7802 twin triode

[Dr. Toobz]

There has been mention of the seemingly rare 7802 pass tube as a possibility for the output position in the Crack, an interesting one since the transconductance of the tube is much higher than a 6AS7 or even 5998 at 20,000 mho! This should lead to a lower output impedance, and possibly, a better damping factor on some "slow" sounding headphones (even my HD 650's seem to like being driven from a lower impedance source than a higher one, but perhaps that's a matter of taste).

There's a few of these tubes "out in the wild" but my concern would be the possibility of oscillation, as I remember PJ bringing up in some buried thread. With a high transconductance tube like this, would simply adding a grid stopper to the circuit prevent the possibly of the tube oscillating, and if so, how might I calculate the appropriate value?

 

[Paul Birkeland]

With a high transconductance tube like this, would simply adding a grid stopper to the circuit prevent the possibly of the tube oscillating, and if so, how might I calculate the appropriate value?

It's not a bad idea, anything from 200 Ohms to 1K should be acceptable. 

If you try it, be sure to let us know how you like it!

 

[Dr. Toobz]

I lucked out and snagged two NOS Tung-Sol 7802's on eBay for about $15/each - try that with a 5998! In any case, first impressions are very positive - huge soundstage, low distortion, and a very crisp sound, similar to the change I saw when going from a 6AS7 to the 2399 (e.g., 5988), but more pronounced. I can verify that the output impedance appears to be much lower than usual, as a pair of AKG's plugged into the Crack actually have bass, and a pair of Grados (32 ohm) play loudly and don't seem to have much different of a sound signature/balance when going between the Crack and an iPhone. (Normally, this is certainly not the case - they sound very strange when driven from 120 ohms). Coupling caps are 100uF JJ electrolytics, which I seem to like a lot, at least more so than the Dayton film caps I had in there previously. Sacrilege, I know....  ;D

Since these are supposedly NOS tubes, the transconductance has to be at or above 20,000 mho, which should mean about a 50 ohm output impedance in the cathode follower arrangement, correct? The 5998 has double the transconductance of a 6AS7 (e.g., 14,000 vs 7,000 mho), and the 7802 is almost three times the 6AS7. Even my 300 ohm Senns sound a bit better than usual, which makes me wonder if lowering the source impedance is either a) lowering distortion, which is already very low, or b) changing the damping factor and/or response curve of the drivers?

I am looking forward to seeing if this really makes a difference once my 600 ohm DT880's arrive, which I am going to give a try after not liking the 250 ohm "pro" variant with my pre-Bottlehead gear.

 

[Paul Joppa]

Just a detail - transconductance is a function of plate current and (to a lesser degree) voltage. At Crack currents (around 30mA) all of these tubes have more like half the quoted transconductance - the spec sheets always give gm at or near maximum current.

Nevertheless, they still rate in the same order.

 

[Dr. Toobz]

Ah, that makes sense. This must be how the "high current Crack" experiment (which I am now assuming is vaporware?) lowered its output impedance. If I were to somehow run the 7802 tube at 60mA (presumably by ditching the Speedball on the power tube, cap coupling the two stages and altering the cathode resistor value), wouldn't this make the amp truly viable for 32 ohm phones? I've yet to hear a transformer I've liked as much as a tube in OTL arrangement 

 

[Paul Joppa]

Ah, that makes sense. This must be how the "high current Crack" experiment (which I am now assuming is vaporware?) lowered its output impedance. If I were to somehow run the 7802 tube at 60mA (presumably by ditching the Speedball on the power tube, cap coupling the two stages and altering the cathode resistor value), wouldn't this make the amp truly viable for 32 ohm phones? I've yet to hear a transformer I've liked as much as a tube in OTL arrangement 

So far, none of us really knows how low an output impedance is necessary to properly damp most 32 ohm phones, so I can't really answer your question.

The higher-current Crack is still an active project; most likely as Crack II or something. The power transformer limits the available current, and so does the heat generated by more current through the cathode resistor. As I'm sure you have noticed, Crack already gets pretty hot. :^)

There is no need to give up the direct coupling; all of these tubes can carry quite a lot of current with only 50-70 volts plate to cathode. But more current means more heat from the power supply resistors as well as the cathode resistor; and lower voltage unless the power supply resistors are lower resistance. And both higher current and smaller resistors means more hum, so then you need another PSU filter stage or larger caps ....  as you can see it's not just a smaller cathode resistor or bigger Speedball!

 

[ice9mike]

I just purchased a Tung Sol 7802WA. I found that it worked in my Crack (with Speedball) without any modifications. Is there any risk in running this setup?

