Pentode current sink

[ABJ]

Hi All (first post),

Going insane trying to work out how to set up a pentode differential amp stage (6AU6) with a purely pentode current sink in the tail (most likely 6EJ7) - I'm emotionally alergic to silicon. My difficulty is primarily in understanding how to force the division of HT across the valves to give sensible operating points, particularly for the 6AU6's.

If for arguments sake if I have B+ of 300V and B- of 250V how can I force say 300V across the diff amp, leaving 250V for the current sink - or better yet for a more general case 350V across the 6AU6's and 200V across the 6EJ7.

I can cope with the process if the diff amp is made up of triodes since the current and grid bias along with the load force the triode part leaving the pentode sink to use whatever voltage is left (fine by me!).

Of course a heads-up to a really good source of understanding would be terrific (I have Morgan Jones V3 but there isn't anywhere near enough detail therein). I have searched the forum but couldn't find much to cover this case....

Thanks,
Andrew.

 

[bobhayes]

Hi,

I'm no expert but wouldn't you simply choose all the ideal operating points for your upper valve, find the total quiescent current flowing at your chosen operating point, then design your constant current sink conditions to meet this current?

Say you design the upper valve operating point at 300v at say 10mA through each side of your diff. pair as normal, then design current sink to sink 20mA?

 

[Sch3mat1c]

Going insane trying to work out how to set up a pentode differential amp stage (6AU6) with a purely pentode current sink in the tail (most likely 6EJ7) - I'm emotionally alergic to silicon.

Welcome! I get a feeling we'll get along well..

My difficulty is primarily in understanding how to force the division of HT across the valves to give sensible operating points, particularly for the 6AU6's.

Bias the grids with a voltage divider from B+. The cathodes will then appear as cathode followers, abruptly setting the current levels. The top tubes are in essence variable CCS's, set by the master CCS in the tail.

If for arguments sake if I have B+ of 300V and B- of 250V how can I force say 300V across the diff amp, leaving 250V for the current sink - or better yet for a more general case 350V across the 6AU6's and 200V across the 6EJ7.

Simply run grid leaks to ground.

Unless you need the zero grid voltage (say for direct coupling) or an extreme amount of drive, you can do this easily from just the +300V. Negative supplies are extra hassle to me...

I have searched the forum but couldn't find much to cover this case....

Because no one's stupid enough to use pentodes standing on a long tail. I personally would love to, and maybe add a current mirror at the top - get SS style mega-gain and thus linearity (after applying NFB, that is!), but pentodes have the unfortunate screen grid to contend with. Screen current can easily be half plate current in these types, and since one controls the other (as plate voltage changing its own current in triodes), you could get some nice problems.

It's not impossible, but it won't be easy perfecting it. I wonder if it would be possible to add a variable current sink, proportional to the screen current, parallel to the tail. Thus the master CCS controls plate current, while another in parallel sinks screen currents.

Say, what's your intended application anyway?

Tim

 

[SY]

ABJ- I'm emotionally allergic to small-signal pentodes!

In any case, as others have said, design the pentode diff amp to give you the operating point you want, ignoring the CCS. Pretend that the cathode voltage will work out OK- after all, it will only be a couple of volts above the control grid no matter what.

Then design the CCS with a trimmer in its cathode circuit so that you can adjust the current to hit the desired operating point.

 

[abj1]

Mmmm, probably is a foolish idea to try to put a pentode diff amp on top of a pentode sink but... I do like pentodes when used appropriately.

My intended application is a 3 stage all differential amp and my real intention is to get some experience with current sinks in the tails (have used classic long-tailed pairs many times before). Plan at this stage is pair of 6J6's on a 6EJ7 sink as input and phase splitter, RC coupled to pair of 6AU6's on another pentode sink (?6EJ7 or 6BX6) and RC coupled to output stage of 6DB5's (may well go all out and sink these too - haven't decided yet). The 6AU6 stage will use an E-linear configuration (my current amp uses it and it is very pleasing (6F8G LTP to 6K7G diff biased away from remote cutoff to 6V6G in E-linear config.) - the high output impedance of the pentode diff seems useful with this type of local partial feedback. Of course I always use VR tube regulated screen supplies - really tames the much-maligned pentode!

