Yet another 12B4 line stage, or is the 12B4 better than the Grounded Grid.....

[ginetto61]

Correct! The O/P impedance of a cathode follower is approx. = 1/gm.

Hi ! thanks again ... so in the case of the 12b4 i see gm=6300 microohm
Then Zout = 159 ohm ... that is low .... very good indeed.

Adhering to the textbook 1:10 rule for a triode cathode follower is easy enough, when working into that 10 Kohm IHF number. A cathode follower O/P impedance <= 1000 Ω does the trick.

Yes and thanks again. I knew about the 1:10 rule for preamp and power amp impedances matching ...
now that Zout=1/gm formula that is very very important lesson to learn
Most triodes then can work as CF ... and actually i remember the dac of a friend with a single 12ax7 used as CF at the output ... one section per channel
He loves is dac deeply
I think that tubes and digital mate particuarly well
Do not know why but the end result is very very nice sounding

For instance, take the triode found in 6SN7s, whose gm is roughly 2.7 mA./V.
Wired as a cathode follower, the 6SN7 triode's O/P impedance is approx. 370 Ω.

Yes ... i see gm=3000 uohms > Zout= 1/0.003= 333 ohm This i have understood. Many tubes CF will be below 1kohm then

Whether or not such a cathode follower can charge and discharge the net capacitance of the downstream load is another matter

this is tricky .... you mean that the power amp input has a certaing capacitance that could be too high for a tube CF ? i know that a power amp input is not a pure resistive load unfortunately
Anyway at this point i have already the basics to try a simple CF and listen. I will do an report with some pics. The formula 1/gm has opened my eyes
The very nice thing is that a CF fwiu has also a decent psrr ... i could try to use one of those very handy dc dc step-up converter. If the noise will be too high i think to put a cap multiplier after the converter.
Thanks a lot again

 

[daanve]

I'd suggest to read a bit more deeply into the subject of cathode followers, as there persistently floats around some misunderstanding.
There is more at hand than simply "output impedance = 1/transconductance".
A very good read to begin with is:
The Tube CAD Journal, August 1999 pg 1
When taking the effort to extend this first read you will find a treasure of useful and reliable information and building suggestions on that same website.

 

[ginetto61]

I'd suggest to read a bit more deeply into the subject of cathode followers, as there persistently floats around some misunderstanding.
There is more at hand than simply "output impedance = 1/transconductance".

Hi ! so you mean that the formula is not correct ? I need a formula that gives me an approximate value of the output impedance ... also like +/- 100%
But thinking about it better, as I said a few posts before, a friend of mine has a dac that has a CF at the output made with an ecc83 ... I don't know how much the Zout is but I think if they used it it could be less than 1000 ohm? I now use for example a tube preamp that could have more than 5kohm of Zout with an amplifier downstream of 20kohm ... the 1:10 rule is not respected I mean

A very good read to begin with is:
The Tube CAD Journal, August 1999 pg 1
When taking the effort to extend this first read you will find a treasure of useful and reliable information and building suggestions on that same website

Thanks a lot for the link. I will look at it hoping that is not for expert designers. I need a book at secondary school level
Better if for dummies

 

[daanve]

"More at hand" is not the same as "not correct".
For a start: transconductance is not a fixed value.

 

[ginetto61]

"More at hand" is not the same as "not correct".
For a start: transconductance is not a fixed value.

Hi i see. But i am correct in thinking that almost any tube around can be when rightly used at least decent CF ? and another great thing of CF is the better power supply noise rejection ? in an anode follower the noise coming from the PS will mix with the signal at the output ... so great care must be taken in the psu design and construction
Anyway i do not see many CF designs ... why ?

 

[TG]

so you mean that the formula is not correct ?

It is almost correct - it's actually ra/(mu+1) rather than ra/mu (=1/gm), but since mu is often high enough, it's roughly the same.

The thing is:
a) as already mentioned by daanve, gm is not constant;
b) you also have to consider the current capabilities of the CF.

Say, you try to drive 10k resistive load with ECC83 CF biased at its "typical" 1.2mA. Output impedance is slightly over 600 Ohm - looks good at first glance.
Now you want to deliver 2Vrms (2.83V peak) to this load, for example. The current swing is +/-0.283mA - almost a quarter of a CF's quiescent current. That'll work, though such a current swing causes substantial gm variation, and hence the output impedance variation too. That leads to additional distortion, since now you basically have the voltage divider with the upper part that changes with the signal.

Now, can you deliver 10Vrms (+/-1.41mA swing) to the same load with the same CF biased at 1.2mA? No.

There are a lot of things to consider apart from the nominal output impedance indeed.

