In the past when I want ed a constant curent source for a tube differential
amp I used solid sate componets in the cathode circuit. For grins and
giggles this time I was thinking about going all tube but have never done it
before and cant seem to find a good reference with this type of circuit.
Best I can figure is to use a small signal pentode in the cathode and
regulate the screen voltage with a vr tube/s. Any thoughts on this or
direction to references would be appreciated.

"Jimmy" wrote In the past when I want ed
a constant curent source for a tube differential
amp I used solid sate componets in the cathode circuit. For grins
and
giggles this time I was thinking about going all tube but have
never done it
before and cant seem to find a good reference with this type of
circuit.
Best I can figure is to use a small signal pentode in the cathode
and
regulate the screen voltage with a vr tube/s. Any thoughts on this
or
direction to references would be appreciated.

Sounds like you have the right idea. It is not difficult to
calculate the anode resistance and voltage. Obviously it will carry
the current of both of the top valves. I would use a simple divider
for the screen, coupled to cathode or ground via a suitable cap,
before adding improvements.

There are a couple of examples in Morgan Jones "valve amplifiers". I
am sure I have seen *loads* on the net.

cheers, Ian

Ian Iveson wrote:

"Jimmy" wrote In the past when I want ed
a constant curent source for a tube differential
amp I used solid sate componets in the cathode circuit. For grins
and
giggles this time I was thinking about going all tube but have
never done it
before and cant seem to find a good reference with this type of
circuit.
Best I can figure is to use a small signal pentode in the cathode
and
regulate the screen voltage with a vr tube/s. Any thoughts on this
or
direction to references would be appreciated.

Sounds like you have the right idea. It is not difficult to
calculate the anode resistance and voltage. Obviously it will carry
the current of both of the top valves. I would use a simple divider
for the screen, coupled to cathode or ground via a suitable cap,
before adding improvements.

There are a couple of examples in Morgan Jones "valve amplifiers". I
am sure I have seen *loads* on the net.

cheers, Ian

Ian is correct about Morgan Jones, who uses two
transistors in series to get a very near perfect constant
current source, ie, a source of current equivalent to
a near infinite value of resistance.

But some CCS tails used with a twin triode LTP
used a pentode, say a 6AU6 with its anode connected to the
commoned cathodes, its screen fed from a B+
large resistance, but bypassed well with a C to the cathode, its cathode
taken via some
resistance to say -200v, and the grid taken
to say -100v, from a resistance divider, with a large value
electro between it and the -200v.

For every tube, U = Gm x Ra
In the case of a pentode like the 6AU6 with Ra = 0.5M at 5 mA,
and Gm = 0.0039A/V, the U will be 1,950.

let us suppose the 6AU6 cathode Rk is 22 k, then the effective
value of Ra looking into the 6AU6 anode
will be Ra + [ (U + 1) x Rk ], which in this case is
500k + 42.9M, or around 43M.

Such a large value of LTP tail impedance will act to
give good balance of output voltages from the LTP,
even with wildly mismatched triode halves, and this balance become
determined chiefly
by the matching equality of the RLs used in the LTP.

I have sometimes used an MJE 340 as a CCS, with a 4.7 k emitter R,
and although it isn't correct to use U, Gm and Rc with transistors,
one does get an extraordinarilly high collector input impedance
or similar largeness compared with a more costly pentode,
and there is no screen voltage and bypass to worry about.
The transistor is like a pentode with vastly higher Gm,
and lower Ra, so the "U" is still very high.

The maths are complex to the beginner, but can be found in the old text
books.

Patrick Turner.

"Patrick Turner" wrote in message
...


Ian Iveson wrote:

"Jimmy" wrote In the past when I want ed
a constant curent source for a tube differential
amp I used solid sate componets in the cathode circuit. For grins
and
giggles this time I was thinking about going all tube but have
never done it
before and cant seem to find a good reference with this type of
circuit.
Best I can figure is to use a small signal pentode in the cathode
and
regulate the screen voltage with a vr tube/s. Any thoughts on this
or
direction to references would be appreciated.

Sounds like you have the right idea. It is not difficult to
calculate the anode resistance and voltage. Obviously it will carry
the current of both of the top valves. I would use a simple divider
for the screen, coupled to cathode or ground via a suitable cap,
before adding improvements.

