New DISPRE preamp, successor to previous popular version

[MaxS]

Hello,

He'll publish a picture of the finished project and we reverse engineering color code of resistors as usual.

What ? This is a secret

PMA makes an honest living on his design.
So, we can understand he doesn't want to give ALL the project. We got the schematic. They're not many ways to design a DC servo and route a board. It is already a good thing to get the input stage schematic which is the masterpiece of this whole preamp to my mind.

But you can feel confortable too share the plan here, they're ajust round thousands of people here but he'll stay confidential, that's sure


Bad weather outside, it is raining, it will kept the temperature at reasonable level ( quite under 20°C )

 

[PMA]

Maybe I would release it, but not at the moment. It is still in a prototyping phase.

 

[jam]

Hi PMA,

Is there a reason not to current source the differentials?

Jam

 

[aparatusonitus]

Higher OLG I guess.

 

[PMA]

Yes.

 

[crt]

for resistor 2K+499, can i replace with 1k5+1k.
and for 2r7 can i replace with 3r3 or near.

very hard to find 3r3 and 499.

for BD139/140, i had tested. but the different between PNP and NPN is quite far 80 vs 160. that's fine?

Thanks

 

[AndrewT]

510//22k~=498r4 +-5r
510//24k~=499r4 +-5r

 

[PMA]

CRT, this thread is about the New Dispre. This preamp has got now a name, and it is Dispre SSE.
SSE means Super Symmetry Enhancement. This new preamp has been ordered 30 pcs within 1 day after announcement (in my country). To speak about older version, please use the older thread

http://www.diyaudio.com/forums/showthread.php?threadid=64843&highlight=

Thanks,
Pavel

 

[lumanauw]

Hi PMA,

Is there a reason not to current source the differentials?

Jam

Yes.

PMA, what is the reason you put resistor instead of CCS for R32-R34 positions?

 

[Lumba Ogir]

PMA, what is the reason you put resistor instead of CCS for R32-R34 positions?

kind of stinginess, considerably limiting the LTP`s functional accuracy.

Bonsai,
unlinear diodes in the signal path and again, low impedance current sources would make this edition less enhanced.

 

[PMA]

The reason of no use of CCS is to get low differential gain. The input stage just "looks" like LTP, but, for the reason of quite high resistor values, it behaves quite different. The main goal was to get speed, same step response for any output amplitude and excellent behaviour when overdriven.

 

[lumanauw]

Do you need extra clean power supply because of this?

 

[PMA]

It is better, yes. But if you have a look at schematic of the old Dispre, you will see RC filter 100R + 1000uF in every supply rail of both L and R frontends. These parts are there just for this reason, and they are here again in the new Dispre SSE.

 

[Lumba Ogir]

Hi,
high impedance CCS would be beneficial also for the topology in post #1, however I would not configure it that way, rather like in post #49. Increased value emitter resistors is a much better method to reduce voltage gain.
Regarding the pre-drivers, I would keep the CCS and insert emitter resistors, which will improve linearity, unlike diodes.

 

[aparatusonitus]

Hi Pavel,

I noticed that you have chosen somewhat highish values for NFB voltage divider, in order to use as least as possible capacitance (C13) over FDB resistor? What about noise injection from such highish value, is it a problem for lineamp like this one or not?

 

[PMA]

Hi,

Regarding T or Pi (H) configuration of emitter resistors, they are equivalent. T is more difficult to tune, but once estimated the correct values, it makes no difference.

Regarding noise, if I change FB network to 2k/1k, I get 15nV/rtHz. For the resistor values shown, it is 19nV/rtHz (at the input).

 

[AndrewT]

I was expecting much lower noise.
At least better than 4nV/rtHz for a discrete, gain of 3, opamp and possibly as low as 1 to 1.5nV/rtHz.

 

[Lumba Ogir]

PMA,
to me the H configuration is inferior. What is the idea behind it?
How do you estimate correct resistor values and how do you make performance tuning?

 

[Bonsai]

Lumbar,

re th elast paragraph in your post 74, see my post 60.

 

[PMA]

I was expecting much lower noise.
At least better than 4nV/rtHz for a discrete, gain of 3, opamp and possibly as low as 1 to 1.5nV/rtHz.

The noise is similar as single AD744. There were different design goals, that go AGAINST noise, unfortunately. These goals were:

- same shape of step response for ALL output amplitudes, i.e. no slew rate limitation

- good behaviour when overdriven, even for high frequencies

- high output current, stability for ANY capacitive load

- very low output impedance, to minimize output cable influence

During years of tests, I found these criteria very important for preamps. To me, the preamp with 1k output impedance is a poor solution, as well as passive "preamps".

Just for the record, step response into 1uF: