• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

IXCP10M45S

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm assuming you are talking about a standard grounded cathode gain stage.
As a plate (anode) load, no you don't need a resistor. You could use one to reduce the power dissipated by the 10M45 since a constant current with a constant resistance produces a constant voltage drop. Just make sure that the voltage drop by the resistor isn't so much that it interferes with the voltage swing of the tube.
You will need a cathode resistor to set the bias point. You can also use a LED, diodes, or a zener diode if you wish.
 
Maybe someone knows the answer to this…

But doesn't using any type of constant-current source maximize the effect of Miller capacitance between the plate and grid producing the lowest pole of (and least desirable, I'd guess) high-frequency roll-off?

If I recall accurately, the cascode topology, where the "amplifying tube" is fed by a constant voltage anode load almost completely counteracts Miller capacitance by limiting the voltage swing of the anode, which in turn can't feed its inverted "signal" back to grid thru grid-anode Miller capacitance.

And ordinary resistor load is intermediate: it modulates the anode with the inverted, amplified signal, far more than the cascode constant-voltage load, and less than any of the various constant-current loads. (Think … 10M45 CCS or 250 H choke load, or interstage transformer load or even gyrator loads), so the effect of Miller capacitance anode-grid feedback is modestly reduced. Indeed: it is reduced directly in relationship to how much gain the stage is designed to have.

Just wondering, is all.

And wondering … if the optimal design “solution” (as long as one is contemplating employing a chip-of-silicon) mightn't just be a MOSFET+resistor cascode anode load. Almost completely bypassing the Miller capacitance hi-frequency roll-off effect. You know?

Just saying,
GoatGuy
 
GoatGuy,

Good to hear from you again. Good points.

Cascode operation significantly reduces the miller effect of the bottom stage. But, if you get the same gain out of the plate of the top stage versus a non cascode stage, the current that the plate has to supply for slewing the grid / plate capacitance is the same (well actually C x gain versus C x (gain +1)). True, the top grid is tied to AC ground, so it is not influenced. But . . . the plate impedance of the top tube in a cascode stage is very high impedance, versus a simple common cathode stage.

So now the high impedance cascode plate has to drive the next stage's grid resistor, and any grid to cathode capacitance, and any grid to screen capacitance if there is a screen, and any miller capacitance if there is no screen.

What is the distributed capacitance of that 250H plate choke load? That might have more effect than the miller effect.

And I used to use interstage transformers, but did not like the square wave look, nor the phase and frequency response (I never felt I could "afford" the really good interstage transformers, so did not even have the chance to use and test them.

But for me, I do like the 900V version of the IXYS CCS in the triode plate load. My tweeters and my ears do not seem to know the difference (do not work well above ~ 20kHz for the tweeters, and about 13kHz for the [one] ear. I do use a lower resistance input potentiometer than many do, to partially swamp out the effect of the miller capacitance. And I do use real resistor plate loads sometimes too, instead of the IXYS CCS.

Not all lunches are free.
 
Last edited:
No, I mispoke. I meant 10 ma and -6 volt bias. I do have another question. I know I need to use the cathode resistor to set the bias voltage so the tube will conduct. Do I then set the CCS to the same current, 10 ma? Or should it be set to more for some headroom?
 
mr2racer,

For your triode with a CCS as the plate load: Set the CCS to 10 mA (many CCS use a single resistor to set the current, calculate for 10 mA, and use that resistance). Connect the grid through its grid return resistor to ground (0V) Calculate the cathode resistor, 6V/0.010A = 600 Ohms. The self bias resistor is 600 Ohms, connect the cathode through a 600 Ohm resistor to ground. Turn the filaments on, and the B+ on. The voltage on the cathode Will Be 6V (600 Ohms x 0.010A) Done.

You may or may not want to use a bypass capacitor across the 600 Ohm self bias resistor. The cathode impedance is as high as it ever can be, because the plate is loaded by an extremely high impedance (CCS). The grid resistor of the next stage is quite often lower than the impedance of a good CCS plate load.

Example gain of the circuit:

Suppose the circuit has the plate and CCS plate load, driving a coupling cap, driving the next stage grid resistor: Suppose the CCS is not real good, so it only has an impedance of 100k Ohms.