My confusion probably comes from thinking about it too much but let me set an example. If I decide to aim for 6AU6 with 150V on screen grid and operating conditions of 30K load resistor with the load line running from 300V, 0mA screens to 0V, 10mA and bias grids at -2.5V for an anode current of 4mA per valve. I now set up a 6EJ7 for 8mA total current (ignoring the screen currents for the sake of simplification). Looks good with the sink dropping 250V. Now even if I set the 6AU6 grids at 0V, what is to stop the loadline from drifting left or right? Right I'd not be too worried about (ignoring max anode dissipation for now) but left is a move towards rising 3rd harmonic distortion.

As the plate curves are parallel to the voltage axis it seems to me that there is not a unique solution to the position of the load line (unlike with triodes) and therefore anode voltage may drift alarmingly as the sink gives or takes up the slack. Put another way setting the 6AU6 current and plate load doesn't seem to pin down a particular location for the loadline.

I hope that I haven't written myself into a corner although it wouldn't be the first time... I would very much appreciate being put back onto the straight and narrow and having what I hope are my misconceptions pointed out! And apologies for the long post.

Cheers,
Andrew.

 

[abj1]

Oops - the trouble with composing a long post - missed SY's reply and don't wan't to appear too bullet headed. I will give it a try, as suggested, ignoring the sink and just see if it really does fall into place

I'm still a bit lost on how it works though (guess that probably means I should stick to triodes, sigh)

Thanks,
Andrew.

 

[SY]

It works by (shhhhh!) feedback. The plate current is set by a (presumably stable) screen grid to cathode voltage and the local feedback controlled control grid to cathode voltage. The plates are fed by resistors, so Ohm's Law will set the plate voltage.

That 2.5V bias voltage is pretty negligible in the big picture so will self-adjust, as in any CCDA.

 

[Gary P]

Using a pentode as a CCS for a differential amplifier stage isn't that crazy.

I think your making things more difficult than they really are. When dealing with either a long tail pair or a CCS loaded differential stage (they are really the same thing... just different performance levels of CCS action) the grids of the differential stage are the reference point. If you are using split supplies and grid resistors referenced to ground the cathode voltage with be just a couple of volts positive.

As the current from the tail CCS is varied the grid to cathode voltage of the differential stage will vary slightly. The actual grid to catode voltage of the differential stage can be roughly determined from the operating curves.

The plate voltage of the differential stage will be determined by the tail CCS - screen current/2.

Keep in mind that when calculating the cathode resistor for the CCS pentode you will need to include the plate current for both halfs of the differential stage, the screen currents for both differential pentodes and the screen current of the CCS pentode. All these currents will be flowing through Rk of the CCS pentode.

A usefull trick for balancing the differential stage is to include the ability to independently adjust the screen voltages. When the screen voltage is varied on a pentode that has a CCS on the cathode all that changes is the grid to cathode voltage. Increasing the screen voltage will increase the grid to cathode voltage as the pentode adjusts it's grid bias to meet the needs of the operating point.


Hope this rambling helps some...

Gary

 

[abj1]

Thank you all for the excellent clarification - not so confused now!

I usually insist on using VR regulation of screen potentials but it sounds like it might be more useful in this case to use a potential divider or series resistor to allow for individual adjustment within each differential pair (perhaps even including the 6DB5 output stage) - I will give this serious consideration since an exploration of balance in diff stages is the major reason for this project (well maybe to improve the sound from our TV as well.....)

Just to extend my understanding a little, is it viable to reference the grid leak resistor for each valve in the diff to a negative potential to increase the available voltage drop across the diff stage? Should this be done with a potential divider and does the return of the signal through the potential divider affect the way it works (or sounds)? Intuitively I would imagine that returning the grid leak to ground would be the way to go but if I wanted to steal a few volts from the sink pentode.....

Ta muchly,
Andrew

 

[SY]

Sure, you can do that as long as you capacitively couple the input and take into account the new diff amp cathode voltage when designing the sink. You can use a divider since the grids won't normally draw current.

Or you could just increase the B+.

