K/2W resistor from plate to ground (4mA), for a total of 10mA.  (90 Ohm R1)

You'll need to adjust the bias trim pot to see 7-8V on the filament so that you can get 100V on the plate of the 3S4 (I believe the "bottom" of the filament, see the datasheet), so the 4.99K resistor by the 10K trim pot should get another 4.99K resistor soldered across it to increase adjustment a bit. 

Obviously, you'll need to increase the overall driver stage current to allow some current to flow through the shunt regulator.  If we assume 10mA through the 3S4 and its plate load, 1mA to bias it, then maybe add 2mA for the 6J6, which leaves you at ~13ma, so a 66 Ohm R1 or thereabouts resistor will work (a little extra current can be sunk through half the 6J6 without any concern).

If you want to make 3V for the 3S4, cut the drain wire going to the "CT" on the 6.3V feed.  Bridge rectify that into a 10,000uF/10V cap, then feed a 3.3V TO-220 regulator with a nice 10uF cap on the output.  (3.3V is kinda high, but the battery tubes are tolerant to voltage shifts).

 

[johnsonad]

Thank you very much for the detailed reply Paul. It seems like a very doable conversion (3S4). Is 100v enough to swing the 300B (for my own understanding)? Are there any other DHT that I might want to consider?

I'm going to exhaust my stash of 2C51's and 5670's before changing driver tubes. So far with the tubes I'm using there seems to be room for improvement.  It could be the iron and amps as a whole breaking in too and I've got plenty of time to give them.

 

[Paul Birkeland]

It could also be that particular 300B operating point, it'd be easy to tell if you had stock Paramounts to compare them to.

At 100VP on the 3S4, you will be in the ballpark to drive the 300B to clipping with your operating point. 

 

[johnsonad]

Good point but so far I'm really impressed by this sound. It does need to work out a few bugs but overall the range seems balanced and the bass is impressive. It really shows the difference in driver tubes, more so than what I remember with my original Paramounts. Maybe a few more years of experience has taught me better what to listen for.  I'm a little surprised at how much of a difference in sound the driver tube makes in this build. I've got about every brand made and am slowly working though them until we find one that meets the overall goal, then we can fine tune after that.

If it doesn't end up working out then I'll consider a different driver tube or an alternate operating point.

 

[Paul Birkeland]

Good point but so far I'm really impressed by this sound. It does need to work out a few bugs but overall the range seems balanced and the bass is impressive. It really shows the difference in driver tubes, more so than what I remember with my original Paramounts. Maybe a few more years of experience has taught me better what to listen for.  I'm a little surprised at how much of a difference in sound the driver tube makes in this build. I've got about every brand made and am slowly working though them until we find one that meets the overall goal, then we can fine tune after that.

If it doesn't end up working out then I'll consider a different driver tube or an alternate operating point.

I think I might be able to get two loctal plugs and make some adapters to try the 3S4 in my Paramounts (though I currently run a different tube with a 10K plate load).

 

[johnsonad]

It goes without saying but please report back with how it works out if you do try them!

 

[Paul Joppa]

If you stay with indirectly-heated dual triodes, here are the limitations:

The high voltage is regulated to 300 volts, so that is the maximum plate voltage rating for the regulator. Most dual triodes will meet that.  (The driver itself operates at around 175v.)

The standard setup has about 3.4 mA through the plate of the driver. The 5670 saturates (zero bias voltage at 3.4mA) at 55v from the published curves, so the peak output is +/-125v so in the stock Paramount with 70v bias, the driver operates at 56% of its peak capability. This margin reduces the driver's distortion contribution to the total, which is one of my design goals. To maintain this situation, you need a tube with similar plate resistance, unless you adjust the plate current to re-optimize.

The bias of the 5670 is about 4 volts, and as designed the bias (from the 431) cannot be adjusted below 2.5 volts. In practice this means you can't get away with a mu greater than around 40-50 because the bias would be less than 2.5 volts. A lower mu is pretty easily accommodated, and the obvious candidates are the 6SN7 and 12AU7. These tubes will work in the stock design if the bias adjust trimpot is changed to 20K from 10K ohms.

Since you are running the amp at lower 300B current, there is more current available for the driver, so tubes with moderately lower plate resistance are possible if the board currents are adjusted. But the board's small heatsinks can't take a lot more current. I would be reluctant to go much lower in current because of the Miller capacitance of the 300B. A current as low as 2.0mA should work (this would be about right for a 6072), though I am not confident it would sound as good.

I don't have detailed designs for other tubes at present.

 

[Paul Birkeland]

I thought you'd chime in and throw something out like an IT transformer with a positive voltage regulator biasing a 3C24 into A2 to drive the 300B...

 

[Paul Joppa]

@PB - nah, this is a Paramount variation so I'll keep it similar in character.

Should have mentioned the Russian 6N1P and the 5965 (a favorite of the late John Camille) as also-suitable tubes. These were considered in the design. But we have in the past experienced early fading of sound quality from many 6N1Ps, and the 5965 is not quite as linear as the 5670, based on the published curves. We did not listen to any of the alternatives in the Paramount design though.

 

[johnsonad]

Thanks Paul and Paul. The list is enough to keep me busy for a while.

I'll probably try the 6sn7 first and see if the gain is sufficient once the cabs are built for these new drivers.  If so, I've got a stash of 6J5's and may try the 6J6 as a regulator. 

 

[johnsonad]

Would a 6C4 be an option as a driver? I don't have access to the curves but it's supposedly similar to the 6J5 though max 250v on the plate. How about a 7A4?