 

[ginetto61]

It is almost correct - it's actually ra/(mu+1) rather than ra/mu (=1/gm), but since mu is often high enough, it's roughly the same.
The thing is:
a) as already mentioned by daanve, gm is not constant;

Thanks a lot. I need to fix some simple concepts to proceed
One thing is now clear to me ... the CF is more indicated to drive a low Zin solid state power amp what i intend to use at least for now.
Fixed the CF solution next step to identify a good tube candidate

b) you also have to consider the current capabilities of the CF. Say, you try to drive 10k resistive load with ECC83 CF biased at its "typical" 1.2mA.

Thanks a lot again. I guess that you are reffering to the anode current Ia mentioned in the tubes datasheets ?
I see Ia=1.2 mA per ecc83 and instead about 11 mA for the ecc 82 and 12bh7 tubes And even 50mA min. for a 6H30
I guess these would be more suitable for a CF ?
Now i wonder why they have selected a ecc83 for the California Audio Lab dac ... a little on the weak side ? or with a tube preamp in mind ?

Output impedance is slightly over 600 Ohm - looks good at first glance.
Now you want to deliver 2Vrms (2.83V peak) to this load, for example. The current swing is +/-0.283mA - almost a quarter of a CF's quiescent current. That'll work, though such a current swing causes substantial gm variation, and hence the output impedance variation too.
That leads to additional distortion, since now you basically have the voltage divider with the upper part that changes with the signal.
Now, can you deliver 10Vrms (+/-1.41mA swing) to the same load with the same CF biased at 1.2mA? No.
There are a lot of things to consider apart from the nominal output impedance indeed.

Great, to sum up i have to look at gm and Ia
Many small power tube have Ia around 10 mA max ... for sure they are not good at current delivery ?
i am thinking ...
Thanks a lot again.

 

[rongon]

Ia of 10mA should be plenty.

If the solid state power amp in question has input resistance of 5k ohms (not uncommon nowadays) then following the 1:10 rule of source impedance:load impedance you can see that most any triode used as a cathode follower will qualify. You need an output impedance (Zout) of 500 ohms. But...

There is also slew limiting.

If you want the cathode follower to be able to drive long cable runs to a power amp input with low input resistance and high input capacitance, then the cathode follower will have to sink enough current to charge that C and R presented by the cable capacitance plus the load at the highest frequencies of interest (let's say 40kHz, to provide some headroom).

From slew rate explained | ken-gilbert.com

If the power amp has input resistance of 5k ohms and an abnormally high input capacitance of 300pF, and you have a 3 meter (10 foot) cable run from the preamp to the power amp, and let's say your cable is not so good and has 500pF capacitance, the cathode follower will have a 5k ohm + 800pF load to drive.

The formula for finding the current required from the cathode follower to drive the load is

I = C * 2 * pi * F * Vpeak

To cleanly swing 40kHz into that load at 2.828V peak signal level will take

800pF * 6.283 *40,000 * 2.828 = 0.57mA

The author of that slew rate article (Ken Gilbert) writes that "(Stephie) Bench points out (that) a nice headroom would be about five times this value, to provide a cushion for the large peak to average ratio of music."

0.57mA * 5 = roughly 3mA

So the cathode follower needs to have an output impedance of 500 ohms or lower, and anode current of 3mA or higher.

If 1/gm gives you your Zout (or close enough) then even an ECC82 should do.

1/0.002 = 500 ohms

Bias that ECC82 to Ia = 3mA or higher and that should just barely make it.

ECC88 would be even better (lower Zout, higher Ia). Or 6N1P, or 6N3P. You have many choices. Unless of course I am totally wrong and everything above is misleading. In that case, I hope someone steps in and sets me straight!
--

 

[ginetto61]

Ia of 10mA should be plenty.
If the solid state power amp in question has input resistance of 5k ohms (not uncommon nowadays) then following the 1:10 rule of source impedance:load impedance you can see that most any triode used as a cathode follower will qualify. You need an output impedance (Zout) of 500 ohms. But...
There is also slew limiting. ... --

Hi thank you very much indeed. I see much more clearly now the all issue
I think that i have now all the info i need to make some decisions.
I can say that ... it is tricky to get current from tubes ?
Anyway i am sorry to have derailed the thread a little. I will search for specific thread treating cathode followers and tube buffers in general
But as i said i was not aware of this complications. Better to use tubes with solid state amps that present an easier load to the preamp
Like maybe ss amps with fet input ? high Zin ?
I am looking for a more universal solution.
I thank you all for shedding some light in a very obscure topic for me.
Kind regards, gino