Need to find - Triode Va=500V, Mu=20

[thehoj]

I've been on the hunt for an IDHT triode, or a pentode that I can run in triode mode that will allow for a 500V anode voltage and have an amplification factor of close to 20..

The 6SN7GTB is close.. But Va max is 450V. I'm also considering the 6S4A which can handle 500V, but has a Mu of 16.2..
I've read that the 6SN7GTB can handle 500V without too much trouble.. But I'll be running it with a CCS anode load, so current will still be up there even when it's near the upper part of the voltage swing.

I've looked at a number of pentodes run in triode mode, but none of the ones I've found have a high enough Mu in triode mode..

Anyone have any ideas?
I found the TDSL Personal edition application, which allows for parametric searches, but I haven't found anything that would be suitable in its database either..

 

[Eli Duttman]

Excessive power dissipation, rather than "mere" volts is what causes destruction. The ECC99 data sheet gives a 400 V. max on the plate. However, the type can dissipate a bit of power. It might be worth trying the mu 22 ECC99 over voltage, but under its dissipation limit.

 

[artosalo]

...I've read that the 6SN7GTB can handle 500V without too much trouble...

If the triode should tolerate 500 V Ua, then you have some 1000 V as +Ub or how much ?

Could you describe your circuit a bit ?

 

[thehoj]

I want to be able to swing up to 400Vp-p.
This will be a driver for an 845 SET amp. The 845 is fed by a mosfet source follower, so I really just need to provide enough voltage swing to the gate of the mosfet.
As configured I will be able to push the amp far into class A2, which will require 350 - 400Vp-p.

I ALSO came into a good number of #27 tubes which I would like to use. These have a mu of 9. I'd like to be able to get close to the swing I need with 2Vp-p on the input to the driver. So I need something following the #27 that has enough voltage on the anode to swing 400Vp-p..
Realistically this means 450 - 500 V supply once you figure out actual swing based on the triode curves.

I can basically make it work with a 6SN7GTB, if I push past the listed max Va (while still respecting the max power dissipation)..

Would be nice to find something for that second stage of the driver that isn't maybe quite so maxed out though.

 

[pmillett]

Well, there is the 2C26... I don't have a full datasheet, but there is this: CV1759 / 2C26A . Mu is a little lower.

The 2C22 might also work, its rated for 500V in ICAS service. Both of these have plate and grid caps ("mickey-mouse" tubes).

Pete

 

[thehoj]

Well, there is the 2C26... I don't have a full datasheet, but there is this: CV1759 / 2C26A . Mu is a little lower.

The 2C22 might also work, its rated for 500V in ICAS service. Both of these have plate and grid caps ("mickey-mouse" tubes).

Pete

Hm.. I like that 2C22!.. I've always been hesitant to have high voltage outside of the chassis though.. Since there will be no cage or enclosure over the tube.. Might still consider it though.

 

[Tubelab_com]

The 6SN7GTB was rated for a plate voltage of 450 volts in vertical amplifier (TV vertical sweep) service. The vertical amplifier in a TV set IS a class A SE audio amplifier working into an OPT. It is just optimized for one frequency, either 60 or 50 Hz depending on which side of the big pond you're on.

It operates into an OPT meaning that the expected plate voltage can swing from near 0 to 2 X B+. This means that a plate voltage of up to 900 volts peak is expected. The tube is rated for up to 2000 volts of peak pulse voltage due to ringing transients in the deflection yoke / OPT combination during retrace.

The 6SN7GTA or 6SN7GTB should work well in your application, but if you need more headroom you can use the 6BL7, but the Mu is only 15.

I built an 845 SE amplifier with a CCS loaded triode for the driver. I used the 45 tube. It's plate voltage spec is 275 volts, but that is also into an OPT. I ran 500 volts of B+ into a 10M45 and directly to the plate of the 45. The voltage swing at full crank is about 400 volts P-P with the positive swing reaching 450 volts or so. The amp still has the same 45's in it that it was built with. There have been no issues there.

 

[kevinkr]

I used 6SN7GTB at high voltages > 400V in several amps and I experienced a significant number of failures over time.. (More than enough to indicate it was more than mere coincidence)

Given the swing required I would go for a choke, from the perspective of swinging 400Vpp, a choke load would allow you to do this comfortably on a supply of <400V, however you will be on the edge in terms of grid bias voltage vs required AC signal voltage on the grid; lower current at high Vp would be the best approach to get the dynamic range, but might suffer from reliability issues.

A triode connected EL34 might work better except for the low mu..

