wrote:

Hi Group:
I should have posted this before but here's a link to the actual circuit:

http://www.nashaudio.com/Public/12AX7_DiffPair_CCS.pdf

It will be the input/phase inverter section of an amp I'm building.
A differential driver will follow this stage and then the output section
(Kt88's)

The schematic above appears to be OK although an MJE340
used for only 2 mA of constant current seems a bit over the top
when a TO92 package smaller bjt or a j-fet would do.

The 5k R6 does not have to be in the circuit but it isolates the cathodes from
stray C in the bjt, and allows you to check the actual CCS current easily as
well as provide some
guard in case a tube shorts between anode and cathode, which wouldn't
hurt the bjt because there is the high value anode load RL there to limit
current.

With 50Vrms of signal out of both anodes, and only
say 1.4Vrms input to one side of the LTP, there should be about
0.7Vrms at the commom cathodes.
if you try to measure the signal voltage across the R6 5k,
you should find it to be an extremely small voltage,
so signal current in the 5k is low because below the 5k the
collector input resistance is very very high, many megohms.

Patrick Turner.



"Patrick Turner" wrote in message
...


wrote:

I'm working on a long tail pair phase inverter using a 12ax7 with a
current
sink
on the tail. I've noticed in some other designs (ARC) they use a resistor
~5k between the 2 tails
and the CC sink. What's the benefit of this resistor? I find very little
difference with or without it.

But what are you saying?

Are there two CCS, one for each 1/2 of the 12AX7?
If so, a 5k between each cathode acts like a current feedback R
to linearize the current change in the two tubes;
there is none in the CCS.

Have you checked the gains of the 12AX7?

What *exactly* is the circuit configuation?




It causes the CCS to work a little harder to make up for the series
resistance but other than that
I see no change in the bias or operating point.

CCS never "work hard"
There is no current change; the current is constant, so no
signal power is liberated in the CCS, only the DC power
of Vdc x Idc across the device.




There's very little written about using a CCS with long tails. I've
searched the web
searched the web, but nothing.
Any Ideas??

A CCS **is** a "long tail", because it acts exactly like a resistor of
huge
value to some mythical
very negative voltage, ( except for stray capacitances ).
A typical CCS would have a **real** finite R = 20Mohms, and if there is
5mA of
current,
then the CCS acts like 20Mohms commom Rk taken to -100,000 volts dc
supply,
which of course is difficult to arrange in a little box.

But you could have an LTP with each grid at 0V bias, cathodes at say +5V
bias
and
a 5mA common cathode current taken to -12V, or taken to some lower
negative
supply
as in the case at
http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm

Patrick Turner.


RonL