All this assumes that the resultant voltages will be within the maximum ratings for the tubes.

 

[abj1]

Thanks SY

I really would prefer a little more voltage on the diff end (better operating points.....) and a little less across the sink - especially across the 6J6 splitter stage. Should be able to get good results from the sink at 120 to 150V across it so it makes sense to allow for this in the design rather than just take the simple option and return the grid leaks via 0V. (Can't really increase the B+ with the trannies I have available for this project).

One more question - if I raise the grid potential above B- for the sink with a potential divider - does this allow me to use a larger cathode resistor and increase the impedance of the sink?

Thanks all, I feel much less confused now and I can begin to put together a final draft of the circuit over the weekend.....

Just love this hobby and the support available on this forum.

Cheers,
Andrew.

 

[SY]

Hmmm, that should work at the cost of a bit of compliance. As a practical matter, getting the impedance ultra-high doesn't gain you much since you end up being limited by capacitance at the cathodes (e.g., cathode-to-heater as well as stray).

Gary P has done some excellent work on current sources- check out his website.

 

[abj1]

Pretty much got this all under control now - thanks everyone.

Question for Gary P I guess as he alluded to it in his post - is it really necessary to send the current for the diff amp screen grids through the CCS. Normally with pentodes (both output and small signal although CCS-less to date) I generally like to use a VR tube to regulate the potential (IMHO this really sorts pentodes out and makes them behave well). Would it be reasonable to do this with a sink in the tail and remove the screen currents from the picture (ie connect the cathode of the VR to 0V rather than using a cathode reference as would be more typical with a normal series resistor and bypass arrangement). The one disadvantage I can see with this would be losing the ability to easily adjust the balance within the pair by adjusting the screen potential. On the upside would be consistent screen postential with changes in B+ (our line voltage varies from around 250V to less than 230V when the whole suburb uses their airconditioning on hot evenings).

Is it also possible to similarly fix the screen potential of the pentode sink (and thereby make the the sink current much less sensitive to line voltage shifts)? Or should I stop asking and just try it - as I will have plenty of current available to 'waste' with shunt regulation.....and just love a tweak or three

Cheers,
Andrew.

 

[Gary P]

Here is an idea that might work well. It's on my project list but hasn't been tried yet.

Set the VR tube up with the cathode on the cathodes of the diff pair, the anode to the screens, and feed the VR tube from a good CCS. Configured in this arrangement the sum of the screen and VR tube current is always constant. The main tail CCS is now set for the current drawn by the diff pair + screen circuit current.

The advantage is the screen circuit current is constant and won't upset the differential pair's operating point. The down side is you need another CCS in the circuit.

For balancing the circuit you can use some series resistance between the VR tube and screens.

With the series resistance on the screens the thought comes up about linearizing the transconducance of the diff pair pentodes. Pentodes, like triodes have increasing transconductance with decreasing grid bias (increasing current). In a pentode when the plate current increases, the screen current will also increase. With series resistance in the screen lead this will lower the screen voltage slightly as plate current increases. Decreasing the screen voltage will decrease the transconductance. With the right amount of screen resistance one should be able to get the overall transconductance to be very linear over the operating current range of the input signal. Another form of local feedback.

The series screen resistance is in the Tabor and 47PP amplifiers but the screen circuit is not running in constant current mode. None of has not been optimized in any way. Just another item on the list of things to optimize...

On the tail CCS you could do the same screen VR tube arrangement but feed the VR tube from a "long tail" resistor to B+. This might get the current change from line voltage fluctuations down far enough, or you could feed it from a CCS also...

Gary

Gary P's DIY page
Link to Tabor schematic. The schematic is not linked from the web page yet.

 

[abj1]

Thanks Gary for some very interesting food for thought.

Particularly like the idea of the series resistance to balance the pentode diff pair via the screen potential with VR tube regulated supply. And hadn't even considered the likely benefits from the partial feedback in linearising the transconductance with the appropriate resistor values (hours of tweaking here alone ) - brilliant - would never have thought of that in a million years. What a bonus - will definitely explore the possiblilities with this.

Thanks also for the opinion on the possibility of controlling the screen current for the sink also.

Cheers,
Andrew.