phofman, not yet, but I got their samples, and they are very accurate. I decided to buy 2000pcs.
Oh, one funny detail about that XLRs, they are not an exact copy of Neutrik, and after Lianzhan provides 3D step models(very good idea to supply steps for me, and for Lianzhan as advertising of their XLRs - you asking after reading "Lianzhan" logo, right?), I did modify my PCB and mechanical parts accordingly. And also, I wish to have a bit more noticeable gold-platting on its pins too, however, XLR is so huge that usually, clear nickel is fine(like my SYS2522 no any gold on XLRs).
Oh, one funny detail about that XLRs, they are not an exact copy of Neutrik, and after Lianzhan provides 3D step models(very good idea to supply steps for me, and for Lianzhan as advertising of their XLRs - you asking after reading "Lianzhan" logo, right?), I did modify my PCB and mechanical parts accordingly. And also, I wish to have a bit more noticeable gold-platting on its pins too, however, XLR is so huge that usually, clear nickel is fine(like my SYS2522 no any gold on XLRs).
Thanks for the valuable info. Low contact quality has direct impact on H2/H3 distortions (showed up clearly in my measurements with no-brand TRS jacks replaced by neutriks). The question is about longevity of the plating, hard to tell without prior experience with the manufacturer or some specialized lab measurements...
oh, trust me, this is not the biggest problem that I try to solve in that design. I see some hump on the 3HD ratio plot at -27db, 2HD, and 4..15HD are smooth. Of course, 9822 ADC isn't for -20-30dbfs measurements due to it has excellent THD at -.1dbfs already(and my front-end has 8 voltage ranges for that) but anyway, I try to realize the root cause of that hump. Probably that coming from my MOSFET attenuator yet.
I would be happy if real life was so simple 
Last week I did fight with some uncompensatable H2, and the reason was found in the power rail modulation. Also, harmonics drifts with the PCB temperature. My ADC has no enough resolution, or perhaps too high input impedance, to care about the THD difference Neutrik vs Lianzhan.
Last week I did fight with some uncompensatable H2, and the reason was found in the power rail modulation. Also, harmonics drifts with the PCB temperature. My ADC has no enough resolution, or perhaps too high input impedance, to care about the THD difference Neutrik vs Lianzhan.I am not saying that Lianzhan is worse compared to Neutrik, but that low-quality connectors cause "random" distortions which do not yield to digital compensation. I would be happy to try Lianzhan in my project, every spared dollar in BOM counts. But the current shipping costs to outside of China tend to equalize the final prices...
You ask Lian directly lian@hireyele.com tell her Ivan E1DA recommended
Actually, I asked her about customized XRLs with +50% more gold or so, she didn't refuse but mentioned that it will affect the price a lot.
Actually, I asked her about customized XRLs with +50% more gold or so, she didn't refuse but mentioned that it will affect the price a lot.
Last edited:
Ok, so I finished studying the op amp choices for the ES9038PRO in mono mode, schematic was attached few pages up. The topology is two groups of 4 outputs in parallel, summed up in the output op amp, for each balanced output, the total is 6 op amps, two duals (I/V conersion) and two singles (summing) for a full balanced output.
The nature of this configuration is the high demand for output current in the I/V conversion op amps. The need to avoid the ESS "hump" requires using a common voltage (Vcm) for the I/V stage that is less than AVCC/2 (I used 1V, AVCC/2=1.65V). On the other side, 4 ES9038PRO outputs in parallel have a source impedance of about 50ohm, so the I/V resistors can't be larger than 50ohm, for an inverting unity gin in the I/V stage. Therefore, the requirements for the I/V op amp are pretty tough; not only ultra low distortions, but also able to drive loads down to 50ohm. OTOH, the shifted Vcm=1V triggers a 0.65V/0.05Kohm = 13mA DC current at the op amp output (it-s actually output sinking current) which adds to the peak 1.65V/0.05Kohm=33mA peak current at 0dBFS output. So the maximum peak output current for the I/V op amps reaches 46mA, which is pretty high.
Out of the TI catalog, there are not many ultra low distortion op amps that would satisfy this output current requirement (also having decent bandwidth and SR). I identified 3 candidates
- OPA1622, this was the original choice. Excellent output current capability, is characterized in the data sheet down to 32ohm load, but is not in particular high spped or very large bandwidth. Comes only in a weird 10 pin VSON case, incompatible with other popular op amps.
- OPA1612, excellent in all DS parameters, but has maximum output current limited to some 50mA, too close to the requirement to my taste.
- OPA1656, also excellent in all DS parameters, and has 100mA output current, looks like a great candidate.
For the first two I/V options, the summing op amp was a OPA1611. For the latter, it was a OPA1641 in MSOP package (because the board real estate did not allow accommodating 4 SOT8 packages). I'll save posting the measurement charts for a day I'll have more time in my hands and show the results and my commets for now. All numbers are given after carefully compensating the 2nd and 3rd harmonics using the ES9038PRO registers. Interesting enough, the register values were not much different when switching op amps, meaning that the ES9038PRO had to compensate more for the PCB layout, rather than the op amps. The I/V and summing stages were fed at +/-10V (I went from -7 to +15 and did not note any performance hit, +/-10V was chosen as a compromise to keep the power dissipation under control). All measurement were at 1KHz.
