Author Topic: Valhalla 2701c schematics + firmware V3  (Read 12608 times)

0 Members and 1 Guest are viewing this topic.

Offline RaymondMack

  • Regular Contributor
  • *
  • Posts: 63
  • Country: us
Re: Valhalla 2701c schematics + firmware V3
« Reply #75 on: August 22, 2021, 08:09:53 am »
I played around with the Tina-TI simulations again and found a few mods that look interesting.

Blue:   C6,7,8 = 1uF, C4 = 0.5uF (original filter, fastest settling time)
Green:  C6 = 1uF, C7= 2uF, C8 = 0.5uF, C4 = 1uF (lowest LF noise)
Maroon: C6,7 = 2uF, C8 = 1uF, C4 = 0.5uF (sharpest attenuation)
Gold:   C6,7 = 2uF, C8 = 0.5uF, C4 = 1uF (slow, but decent LF noise and attenuation)

Both the LF356 and OPA189 have the lowest LF noise with the "Green" filter (swapping C8 and C4 values and increasing C7 to 2uF). This also has equal or better attenuation than the original filter but has about 67% longer settling time.

The "Maroon" filter increases C6,7 to 2uF and has slightly (72%) longer settling time and more LF noise but the sharpest attenuation.

The Gold filter has the longest settling time (twice the original filter) but sharper attenuation and nearly as good as noise performance as the "Green" filter.

The primary downfall of each filter mod is the increased settling time, but since the 2701Cs are so noisy I think they might be worth testing out.

Time to reach 99.999% of input
Blue:   2.45s (original filter)
Green:  4.08s, delta = 1.63s
Maroon: 4.20s, delta = 1.75s
Gold:   4.89s, delta = 2.44s

Filter |  -10dB   |  -20dB   |  -40dB   |  -80dB   |
Blue   | 2.32 Hz  | 5.74 Hz  | 13.03 Hz | 54.97 Hz |
Green  | 1.30 Hz  | 3.54 Hz  | 11.43 Hz | 48.85 Hz |
Maroon | 1.19 Hz  | 2.75 Hz  | 7.68  Hz | 36.14 Hz |
Gold   | 1.03 Hz  | 2.54 Hz  | 8.38  Hz | 40.19 Hz |
« Last Edit: August 22, 2021, 08:20:24 am by RaymondMack »

Offline Electrole

  • Contributor
  • Posts: 40
  • Country: dk
Re: Valhalla 2701c schematics + firmware V3
« Reply #76 on: August 22, 2021, 10:32:56 am »
The noise simulations are very interesting. They definitely show that the filter design is not just about suppressing PWM-noise, but one should also keep an eye on the noise generated by the filter itself. The simulations also suggest that despite the low current noise of the OPA134 the voltage noise kicks in below some 10 Hz. The LF356 seems to have more noise above 10 Hz which could explain the improvements observed when replacing the LF356 with the OPA134. The OPA189 is a cool op-amp, but it's not available in a leaded 8-pin package, so I don't know if it's worth the effort required to put it into use in the 2701C. Personally, I would just settle for the OPA277.

It's also interesting that Le_Bassiste's 2701C wasn't equipped with any EMI measures. In my unit there were already toroidal inductors installed for each of the 8 terminals. There are inductors shown in the Rev J mechanical drawing, so these are presumably added at one point. Unfortunately the assembly drawing reveals no details on these ferrites so I took out one and measured it: It reads close to 5.39 µH + 0.10 Ohm at 100 kHz, and 5.20 µH + 7.55 Ohm at 1 MHz. The 50 µH ferrites used by Le_Bassiste may be an even better choice. If others have a 2701C with no ferrites added, it will be a good idea to add some. I don't think the exact value is essential, but I would go for ferrite inductors that are lossy over a wide frequency range rather than inductors with a high Q. At one point I also considered adding common-mode inductors inside the 2701C but eventually I went for an external filter.

Regarding the buzzing noise from the high-voltage transformer: I have a bit of that too, and the level of noise depends on the voltage setting, which changes the current draw and the firing angle of the switching controller. I do not notice the noise, not even with the lid off, so the transformer in my unit may simply be more quiet.

It may be an idea to replace the ICL7650S with an LTC1052. Again, it will not do anything to the wideband noise issues, but if the stability and the low-frequency noise are improved, then why not. I would also consider replacing the socket with a better model, or solder IC5 directly into the board. I will not have time to modify the 2701C any further right now, so I will be looking forward to seeing results from a shoot-out between ICL7650S and LTC1052  :-)

Finally, a note on CE-marking and EMI/EMC: My 2701C does not have any CE-marking, neither does the unit shown on Valhalla's home page. The question is then how much consideration Valhalla gave EMC when designing the 2701C, and how much testing of EMC Valhalla has carried out. I suspect there could be a link between the noise issues we see and potential EMC issues. Without the CE-marking Valhalla is not even allowed to market or sell the 2701C inside the European Union, as far as I understand the legislation. Perhaps that's why there are no links to sales representatives abroad on the Valhalla home page?

Offline 1audio

  • Regular Contributor
  • *
  • Posts: 176
  • Country: us
Re: Valhalla 2701c schematics + firmware V3
« Reply #77 on: August 23, 2021, 04:23:42 am »
RE EMC/EMI- I have a very early 2701 (pre-a). and it has almost no bypass caps on the digital circuits. Mine had chunks of drafting vellum as insulators for the brackets at each end of the PCB. They had broken down and were arcing at higher voltages. I get the feeling that this may have been someone's PHD or MSEE project that morphed into a business. There are aspects that even a junior engineer would have done (like the caps) that were not. It also has plastic top, bottom and front panels. Nevertheless it seems to work pretty well.

Maybe they were reluctant to make too many changes which would have needed a lot of verification on performance.

Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo