ThermalTrak+TMC amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
triple territory.

I have just a tad of ringing on my triple (not simmed). So you might want to simulate and leave your real world design open to a few stability enhancements.

Besides the pre/driver RC decoupling Cordell advises on , here are a few additional "tricks".

The B/C driver stopper caps are from the APT1 and the parallel L/R network is a Harmon Kardon "trick"... both reduce the VHF bandwidth of the stage without any degradation. The HK L/R works similar to the main OP inductor typical of most amps. Both mitigate negative resistances at the OP bases , augmenting the basestoppers action.

PS - make your own "L/R" by wrapping 20-25 turns of thin wire around a 100R 1/2w resistor.

OS
 

Attachments

  • triple tricks.gif
    triple tricks.gif
    55.2 KB · Views: 1,710
Last edited:
I have just a tad of ringing on my triple (not simmed). So you might want to simulate and leave your real world design open to a few stability enhancements.

Besides the pre/driver RC decoupling Cordell advises on , here are a few additional "tricks".

The B/C driver stopper caps are from the APT1 and the parallel L/R network is a Harmon Kardon "trick"... both reduce the VHF bandwidth of the stage without any degradation. The HK L/R works similar to the main OP inductor typical of most amps. Both mitigate negative resistances at the OP bases , augmenting the basestoppers action.

PS - make your own "L/R" by wrapping 20-25 turns of thin wire around a 100R 1/2w resistor.

OS

Thanks OS I'll try that.
 
With small changes in TMC is possible to get below ppm at 20kHz and 80W/6R.
Input 0.12v 20kHz
1W/6R Total Harmonic Distortion: 0.000026%

Input 1V 20kHz
80W/6R Total Harmonic Distortion: 0.000093%

Diagrams in this order:
schematic, FFT 1W, FFT 80W, OLG and CLG

Someone can ask why I use local VAS feedback R15//C22. Whitout it disstortion is even lower, but the phase is passing dengerously close to 180 degree, and with this arrangement it is much smoother.
dado
 

Attachments

  • DADO-TT-TMC-3-1_015-triple-fina2.sch.jpg
    DADO-TT-TMC-3-1_015-triple-fina2.sch.jpg
    145.8 KB · Views: 1,682
  • 1W.jpg
    1W.jpg
    210.3 KB · Views: 1,547
  • 80W-6R.jpg
    80W-6R.jpg
    228.4 KB · Views: 1,298
  • DADO-TT-TMC-3-1_015-triple-fina-OLG.jpg
    DADO-TT-TMC-3-1_015-triple-fina-OLG.jpg
    373.9 KB · Views: 1,223
  • DADO-TT-TMC-3-1_015-triple-fina-CLG.jpg
    DADO-TT-TMC-3-1_015-triple-fina-CLG.jpg
    363.8 KB · Views: 265
Finally I am going to test this amp.
Last simulation are here and this is going to be the schematic I will use for real life test.
Some simulated data are here:

THD20k at 100W//8ohm is 0.72ppm
THD20k at 1W//8ohm is 0.11ppm

Slew rate with input filter 150V/uS
Slew rate without input filter 280V/uS

FFT atached are:
40W 1kHz
1W 20kHZ
1W 1kHz
dado
 

Attachments

  • shcematic.jpg
    shcematic.jpg
    156.6 KB · Views: 754
  • 100W-8ohm-20kHz.jpg
    100W-8ohm-20kHz.jpg
    200.9 KB · Views: 710
  • 1W-8ohm-20kHz.jpg
    1W-8ohm-20kHz.jpg
    211.2 KB · Views: 298
  • Square-wave.jpg
    Square-wave.jpg
    170 KB · Views: 257
  • FFT-40W-1kHz.jpg
    FFT-40W-1kHz.jpg
    166.7 KB · Views: 302
  • FFT-1W-20kHz.jpg
    FFT-1W-20kHz.jpg
    155.4 KB · Views: 243
  • FFT-1W-1kHz.jpg
    FFT-1W-1kHz.jpg
    171.7 KB · Views: 211
Finilly I've powered my amp. I did bias adjustment and then noticed that output jumps when I touch drivers heathsink. Adding 220pF between base and collector of the drivers cured that(probably some oscilation). After that I did some measurement with sinus and square wave signals using 8ohm load. All looked OK. A power supply was temporary one with ordinary diode bridge and 2x4700uF.
I was curious to hear it and I connected it to my listening setup, one PCB only.
First impression was very good, crystal clear hights and phenomenal bass.
I will present more measurings and listenings soon.
dado
 
I've got idea to make universal PCB which I can use for more then one configuration of this amp. In this case it can be high Feedback amp, and low Feedback amp.
In this forum there are some discutions about high Feedback with very low THD, and low Feedback and higher THD but with more "benign" harmonics distribution.
Here is the PCB and simulation of two low Feedback amps, one with NJF LTP and other with BJT LTP. I think I am going to make one of both type, listen and hope to hear the differences.
dado
P.S. Important feature of the low Feedback amp is better phase and gain margin.
 