There are a couple of examples in Morgan Jones "valve amplifiers". I
am sure I have seen *loads* on the net.

cheers, Ian

Ian is correct about Morgan Jones, who uses two
transistors in series to get a very near perfect constant
current source, ie, a source of current equivalent to
a near infinite value of resistance.

But some CCS tails used with a twin triode LTP
used a pentode, say a 6AU6 with its anode connected to the
commoned cathodes, its screen fed from a B+
large resistance, but bypassed well with a C to the cathode, its cathode
taken via some
resistance to say -200v, and the grid taken
to say -100v, from a resistance divider, with a large value
electro between it and the -200v.

For every tube, U = Gm x Ra
In the case of a pentode like the 6AU6 with Ra = 0.5M at 5 mA,
and Gm = 0.0039A/V, the U will be 1,950.

let us suppose the 6AU6 cathode Rk is 22 k, then the effective
value of Ra looking into the 6AU6 anode
will be Ra + [ (U + 1) x Rk ], which in this case is
500k + 42.9M, or around 43M.

Such a large value of LTP tail impedance will act to
give good balance of output voltages from the LTP,
even with wildly mismatched triode halves, and this balance become
determined chiefly
by the matching equality of the RLs used in the LTP.

I have sometimes used an MJE 340 as a CCS, with a 4.7 k emitter R,
and although it isn't correct to use U, Gm and Rc with transistors,
one does get an extraordinarilly high collector input impedance
or similar largeness compared with a more costly pentode,
and there is no screen voltage and bypass to worry about.
The transistor is like a pentode with vastly higher Gm,
and lower Ra, so the "U" is still very high.

The maths are complex to the beginner, but can be found in the old text
books.

Patrick Turner.


Thanks guys, Its been a long long time since I've done this kind of math.
I'm not so sure I want to do it now or go thru a relearning process. It
occured to me that some of the older tube oscopes used this kind of circuit.
I may look around at some of these before I decide whether or not I really
want to do this since the solidstate constant current source is so easy to
impliment. This may be a case of good sense winning out over nostalgic
urges.

The maths are complex to the beginner, but can be found in the old text
books.

Patrick Turner.


Thanks guys, Its been a long long time since I've done this kind of math.
I'm not so sure I want to do it now or go thru a relearning process. It
occured to me that some of the older tube oscopes used this kind of circuit.
I may look around at some of these before I decide whether or not I really
want to do this since the solidstate constant current source is so easy to
impliment. This may be a case of good sense winning out over nostalgic
urges.

I use SS CCS inn my tube amps because they
are easy, and effective as tube, and it makes sense to use
an iddy biddy transistor as a slave to serve the tubes
appetite for current.
The maths are not always useful to build a simple CCS,
if in less time you can apply the concept by the cut, measure and try method.

One could use a choke as a high impedance cathode
impedance, but that introduces bandwidth limitations and distortions.

Patrick Turner.

Dear Jimmy,

In a moving coil head amplifier circuitry published in about 1980 or a
little earlier in "Hi-fi News and Record Review" (I think it was), it
utilized an FET as a courrent source as the Anode loading for the E88CC to
achieve high gain. The B+ was at about 30Volts in this case. The input
aand output signal here was also very small, of the order of 0.2mV amplified
to 5mV, approximately.

Wonder if this suits you?


Herman

"Jimmy" ¦b¶l¥ó
. com ¤¤¼¶¼g...
In the past when I want ed a constant curent source for a tube
differential
amp I used solid sate componets in the cathode circuit. For grins and
giggles this time I was thinking about going all tube but have never done
it
before and cant seem to find a good reference with this type of circuit.
Best I can figure is to use a small signal pentode in the cathode and
regulate the screen voltage with a vr tube/s. Any thoughts on this or
direction to references would be appreciated.

Yves Monmagnon wrote:

"Patrick Turner" wrote in message
...

...................;;
I have sometimes used an MJE 340 as a CCS, with a 4.7 k emitter R,
and although it isn't correct to use U, Gm and Rc with transistors,
one does get an extraordinarilly high collector input impedance
or similar largeness compared with a more costly pentode,
and there is no screen voltage and bypass to worry about.
The transistor is like a pentode with vastly higher Gm,
and lower Ra, so the "U" is still very high.

