Will ccs work with a DHT?

[sgerus]

Will ccs work with a DHT?

Please forgive me if this is a stupid question.

I was wondering if a constant current sink could be used in place of the cathode resistor with a DHT (2A3)?

I just made a cc source for my 6sl7 input and it made a huge improvement over the standard plate loaded setup.

The chip I have in mind (10M45) can handle the current and power.
My thinking is to just replace the cathode resistor with a ccs.
Right not the cathode resistor is off the heater transformer center tap.

What do you think?

 

[isaacc7]

I think I read in Morgan Jones' book that you can't do it in a SE topology but that it can work very well in PP. Something about 100% feedback with SE. I'll have to go look it up...

Isaac

 

[timpert]

Hi Isaac,

No, it is not a stupid question. It will work, if you replace the cathode resistor by a CCS set at the normal standing current of a 2A3, then the tube should bias normally. You may see a somewhat larger spread between tubes from the nominal bias point than with a cathode resistor, but I wouldn't worry about that. Of course, you shouldn't omit the cathode decoupling capacitor or otherwise you'll hear nothing.

I understand from the IXYS datasheet that the xx in xxM45 indicate the nominal current, so you'll have to use the 50M45 or 60M45.

I cannot say anything about the sound, other than that the cathode decoupling capacitor carries the vast majority of the AC current in the cathode resistor case, and all of the AC current in the cathode current-source case. There seems to be a consensus about this cap being much more important to the sound of the amplifier than the cathode resistor. But if you're willing to experiment, by all means do so and please share the results!

Regards,
Jurgen

 

[sgerus]

Jurgen,

I a bit confused by 10M45 datasheet. The top of the sheet shows a 2-100mA range
But lower in the datasheet they show a 7-15 mA range!

My chip seems to be working fine at 3mA for the 6sl7

Anyway, would I want to run the 2A3 at a lower current, say around 35mA, to keep the power under the 15W level when the voltage swings way to the right (on the load line chart)?

 

[Tubelab_com]

I have used the IXYX 10M45 as the plate load for a 45 in two different circuits. One is used as the driver for an 845 and the other was used in a chokeless parafeed. Both of these worked quite well. I have not tried it in the cathode circuit, but it should work well.

The IXCP10M45S is an adjustable version that can be set for any current from 2 to 100 mA. They are available at DigiKey.

 

[sgerus]

Hello Mr Tubelab,

I got the ccs idea from YOUR SITE.... thank you for providing all that info for the rest of us!

I will let you know if / how it works on a DHT
(in a few days)

-Thanks, Scott

 

[Tubelab_com]

I a bit confused by 10M45 datasheet. The top of the sheet shows a 2-100mA range But lower in the datasheet they show a 7-15 mA range!

The chip will adjust from 2 to 100 mA. The 7 to 15 spec shows the chip to chip variation with a 300 ohm resistor. This illustrates what several of us have found out, these chips are all different, and you have to adjust each one individually.

I have found that each chip in the batch that I got from DigiKey were reasonably similar, but each batch was different from the batch before it. I have used them at 30 mA with a 45 but the resistor value was different each time.

 

[isaacc7]

Hi Isaac,

No, it is not a stupid question. It will work, if you replace the cathode resistor by a CCS set at the normal standing current of a 2A3, then the tube should bias normally. You may see a somewhat larger spread between tubes from the nominal bias point than with a cathode resistor, but I wouldn't worry about that. Of course, you shouldn't omit the cathode decoupling capacitor or otherwise you'll hear nothing.

Aha! I know I remembered something like that So I'm not completely insane after all...


Isaac

 

[lt_texan]

Of course, you shouldn't omit the cathode decoupling capacitor or otherwise you'll hear nothing.

Regards,
Jurgen

Jurgen,

Are you saying that the unbypassed cathode ccs will cause the circuit to operate under 100% negative feedback if not bypassed?

I thought Allen Wright used unbypassed ccs in some preamp circuits without cathode decoupling capacitors?

Thanks
Dan

 

[Giaime]

It seems so strange to me. I'm using a CCS-loaded cathode follower in my preamp, unbypassed. It works, so? You don't need to bypass the cathodes of SE output tubes if you use a CCS.

Imagine of the tube being a resistor, the value controlled by the grid. The CCS sets a current throught this chain (OPT+tube+CCS), but single values of "resistors" (aka. the voltages at the cathode or the anode) can change, keeping the sum constant. If you inject a positive signal at the grid, the current try to increase, the CCS keeps it constant but you have shifted down the voltage at the anode (because "CCS" seen as a resistor has it's value increased).

