Replacing a 12AU7 in a cathode-follower.

[beppe61]

Originally posted by cerrem
The 5814A and the 12AU7A are the same tube...they come off the same factory run .....
The difference is that the 5814A tubes usually measure stronger than 12AU7A spec and are seperated and sorted....
They are the cream of the crop 12AU7A tubes...
SO this means the gm will be higher and the plate resistance a bit lower which makes for better low frequency reponse compared to a plain 12AU7A...
The spec sheet for the 5814A read exactly like a 12AU7A but that is just for satisfying MIL requirements...
in reality they measured better than the data sheet...
Chris

Dear Mr Chris, thank you very much again for the very kind and precious advice.
They are the cream of the crop 12AU7A tubes...in reality they measured better than the data sheet...
Assuming that the stage being that of a commercial unit is optimized, my idea would be to try, as last attempt, a nice 5814A (suggestion of any particular brands ?).
Then if the sound will be still not convincing I would start with a DIY project about a ECC88 cathode-follower.

Thank you very much again.
Kind regards,

beppe

[SY]

Chris, did you note the 5k load?

[beppe61]

Dear Sirs,
an old review of the preamp in question:
http://www.eminentaudio.co.uk/Fronti...ews/Micro2.htm

In the specifications at the end of the page I see 500 ohm output impedance , maybe just too much for a 5 K load.
Anyway, before throwing it away, I would strongly like to try a 5814A tube in the buffer position in place of the 12AU7.
I am optimistic.
Thank you very much to all of your Sirs for the extremely kind and precious advices.
Kind regards,

beppe

[kevinkr]

Regardless of source impedance the 12AU7A is not going to do that great a job driving a 5K load impedance, although in this case the power amplifier's high input sensitivity of 0.5Vrms will help. I would expect very mediocre linearity at best.

For an application like this the 5687 would be much better, but the much higher filament and plate currents required would probably make it a non-starter in the Croft line stage. You could however build something using a 5687 to replace the Croft..

Your low frequency corner at 3.2Hz seems low enough, but even here a lower corner can't hurt in terms of phase shift, and could improve the subjective bass quality of the pre-amplifier. I'd try increasing the capacitance further and seeing whether or not there is an audible improvement.

[beppe61]

Quoting Mr Kevin

Regardless of source impedance the 12AU7A is not going to do that great a job driving a 5K load impedance,
although in this case the power amplifier's high input sensitivity of 0.5Vrms will help.
I would expect very mediocre linearity at best
Thank you very much Sir for your very useful advice.
It seems to me a game lost in advance.
I am thinking to going directly with a ecc88 buffer instead.
From what I read a much more powerful driver.
After all these amps are a very tough load.

For an application like this the 5687 would be much better, but the much higher filament and plate currents required would probably make it a non-starter in the Croft line stage.
You could however build something using a 5687 to replace the Croft..
Thank you for your advice.
But I would like to stay with a much more common tube, like ecc88.
By the way I see that the last preamps from Audio Research are getting extremely favourable review.
They use a 6h30 tube, a stronger version of the ecc88.
Do you know it ?
Maybe one of this used as a buffer could drive even my amps.

Your low frequency corner at 3.2Hz seems low enough, but even here a lower corner can't hurt in terms of phase shift, and could improve the subjective bass quality of the pre-amplifier
I'd try increasing the capacitance further and seeing whether or not there is an audible improvement

Thank you very much again but it seems to me that this ecc82 is not the right tube for my purpose.
I am getting weak bass even in the 100 Hz region.
The piano is weak, drums are weak without the right impact.
I am sorry because all in all the sound is nicely natural and not fatiguing. But also boring and not emotionally involving.

Thank you very much again.
Kind regards,

beppe

[SY]

A single ECC88 as cathode follower will drive a 5k load, and do it far better than a 12AU7-oid. But the distortion performance will not be optimal. A 10k load would be kinder. Hmmmm, what to do, what to do...

Here's what: if you parallel two section of the ECC88, it's the same as running a single section with a 10k load. The distortion performance will be pretty good, too- with a 10k plate load, 8mA current and 2.5VRMS out, an ECC88 shows about 0.2% THD as a grounded cathode amplifier, dominated by second harmonic. So the distortion of the same tube and same operating conditions as a cathode follower will be divided by the gain; in this case, gain is about 25, so the distortion of a cathode follower with that load will be 0.2%/25 = 0.008%. Not bad at all!

The main precaution to doing this is that each section of the ECC88 will need to have its own grid-stopper resistor- compared with 12AU7, the ECC88 has a greater tendency to oscillate, but it's easy to handle if you know this going in. With separate 1k resistors wired right at the grid pins (old fashioned carbon composition do very well in this spot), you'll probably have no trouble at all.

[beppe61]

Originally posted by SY
A single ECC88 as cathode follower will drive a 5k load, and do it far better than a 12AU7-oid. But the distortion performance will not be optimal. A 10k load would be kinder. Hmmmm, what to do, what to do...
Here's what: if you parallel two section of the ECC88, it's the same as running a single section with a 10k load.
The distortion performance will be pretty good, too- with a 10k plate load, 8mA current and 2.5VRMS out, an ECC88 shows about 0.2% THD as a grounded cathode amplifier, dominated by second harmonic. So the distortion of the same tube and same operating conditions as a cathode follower will be divided by the gain; in this case, gain is about 25, so the distortion of a cathode follower with that load will be 0.2%/25 = 0.008%. Not bad at all!
The main precaution to doing this is that each section of the ECC88 will need to have its own grid-stopper resistor- compared with 12AU7, the ECC88 has a greater tendency to oscillate, but it's easy to handle if you know this going in.
With separate 1k resistors wired right at the grid pins (old fashioned carbon composition do very well in this spot), you'll probably have no trouble at all.

Thank you very much Sir for the kind and valuable advice.
So the best solution would be parallelling two section of a ECC88 per channel .
I understand that 5K are indeed a very tough load for a tube without recurring to expensive coupling transformers.
What is your opinion of the 6H30 tube used as cathode follower ?
Or better, what would be in your opinion the ultimate cathode-follower (if such tube exists of course) ?

Thank you very much again.
Kind regards,

beppe

[SY]

I've never used a 6H30 so I can't comment.

My "ultimate" CF would not be much different- I'd put my attention toward raising the load impedance.

[beppe61]

Originally posted by SY

I've never used a 6H30 so I can't comment.
My "ultimate" CF would not be much different- I'd put my attention toward raising the load impedance.

Thank you very much again.
I have no question left.
By the way I will try to get the schematic at least of the input stage of the amps in order to check if it is possible to raise that 5K very low value.
A design choice that I do not understand completely.
It makes the matching with a preamp a quite difficult task.

Thank you very much indeed.
Kind regards,

beppe

[billbo]

If you want very low O/P Z, Try an IRF710/IRF720. It is good for up to 100Khz bandwidth and you can use the same power supply (up to 300 volts) with little mods. O/P Z about 25 ohms. Fets are the next best thing to tubes. I have a preamp that uses 12AU7 as an amp/cathode follower and it works very well. The only downfall is that it is a little noisy (shot noise) which I think it is inherent in these tubes since they are switching tubes. I would like to substitute to an ecc88 to see if the noise level reduces. One job
for me to do on a cold winter's day.
Regards
Billy...