All connections were made with solid 20 awg, I connected the ground from the nearest 1 uF film capacitor hole as you suggested. 3.3 V is ran from the AVCC regulator. AVCC bias left at the stock AVCC/3. Connections are definitely rather short but I think if I was going to do something final with a board I would mount the I/V boards above the DAC board to keep all the connections as short as possible.
Will definitely be interesting to see what the noise looks like when my Cosmos ADC arrives.
Michael
Will definitely be interesting to see what the noise looks like when my Cosmos ADC arrives.
Michael
Attachments
Yes, the COSMOS ADCs are very impressive, but Ali only has the B's in stock. Will be great for AP level measurements anyway.
The opa1612 seem to be really great, I had replaced the LM4562 with these for the SE ouputs.
Good to hear Michael!
@marcelooms not that its not necessary with opa1632, its that its not a straight forward option, as they dont have an exposed non-inverting input to achieve it easily. it is doable with them, but TP probably just left it out.
I would not say that -101dB is exactly close to -110dB, but it is much better. I dont have info (for 9016) on what the pure 2nd H THD (without noise) of -119 is close either, but I wouldnt think so. typically these would also all be minimum 10dB better and those will not be so limited, without noise.
@marcelooms not that its not necessary with opa1632, its that its not a straight forward option, as they dont have an exposed non-inverting input to achieve it easily. it is doable with them, but TP probably just left it out.
I would not say that -101dB is exactly close to -110dB, but it is much better. I dont have info (for 9016) on what the pure 2nd H THD (without noise) of -119 is close either, but I wouldnt think so. typically these would also all be minimum 10dB better and those will not be so limited, without noise.
actually no, its not that much better is it Michael? its about the same total, just with a more suitable output voltage and some harmonics higher, some lower and a lower pop on turn on/off. it is much better than the 9038 board, but it always was.
OK went back and looked. ~5dB improvement, not bad. wasnt much difference between the 2V and 4V output measurements.
OK went back and looked. ~5dB improvement, not bad. wasnt much difference between the 2V and 4V output measurements.
Last edited:
What are the supposed benefits of AVCC bias? I thought it was to take off strain from the DAC, and thus lower distortion?
Without any AVCC-bias, the DC offset is exactly 1.3V on all DAC lines, using the IVY III board. I wonder therefore, what good an AVCC/3 - i.e. 1.1V - can do? In my un-electronically educated mind that would mean that the DAC output fets would have to provide even more current to fight the lower voltage from the voltage divider ??
Yeah, I would have said 5-6 dB THD improvement which in my books isn't a huge improvement. I think that elevated 3rd harmonic after the removal of the feedback resistors may have been the result of the ADC and the higher input voltage, when I reduce the level in REW by 6 dB to get a 2 vrms output I still end up with a -130 dB 3rd harmonic and the overall THD is just a tad better.
Will be interesting to see if / how the noise performance changes. I also popped in some op amps I had laying around and measured THD at 1 kHz at 2 V (-6 dB) and 4 V, here are the results.
OPA1612
4V: -111
2V: -110
NJM2114D
4V: -113
2V: -107
LM4562
4V: -110
2V: -109
NE5532:
4V: -108
2V: -105
NJM2068DD
4V: -106
2V: -107
Michael
it
appears you still have no idea what this is all about and i've had about enough time trying to educate you.
Thanks for the lovely complement so nice of you. I know the dac itself is biased. Please read. Or don't. This is a hobby for me. I have no desire to deal with your rude ways. Bye
All of the answers to your questions have been posted multiple times. that you either dont understand, or wilfully misunderstand, because it doesnt suit your feelings of everything being fine; really isnt my problem.
I've spent literally HOURS here outlining the problems, correcting misunderstandings, even explaining ohms law and trying to help solve them where possible, even though IMO its a lost cause; yet you still act like a kid whos been told his toy car isnt as fast as the new model.
I've spent literally HOURS here outlining the problems, correcting misunderstandings, even explaining ohms law and trying to help solve them where possible, even though IMO its a lost cause; yet you still act like a kid whos been told his toy car isnt as fast as the new model.
Last edited:
InspectorGadget did describe Syn08 findings about ESS dac output stage bias earlier in this thread.
To summarize, if left floating ESS dac outputs are ramped up to AVCC/2 by the dac chip. ESS uses op-amp I/V to hold the dac outputs at that voltage. Other people have found that hump distortion changes and can be improved by reducing Vref. Doing so introduces a DC offset current in the dac output that the I/V amp must supply. For such reasons some ESS dac designs set Vref to AVCC/3. For ES9038PRO, Syn08 found that going a bit lower offered some further improvement in measurements. I have not seen any discussions at to whether or not SQ is at all affected.
To summarize, if left floating ESS dac outputs are ramped up to AVCC/2 by the dac chip. ESS uses op-amp I/V to hold the dac outputs at that voltage. Other people have found that hump distortion changes and can be improved by reducing Vref. Doing so introduces a DC offset current in the dac output that the I/V amp must supply. For such reasons some ESS dac designs set Vref to AVCC/3. For ES9038PRO, Syn08 found that going a bit lower offered some further improvement in measurements. I have not seen any discussions at to whether or not SQ is at all affected.
With a DIYINHK 9016 board on its way to me I have been doing some digging on the subject. Yes, I realise that, justifiably, its not seen as a well implemented multi-channel DAC here, but its certainly the cheapest (I already everything else other than the DAC/XMOS boards) if not the only option available to me at the moment. Maybe a Motu Ultralite in my future but not for a while.
So regarding the output stage (I won't call it an I/V stage), from what I have read I believe that DIYINHK is treating the DAC as voltage output and the output stage is intended simply to convert the differential voltage into a single-ended output. Whether or not this is achieved in the best manner with this implementation is another matter. I have found circuits for such a stage and they pretty much match what is being done here, won't bore you with the circuits I found.
Cut and pasted from another DIYAUDIO posting:
So it is valid to treat the 9016 as voltage output but only with a degraded THD+N. This is supported by Michael's measurements with the Ian Canada I/V stage which shows 5-10 db improvement with a true I/V stage.
Do you think this is a reasonable assumption (or has this assumption already been made)? I have an unused DIYINHK ES9018K2M board lying around and it has the same output stage, I also see other ESS9018 family DACS out there (cheaper Chinese ones) with similar output stages so not that unusual.
Apologies in advance for my lack of technical knowledge
So regarding the output stage (I won't call it an I/V stage), from what I have read I believe that DIYINHK is treating the DAC as voltage output and the output stage is intended simply to convert the differential voltage into a single-ended output. Whether or not this is achieved in the best manner with this implementation is another matter. I have found circuits for such a stage and they pretty much match what is being done here, won't bore you with the circuits I found.
Cut and pasted from another DIYAUDIO posting:
So it is valid to treat the 9016 as voltage output but only with a degraded THD+N. This is supported by Michael's measurements with the Ian Canada I/V stage which shows 5-10 db improvement with a true I/V stage.
Do you think this is a reasonable assumption (or has this assumption already been made)? I have an unused DIYINHK ES9018K2M board lying around and it has the same output stage, I also see other ESS9018 family DACS out there (cheaper Chinese ones) with similar output stages so not that unusual.
Apologies in advance for my lack of technical knowledge
I have some updated measurements with the Cosmos ADC -> https://www.audiosciencereview.com/...hk-dac-measurements.23629/page-2#post-1068777. The stock output stage actually gave surprisingly good performance but Ian Canada I/V + OPA1612 was even better.
Michael
Michael