The maths are complex to the beginner, but can be found in the old text
books.

Maths for SS CCS are so simple !
Posted an article at ABSE : Easy SS Current source

I sometimes place a 1k R between the common cathodes and the
collector of a CCS.
Then I have a signal applied to one side of the LTP,
with the other side grounded, so about half the input voltage appears at
the commom cathodes.
The signal voltage across the 1k is neglible, proving
an extremely low current flow, and if this voltage is say 1mV,
then the current is 0.001 mA, and if the cathodes' voltage is
say 4 vrms, then the impedance looking into the
CCS with 1 k in series is 4 / 0.000001 ohms, or 4 Mohms,
at least.

Patrick Turner.



Cheers, Yves

Patrick Turner.

"Patrick Turner" wrote in message
...

...................;;
I have sometimes used an MJE 340 as a CCS, with a 4.7 k emitter R,
and although it isn't correct to use U, Gm and Rc with transistors,
one does get an extraordinarilly high collector input impedance
or similar largeness compared with a more costly pentode,
and there is no screen voltage and bypass to worry about.
The transistor is like a pentode with vastly higher Gm,
and lower Ra, so the "U" is still very high.

The maths are complex to the beginner, but can be found in the old text
books.

Maths for SS CCS are so simple !
Posted an article at ABSE : Easy SS Current source

Cheers, Yves

Patrick Turner.

"Herman Lo" wrote in message
...
Dear Jimmy,

In a moving coil head amplifier circuitry published in about 1980 or a
little earlier in "Hi-fi News and Record Review" (I think it was), it
utilized an FET as a courrent source as the Anode loading for the E88CC to
achieve high gain. The B+ was at about 30Volts in this case. The input
aand output signal here was also very small, of the order of 0.2mV
amplified
to 5mV, approximately.

Wonder if this suits you?


Herman

I know how to do it any number of ways using SS but I was pondering a way to
do constant current source for a differential amp with all valves. Since I
posted this I have come across a couple of ways using triodes. This suprised
me s I figured the best way would be to use a screen tube since the current
in these varies less with plate voltage than does a triode.


"Jimmy" ¦b¶l¥ó
. com ¤¤¼¶¼g...
In the past when I want ed a constant curent source for a tube
differential
amp I used solid sate componets in the cathode circuit. For grins and
giggles this time I was thinking about going all tube but have never
done
it
before and cant seem to find a good reference with this type of circuit.
Best I can figure is to use a small signal pentode in the cathode and
regulate the screen voltage with a vr tube/s. Any thoughts on this or
direction to references would be appreciated.

You are right that a pentode makes a more constant CCS, as a glance
at the anode characteristics shows.

But a triode is better than a fixed resistor.

Patrick Turners remark following my post is misleading. Morgan shows
several pentode-ccs LTP. One amp shows a BJT ccs LTP followed by a
pentode version, with an EF184 underneath a E182CC, which is used to
drive a pair of KT88. The configuration allows the two LTP to be
direct-coupled, and accepts balanced input.

Or how about an EL84 underneath a 6BX7? Needs 900V though.

cheers, Ian.

"Jimmy" wrote in message
. com...

"Herman Lo" wrote in message
...
Dear Jimmy,

In a moving coil head amplifier circuitry published in about
1980 or a
little earlier in "Hi-fi News and Record Review" (I think it
was), it
utilized an FET as a courrent source as the Anode loading for
the E88CC to
achieve high gain. The B+ was at about 30Volts in this case.
The input
aand output signal here was also very small, of the order of
0.2mV
amplified
to 5mV, approximately.

Wonder if this suits you?


Herman

I know how to do it any number of ways using SS but I was
pondering a way to
do constant current source for a differential amp with all valves.
Since I
posted this I have come across a couple of ways using triodes.
This suprised
me s I figured the best way would be to use a screen tube since
the current
in these varies less with plate voltage than does a triode.


"Jimmy" ¦b¶l¥ó
. com ¤¤¼¶¼g...
In the past when I want ed a constant curent source for a tube
differential
amp I used solid sate componets in the cathode circuit. For
grins and
giggles this time I was thinking about going all tube but have
never
done
it
before and cant seem to find a good reference with this type
of circuit.
Best I can figure is to use a small signal pentode in the
cathode and
regulate the screen voltage with a vr tube/s. Any thoughts on
this or
direction to references would be appreciated.

Ian Iveson wrote:

You are right that a pentode makes a more constant CCS, as a glance
at the anode characteristics shows.

But a triode is better than a fixed resistor.

Doesn't it depend on the triode chosen?
and the way it is used?

In a typical LTP using say 6SN7, one can use around 22k for the tail R,
if the grids are biased from the input tube anode, at about +120v.
But when driven from one side only, the anode RL values have to be
considerably different to achieve balance, and some 2H will sneak
through.

A triode, say a 1/2 6SN7 used as a current source with
a simple unbypassed Rk of 2.2k will have
a dynamic R of around 56k,
and still this is nowhere near what could be achieved
with one MJE340, or a pentode, with a large value Rk taken to a -ve
voltage supply.



Patrick Turners remark following my post is misleading.

What could you expect? that I copied all of what the good book said?

Morgan shows
several pentode-ccs LTP. One amp shows a BJT ccs LTP followed by a
pentode version, with an EF184 underneath a E182CC, which is used to
drive a pair of KT88. The configuration allows the two LTP to be
direct-coupled, and accepts balanced input.

Or how about an EL84 underneath a 6BX7? Needs 900V though.

EL84 has a lowish Ra and lowish U of 400 for a CCS.
But a transistor used in the cathode circuit of the EL84 would
make it into a CCS, but then why? the transistor
would easily do what the EL84 does, and use less voltage headroom up.
I do this at
http://www.turneraudio.com.au/websch...ma550w335h.gif
where an MJ340 is used to provide a CCS to a pair of EL84 in triode
as a low thd LTP driver.

Another CCS way is to cascode the triodes, and with 2 x 1/2 6SN7,
with unbypassed 2.2k Rk on the bottom tube you get
about 1.2M as the dynamic impedance, and you use
200v of headroom.
With 300v of headroom, and say 18k as the bottom tube Rk, you get
about 8 M ohms, still well short
of a pentode or bjt.

Patrick Turner.


cheers, Ian.

"Jimmy" wrote in message
. com...

"Herman Lo" wrote in message
...
Dear Jimmy,

In a moving coil head amplifier circuitry published in about
1980 or a
little earlier in "Hi-fi News and Record Review" (I think it
was), it
utilized an FET as a courrent source as the Anode loading for
the E88CC to
achieve high gain. The B+ was at about 30Volts in this case.
The input
aand output signal here was also very small, of the order of
0.2mV
amplified
to 5mV, approximately.

Wonder if this suits you?


Herman

I know how to do it any number of ways using SS but I was
pondering a way to
do constant current source for a differential amp with all valves.
Since I
posted this I have come across a couple of ways using triodes.
This suprised
me s I figured the best way would be to use a screen tube since
the current
in these varies less with plate voltage than does a triode.


"Jimmy" ¦b¶l¥ó
. com ¤¤¼¶¼g...
In the past when I want ed a constant curent source for a tube
differential
amp I used solid sate componets in the cathode circuit. For
grins and
giggles this time I was thinking about going all tube but have
never
done
it
before and cant seem to find a good reference with this type
of circuit.
Best I can figure is to use a small signal pentode in the
cathode and
regulate the screen voltage with a vr tube/s. Any thoughts on
this or
direction to references would be appreciated.

Couldnt realy find any Morgan Jones circuits on the net, just references to
his books

"Ian Iveson" wrote in message
...
"Jimmy" wrote In the past when I want ed
a constant curent source for a tube differential
amp I used solid sate componets in the cathode circuit. For grins
and
giggles this time I was thinking about going all tube but have
never done it
before and cant seem to find a good reference with this type of
circuit.
Best I can figure is to use a small signal pentode in the cathode
and
regulate the screen voltage with a vr tube/s. Any thoughts on this
or
direction to references would be appreciated.

Sounds like you have the right idea. It is not difficult to
calculate the anode resistance and voltage. Obviously it will carry
the current of both of the top valves. I would use a simple divider
for the screen, coupled to cathode or ground via a suitable cap,
before adding improvements.

There are a couple of examples in Morgan Jones "valve amplifiers". I
am sure I have seen *loads* on the net.

cheers, Ian

Couldnt realy find any Morgan Jones circuits on the net, just references to
his books

I don't think he does anything on the net. He's a pretty private person and
though very pleasant, values his 'consultant' status and doesn't like being
bothered by questions. I think you just have to buy his latest book - ed.3 -
which is well worth it. He's a damn good writer.

=== Andy Evans ===
Visit our Website:- http://www.artsandmedia.com
Audio, music and health pages and interesting links.

I had good success with a pentode current source. I've used a 7-pin 6DS5
for this purpose with a 0A2 150 voltage regulator to reference the screen
to the cathode. The 6DS5 plate can be used over a wide voltage range, including
at values lower than the screen. The high plate resistance is a bonus in
this application.

The circuit is quite simple. A 1 meg resistor is connected from the grid
to a cathode resistor to set the bias. (I recommend using a pot to
adjust for the desired current.) A cap from grid to ground completes it.

I've also seen a variation where the grid cap is attached to the
mid-point of two 1 meg resitors, each connected to a plate of the
differential amp. It's supposed to improve balance but I'm not sure how
this works.

The power supply can be a pain and this certainly takes more real estate
than a SS current source, but I found it a fun challenge. It looks cool
and sounds good to boot!

Good luck!

"Paul D. Spiegel" wrote:

I had good success with a pentode current source. I've used a 7-pin 6DS5
for this purpose with a 0A2 150 voltage regulator to reference the screen
to the cathode. The 6DS5 plate can be used over a wide voltage range, including
at values lower than the screen. The high plate resistance is a bonus in
this application.

The 6DS5 is a small power pentode with Ra = 28k, with Ia = 27 mA,
Ea = 250, and Eg2 = 200v.

The 28k is far to low to make the tube qualify as a CCS,
but used at lower Ea and lower Ia, the Ra usually goes a lot higher,
but still nowhere near as high as a signal pentode such as the 6SH7,
or 6AU6, etc.



The circuit is quite simple. A 1 meg resistor is connected from the grid
to a cathode resistor to set the bias. (I recommend using a pot to
adjust for the desired current.) A cap from grid to ground completes it.

So it appears the cathode is just grounded, and the tube develops a negative bias
by way of the small grid current, ie, the tube is grid leak biased,
which was done commonly in radio sets for the first AF tube which needed a
grounded cathode since the detector diodes used the same cathode,
and cathode bias is not used.



I've also seen a variation where the grid cap is attached to the
mid-point of two 1 meg resitors, each connected to a plate of the
differential amp. It's supposed to improve balance but I'm not sure how
this works.

Any unbalance in the anode outputs is applied to the
"CCS" with effectively makes the CCS an active part of the circuit.

To make the circuit have a far truer "passive" CCS characteristic, the effective Ra
of the pentode can be increased enormously by simply
having the grid biased up at a fixed +20v, and a cathode resistor
left unbypassed, so around 2.2k would be about right
for constant 10 mA.
The voltage reg isn't really needed, and its feed resistor will be too low,
and in parallel with Rk, so a plain R from the B+ to the screen will do,
and a large value electro to bypass the screen to the cathode to
make the tube work as a true pentode, which depends on the screen being properly
bypassed to the cathode, so the signal voltages applied to each are the same,
except perhaps at 0.0 Hz, when the cap has a high impedance.
Since the U of the 6DS5 is Ra x Gm = 160 approximately,
the effective Ra will be Ra + ( 2.2k x 161) = 382k, which still is a longway short
of
a true CCS, but a lot more than 28k, which is little better than a resistor tail for
an LTP.


The power supply can be a pain and this certainly takes more real estate
than a SS current source, but I found it a fun challenge. It looks cool
and sounds good to boot!

What is the balance of a differential amp like with your existing set up?