 

[Sch3mat1c]

The ratings are design-center values, meaning one normally expects variations due to line voltage and component tolerances. The absolute rating is usually 10-50% higher. Note also that the voltage rating is in DC, i.e., they expect that you'd use a transformer so that the peak voltage can be almost 450 * 2 volts (and in vertical output service, the peak is considerably more than that, with the consideration that the tube is fully in cutoff under that condition).

Are you saying your B+ is 500V, or that you intend to bias it at 500V, with the ultimate B+ being much higher?

If the former, you're fine, it doesn't matter in the least. If the latter, then your bias point is quite arbitrary within the available range, and you could set it for 400V instead

Tim

 

[thehoj]

The 6SN7GTA or 6SN7GTB should work well in your application, but if you need more headroom you can use the 6BL7, but the Mu is only 15.

If I'm going for a similar design to yours, is there any reason to be concerned about my driving stages running at somewhat low current? Shouldn't be.. right? If current is supplied via the mosfet source follower.. It's just about providing enough voltage swing isn't it?

I kind of wondered about that because you went with that 45 in your design.. Did you do that just because of the super linearity of the 45?

A triode connected EL34 might work better except for the low mu..

I know a lot more options open up if I can do with a mu of ~15 or less..
I'd just really love to work with these #27 tubes I got..

The ratings are design-center values, meaning one normally expects variations due to line voltage and component tolerances. The absolute rating is usually 10-50% higher. Note also that the voltage rating is in DC, i.e., they expect that you'd use a transformer so that the peak voltage can be almost 450 * 2 volts (and in vertical output service, the peak is considerably more than that, with the consideration that the tube is fully in cutoff under that condition).

Are you saying your B+ is 500V, or that you intend to bias it at 500V, with the ultimate B+ being much higher?

If the former, you're fine, it doesn't matter in the least. If the latter, then your bias point is quite arbitrary within the available range, and you could set it for 400V instead

Tim

I'm saying that my B+ is 500V, biased at -11V / 4mA. That puts the Q voltage at around 250V.. Though it would be with a CCS anode load, so when the voltage swings up to ~475V, the current will still be 4mA. Pd will still be safe @ 1.9W.

 

[thehoj]

I did find the 6CM7 which is a pair of dissimilar triodes in one envelope.. Section 2 has a mu of 18, max Va is 500V. I can't find a model for LTSpice though, so haven't done any simulations with it.
Biased at -12V / 10mA or 12mA it looks like it would work pretty good with CCS.. (though 12mA would probably be pushing the plate dissipation of 5.5W)

 

[MelB]

Have you thought of paralleling two sections of the 6SN7? You could run 6 mA through each side. ~3 watts dissipation per side.

 

[Tubelab_com]

is there any reason to be concerned about my driving stages running at somewhat low current?

No, as you said the tube provides the voltage swing while the mosfet provides the current. Pick the operating point that provided the most linear voltage swing over the range you need.

I'm saying that my B+ is 500V, biased at -11V / 4mA. That puts the Q voltage at around 250V.. Though it would be with a CCS anode load, so when the voltage swings up to ~475V, the current will still be 4mA. Pd will still be safe @ 1.9W.

If your B+ is 500 volts, and you are using a CCS load, the instantaneous plate voltage will never go over 500 volts, and in reality won't get that high because the CCS needs some headroom.

Plate dissipation in a tube is the measure of how much heat energy must be shed (dissipated into the surroundings) by the plate. The plate has considerable thermal mass compared to rate of change in the audio signal. In this case we only need to be concerned with the AVERAGE plate dissipation since the peaks are shorter than the time needed to heat the plate, and averaged down by the valleys. If you run 4 mA through the tube and your average (Q point) voltage is 250 volts, then Pd is 1 watt.

Peak plate dissipation is considered when the operating frequency is very low like in a servo driven shaker table that runs down to DC. Ever see a shaker table rattle an old tube type mobile two way radio at 1/2 Hz? Ours was solid state, but it had a program for simulating "military Jeep" that shook the whole building.

The 6SN7GTA (or B) has a plate dissipation rating of 5 watts per plate, or 3.75 watts per plate if both plates are run at max. 2.5 watts is a conservative place to run them, so I would find the point where the tube produces the lowest distortion in your circuit at the drive level required. The 6SN7 seems to prefer a bit of current.

After the amp is built, go back and tweak things to your musical tastes by listening. Maybe you set the CCS for best performance at 200 volts P-P, but your speakers see 1 watt most of the time. Sometimes better listening performance can be had by optimizing things for what you listen to and letting the absolute specs slide a bit.

I kind of wondered about that because you went with that 45 in your design.. Did you do that just because of the super linearity of the 45?

