New DISPRE preamp, successor to previous popular version

Jump to Latest
#361
Quite large number of units were assembled and produced, and a few of them had strange behaviour. There was a hum and noise in the right channel (designation according to documentation). Also, DC component was unstable when turning the pot.

The reason was oscillation in a diamond buffer, about 200MHz. When measured THD, mains line components and LF noise increased.

The solution is to add 100R resistor before input of the diamond buffer. And 2 PCB tracks must be cut. Please see the images. I am sorry for the problem.

http://web.telecom.cz/macura/uprava.jpg

http://web.telecom.cz/macura/uprava_zespoda.jpg

http://web.telecom.cz/macura/uprava_100R.jpg
 
#363
johnYks said:
http://www.lyraconnoisseur.com/Products/Lyra_Connoisseur/files/blocks_image_1_1.png

Inspired by the Lyra Connoisseur?🙂
Click to expand...

Yes Sigurd sent me a photo, but I left out the dovetail since it looks like an old drawer, I used tap and groove instead, looks a lot nicer than seeing the joint.


Sigurd Ruschkow said:
That is a heck of a nice box you have for the preamp, Nico! I am glad I sent you the photos of the Lyra Connosieur.

Nico, I think that PMA likes low open loop gain and little feedback but your design is a high open loop design with much FB, IMO.

My version of our line amp has about 30 dB of OL gain and 1 MHz OL freq resp (-3 dB) which I think is more inlie with PMA:s design philosphy.
PMA - correct me if I am wrong here.

My THD levels are around 0.002% up to about 100 kHz at 1Vrms out and 1k load.

SIgurd
Click to expand...
BTW have you completed one, the PCB layout you did was really very good.

And yes, I remember your preference of low OGL Sigurd. My ultimate goal was zero THD.

Finally, I added only the CCS and optimised the component values of the the original design I did in 1984 for a DC - 15 MHz signal/function generator. Simulators are great design aids.

I have since replaced the transistors with the few original HP branded types I kept.

Even with the common transistors in the prototype, only H2 is bearly visible above the noise floor on the spectrum analyser in 1 Hz BW -140 dB down into 50 ohms. For all practical purposes H3 and higher order products were theoretical.
 
#364
Hi Nico,

we have some new results as well. The feedback is 12dB.
 

Attachments

  • d04_2k.png
    d04_2k.png
    68.4 KB · Views: 1,040
#368
PMA said:
Yes it is, about 4x higher. It is all about distortion cancellation in input complementary-differential stages, exact BJT PNP-NPN matching and emitter (degeneration) and bias resistor sellection. I cannot get this with JFEts.
Click to expand...

See if the design can work safely with 1.5 - 2 x the current rail voltage. THD is indirectly related to supply voltage. This is because distortion is related to second order intercept.
 
#369
I see an interesting phenomina, I did not look carefully before. You have no even order products and only H3 and H5.

If this is a real signal then look for the problem in the signal generator, it may be generating these harmonics and not the amplifier. Try putting a very high Q passive filter at the output of the signal generator.

Nico
 
#373
I had been fighting against HF oscillations in the output diamond buffer of some of the modules, and found the solution as described in

http://www.diyaudio.com/forums/showthread.php?postid=1690502#post1690502

but a 68pF cap must be added as well. Please see the photo. The modification has been tested since December 22 on several pieces successfully.

I will produce and evaluate version V3 PCB in January, with all the modifications. I will also rewrite my Dispre 2 web page.
 

Attachments

  • mod_100r_68pf.jpg
    mod_100r_68pf.jpg
    90 KB · Views: 789
#374
PMA said:
I had been fighting against HF oscillations in the output diamond buffer of some of the modules, and found the solution as described in

http://www.diyaudio.com/forums/showthread.php?postid=1690502#post1690502

but a 68pF cap must be added as well. Please see the photo. The modification has been tested since December 22 on several pieces successfully.

I will produce and evaluate version V3 PCB in January, with all the modifications. I will also rewrite my Dispre 2 web page.
Click to expand...

Make sure you also redo the measurements; if the cap is placed where I believe it is, it's going to be some significant impact on OLG, distortions and SR. I'm also curious if this change will affect the sound?
 
#375
Both simulations and measurements were already done, as well as step response measurements. And listening tests also.

IMHO the circuit is still fast enough for audio use and has flat OLG above audio band.

THD spectral analysis shows almost no difference in audio band; step response is longer (rise time 880ns) and preserves same shape for large and small signals.

Simulation shown is without capacitor in input RC filter.
 

Attachments

  • v3_plots.png
    v3_plots.png
    69.8 KB · Views: 681
#376
I would like to show the difference between oscillating and non-oscillating status in the same channel. It was pretty tough problem, not seen by scope at the output.

The 68pF cap position may seem strange, but remember they are the PCB parasitic impedances together with fast buffer driver transistors that made oscillation condition to fulfill. It oscillated locally above 200MHz, with almost no change in THD. The cap must be placed directly where it is.
 

Attachments

  • oscillations.png
    oscillations.png
    36.8 KB · Views: 1,028
#377
PMA said:
Both simulations and measurements were already done, as well as step response measurements. And listening tests also.

IMHO the circuit is still fast enough for audio use and has flat OLG above audio band.

THD spectral analysis shows almost no difference in audio band; step response is longer (rise time 880ns) and preserves same shape for large and small signals.

Simulation shown is without capacitor in input RC filter.
Click to expand...

I have no doubt that your preamp is still very good. Could you though describe the sound of the modified preamp? The impact of the two caps should be clearly audible.
 
#379
Hello all,

I have tried to describe all the modifications in my web, so please kindly visit

http://web.telecom.cz/macura/dispre2_en.htm

I am recently having problems to reply to e-mails to some of you (johnYks, burbeck, JP ....), my mails are refused by destination servers. I am trying from another address (same server as in the address), but still having problems. Probably too high security setting, sorry to be unable to reply.

Caps 68pF are NPO ceramic.
 
#380
PMA said:
Hello all,

I have tried to describe all the modifications in my web, so please kindly visit

http://web.telecom.cz/macura/dispre2_en.htm

I am recently having problems to reply to e-mails to some of you (johnYks, burbeck, JP ....), my mails are refused by destination servers. I am trying from another address (same server as in the address), but still having problems. Probably too high security setting, sorry to be unable to reply.

Caps 68pF are NPO ceramic.
Click to expand...

Something is not right... http://web.telecom.cz/macura/step_s.gif only 5V/uS? It says 0.3V/div and 400nS/div.

The cap is where I was assuming it has to be... But the 22k resistor, if installed, doesn't have a dramatic impact on the OLG (and hence THD)?
 
You must log in or register to reply here.
Top Bottom