To get good balance with a differential amp, using equal RL to each
anode, and with the input going to one side only, the cathode "tail" needs to have a

high value, and 10meg ohms is easy to get using almost
any pentode, as long as the cathode resistor is of high enough value,
which means to get it to be a high value, and have the required current flow,
one has to have a -200v supply, and take the cathode R down to this -ve voltage.
The pentode's grid can be simply grounded, but to have
stable constant current, the -200v supply must be well
bypassed to 0V, and a CRC filter from -350v will do, with high value electros,
say 470 uF.
Such a negative supply is easily arranged using two silicon diodes from
a full wave rectifier centre tapped winding on a power supply,
or from some small power tranny with a 120 volt winding
from which -312v can be derived using a voltage doubler.

An easier way is to replace the 2.2k Rk I mentioned with a transistor,
which can be a small normal npn type CCS, and the CCS
with the pentode you have will have an enormously high real value.
But for you that might be cheating.

Just my 2.83 chrissy bells' worth,

Patrick Turner.


Good luck!

Patrick Turner wrote:

So it appears the cathode is just grounded, and the tube develops a negative bias
by way of the small grid current, ie, the tube is grid leak biased,
which was done commonly in radio sets for the first AF tube which needed a
grounded cathode since the detector diodes used the same cathode,
and cathode bias is not used.


This is what happens when I write these postings late at night. I left
out some key information.

The plate of the 6DS5 is at about 100v (the cathodes of the diff-amp
6CG7). The cathode of the 6DS5 is fed from a -250v supply. There is a
5.76K resistor from the negative supply followed by a pot that sets the
bias for the 1M grid resistor at a level to draw 17ma through the 6CG7.

The screen is fed from a separate power supply tap at about 65v. The 0A2
regulator tube attaches to the junction of the pot and 5.76K resistor at
-85v.

Incidentally, I tried a zener in place of the 0A2. It added a hashy edge
to the sonic result. (I'm not sure how. Any noise in the common cathode
should cancel out in a diff-amp, but my ears told me otherwise.) A cap
worked well, but I thought the regulator tube sounded tighter. Besides,
as others have noted, I liked the neon like glow from the 0A2.

Since the U of the 6DS5 is Ra x Gm = 160 approximately,
the effective Ra will be Ra + ( 2.2k x 161) = 382k, which still is a longway short
of
a true CCS, but a lot more than 28k, which is little better than a resistor tail for
an LTP.


So I get (5.76K x 161) + 28K = 955K. Not amazing, but decent for a
little 7 pin pentode. Knowing more now I could do better next time, but
this made for a fun challenge.

"Paul D. Spiegel" wrote:

Patrick Turner wrote:

So it appears the cathode is just grounded, and the tube develops a negative bias
by way of the small grid current, ie, the tube is grid leak biased,
which was done commonly in radio sets for the first AF tube which needed a
grounded cathode since the detector diodes used the same cathode,
and cathode bias is not used.


This is what happens when I write these postings late at night. I left
out some key information.

The plate of the 6DS5 is at about 100v (the cathodes of the diff-amp
6CG7). The cathode of the 6DS5 is fed from a -250v supply. There is a
5.76K resistor from the negative supply followed by a pot that sets the
bias for the 1M grid resistor at a level to draw 17ma through the 6CG7.

Aha!, now we got a nicer picture when you filled in the detail.

Se how essential it is to fully decribe what you got there!



The screen is fed from a separate power supply tap at about 65v. The 0A2
regulator tube attaches to the junction of the pot and 5.76K resistor at
-85v.

Incidentally, I tried a zener in place of the 0A2. It added a hashy edge
to the sonic result. (I'm not sure how. Any noise in the common cathode
should cancel out in a diff-amp, but my ears told me otherwise.)

I have a few zeners in my circuits, but I have tried to eliminate them
or have an RC filter after them, and thast takes care of the zener noise.

Noise would get into a diff amp if the CCS is noisy, and the amp has one side at signal
0V.
But then if the impedance of the CS is really high,
the noise just should not appear a sa voltage at the cathodes,
because the Rin to the common cathodes is very low, less than 1k,
yet the CCS has a finite value far higher, and you have an effective
resistance divider, preventing noise entry.

A cap
worked well, but I thought the regulator tube sounded tighter. Besides,
as others have noted, I liked the neon like glow from the 0A2.

Aha, the glow.
I have never used tube voltage regulators, maybe I should try one some day,
but so far a 470 uF cap does the trick if I want stable screen voltages
at signals above 10 Hz.



Since the U of the 6DS5 is Ra x Gm = 160 approximately,
the effective Ra will be Ra + ( 2.2k x 161) = 382k, which still is a longway short
of
a true CCS, but a lot more than 28k, which is little better than a resistor tail for
an LTP.


So I get (5.76K x 161) + 28K = 955K. Not amazing, but decent for a
little 7 pin pentode. Knowing more now I could do better next time, but
this made for a fun challenge.

You got the idea.
You can measure the value of the CCS, by placing a 1k R
in series with it, and have a large signal applied to the diff amp,
but without going into clipping.
Measure the signal across the 1k, and measure the signal at the CCS,
and the value of the CCS id Vk / V across 1k in kohms.
If you have 4vrms at the commoned cathodes, and 1 mV across the
1k, at say 1 kHz, then the CCS value is 4 / 0.001 kohms, = 4 megohms.

Diff amps are strange things. You can have a 12AU7 on one side, and 12AT7
on the other, and as long as the anode loads are equal, and the CCS
has a very high real value, the outputs are equal.
But the 2H from each tube won't cancel fully, and there will be some
in the output of each unequal tube.
So for the least thd from a diff amp, matching tubes are required.
You do get 3H.
But a diffamp using 6CG7 with 3 mA in each, with a high value pair of loads to say
+500v should make 100vrms from each side at less than 1% thd, mainly 3H.
At 1vrms, its extremely low.

Patrick Turner.

I thought this type of bias was called contact potential bias instead of
grid leak.


So it appears the cathode is just grounded, and the tube develops a
negative bias
by way of the small grid current, ie, the tube is grid leak biased,
which was done commonly in radio sets for the first AF tube which needed a
grounded cathode since the detector diodes used the same cathode,
and cathode bias is not used.



I've also seen a variation where the grid cap is attached to the
mid-point of two 1 meg resitors, each connected to a plate of the
differential amp. It's supposed to improve balance but I'm not sure how
this works.

Any unbalance in the anode outputs is applied to the
"CCS" with effectively makes the CCS an active part of the circuit.

To make the circuit have a far truer "passive" CCS characteristic, the
effective Ra
of the pentode can be increased enormously by simply
having the grid biased up at a fixed +20v, and a cathode resistor
left unbypassed, so around 2.2k would be about right
for constant 10 mA.
The voltage reg isn't really needed, and its feed resistor will be too
low,
and in parallel with Rk, so a plain R from the B+ to the screen will do,
and a large value electro to bypass the screen to the cathode to
make the tube work as a true pentode, which depends on the screen being
properly
bypassed to the cathode, so the signal voltages applied to each are the
same,
except perhaps at 0.0 Hz, when the cap has a high impedance.
Since the U of the 6DS5 is Ra x Gm = 160 approximately,
the effective Ra will be Ra + ( 2.2k x 161) = 382k, which still is a
longway short
of
a true CCS, but a lot more than 28k, which is little better than a
resistor tail for
an LTP.


The power supply can be a pain and this certainly takes more real estate
than a SS current source, but I found it a fun challenge. It looks cool
and sounds good to boot!

What is the balance of a differential amp like with your existing set up?

To get good balance with a differential amp, using equal RL to each
anode, and with the input going to one side only, the cathode "tail" needs
to have a

high value, and 10meg ohms is easy to get using almost
any pentode, as long as the cathode resistor is of high enough value,
which means to get it to be a high value, and have the required current
flow,
one has to have a -200v supply, and take the cathode R down to this -ve
voltage.
The pentode's grid can be simply grounded, but to have
stable constant current, the -200v supply must be well
bypassed to 0V, and a CRC filter from -350v will do, with high value
electros,
say 470 uF.
Such a negative supply is easily arranged using two silicon diodes from
a full wave rectifier centre tapped winding on a power supply,
or from some small power tranny with a 120 volt winding
from which -312v can be derived using a voltage doubler.

An easier way is to replace the 2.2k Rk I mentioned with a transistor,
which can be a small normal npn type CCS, and the CCS
with the pentode you have will have an enormously high real value.
But for you that might be cheating.

Just my 2.83 chrissy bells' worth,

Patrick Turner.


Good luck!