Sonically, it sounds clearer and louder into my DT880 Pros (250 ohm). Bass has more impact. I have run various 6080 (GE, RCA, Raytheon) and 6AS7G (RCA, GE), but the 7802 is by far my favorite power tube.

 

[Paul Birkeland]

What DC voltage is present on the O pads of the larger Speedball PCB?

 

[ice9mike]

I did the full voltage measurement using my Fluke 87III with an RCA 6080 and the Tung Sol 7802:

Terminal    6080    7802

SB-O1        105.4    83.4
SB-O2        104.9    63

1                77.5      76
2                168.3    176.4
3                0          0
4                168.8    176.6
5                79.7      78.6
6                0          0
7                102.6    66
8                0          0
9                104      83.3
10              0          0
11              0          0
12              0          0
13              170.2    176.7
14              0          0
15              189.9    195
21              208.7    213.4

A1              78.8      78.1
A2              0          0
A3              1.55      1.54
A4              0          0
A5              0          0
A6              76        75.4
A7              0          0
A8              1.55      1.50
A9              0          0

B1              75.6      75.4
B2              175.5    176.2
B3              104.2    82.2
B4              77.8      75.9
B5              175.4    176.2
B6              105.4    84
B7              0          0
B8              0          0

I would have thought that O1 and O2 would match better in the 7802 but they don't. O2 also oscillated between 56V and 65V. Does this indicate a problem?

Thanks.

 

[Paul Birkeland]

Looks good, the cathode voltages are a bit lower, so the dissipation across the Speedball heatsinks will drop a little.

 

[ice9mike]

Thanks PB for checking it out and replying late on a Saturday. This is a good community that I have learned a lot from by just reading old posts. I hope to add some wisdoms of my own. I just posted my solution to fastening the top plate to the frame. I think it works great, but there are some trade offs. The great things about this amp is that it is not limited to one design.

ice9mike

 

[Snarii]

An update on this ancient thread that might be of interest for anybody interested by a 7802.

I've just acquired a very nice 7802WB from a fellow forumer. He didn't have any problem with it, but on my stock crack+speedball it had an huge hum centered on both channels. I had to try something...
According to Paul advice a few posts ahead, I make a quick try with a 220ohm (1/4W that's only what I had on hand, no idea if it's suitable) on one grid of the 7802. Of the two, the pin 4 grid was the most accessible...
And voila ! the hum is gone, only very very faint distant hum. I of course intend to suit the symetrical resistor on pin 1, but that single grid stopper resistor was enough to do the trick, as if the oscillation was internaly induced in the 7802.
I still have to check the sonic impact of the modification on my other output tubes...

I didn't have my best input tube on for the first run, but as far as I can say, sonically the 7802WB is a very nice tube, very dynamic !

 

[hpjun]

I have the 7802 Tung-Sol tube and it sounds amazing.

Now I would like to get the output impedance down even further. I heard for cathode followers, lowering the cathode resistor will drop the output impedance and also the gain as well. But the crackatwoa is CCS, is there a resistor on the CCS that I can adjust to drop the output impedance and gain as well?

So far, none of us really knows how low an output impedance is necessary to properly damp most 32 ohm phones, so I can't really answer your question.

The higher-current Crack is still an active project; most likely as Crack II or something. The power transformer limits the available current, and so does the heat generated by more current through the cathode resistor. As I'm sure you have noticed, Crack already gets pretty hot. :^)

There is no need to give up the direct coupling; all of these tubes can carry quite a lot of current with only 50-70 volts plate to cathode. But more current means more heat from the power supply resistors as well as the cathode resistor; and lower voltage unless the power supply resistors are lower resistance. And both higher current and smaller resistors means more hum, so then you need another PSU filter stage or larger caps ....  as you can see it's not just a smaller cathode resistor or bigger Speedball!

Now that the Crackatwoa exists, I want to know if I can modify it for high current? Is higher current more linear?

 

[Paul Birkeland]

So, looking at the datasheet, running the 7802 in place of the 6080 will drop the output impedance to about 100 Ohms.  This is insignificant.  (The tube may still sound different/better, I won't argue that)

Lowering the cathode resistor value on a cathode follower changes the operating point, this is sort of like advancing the timing on an engine.  When you go too far, you get pinging and the engine explodes.

If you want to lower the output impedance, you can add loop feedback.

 

[hpjun]

would running 2 cathode followers in parallel work without messing up the amp on the crackatwoa?