Attachments

  • DADO-TT-TMC-5.LAY].jpg
    DADO-TT-TMC-5.LAY].jpg
    295.9 KB · Views: 704
  • WideOLG-NJF-LTP.jpg
    WideOLG-NJF-LTP.jpg
    195.7 KB · Views: 762
  • WideOLG-BJT-LTP.jpg
    WideOLG-BJT-LTP.jpg
    193.4 KB · Views: 368
  • WideOLG-NJF-LTP-OLGgraph.jpg
    WideOLG-NJF-LTP-OLGgraph.jpg
    386.4 KB · Views: 269
  • WideOLG-BJT-LTP-OLGgraph.jpg
    WideOLG-BJT-LTP-OLGgraph.jpg
    390.2 KB · Views: 182
Dilemma

After listening to the triple and problem to stabilize it I am not sure if a triple is worth a hassle. I simulated here a triple and standard EF OPS.
IPS is the same for both amps, with NFET cascoded LTP, VAS is emitter follower before and emitter follower as buffer after VAS transistor and TMC is used. Global Negative Feedback is the same for both.
What you think looking at FFT for both of them? Distortion of the triple is lower but it is so low for both of them and I think no one could hear difference. Standard EF shows dominant second harmonics mostly.
dado
 

Attachments

  • dilemma.jpg
    dilemma.jpg
    260 KB · Views: 331
Yes I know it works, my worked too, read the whole thread. The question is does it give so much benefit, if we can get similar result with simpler circuit.
dado

Hi Dadod,

If there was a device with the gain and characteristics we wanted, we would most probably end up with a one stage amplifier. The whole amplifier would be made up of just one device. :)

Whether the output its 2 stage or 3 stage or single stage mosfet depends on what one is looking for, so long as it gets the job done. ;)

kind regards,

Harrison.
 
Dadod,
can you build symef and make listening tests vs your amp?
It seems like comparsions ffts gives no all knowledge of the performance.
Anyway I am impressed of thd number you got:happy1:

padamiecki,
I build only my amps now as there is no time to build all different amps presented here. Symef is with laterals very ordinary circuit(where is the formula there???), and I have JLH 80W amps(six channels modified by me) and no intention to buy expensive laterals any more.happy1
 
Hi

Sometimes in these threads the same problems come around and around ...
one reason Self's slew rates were asymmetrical was because of the CCS in the VAS stage. The trouble is with the resistor in series with the base, in dadod's circuit it is R38. This seems to have been a derivative of an old RCA circuit (used in their hybrid circuit HC1000) but was always a cheap fix to the problem of the CCS shorting the current source to the input stage if it were to saturate. Far better to use two reference diode chains and separate the bases of the CCS and the CCS for the input stage.
Although Cherry pointed this out, as well, dadod has a capacitor across the base of the CCS which offsets the speed issue to some extent. But parasitic capacitances on the base of a transistor can sometimes lead to oscillations. Better to use a low impedance (low resistance) - say less than 50 ohms - as might be provided by two forward biased diodes.
As for the VAS stage, the use of 2.2 ohm resistor is curious. To improve the slew rate it is necessary to use greater local feedback (10 ohms or more) or, preferably, to use cascodes in the VAS stage. Most VAS stages even with emitter resistors of 10 ohms or so do not seem to be able to offer the speed improvement nor linearity improvement of a cascode. Most distortion seems to arise in the VAS given other stages are optimised, and with the traditional VAS this is not satisfactorily addressed. Cascodes on the other hand offer at least double the frequency performance, and reduce Early effect distortion to negligible proportions.

DOn't know whether this helps at all, but maybe some of the old problems won't keep going round...

John
 
Hi John,
Thank you for your comments.
I would like to know about what schematic you are tollking about as I can't connect it to my schematics(problem could be that I changed the schematics during period of time).
Regarding CCS, I used simple two transistors CCS, usud by meny, JLH between them. I have JLH 80W mosfet amp were this kind of CCS was used and it sounds very good.
Please tell me the schematic you comment and I will be hapy to use your advise for improvements. I am not sure that CCS is so critical to the sound, but I may be wrong. I know for better CCS and I've got Cherry article about CCS.
dado
 
By model do you mean transistors models, or LTspice schematic and transistors models?If you whant a schematic, which one as I put here some different ones, triple, standard EF or low GNFG?
dado

I mean the LTSpice schematic. I noticed that you've gone through several iterations and would like to see your latest and greatest. The THD numbers on your circuit are outstanding.

In particular your JFET input stage and VAS topology are very interesting. The use of ThermalTrak makes this a very nice amplifier design!

How does your prototype sound? Have you designed a PCB?

/Mason
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.