CCS vs mu-follower. What's the difference?

[Radames]

Hi,

I am trying to squeeze some information from all of you again.

The question is: what are the advantages of using Gary Pimm's CCS to load a tube compared to a mu-follower?

Gary's CCS shows a load of over 100MegaOhms, while a mu-follower could probably get up to 2Megas or may be a bit more. Is that the only difference?

I would like to hear your experience if you tried one, the other or both.

I was thinking on using a CCS to lead a 76 tube (input stage), capacitively coupled to a 211 grounded cathode (output stage). The change from 100Megas to 2 Megas does not change much the slope of the AC loadline (considering a 100K grid resistor for the 211).

Any thoughts and comments will be very appreciated.

Regards,

Rada

 

[SY]

Don't be too impressed with ultra-high CCS impedances. First, once you get more than a decade above the plate resistance, you've hit Diminishing Returns territory. Second, you'll find that stray capacitances start dominating, which powerfully levels the playing field.

 

[EC8010]

Once you exceed 50r_a there's no change in linearity. You'll find that Gary Pimm's efforts focussed on reducing the stray capacitance of his CCSs in an effort to maintain performance at high frequencies. It's then up to you to ensure that you don't throw away that performance by inadvertently adding some shunt capacitance.

Practically, not needing an elevated heater supply might be a powerful plus for using a semiconductor CCS. Of course, some people find it politically incorrect to mix sand with vacuum...

I haven't tried Gary's designs, but would expect them to be very good.

 

[rdf]

Does a CCS (as implemented by Gary Pimm) lower the stage's output impedance?

 

[Radames]

If you take the output from the mu-follower, yes it does reduces Rout.

My interest is, however, not in the Rout reduction but in the maximization of plate swing due to a flatter loadline. I understand that above 50*ra the improvement in distortion is little or none (I have done the simulated experiment in the computer).

I don't understand what the stray capacitances are and how do you add them or reduce them. Morgan Jones comment very briefly on them, but I don't fully understand where do they come from.

Thanks for the comments!

Rada

 

[Shoog]

Hi there,
This information might not be of any use to you at all but here goes.
I built a 807 parafeed amp and loaded the output stage with a TT21 CCS based on Garys straight Pentode design. Due to issues with the internal DC resistances I couldn't get it working as the TT21 never dropped enough voltage. I then went to a Mu follower and got this working by dropping 60V over the sensing resistors. Works fine, but due to the fact that I was running parafeed I couldn't cope with really low output impedances so I took the ouput from the plate of the 807's.
If you have spare voltage to throw away (as I did) I think a mu follower can perform really well.

I also read on TubeCAD that mu-follwers perform best into a fixed load.


As I said, probably not relevant to your situation.

Shoog

Shoog

 

[EC8010]

The stray capacitances are any capacitors down to ground from the lower valve's anode. If you look in the valve's data sheets, you will often find a value for C_out which is the capacitance from anode to all electrodes except control grid. Sometimes, rather than lumping these capacitances together, the data sheet gives individual capacitances. It is common for C_out to swamp the strays of a good CCS. However, it's easy to accidentally raise C_out. The CCS needs to be close to the anode pin so that it can use as short a wire as possible. Valve manufacturers tended to specify capacitances including the valve base - but didn't say which type. Use PTFE or ceramic in preference to phenolic. Another issue that people tend to forget is that although not bothering to bypass the lower cathode resistor in a mu-follower doesn't cause noticeable loss of gain at 1kHz, it raises r_a, compromising its ability to drive stray capacitance; this shows up as HF loss and can be highly significant if the lower valve already has high r_a (such as a 6SL7).

 

[rdf]

If you take the output from the mu-follower, yes it does reduces Rout.

Rada

My interest was avoiding the extra stage. Thanks Rada.

 

[Sch3mat1c]

Re: Capacitance

Don't forget our old friend Miller either. Remember it's not just grid loading but actual capacitance from grid to plate. Loads the plate too!

Tim

 

[EC8010]

Backwards Miller

Good point. It won't be much of an issue with a pentode, but could be significant with a triode even though it will reduced by a factor of A/(A+1) - (oh, wow).

 

[Radames]

Miller and friends

Oh!

And that's why a pentode is a better choice than a triode for the upper tube tube of a mu-follower?

Or is it just the high mu?

 

[EC8010]

The previous discussion was about the lower valve.

In theory, a pentode makes a good upper valve in a mu-follower. In practice, it's rarely any better than a good triode.