I need help with ccs based LTP

[pauldune]

Oh my God! I had a revelation this morning when I thought about how the AC signal sees the load. I always try to think in AC current loops to understand what is going on in a schematic. Until now I couldnt quite understand the LTP, CCS or not.
I thought about how the currents flow and then it became clear.
So I sat down and started drawing. The cool thing is that everything is explained now, and measurements correlate with calculations. Here is what I came up with:

First the standard LTP with only CCS in tail.

Drawing 1 (step 1+2):
Ra is Ranode; the external resistor connected between plate and B+.
Ri is internal resistance of tube (plate/anode resistance (d3a->2k)
Rl is Rload.
In the drawing I try to explain that the powersupply capacitor is a shortcircuit for AC, and the tail CCS is an open circuit for AC.

Drawing 2 (step 3):

Here you can clearly see both current loops; Ra parallel to Rload, parallel to Ri+Ri+Ra. Output impedance with Ri=2K and Ra=4K is 2.7K. I measured 2.7K

Now the CCS version:

Drawing 3 (step 4+5)

I try to explain that Ra is now a CCS, wich is an open circuit for AC.
As you can see in step 5, There is no current loop any more, it is an open circuit, so output impedance is infinite. Because I used the current spilling resistors in my real schematic, and the CCS's aren't perfect, I measured 30K.

Drawing 4 (step 6+7)

As proposed by Michael Koster and Godfrey, I added Cx from the - output to ground. As Cx and the powersupply capacitor are both shortcircuits to AC, I drew the current loop, and that led to step 7, with the extra capacitor Cx marked with green and red. Essentialy it closed the current loop.
Rout is now Ri+Ri: 2K+2K-> 4Kohm. I measured 5Kohm.
I suspect that the real Ri at my chosen setting for the tube (180v Va, 15mA)
is closer to 2.5K.

Anyway, it looks like it explaines everything I measured.

Please take a look at it, and let me know if i'm on the right track!
Or not

Paul

[pauldune]

I see now that step 6 is drawn wrong, but the conclusion (step 7) remains the same.
Because Rload and Cx are both connected to ground, ac flows only through Cx. (and not C-powersupply)

[pauldune]

If I apply the same logic to the standard 1 CCS version, I get the following:

If I add Cx there, output impedance changes from :Ra//Ri+Ri+Ra to Ra//Ri+Ri.

Calculation of Cx looks simple to me; I suspect: Fc->10Hz = 1/ (2 x PI x (Ri+Ri) x C)
C=~4uF -> 4.7 uF should be good for -3dB below 10Hz.
Am I right?

Paul

[pauldune]

Before I forget it: Here are the distortion figures for this pre-amplifier;
It measures extremely well!
distortion at maximum output ~3v pp: 0.004% without any feedback in a single stage!
My soundcards loopback distortion is at 0.003%, so it is possible that it is even less.

I still have some work in the powersupply section, but it measures and sounds already so good, that I intend to make a thread in the digital line level forum, with all the relevant information, including all schematics.
Maybe even PCB.

[pauldune]

I found out some interesting facts:
The noise in the 3 CCS version was coming from the top CCS (10m45s)
I just made a shunt powersupply with a 10m45s, and before I connected a capacitor to ground from the output, I had the same noise (hiss) as the 3 CCS LTP.
After the capacitor was placed, the pre-amp was dead quiet, no noise and no hum.
But i'm still thinking about a good actively regulated powersupply instead of the passive c-r-c-r-c-r-c. If only I could get LT3080, I would use the "21 century Maida Regulator" from Tomchr:
21st Century Maida Regulator

2nd thing I found out:
Remember the capacitor that Michael Koster and Godfrey advised? (from the second output (-) to ground)
In the 3 CCS version it dropped the noise, hum and output impedance and increased gain, but I did no distortion measurements, because it sounded less good.
Now with the standard 1 ccs version, I decided to give that same capacitor a try, as I thought it would lower output impedance a bit, and maybe decrease hum.
After connecting it DID reduce hum, but again it sounded "less".
Then I decided to measure distortion, and what do you think:
Without capacitor: some hum but distortion ~0.005%
With capacitor: Far less hum BUT distortion rose to ~0.04%! that is almost 10 times as much! It was mostly 2nd order and a little third.
So maybe my ears arent so bad as I feared

[pauldune]

The shunt powersupply from the previous post was an attempt to lower hum AND maintain low distortion figures. It succeeded, no noise, no hum (at least much less then my SE poweramp wich has some residual hum if your ear is near the speaker.)
And the distortion stays the same low 0,005%.

What did I learn:
-CCS 10m45s can be noisy
-Somethimes at first glance some improvements (less hum and noise) arent improvements at all (more distortion)
- Trust my ears!

I wonder if there are still some people reading this thread

Paul

[pauldune]

I think this is going to be the last post in a while; results so far:
-Latest schematic attached (stereo)
-THD 1kHz ~0.005%
-Gain ~23x: Buffalo3 DAC 8mA into 2x4.7ohm= vout pp 1.68v
-Output impedance ~3Kohm
-CMRR excellent (just tested)
-PSRR not so good; needs regulated power supply


Things to do:
-Change I/V resistors from simple metal film to shinkoh tantalum.
-Increase I/V resistors to 10ohms to increase output to ~3.5v pp.
-Evaluate different power supplies.
-Build some nice enclosures; 1 for DAC preamp, and 1 for powersupply.

[pauldune]

One last thing:
IMD (intermodulation distortion) seems really low too: see attached image of a 1Khz+5Khz signal. I see almost no additional "spikes"; to bad my program can't measure it.

Paul