The real reason for the 45 was far simpler. I had built the "Poor Mans Ongaku" from the schematic published in an old Sound Practices magazine. It used a 7044 tube for the driver, which is not too far from a 6SN7. I did not like the way it sounded especially when pushed hard. Poor overload recovery and transient response was obvious on percussive music, so it was disassembled.

I already had a TSE with 45's in it and I liked the way it sounded, so I measured the plate voltage swing on the 45 and it was enough to feed an 845 or a 211 or an 833A, so I added a mosfer buffer between the 45 based TSE and the 845SE was made.

My original TSE is still being used. It has NX-483 tubes in it that were pulled from a scrap Spartan radio. Those tubes were installed in the radio in 1929 according to the sticker on the radio. The NX-483 is essentially a 5 volt 45. I run them at 325 volts and 30 mA in that amp. It produces 2 WPC and sounds excellent. The tubes are still going strong and are now at least 86 years old! I hope I make it that long!

The 45's in the 845SE amp are two different brands and were both used tubes, obtained from Ebay way back before the 45 feeding frenzy. I don't remember the exact current setting. They were set by listening to dynamic music and also testing the harmonic distortion with an FFT. I think they were in the 20 mA range, and there was several mA difference between the two channels.

I still have the amp, although it hasn't been powered up in several years due to it's ability to heat up my small room in Florida. I now have a 2000 square foot basement in a cold climate, so it will be rebuilt and given a new lease on life.

I don't plan on changing anything in the amp itself, but the power supply could use a rebuild, t was a Rube Goldberg design. The whole thing needs a serious safety upgrade. The plate terminal for the 845 is exposed and above the deck on the power amp. I know better, but now there are grandkids with curious fingers in the house.

I did find the 6CM7 which is a pair of dissimilar triodes in one envelope..

I haven't used that one, but I used the 6EM7 for a high voltage driver in a cathode follower amp. Mu of the second stage is quite low, but the first stage is quite high. I found considerable variation in total gain from tube to tube with those.

Another choice is the 6DN7. I built a headphone amp for a friend with them several years ago.

The advantage of the 6EM7, 6DN7, 6BX7, 6BL7 is that they all share the same pinout as the 6SN7 so you could try them ALL without too much trouble.

The 6SN7 was used for vertical sweep in early B&W TV sets. all the other tubes I mentioned were improvements over the 6SN7 for vertical sweep duty. Since the vertical section of a TV set is a linear amp, all of these have been used for audio applications.

I used a 45 in my design. These have become rare and expensive. I have found that the closest common tube to the 45 in sound characteristics is the triode wired 6V6. The Mu there is about 10 though. That's enough if you have good gain in the first stage.

 

[cogsncogs]

I did find the 6CM7 which is a pair of dissimilar triodes in one envelope.. Section 2 has a mu of 18, max Va is 500V. I can't find a model for LTSpice though, so haven't done any simulations with it.
Biased at -12V / 10mA or 12mA it looks like it would work pretty good with CCS.. (though 12mA would probably be pushing the plate dissipation of 5.5W)

Federico Scarpa posted this model some time ago here at diyAudio. I haven't tested this model, so...

* 6CM7 Federico Scarpa @diyAudio
*
* 6CM7 section 1
.SUBCKT K6CM7_1 A G K
+PARAMS: MU=22.33 ERP=1.5
+ KK1=2100  KP=150  KVB=24  vg0=-0.85 va0=0
+ CGA=3.8p CGK=2.0p CAK=0.5p RGI=1000

.param V_6={KP*( (1/MU)+((V(G,K)-vg0)/sqrt(V(A,K)**2+KVB**2)) )} 
******************************************************************
E8    8  0  VALUE={(V(A,K)-va0)/KP*LOG(1+EXP(V_6))}
Eam am 0 VALUE= {2*Pow(V(8),ERP)/KK1}
GA A K VALUE={V(am)}
D3 5 k DX ; FOR GRID CURRENT
R1 g 5 {RGI} ; FOR GRID CURRENT
Rak A K 1G
Rgk G K 1G
C1 G K {CGK}
C2 G A {CGA}
C3 A K {CAK}
.MODEL DX D(IS=1N RS=1)
.ENDS

* 6CM7 section 2
.SUBCKT K6CM7_2 A G K
+ PARAMS: MU=20.2 ERP=1.4
+ KK1=1000  KP=96  KVB=35  vg0=-0.4 va0=0
+ CGA=3.0p CGK=3.5p CAK=0.4p RGI=1000