Suppose the grid resistor of the next stage is 100k Ohms. The plate load is effectively 50k Ohms. Suppose the triode has a Mu (u) of 20. Suppose the triode has a transconductance, Gm, of 1000 micro mhos. That is 1000 Ohms cathode impedance. With a u = 20, and a transconductance of 1000uMhos, rp = 20,000 Ohms (u/Gm = rp)

But the 1000 Ohm cathode impedance is raised by the 'plate load/u' 50k/20 = 2500 Ohms. The total cathode impedance is 1000 Ohms + 2500 Ohms = 3500 Ohms.

3500 Ohm cathode impedance is in parallel with 600 Ohms = 512 Ohms.

If you use a capacitor that has 51 Ohms of capacitive reactance at 10 Hz, the low frequency will be no more than -3dB at 10 Hz, and only -1dB at 20 Hz (single pole roll off). 51 Ohms = 2 x pi x f x C C = 1 / (51 x 2 x pi x 10Hz) C = 312 uF With the bypass cap, the gain will be 20 x (50k/(20k + 50k)) = 14.3

If the CCS impedance is 400k, and the next stage grid resistor is 400k, the plate load is 200k. Now the gain is 20 x (200k/(20k + 200k)) = 18, much closer to u.

And do not forget that the impedance at the cathode is 600 Ohms in parallel with the cathode impedance (1000 Ohms in series with 200k/20) = 1,000 + 10,000 = 11,000 Ohms. You can see that 11,000 Ohms is much much greater than 600 Ohms, so the bypass capacitor is Not needed in this case (and some people are very happy when they do not need to use a bypass capacitor).
 
It *really* helps to work this out on the tube curves.

I have found that most tubes rarely match the curves. High Gm triodes like the 5842 are worse case, that's why there is a pot in the cathode on the TSE amp. Every tube is different.

As said previously, the CCS sets the tube current. A 10M45 wants about 330 ohms for 10 mA, but there is a pretty big variation among these guys too. The cathode resistor sets the plate voltage. Start with something around 600 ohms, then tweak for lowest distortion, or maximum headroom, whichever is most important for your application.
 

PRR

Member
Joined 2003
Paid Member
On the curves a trial-solution is clear. *Assume* the "10mA" is a Given. Plot 10mA line. There's always a limit how high B+ you can get, plot that limit. Each grid voltage curve implies a plate voltage. Since a hi-fi line amp is not a big-signal thing, we can set idle plate voltage low, or high, but generally mid-way is best (and gives more leeway for bad data). Any of the RK values plotted will work.
 

Attachments

  • 6H6pi-10mA-CCS.gif
    6H6pi-10mA-CCS.gif
    21 KB · Views: 261
Thanks, everyone

I wasn't far off. At this point cathode resistors are 620 ohms. Digikey didn't have one watt 600 through holes. And for the set resistor for the CCS I have both 300 and 330 ohms. I had to replace the power transformer as the Antek I was using was too high. The one I'm using now gives a B+ of 295 volts with the tubes conducting. I'm also using an output transformer that is AC coupled. I listened to the amp a while back. it sounded alright but had a buzz and not enough gain. But at that time the B+ was 385 volts, too much for the tubes and the CCS. I'm waiting on some etch to build a new circuit board. In the original design I used one CCS for the voltage amp and the output tube. That allowed too much plate to plate feedback and killed the gain. Each tube will now have it's own CCS. That should, if I undertand correctly, should maximize gain. And I will be able to use global negative feedback from the secondary of the OT.
 
PRR, I'm assuming because I am using an output transformer I will need a large voltage swing. The primary is 10k and the secondary 150 ohm.

Another CCS question. If I am understanding MJ the CCS provides a very high ripple rejection. I'm assuming by ripple that means the AC signal. So when calculating the gain of the tubes I should use a very high value for the "plate resistor" shouldn't I?
 
Thanks everyone. I've gotten all four tubes within .2 ma. I think that's close enough. When I made the new board and added one IC for each tube at first the first stage tubes were way down on current. It turns out the the aC buzz was coming in through the input and throwing off the grid voltage to -11 instead of the -6 I wanted. Now all I have to do is get rid of the AC in the circuit. Thanks again!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.