ThermalTrak+TMC amp

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

congratulations on your achievements in compensation scheme (post #623), build (post #635) and sound quality (post #653)!

I will probably build a smaller, completely different amp with comparable distortion performance at low to medium power -- with one notable difference: your high-frequency spectrum above some 100kHz is really clean (sim of asc in post #634), while my circuit will produce remarkable noise.

You report some sound improvements with the last design step. Puzzling again and again: how can one hear these small differences? Edmond Stuart reports on his Web page listening tests, where some listeners complained about certain treble harshness in one stereo channel. This then could be traced back to a different small compensation capacitor in this channel.
I can talk of similar experience with my ppm-performance design (low-power, nested MC amp with many, many stages ;-) with respect to small comensation or power supply details, e.g. snubber circuits at rectifiers.

So, my question is: You reported certain small, but clear sound quality progress over the design steps. If one rules out, that only those tiny THD figures determine sound quality: Do you have some hint or suspicion, which other reasons for the improvements one could take into account?

Congrats again and happy listening,
Matthias

PS. Some notable experts like Bob Cordell talk of clipping behaviour. Cannot really believe that; differences already show up at quite low listening levels.

Hi Matthias,
Thank you very much. I don't remember I reported some sound improvements. I said I was very satisfied with the sound. I don't think that so small differences in distortion, specially when so low, could be detected at least not by me.
The main reason I changed compensation from TMC to this kind of output included TPC was better phase and gain margin with even a bit lower distortion. First time I used that compensation in my CHA and it showed as very stable and not sensitive to the components tolerances.
I don't use exotic components, even in this case I used cheap C0G ceramic caps (previously I used silver mica) in the compensation and still the sound I expected was there.
I am trying to get nice clipping behavior but don't think it is paramount to the good sound, not in home listening.
I wish you success in your amp, I followed it but not very carefully. I suggest you to change the background color in your plots from black to white. When I see so many black rectangular and I need to open it to see what it is I lose interest.

Best wishes
Damir
 
Hi Damir,
The main reason I changed compensation from TMC to this kind of output included TPC was better phase and gain margin with even a bit lower distortion.
the straight slope with right flattening at ULGF really looks impressive.
I don't use exotic components, even in this case I used cheap C0G ceramic caps (previously I used silver mica) in the compensation and still the sound I expected was there.
Fully agree with this approach. Compensation capacitors are inside the main loop, so no reason to worry here. They need not be much better than the nonlinear transistors.
For the input circuitry with C1 and C2, *at least* good polyprops with foil-dielectric-foil construction, e.g. Wima FKP, should be used.
I'm wondering about the electrolytic cap in NFB network. Assume that offset isn't stable enough due to power dissipation in the JFET. I would try to use a decent servo.
But very probably carrying owls to Athens ;-)
I wish you success in your amp, I followed it but not very carefully ...
Thank you. Cross fingers that it will sound as nice as the ridiculous proof-of-concept design, only capable of playing a bit louder.
I will follow your background colour suggestion.

Kind regards,
Matthias
 
... Compensation capacitors are inside the main loop, so no reason to worry here. They need not be much better than the nonlinear transistors...

Bruce Hofer (the principal of Audio Precision) has tested and reports NP0 are probably better than silver mica anyway.
More than adequate even for the main loop, let alone inside it.
He has posted a nice presentation on the AP website with his comments on low distortion components.
He is one of the few people whose comments are based on serious experience at PPM or less distortion, not audiophile waffle.
So we are all in accord that Damir's NP0 capacitors are an excellent choice.

Best wishes
David
 
Compensation capacitors are inside the main loop, so no reason to worry here. They need not be much better than the nonlinear transistors.
Just want to weaken this a bit. If compensation capacities were voltage-dependent, then loop characteristics like margins could change with signal swing. Probably not a good thing.

But as pointed out, these ceramics are quite good.

Matthias
 
Bruce Hofer (the principal of Audio Precision) has tested and reports NP0 are probably better than silver mica anyway.
More than adequate even for the main loop, let alone inside it.
He has posted a nice presentation on the AP website with his comments on low distortion components.
He is one of the few people whose comments are based on serious experience at PPM or less distortion, not audiophile waffle.
So we are all in accord that Damir's NP0 capacitors are an excellent choice.

Best wishes
David

Hi David,
Could you post the link to Bruce Hofer presentation, I can't find it.
BR Damir
 
... then one really can be tempted to use florid style.

I didn't mean you use audiophile waffle, of course!
I just meant that there was so much unsubstantiated talk about capacitors.
The best other data I have seen was Samuel's tests of capacitors for ultra low distortion oscillators.
It showed much the same results as Bruce's, with NP0 down around -135 dB or thereabouts.
He had the best film capacitors a bit lower but it may depend on the particular NP0 manufacturer, there are some small differences, apparently.

The official AP link is >Here< but you do need to establish an account.

Best wishes
David
 
Last edited:
Hi David,
I didn't mean you use audiophile waffle, of course!

I'm sorry to have not been clear enough. I really didn't assume that.

But sometimes, I cannot stand the opinions about audiphiles expressed in our community. Seemingly unsubstantiated arguments can be discussed, but labeling them with "audiophile" is not helpful, in my opinion.

Sorry again for the ambiguous off-topic remark.

Kind regards,
Matthias
 
Nice work dadod! Stability in audio circuits is underrated, especially in the diy space. It's refreshing to see refinements based on an in depth analysis. Too much tinkering with audio circuits has become swapping in boutique parts and claiming audible benefit. I much prefer what you've done here. If anyone can make this amp <1ppm THD+N, it's you. Can you point me to IMD numbers/parts list?
 