Was thinking of something like this: maybe using 2 6080 or 2 7802 or 2 6BL7.
http://m.ebay.com/itm/1pc-Gold-Dual-6SN7-6BL7-TO-6AS7-tube-adapter-/201511879930


Also I am still interested in experimenting on the cathode follower current and resistor values. Goal is still to lower the output impedance more. If the cathode follower was operated at a higher current it would also have lower impedance right? Also If I could I would trade away gain for lower impedance I would do it. If I had to run 2 cathode followers I would do it. Just want to get the impedance below 50 ohms without global feedback or transformer. If you have any ideas let me know. I just wouldn't want to make it sound bad.

 

[Paul Birkeland]

To run two cathode followers, you need twice the high voltage current and twice the heater current. To do that, you need a second power transformer and another regulated B+ supply.

You can raise the current a bit, but you'll end up making compromises elsewhere that will limit power output, and you'll have to rebalance the shunt regulator for both current and thermal load, which is not a trivial process.

A transformer isn't the end of the world. A pair of 600:60 units of an appropriate size would be small and inexpensive. Four resistors for feedback will be even less expensive and easier to try.

You also should consider that you'll give up a fair amount sonically (in my opinion at least) by using paralleled cathode followers. This isn't a particularly well thought of method by those who have tried it.

If you want to make a cathode follower with no negative feedback and a single cathode flower triode per channel, the 6C33 would be a good place to start. Since you'd be running 150-200mA of high voltage current per channel, shunt regulating will be very difficult, and I would expect such an amp to not compare well to a Crack-a-two-a on higher impedance headphones.

What you would get is an otl amp with an output impedance well below 50 Ohms.

 

[hpjun]

Could I also bias the Cathode Follower with higher voltage to lower the output impedance? I was thinking of using a tube that runs higher voltage like an e80cc to bias the cathode follower with higher voltage, but I am not sure if its safe to do that or if it sounds good.

6c33 or 6c33c? Looking at the spec sheet and that's two triodes connected in parallel internally. So I would need a pair of 6c33c. The higher current doesn't affect the gain right? cathode followers have a gain less than 1 right? So current lowers impedance by increasing gm.

I think the sensible approach for me would be to increase the current in the cathode follower on my 7802 tube. I don't mind losing gain or power since I can barely turn the pot past 25% before it gets too loud. Not sure how to do this though, is it more complicated than changing a resistor?

 

[hpjun]

So I think I may be getting high frequency oscillation. Start up has a low buzz that goes away and likely there is out of band, high frequency noise. I don't hear it but I feel my ear stiffen up like my ear doesn't want to hear something an uncomfortable feeling. Also to further confirm I noticed the treble is a bit veiled probably from low level higher harmonic distortion products.

So, I think I'll put a grid stopper resistor. Is there draw backs to using a higher ohm resistor more than 1k, I assume it will only just atteunate the input to grid, so less gain and maybe some distortion if using bad resistors.

 

[Paul Birkeland]

Could I also bias the Cathode Follower with higher voltage to lower the output impedance? I was thinking of using a tube that runs higher voltage like an e80cc to bias the cathode follower with higher voltage, but I am not sure if its safe to do that or if it sounds good.

You'll need to study datasheets that graph transconductance.  The 7802 datasheet has such a graph on the final page.  Note the relationship of Eb (plate voltage) to tranconductance.  Also note the relationship between grid voltage and transconductance.

6c33 or 6c33c?

Same tube.

cathode followers have a gain less than 1 right?

Yes, don't worry about the gain of the cathode follower.  It's always less than unity, and it's generally very close to unity.

So current lowers impedance by increasing gm.

There is a governing relatinoship that mu=gm*rp.  In a cathode follower, we can throw 1 in for mu, so we get 1=gm*rp, or rp=1/gm (rp is output impednace in this case).  To maximize the transconductance of something like the 7802, you would ground the cathode and have the grid at ground potential, with about 100V on the plate, to get 30,000umhos of transconductance.  Unfortunately, if you do this, the tube will very likely draw a bit of grid current, and you'd have to have a very beefy circuit to drive the tube running in this manner.  (you'd need another cathode follower behind such a circuit to handle the grid current/impedance demands)

Without those helpful curves, you can look at the simple current/voltage graph with lines based on grid voltage and seek to pick an operating point where the lines are as steep as possible.

I think the sensible approach for me would be to increase the current in the cathode follower on my 7802 tube. I don't mind losing gain or power since I can barely turn the pot past 25% before it gets too loud. Not sure how to do this though, is it more complicated than changing a resistor?

Again, you can't just increase the current in the output stage of the Crack-a-two-a without a major redesign, which means starting over.  If you're going to start over, there are much better tubes than the 7802 to achieve your goal.  Also, increasing the cathode follower current will not affect gain, but you may end up inadvertently adjusting the grid bias the point where you get very early clipping of the first stage.