Again, please read carefully:
"Using 2 separate ladders for each channel (one for the positive and one for the negative rails, that's the sign magnitude notation), you can add 1 bit of precision, to reach 14 bit."
So with sign magnitude notation you add 1 bit of accuracy (or monotonicity if you want) but the math is the same.
Please, take a look at R2R DAC INL and DNL.
But if you are so sure, you shouldn't had any difficult to mathematically demonstrate what you are claiming.
You can calculate the THD using a spreadsheet.
The DAM1021 ladder schematic could help you to do the math.
"Using 2 separate ladders for each channel (one for the positive and one for the negative rails, that's the sign magnitude notation), you can add 1 bit of precision, to reach 14 bit."
So with sign magnitude notation you add 1 bit of accuracy (or monotonicity if you want) but the math is the same.
Please, take a look at R2R DAC INL and DNL.
But if you are so sure, you shouldn't had any difficult to mathematically demonstrate what you are claiming.
You can calculate the THD using a spreadsheet.
The DAM1021 ladder schematic could help you to do the math.
Attachments
Last edited:
Soekris uses 0.01-0.02% resistors,
so in a pure R2R DAC that gives about 13-14bit precision (1/2^13=0.00012*100=0.012%)
13 bit has SNR of about 78dB
Still Soekris manages to get about -110db on the analog output, which is 18bit.
This should be impossible according to you reasoning with 0.01 resistors.
This extra resolution that you get is due to the sign magnitude design.
so in a pure R2R DAC that gives about 13-14bit precision (1/2^13=0.00012*100=0.012%)
13 bit has SNR of about 78dB
Still Soekris manages to get about -110db on the analog output, which is 18bit.
This should be impossible according to you reasoning with 0.01 resistors.
This extra resolution that you get is due to the sign magnitude design.
I'm still waiting for your math to demonstrate what you claim, words are not math.
As I said you can do the THD math with a spreadsheet and the schematic I have attached can help you to understand how a Collinear DAC works.
As I said you can do the THD math with a spreadsheet and the schematic I have attached can help you to understand how a Collinear DAC works.
I'm waiting for an explanation of your claim that the DAM1021 with 0.01-0.02 resistors should only have 13bit 78db resolution, bust still manages 17bit 100db
You are still confusing resolution wit precision (or accuracy or monotonicity), which are two totally different concepts.
As I said previous, please take a look at DAC INL and DNL to understand.
As I said previous, please take a look at DAC INL and DNL to understand.
Déjà vu.
The topic stuck so if I may..
Maybe because of 64 times oversampling which gives additional 3 bits. 14+3=17
That’s why I ask Soren about dam1021 rev.7 FIR2 and shift registers sample rate change.
Oversampling - Wikipedia
Not yet tested, we are still working on the firmware and on the software of the FIFO Lite.
It is a long and complex job due to the versatility of the front end.
It is a long and complex job due to the versatility of the front end.
I have a Soekris 1321 DAC that is not making output. I thought I'd try updating the firmware. The micromanager on my DAC is 1.14 and I believe current is 1.20. When I try to update using the version from the Soekris website, I am able to download but get "Invalid Image" error when attempting to update. Any suggestions? Thanks.
Hey Søren! I am really keen to get my hands on one of the new rev. 7 boards, but the US/Canada web store has not had stock for quite some time. Any word on when they will be available to purchase?
Thanks so much!
I had rev7 boards, and the firmware is done. But to reduce risk I didn't make that many in the first series, and then a couple of my volume customers needed boards and emptied my stock....
I'm about to make the next production series, and this time there should be be enough for everybody. Shipping in three weeks, keep an eye on the soekris website....
But it is also getting harder to source parts, as everybody should know now there is a worldwide semiconductor shortage, official Xilinx leadtime for the Spartan-6 is now at 53 weeks !!!
It's more to do with the oversampling. Same idea behind sigma-delta, just applied differently. Also, there's THD, IMD, SNR, INL, DNL, that all contribute to ENOB. I assume the last two aren't too great for this design, but for audio (no DC levels), that's no concern.
Last edited:
I got a rev 4 and ordered some polymer caps today. So, to confirm. Danny_66 showed how to easily place them on the 3,3V Vout, on J9. https://www.diyaudio.com/forums/ven...magnitude-24-bit-384-khz-482.html#post6496437
Then there is also the cap on Vin, on J2. But where exactly have you connected this? On J2 Pin 1, 4, 7 and 10 are GND, pin 2 is PWR A-, 3 is PWR A+, 5 is PWR -5V, 6 is PWR +5V, 8 is PWR 3.3V and 9 is PWR 1.2V. I am quite sure it is to PWR A+ (and gnd)?
many thanks, Erik