Last edited:
Nice work dadod! Stability in audio circuits is underrated, especially in the diy space. It's refreshing to see refinements based on an in depth analysis. Too much tinkering with audio circuits has become swapping in boutique parts and claiming audible benefit. I much prefer what you've done here. If anyone can make this amp <1ppm THD+N, it's you. Can you point me to IMD numbers/parts list?

Hi nania,
Attached last schematic and BOM if it help you.
Damir
 

Attachments

  • DADO-TT-OITPC-triple_sch.jpg
    DADO-TT-OITPC-triple_sch.jpg
    296.8 KB · Views: 891
  • DADO-TT-OITPC-BOM.txt
    1.6 KB · Views: 170
Damir
Looking over the BOM I have some questions:
1. Is the use of the mosfet @M1 optional?
2. The C6 value indicates 47ppF, shall I assume you meant 47pF?
3. If the 0.1uF values are not typos, why didn't you use 100nF to describe them?
4. Is the BOM only for the amp or does it also include parts for your supply/protection board?
 
Damir
Looking over the BOM I have some questions:
1. Is the use of the mosfet @M1 optional?
2. The C6 value indicates 47ppF, shall I assume you meant 47pF?
3. If the 0.1uF values are not typos, why didn't you use 100nF to describe them?
4. Is the BOM only for the amp or does it also include parts for your supply/protection board?

1. You can use 15 kohm instead M1, R34, R42
2. 47 pF
3. 100 nF is the same as 0.1 uF. 1 uF is 1000 nF.

This is the amp BOM only.
 
Thanks. On very fast circuits, the grounding scheme becomes at least as important as component placement. Have you tried the following ground plane sequence?
1. tie signal (input) ground to the load (output) ground
2. tie the load ground to the power ground
3. tie the power ground to the chassis (common) ground
The idea is to keep the inputs as far from the rumbling of the caps as possible without introducing unnecessary parasitics by nature of its trace length. In order to facilitate this scheme, you may need to position your inputs on a different side of your outputs on the board which may make the final product unorthodox but the benefit of this grounding scheme has been proven to work well with chip amps. You can simulate the effect of this ground configuration by inputting the expected trace parasitic values for the length of each connection. Attention to details like this could be what gets this amp completely <1ppm if you haven't already used this scheme.
 
Thanks. On very fast circuits, the grounding scheme becomes at least as important as component placement. Have you tried the following ground plane sequence?
1. tie signal (input) ground to the load (output) ground
2. tie the load ground to the power ground
3. tie the power ground to the chassis (common) ground
The idea is to keep the inputs as far from the rumbling of the caps as possible without introducing unnecessary parasitics by nature of its trace length. In order to facilitate this scheme, you may need to position your inputs on a different side of your outputs on the board which may make the final product unorthodox but the benefit of this grounding scheme has been proven to work well with chip amps. You can simulate the effect of this ground configuration by inputting the expected trace parasitic values for the length of each connection. Attention to details like this could be what gets this amp completely <1ppm if you haven't already used this scheme.

Sorry for late answer. The layout for this amp was not very good made, later I changed approach with my CFA, but I never used it in this amp. I just used old board and changed the compensation only. I like symmetric complementary circuits now more. You can look in my CFA thread to see and comment on that layout.
 
It’s time to sell some of my power amps boards I built in the past. I have to much different power amp and I need space for new design.
I would like to offer six boards of my TT power amp. It is improved from first version with the TMC to the OITPC with corrected layout (see the photo of the backside). All tested and used for some time. The sound is excellent, powerful and clear. I used bipolar electrolytic capacitors in the feedback (C3 - 470uF) to lower THD specially at low frequencies.
Other caps used are nichicon MUSE, wima polypropilene, polystirol and ceramic NP0. ThermalTrak transistors used in four boards are NJL3281D/NJL1302D and two with NJL0281DG/NJL0302DG.
The price is 100 EUR per board plus shipping, selling in pairs only.
Damir
 

Attachments

  • DADO-TT-OITPC-triple_sch.jpg
    DADO-TT-OITPC-triple_sch.jpg
    296.8 KB · Views: 630
  • IMAG2597.jpg
    IMAG2597.jpg
    678.7 KB · Views: 592
  • IMAG2599.jpg
    IMAG2599.jpg
    604.3 KB · Views: 543
Some specification:
ULGF = 1.7 MHz
PM = 63 degree
GM = 35 dB
SR = 150V/usec

THD20k less than 1 ppm at 100W/8
THD1k less than 0.1 ppm at 100W/8
 

Attachments

  • DADO-TT-OITPC-triple_FFT20k_1W.jpg
    DADO-TT-OITPC-triple_FFT20k_1W.jpg
    173.6 KB · Views: 145
  • DADO-TT-OITPC-triple_FFT20k_100W.jpg
    DADO-TT-OITPC-triple_FFT20k_100W.jpg
    162 KB · Views: 155
  • DADO-TT-OITPC-triple_FFT1k_1W.jpg
    DADO-TT-OITPC-triple_FFT1k_1W.jpg
    163.8 KB · Views: 180
  • DADO-TT-OITPC-triple_FFT1k_100W.jpg
    DADO-TT-OITPC-triple_FFT1k_100W.jpg
    160.4 KB · Views: 268
  • DADO-TT-OITPC-triple_LG.jpg
    DADO-TT-OITPC-triple_LG.jpg
    266.3 KB · Views: 434
  • DADO-TT-OITPC-triple_FFT5k_100W.jpg
    DADO-TT-OITPC-triple_FFT5k_100W.jpg
    165.5 KB · Views: 150
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.