.param V_6={KP*( (1/MU)+((V(G,K)-vg0)/sqrt(V(A,K)**2+KVB**2)) )} 
******************************************************************
E8    8  0  VALUE={(V(A,K)-va0)/KP*LOG(1+EXP(V_6))}
Eam am 0 VALUE= {2*Pow(V(8),ERP)/KK1}
GA A K VALUE={V(am)}
D3 5 k DX ; FOR GRID CURRENT
R1 g 5 {RGI} ; FOR GRID CURRENT
Rak A K 1G
Rgk G K 1G
C1 G K {CGK}
C2 G A {CGA}
C3 A K {CAK}
.MODEL DX D(IS=1N RS=1)
.ENDS

 

[thehoj]

Tubelab, I appreciate the detailed post. I had kind of wondered if the average plate dissipation worked like that with a CCS load, just because current isn't approaching 0 as the the voltage swings up.. But now that I think about it, conversely it also isn't swinging up really high as the voltage goes down, so using the Q voltage to calculate dissipation makes sense.

I guess if I'm dead set on using the 27 tube in my design, I'm kind of limited to the 6SN7 as a second stage.. It's not really looking like there are a lot of other options if I want to use a low gain first stage. Looking at the curves, it looks like maybe somewhere around 8mA with CCS, biased at -10 might be pretty linear and work well. At that current it looks like swinging between 0V and -20 on the grid would swing between 76V and 490V on the anode (414VP-P with 20Vp-p on the input to the 6SN7, which the 27 can do with pretty close to 2vP-P on it's input).. My Q voltage would be at 282V, so power dissipation would be about 2.25W per triode.

Since I'd like a bit of headroom for the CCS load (as you mentioned), I could run 510V B+ with the power supply I'm looking at putting together.

You don't think this would be pushing things too far with a 6SN7 GTA or GTB?

The 6CM7 is interesting, though I don't have any of those at the moment.
I've got 6BL7's, but the gain is a bit on the low side..

Any other suggestions for tubes to consider which are similar to the 6SN7 (primarily gain wise, and capable of 400Vp-p), but perhaps even more linear?

 

[Sheldon]

If you can use a little more gain, another option is a 6N7 or it's equivalents. http://www.mif.pg.gda.pl/homepages/frank/sheets/021/6/6N7.pdf

Any current between 1 to2 mA per tube (sections in parallel) will give you a very linear load line with a CCS, and easily swing 400V. As already mentioned, the max rating is for the quiescent plate voltage, not the voltage at the plate under signal swing.

Sheldon

 

[Chris Hornbeck]

McIntosh's venerable model Mc240 used a Telefunken 12AX7/ECC83 at 430 volts plate to cathode (as the followers driving the output stage) and original examples aren't rare. So there's some leeway in voltage ratings.

My 845 amplifier uses a Soviet-era 6SN7 with sections paralleled per channel, 25K resistor load from +700 volt supply. Same original tubes from 1994 on.

200 volts peak swing seems a little high. Type 845 at a typical 1000 plate volts and 80mA needs -160 volts bias, and so peak voltage swing A1.

All good fortune,
Chris

 

[thehoj]

If you can use a little more gain, another option is a 6N7 or it's equivalents. http://www.mif.pg.gda.pl/homepages/frank/sheets/021/6/6N7.pdf

Any current between 1 to2 mA per tube (sections in parallel) will give you a very linear load line with a CCS, and easily swing 400V. As already mentioned, the max rating is for the quiescent plate voltage, not the voltage at the plate under signal swing.

Sheldon

I did consider the 6N7, I like the look of this tube, but is there an issue with say mismatched sections? I don't know how I feel about paralleling sections.

For some reason I was thinking that with a CCS load that the max voltage rating would work differently, just because the current isn't going down.. I get it now though that it's the same as with a resistive load.

McIntosh's venerable model Mc240 used a Telefunken 12AX7/ECC83 at 430 volts plate to cathode (as the followers driving the output stage) and original examples aren't rare. So there's some leeway in voltage ratings.

My 845 amplifier uses a Soviet-era 6SN7 with sections paralleled per channel, 25K resistor load from +700 volt supply. Same original tubes from 1994 on.

200 volts peak swing seems a little high. Type 845 at a typical 1000 plate volts and 80mA needs -160 volts bias, and so peak voltage swing A1.

All good fortune,
Chris

Interesting about the ECC83.

As for the amount of swing, I will be running into class A2, pretty heavily, potentially up as high as +55V on the grid. Total swing of 380Vp-p.

 

[jane]

6S4A is close but a bit low on mu (16.5).

 

[Chris Hornbeck]

As for the amount of swing, I will be running into class A2, pretty heavily, potentially up as high as +55V on the grid. Total swing of 380Vp-p.

What plate load are you matching to? It's surprising to see a type 845 being driven into A2 at all, given their wide grid pitch.

All good fortune,
Chris