Do you still need matched tubes if running a CCS ?

[Nihilist]

My heading pretty much sums up my question, but here's the details :

PP triode strapped pentode/tetrode tubes

parallel tubes (a quad per channel)

LM317 run as CCS in cathode of EACH tube



If the point of matched tubes is current draw being the same, then isn't this accomplished via the CCS supplying each tube ?

Is it more than that ? Does tube matching have to do with the individual tubes having differing amounts of gain at different current levels ?

Do I need to have matched tubes AND CCS for best performance ?

Thanks....................Blake

[Matt.B.H.]

Great question,
In general I would say no, no need to go mad matching tubes. They are kept in class A by the CCS so you should be fine. Of course getting wildly mismatched tubes to work will be hard but you could even get two totally different types to work eg a 6L6 and an 807 (both quite similar) or even an EL34 with 6L6/807.
Matched tubes is a bit of a misnomer anyway. What did they match them with/at?....... or do they look the same? Some sellers claim they sell matched tubes but they are not matched for the parameters in your amp and wont be after several hours use.
Im a bit torn at the moment with my 6L6ish amp whether to go CCS class A or mental AB2 with bias pots for each valve. A2?
Cheers Matt.

[rotaspec]

I just finished the first iteration of my EL84 PP UL amp, and it has a real mixture of old power tubes. I call it the United Nations amp The next thing I was going to do was try 317 CCS on each cathode to see what effect it has.
It worked great on a little Chinese 6P1 PP (not UL) amp, but as I don't have a tube tester I couldn't tell how mismatched the original tubes were. However, one was odd enough to go into runaway with 270 ohm cathode resistor bias if the B+ rose over the 250v tube limit.

I would love to see some other expert opinion on "flattening" tubes with CCS bias.

[kavermei]

I don't understand the whole CCS-in-the-cathode-of-a-common-cathode-stage thing. You would want a constant voltage there. If you put in the cathode a true CCS (CCS behavior over the entire audio band), then the cathode voltage will just 'track' the grid voltage. Ugk and Ip remain the same, so the AC plate voltage will be the same as the AC grid voltage The end result is a stage with unity amplification, a "plate follower" as it were.

Can someone, please, enlighten me or tell me what is wrong in my above deduction? Or is it the point that we must use CCSes which are CCS for DC only and bypassed for AC?

Thanks
Kenneth

[cerrem]

CCS is usefull for when you have a differential pair amplifier..
The CCS idealy is for DC biasing purpose...but should idealy be as high impedance as possible, as if it weren't there...otherwise, it would act as an AC load which would be not ideal...
The differential coupling takes place in the tail....the follower output will drive the cathode of the other side...you want ALL the AC current to be used to drive the other side of the pair....if the CCS has low impedance it will shunt some of this AC current making a current divider, thus diminishing good AC coupling of the differential halves... Throw in some parasitics such as the CCS capacitance and you can see the high frequency pole introduced...

Having a CCS does not gaurantee or improve matching... The tubes still need to be AC matched for transconductance....A CCS will help in stabilizing the gm in that it reduces this variable's dispersion...

Chris

[kavermei]

Chris, yes but this is not about a long tailed pair with a common CCS, it is about a CCS in the cathode of every tube...

[Arnulf]

Originally Posted by kavermei

Can someone, please, enlighten me or tell me what is wrong in my above deduction? Or is it the point that we must use CCSes which are CCS for DC only and bypassed for AC?

There is nothing wrong with your reasoning. As I pointed out in another thread this is plain silly: CCS should easily have impedance in the megaohm range, right ? That just doesn't fit the U = R * I at required U for desired I for any sane combination of the two. Once bypassed, it is just a weird way of stabilizing voltage (bias).

If there was a real CCS in cathode part of the circuit and the output is taken from the anode, the amplifier wouldn't work, just as you pointed out.

[Merlinb]

Originally Posted by kavermei

it is about a CCS in the cathode of every tube...

In that case the CCS is bypassed with a cap, so it only controls the idle current, in which case you would still need matched valves for best performance.

[Nihilist]

Thank you , gentlemen !

I AM in fact talking about a differentially loaded (self phase splitting) output run Class A. Sorry I forgot to mention that.

What I am gathering from this is that if I use a single CCS in the tail, un-bypassed , that the tubes need not be matched if they are a new run from a single manufacturer , presumably close manufactured runs/tolerance ? This is because ALL current , AC or DC is loaded through the CCS ?

If I were to use a CCS in each tail, this would set DC bias/idle current only, and the bypass cap allows for AC current . In this case I would be better off to use a matched set that is matched for transconductance and not just idle current ?

Do I seem to have the gist of this ?

......................Blake

[cerrem]

If your refering to Class A P-P pair with a CCS in the tail... Then theoretically you do not need any AC bypass if you have perfect gm matched pair... This is because the AC currents, if perfectly matched, are 180 degrees of each other and would perfectly cancel.... This would leave a net of ZERO AC current in the tail... Since we do not have perfect matching, there will be degeneration associated with the AC imbalance...this is why a bypass cap my be added..
If it is a diff pair using one input such as a phase inverter, then you don't bypass, since any bypass will diminish the coupling that is critical...

Chris