Alternative Tubes

[DaveP]

I was shocked recently by the price of common tubes (12AX7 ECC83 etc), even the Chinese versions are very expensive compared to what they used to cost.

The law of supply and demand is one thing, but blind prejudice is another. This spurred me to start an experiment which I had been meaning to do for some time. I set up an experiment to test my collection of B7G pentodes connected as triodes.

In order to compare like-with-like, I fixed the supply voltage at 250V and the anode/plate voltage at 100V with the anode resistor at 100k so they all took 1.5mA, which is a good ballpark current. I adjusted the cathode resistor to achieve this.

I fixed the input voltage at 100mV @ 1kHz and then measured the output with no load and then with 1M. With a good DVM the load can be taken as almost infinite (10M) with a 1M load in parallel with the 100k RA, the load becomes 90.909k.

The standard gain equations can be used to determine the tubes parameters.


Gain1= mu x RA/ ra + RA and Gain2= mu x RL/ ra +RL

Re-arranging the equations we get:-

mu = (Gain1 x ra + RA)/ RA and mu= (Gain2 x ra + RL)/RL

As they both equal each other we can solve for ra by treating them as simultaneous equations.

When we have ra, we just put that back into the first equation to find mu. gm is then mu/ra.

I put all the equations and data into Excel to make life easier and the results are shown below.



The results are arranged in descending mu order and they show that it's possible to achieve gains from an ECC83/12AX7 level through to an ECC82/12AU7 and everything in between. I am not saying that any of the tubes were set up in their optimum operating points, rather it was meant to show the relative differences between them. These radio frequency pentodes sell for $3-$4 apiece and they come in a useful range of heater currents too. I would encourage every tube builder on GroupDIY not to mindlessly follow the herd but to experiment with the vast number of available radio and TV tubes sitting on the shelves in tube shops. I haven't even mentioned the B9A range!

Best
DaveP

 

[ruffrecords]

The 6AU6 was commonly used in early tube mixers by the BBC wired as a triode. It has a handy pin out that allows you to ground the pins on either side of the heaters if you use the screen grid as a plate. I used these in my very earliest tube mic pre designs.

Cheers

Ian

 

[DaveP]

Yes, the pin-outs vary quite a bit with these tubes and some connect g3 to k and some don't, I had to keep my wits about me when changing tubes!

Maybe some of them are not so linear as others, I didn't check, but for something like a guitar amp that might be an asset soundwise.

Best
DaveP

 

[DaveP]

The other noteworthy thing is that the anode/plate resistance is lower with these tubes, so that they can drive lower loads than an ECC83.

best
DaveP

 

[ruffrecords]

The other noteworthy thing is that the anode/plate resistance is lower with these tubes, so that they can drive lower loads than an ECC83.

best
DaveP

Yes, they tend to have a higher gm which means rp tends to be lower (ECC83 gm is about 1.5mA/V IIRC). In the B9A group the EF86 makes quite a good triode.

If you want a B7G power pentode try the N78, mu of 24 in triode mode, rp a nice low 2.1K and gm a whopping 11.4mA/V

Cheers

Ian

 

[john12ax7]

From a guitar perspective, how do they distort? I've heard some 12ax7 alternatives in guitar amps and they can be interesting, but usually don't work as well in heavily overdriven applications.

 

[DaveP]

The Blues de Ville is a very high gain amp and a common mod is to sub a ECC81/12AT7 for an ECC83 to calm it down a bit. This works but the ECC81 is obviously not set-up for its optimum operating point. We can set most tubes up to operate in linear areas or in non-linear areas of the curves, this can be checked with bench testing. A common error guitarists make is to use banks of pedals in a high gain amp, the amp starts to squeal through capacitive coupling and they wonder why! Maybe the best way to set up a guitar amp tube is to put it on the edge of the linear region so it can drift into distortion (crunch) when the gain goes up?

Best
DaveP

 

[Rusan]

Wow, cool thread DaveP!

I've not delved really deep into the realm of "non-audio" tube types, but the few I've used have yielded some amazing results. The amplifier I use in my living room system is something I came up with off the top of my head, as an experiment in achieving the cleanest possible signal path and seeing what it sounded like. It's a triode-strapped EF86 DC-coupled to a 12B4A, which is a small (5 watt, IIRC) power triode that was commonly used as the vertical deflection amplifier in small CRTs back in the day. Back around 2005, I bought 8 "computer grade" NIB General Electric 12B4s that were branded Hewlett-Packard, for something like $35 with shipping.

The 12B4's characteristic curves are a beauty of linearity to behold, reminiscent of a 2A3 or 300B. I chose the EF86 driver for its low Gm and internal shield, with the hopeful intention of getting away with no grid stopper resistor. And, the 12B4's relatively low Miller capacitance allowed the triode-configured EF86's low current and not-so-low Ra to drive it satisfactorily into clipping at full bandwidth.

The direct signal path has zero caps or resistors, and is rock-solid stable and free from oscillations at any frequency that I could find. The sound is amazingly microdetailed and pure, as you'd expect. The caveat? It produces less than one watt per channel at its full, unclipped power, which isn't a problem at all with my moderate listening level and 99dB/1W/1m speakers.

 

[Rusan]

Yes, they tend to have a higher gm which means rp tends to be lower (ECC83 gm is about 1.5mA/V IIRC). In the B9A group the EF86 makes quite a good triode.

If you want a B7G power pentode try the N78, mu of 24 in triode mode, rp a nice low 2.1K and gm a whopping 11.4mA/V

Cheers

Ian

Wow, that sounds similar to EL84 and EL34 territory in Ra and Gm. Might be a "poor man's giant killer" in small guitar amps!

 

[ruffrecords]

Wow, that sounds similar to EL84 and EL34 territory in Ra and Gm. Might be a "poor man's giant killer" in small guitar amps!

Back in the 60s I seem to remember it was often used in single ended radio output stages.

Cheers

ian

 

[rock soderstrom]

The 12B4's characteristic curves are a beauty of linearity

Nice tube, you could certainly build a real good headphone amplifier with it.

 

[Rusan]

Yep, it would surely make a fine headphone amp! The only drawback I've found with them is that they're a bit microphonic, but placing a few silicone O-rings over the bottle works quite well.

 

[DaveP]

The other thing I want to do is measure the comparative noise of these tubes, any ideas on a suitable set-up?

best
DaveP

 

[ruffrecords]

The other thing I want to do is measure the comparative noise of these tubes, any ideas on a suitable set-up?

best
DaveP

About 20 years ago when I first got back into tubes I noise tested over 100 6AU6 tubes and later a similar number of 6CG7 (B9A version of the 6SN7). I built a simple rig in an aluminium box. I powered the heaters from 4XAA batteries. I had an input and output connector. The input one I used for measuring the gain and fitted a shorting plug to measure noise. I made the measurements on a Ferrograph RTS2 which has a 1Meg input impedance. For the 6AU6 I strapped it as a triode. IIRC the cathode resistor (bypassed) was 390 ohns and the plate resistor was 39K (I did not calculate these, they were given to me by a guy on rec.audio.pro). I used BNC connectors because that is what the Ferrograph used which meant I could also look at the output on a scope. This is when I discovered 1N4007 switch off spikes and also Scroggies article on the benefits of cascaded RC circuits for smoothing HT - important because both tube topologies have a poor power supply rejection ratio so a low noise HT supply is a must (3 stages of 1K and 470uF will give you nearly 150dB of ripple attenuation)

Later I discovered the mu follower and used a very similar set up to measure the 6CG7. Because the mu follower has a much lower output impedance you can feed it straight into an interface and use REW or a similar software package for measurement.

Cheers

Ian

 

[DaveP]

Thanks Ian, I knew you would have the answer!
best
DaveP

 

[DaveP]

Ian, can you remember the approximate noise level of the tubes you tested?

best
DaveP

 

[Walter66]

Some of those russian electron tubes beware, they are noisy as hell.

 

[ruffrecords]

Ian, can you remember the approximate noise level of the tubes you tested?

best
DaveP

A lot depends on how you measure. At the time I kept meticulous records of the measurements but I am blowed if I can find them now. It is not the sort of thing I would have thrown away so I will keep looking. On the Ferrograph the maximum sensitivity is -60dBV (1mV) and I remember the typical readings were around one tenth of full scale or less so in the 50uV to 100uV region. So with a typical tube gain of 30dB this represents and EIN of -110dBV which I seen to remember is fairly typical of a quiet tube.

Cheers

Ian

 

[DaveP]

Thanks Ian, that's the kind of level I expected.

Best
DaveP

 

[ruffrecords]

Thanks Ian, that's the kind of level I expected.

Best
DaveP

If you can it is worth attaching a small audio amp and speaker to the output so you can listen to the noise and also test easily for microphonics. It is also instructive to listen as the tube first warms up. The noise increases to quite a large value then, as the space charge develops it quietens down. You can often also hear pings where the internal metal parts heat up at turn on and cool down at turn off.

Cheers

Ian