Or am I wrong?

 

[andyjevans]

I seem to remember a PP design from somebody (?) in a recent thread where the output tubes had a CCS under them. When a capacitor was put between the cathodes and the centre tap of the OPT primary the sound was about equally good but louder. Can't remember the reference here.

 

[Circlotron]

Imagine of the tube being a resistor, the value controlled by the grid. The CCS sets a current throught this chain (OPT+tube+CCS), but single values of "resistors" (aka. the voltages at the cathode or the anode) can change, keeping the sum constant. If you inject a positive signal at the grid, the current try to increase, the CCS keeps it constant but you have shifted down the voltage at the anode (because "CCS" seen as a resistor has it's value increased).

Or am I wrong?

If the cathode current is constant then the anode current is constant so the voltage drop across the anode load resistor is also constant. = no output signal. The ccs will appear as a resistor of almost infinite resistance from the point of view of *change* of voltage versus *change* of current. Hope this helps?

 

[Circlotron]

That Ixys thing has a slope resistance of about 10k according to the datasheet. A LM317 wired as a constant current sink has a slope resistance of 250k. That's 25 times more constant. Here's someting I made a while back that has a slope resistance of at least 400 megohms.

http://www.diyaudio.com/forums/showthread.php?postid=67225#post67225

 

[timpert]

@Dan and Giaime,

Circlotron is right about the constant current source, its infinite dynamic resistance effectively prevents any AC (and thus signal) current to flow in the current loop formed by supply, transformer, triode and cathode circuit. The SE power stage is typically plate loaded, this configuration has all components (including the load) mentioned above wired in series.

I'm using a CCS-loaded cathode follower in my preamp, unbypassed. It works, so? You don't need to bypass the cathodes of SE output tubes if you use a CCS.

The cathode follower has the load in parallel with the CCS, so the infinite dynamic resistance of the CCS is shorted by the load, allowing for AC to flow through the loop. The cathode follower is a different topology from a "normal" SE output stage, so what works in one case doesn't necessarily work in the other.

Jurgen

 

[sgerus]

Jurgan,

Thank You,

Now I understand why ccs WILL NOT work with SE output.
(At least when the output trans is in series with the triode)

Just for kicks, I have attached a pdf document on ccs. On page 10 they show a DHT with the output trans in parrallel with the triode, via a coupling cap.....

I don't know if that's still considered a SE amp, but it looks like that's a way to use ccs with a DHT.

 

[timpert]

Hi,

The circuit on page 10 is called parallel feed, or parafeed (do a search on that last term, you'll find a lot of info). It is 100% SE, and because it doesn't have any DC flowing through the primary, considered by many to be superior to the traditional approach.

Jurgen

 

[Giaime]

Ah, so I was wrong. Well, I learned something.

Could you use cathode output then? SE output tube as cathode follower.

But I think that's better to develop a servo biasing system, that senses the current throught the tube and sets a negative grid voltage.

 

[timpert]

Hi,

In essence, the classic cathode resistor circuit does quite a good job as a servo circuit! It tries to preserve the bias point as much as possible, providing an excellent compromise between constant-current operation and fixed bias voltage operation, which both can be a tad more problematic when the tube ages. It does so by means of a very easy to understand and calculate first-order control loop, which by its very nature is unconditionally stable. Its simplicity makes it virtually bulletproof to boot!

So I very much doubt that any active control circuit will be able to beat the cathode resistor. Okay, you could save some dissipation, but if you're worried about that, maybe SE class A isn't for you...

But yes, using a DHT as a cathode follower is a very real option. You'll have to cope with the very high grid voltage swing this topology requires to achieve full power, but once you've nailed that issue (it has been done), a very low output impedance (and thus high damping factor, many speakers like that) will be yours without having to resort to global NFB.

Jurgen

 

[sgerus]

Jurgan,

If the ccs could replace the cathode resistor, and the cathode capicator
could be left in (in parallel with the ccs), that would allow the ac to flow, right.

The 10m45 ccs chip gets wired with just an input and an output.

Then maybe this could work after all?

 

[timpert]

Hi,

That should work just fine. Indeed, the decoupling cap should be left in parallel with the CCS. For a 2A3, the CCS should be adjusted to about 60 mA. Good luck!

Jurgen