OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #41
Excellent! What changes did you make to the caps? Would love to incorporate any improvements into the design.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
DIY amp testing (with AES17 filter)
- Thread starter mcdn
- Start date
I will organize a boom later tonight. I honestly don’t know why our measurements vary so much. I don’t think I’m using any supa-premium component. It costs less than $40.
I was pretty worried about EMI issue from inductances but with measurement at this level, not so much anymore. I will probably make two of these with an aluminum case at some point.
I measured a 3e audio tpa3255 today using this LPF. Without the LPF, it looks like the second graph(the point is the LPF is doing its job).
Attachments
OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #43
Why do you say our measurements differ? We both have approximately -120dBA N+D, which is the limit of my source (Topping D10 balanced).
Sorry, I'm super new here, but am I missing something? REW has AES17 2015 notch.. why the need for the DIY filter?
I've been doing amateur measurements with "Distortion Factory" on Facebook. but haven't posted any Class D amps yet.
I have been using voltage divider resistors with 10uf film caps in front of that with great results I think.
(unrelated picture of my 2i2 loopback. realizing just now I had my microphone cal activated. maybe that's why -6db was best)
I've been doing amateur measurements with "Distortion Factory" on Facebook. but haven't posted any Class D amps yet.
I have been using voltage divider resistors with 10uf film caps in front of that with great results I think.
(unrelated picture of my 2i2 loopback. realizing just now I had my microphone cal activated. maybe that's why -6db was best)
Attachments
OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #45
All good, learning and fun is what we are here for!
The AES17 standard describes how to measure digital audio equipment. It gets down to details like using 997Hz instead of 1KHz for test tones, and specifies standard centre frequencies for octaves (e.g. octave 3 is 125Hz). It also defines a couple of different standard filters. One is a "notch filter", the other is a "low pass filter".
The notch filter is a filter centred on 997Hz that attenuates the test signal and its inevitable "skirts" due to the FFT windowing. This increases the accuracy of broadband THD+N measurements by taking the relatively large test signal out of the equation. This is the AES17 2015 notch option in REW (2015 being the date of the standard doc used - it was updated in 2020)
The low pass filter is there to remove high frequency noise generated by digital amps from the audio band measurements. This is what the circuit described in this thread does. It has to be very flat in frequency response from 20Hz to ~30KHz, and then have a huge -60dB drop in FR by say 250KHz. This is in order to remove any influence of the typical 400KHz switching noise from digital amps.
Texas Instruments has a nice white paper on the topic: https://www.ti.com/lit/an/slaa114/slaa114.pdf?ts=1705717439615&ref_url=https%3A%2F%2Fduckduckgo.com%2F
In short, the notch filter in REW and this filter are doing different jobs.
I made few boards of DIY AUX0025 last month and also a LPF designed by Ivan. So here is a comparison.
My setup is APU-Scalar-ADCiso. The first FFT is SMSL D6s without any LPF. Second one is with Ivan’s LPF at 0dB attenuation level. The third FFT is with DIY AUX0025. The ADCiso is set to 6.7V level.
It looks like diy aux0025 adds considerable amount of H3. Also, it seems to be very sensitive about EMI, the results vary a lot from different positioning.
My setup is APU-Scalar-ADCiso. The first FFT is SMSL D6s without any LPF. Second one is with Ivan’s LPF at 0dB attenuation level. The third FFT is with DIY AUX0025. The ADCiso is set to 6.7V level.
It looks like diy aux0025 adds considerable amount of H3. Also, it seems to be very sensitive about EMI, the results vary a lot from different positioning.
Hipocrates
Active Member
I was working on a compact version the other day, I forgot to finish it
Hipocrates
Active Member
It was just a practice with Kicad, but if there's some interest I could post a BOM and gerbers here, just need to finish some details.
I can send you one of these pcbs for free(but you pay shipping. I'm in NY). My pcb will need one jumper though. I did a sloppy work when changing the size of the mounting hole of those bigger caps and forgot to change something.
OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #51
Hi! Thanks so much for building this, the more people do the more we can improve it! Is a KiCad project available for Ivan's in-line design?
In my testing there's negligible difference between the filtered and unfiltered distortion. H3 is actually 1dB lower. But I may be limited by the performance of the D10 balanced DAC, perhaps a DAC with lower distortion would show a difference? It's all a bit academic anyway, as the best class D amps are more like 110dB SINAD.
BTW it would be helpful if you could use dBr in posted graphs, and limit to -160dBr, it helps with comparisons and keeping visual scaling consistent!
Last edited:
Here is a set of test.
The first three are measured with D6s-APU-Scalar-ADCiso, directly, with Ivan’s LPF, and with DIY AUX0025 respectively.
The second set is measured with D6s-ADCiso, directly, with Ivan’s LPF, and with DIY AUX0025 respectively.
I have to set the lower bound lower (-180dB) otherwise it’s not showing much of the noise floor.
I will ask Ivan if it’s fine to share it here since it’s all his work.
Edit: Ivan authorized me to share the schematic and gerbers.
Attachments
Last edited:
Thank you and @IVX for sharing. Great work! It would be great if you could share the BOM for the actual layout because of the best part fit and reproducing the test results.
OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #54
Ah I see now, that’s a nicely designed LC filter but it doesn’t have the steep cutoff of an AES17 filter. You can’t really compare the two.
OP
- Joined
- Mar 7, 2020
- Messages
- 1,210
- Likes
- 1,928
- Thread Starter
- #55
That’s really nice!
Hipocrates
Active Member
I will ''OCD" it and put the schematic, BOM and Gerbers tonite.
I forgot to add that this is a great and fun project @mcdn, thanks!
Last edited:
Is it possible to add a switch and few resistors before the LPF circuit to attenuate the voltage? (Adding resistors after the LPF will bring more thermal noise due to higher resistor values?)
I will share it tonight. I think resistors do not matter much. It’s the capacitors that matter.
Hipocrates
Active Member
Sure, can you be more specific? Tell me more.
I honestly don't know, but if you elaborate a little bit I will and fix it
I'm not much a designer and more an enthusiast of build and fix stuff.
Something like a U-Pad attenuator. I think a series of 200ohm with a switch for R1 and 200ohm for R2 like what Ivan did will be good? I never simulated it though.
I think Ivan’s LPF is good enough. Although it’s not as steep as aux0025, most of the modern Class D AMP doesn’t need such a steep attenuation anyway.
Ivan’s filter is also designed to be used to measure Dynamic Range of DACs with super high precision with Cosmos APU. Without the filter, when measuring some DACs, APU behaves crazy because of the slight amount of HF noise.
For the AUX0025, I’m more worried about its EMI sensitivity, I need to change its position every time to capture a FFT without much lower frequency noise, I’m suspecting that the inductance used is causing the issue.
Attachments
Similar threads