OPA1612
Output THD was around -122dB and SNR -112dB, and excellent result, these were achieved at an output of about -10dBFS, that is, a peak output current (sinking) of about 25mA. Over -10dBFS the THD starts to slowly increase, but is still under -100dB up to about -1dBFS where they sharply increase to -70dBFS. This maps to an output peak current (sinking) of about 43mA, which explains it. This current is at the very limit of this op amp capability, in particular at the temperature of about 80C (power dissipation is quite high too, at those output levels.
OPA1656
Output THD was around -118dB and SNR around -111dB, at an output level of about -15dBFS, that is, a peak output current (sinking) of about 18mA. Over -15dBFS, the THD increases rather monotonically, up to about -100dB at 0dBFS. Although this op amp is claimed to deliver 100mA output current, the data sheet does not characterize this op amp with low output loads (like 50ohm here). Unfortunately, it appears the distortions are significantly increasing with the output current.
OPA1622
Output THD was around -122dB and SNR -112dB, and excellent result, achieved at an output level of about -5dBFS, that is, a peak output current (sinking) of about 33mA. Over this level, the distortions were slowly increasing, up to about -105dB at 0dBFS. The DS doesn't show a distortions vs. output level at lower that 600ohm load, ut by interpolating the THD plots it can be inferred that what I got is to be expected from this op amp.
A few conclusions:
- OPA1622 was, not unexpected, the clear winner. I am not aware of any other op amp on the market that has the same combination of parameters, making it an almost ideal candidate in this application. It's VSON casing, although incomfortable to work with in an DIY environment ahas the advantage of a cooling pad, helping with the power dissipation.
- OPA1612 was the runner up; excellent performance, if it would have an extra 10-15% output drive capability, would be the winner. This op amp is the ideal candidate, if the toplog would be cnaged to paralleling two outputs pe channel, followed by two summing stages per balanced output. That's a total of no less than 14 op amps per balanced output (7 chips of dual op amps) which is a little crazy cost wise, and to lay out on a PCB (in my opinion).
- OPA1656 behavior was rather disappointing; I was not expecting an op am with 100mA output capability to start raising its distortions at as low as some 33mA output current. Makes me wonder what was the purpose of including a beefy output stage on this chip, if one could not fully benefit from it's high output capability? OPA1656 could though be considered as ideal for the much less demanding ES9038Q2M I/V conversion; although it would likely be at par with the OPA1612 (and probably others from the TI audio op amps offer) it s really low cost (3 times cheaper than the OPA1622) which is a competitive advantage.
Funny enough, I've seen on EBay the Chinese offer of dual mono ES9038PRO DACs (for $500 including shipping, I would not call them cheap), fitted with socketed JRC5532 op amps (38mA short circuit output current) in the I/V stae (same topology as above). Easy to imagine what level of performance would this provide, confirming once again these chinese guys have no ideea what they are doing on these DAC boards. To add insult to injury, I am not aware of any decent op amp for this application, coming in DIP8, hence messy adapters are required, etc... And duuble funny, the manufacturers are advertising "film caps decoupling", like that would fix the poor op amps choice in this application (I used exclusively X7R ceramics for decouplings, both on the analog and digital sides).
I'll keep updating here, time permitting. The next step will be a huge integration work with XMOS and a programmable clock source.
The nature of this configuration is the high demand for output current in the I/V conversion op amps. The need to avoid the ESS "hump" requires using a common voltage (Vcm) for the I/V stage that is less than AVCC/2 (I used 1V, AVCC/2=1.65V). On the other side, 4 ES9038PRO outputs in parallel have a source impedance of about 50ohm, so the I/V resistors can't be larger than 50ohm, for an inverting unity gin in the I/V stage. Therefore, the requirements for the I/V op amp are pretty tough; not only ultra low distortions, but also able to drive loads down to 50ohm. OTOH, the shifted Vcm=1V triggers a 0.65V/0.05Kohm = 13mA DC current at the op amp output (it-s actually output sinking current) which adds to the peak 1.65V/0.05Kohm=33mA peak current at 0dBFS output. So the maximum peak output current for the I/V op amps reaches 46mA, which is pretty high.
Out of the TI catalog, there are not many ultra low distortion op amps that would satisfy this output current requirement (also having decent bandwidth and SR). I identified 3 candidates
- OPA1622, this was the original choice. Excellent output current capability, is characterized in the data sheet down to 32ohm load, but is not in particular high spped or very large bandwidth. Comes only in a weird 10 pin VSON case, incompatible with other popular op amps.
- OPA1612, excellent in all DS parameters, but has maximum output current limited to some 50mA, too close to the requirement to my taste.
- OPA1656, also excellent in all DS parameters, and has 100mA output current, looks like a great candidate.
For the first two I/V options, the summing op amp was a OPA1611. For the latter, it was a OPA1641 in MSOP package (because the board real estate did not allow accommodating 4 SOT8 packages). I'll save posting the measurement charts for a day I'll have more time in my hands and show the results and my commets for now. All numbers are given after carefully compensating the 2nd and 3rd harmonics using the ES9038PRO registers. Interesting enough, the register values were not much different when switching op amps, meaning that the ES9038PRO had to compensate more for the PCB layout, rather than the op amps. The I/V and summing stages were fed at +/-10V (I went from -7 to +15 and did not note any performance hit, +/-10V was chosen as a compromise to keep the power dissipation under control). All measurement were at 1KHz.
OPA1612
Output THD was around -122dB and SNR -112dB, and excellent result, these were achieved at an output of about -10dBFS, that is, a peak output current (sinking) of about 25mA. Over -10dBFS the THD starts to slowly increase, but is still under -100dB up to about -1dBFS where they sharply increase to -70dBFS. This maps to an output peak current (sinking) of about 43mA, which explains it. This current is at the very limit of this op amp capability, in particular at the temperature of about 80C (power dissipation is quite high too, at those output levels.
OPA1656
Output THD was around -118dB and SNR around -111dB, at an output level of about -15dBFS, that is, a peak output current (sinking) of about 18mA. Over -15dBFS, the THD increases rather monotonically, up to about -100dB at 0dBFS. Although this op amp is claimed to deliver 100mA output current, the data sheet does not characterize this op amp with low output loads (like 50ohm here). Unfortunately, it appears the distortions are significantly increasing with the output current.
OPA1622
Output THD was around -122dB and SNR -112dB, and excellent result, achieved at an output level of about -5dBFS, that is, a peak output current (sinking) of about 33mA. Over this level, the distortions were slowly increasing, up to about -105dB at 0dBFS. The DS doesn't show a distortions vs. output level at lower that 600ohm load, ut by interpolating the THD plots it can be inferred that what I got is to be expected from this op amp.
A few conclusions:
- OPA1622 was, not unexpected, the clear winner. I am not aware of any other op amp on the market that has the same combination of parameters, making it an almost ideal candidate in this application. It's VSON casing, although incomfortable to work with in an DIY environment ahas the advantage of a cooling pad, helping with the power dissipation.
- OPA1612 was the runner up; excellent performance, if it would have an extra 10-15% output drive capability, would be the winner. This op amp is the ideal candidate, if the toplog would be cnaged to paralleling two outputs pe channel, followed by two summing stages per balanced output. That's a total of no less than 14 op amps per balanced output (7 chips of dual op amps) which is a little crazy cost wise, and to lay out on a PCB (in my opinion).
- OPA1656 behavior was rather disappointing; I was not expecting an op am with 100mA output capability to start raising its distortions at as low as some 33mA output current. Makes me wonder what was the purpose of including a beefy output stage on this chip, if one could not fully benefit from it's high output capability? OPA1656 could though be considered as ideal for the much less demanding ES9038Q2M I/V conversion; although it would likely be at par with the OPA1612 (and probably others from the TI audio op amps offer) it s really low cost (3 times cheaper than the OPA1622) which is a competitive advantage.
Funny enough, I've seen on EBay the Chinese offer of dual mono ES9038PRO DACs (for $500 including shipping, I would not call them cheap), fitted with socketed JRC5532 op amps (38mA short circuit output current) in the I/V stae (same topology as above). Easy to imagine what level of performance would this provide, confirming once again these chinese guys have no ideea what they are doing on these DAC boards. To add insult to injury, I am not aware of any decent op amp for this application, coming in DIP8, hence messy adapters are required, etc... And duuble funny, the manufacturers are advertising "film caps decoupling"
, like that would fix the poor op amps choice in this application (I used exclusively X7R ceramics for decouplings, both on the analog and digital sides).I'll keep updating here, time permitting. The next step will be a huge integration work with XMOS and a programmable clock source.
Considering the performance Gustard managed to squeeze out of the ES9068AS with their X16 DAC, I am wondering if/when that part will become available (with a full datasheet). It already has a product page on mouser (although it is still somewhat empty), so it might only be a matter of time? ES9028/38 Pro are just so expensive.
Only 50mA short circuit current, not good enough. But actually using a diff in/out opamp for I/V conversion could be an option, but comes at extra complexity and cost (these diff in/diff out are not in particular cheap, too). Still, two stages are required to implement the output filter in between.
Thanks, that's the first time I've seen the price for this new chip. Any idea about the performances? I don't see anything mentioned in the product brief. But the "33mW operating power consumption" clues toward a chip primary intended for mobile devices, meaning it won't be a top performer. However, I see the X16 performances as excellent, so then the next question is, why would ESS sabotage their own flagship product, ES9038PRO?
I guess we have to wait until the data sheet will leak or made available (not holding my breath). Anyways, "the hump" is present in this chip, so ESS couldn't care less about.
I guess we have to wait until the data sheet will leak or made available (not holding my breath). Anyways, "the hump" is present in this chip, so ESS couldn't care less about.
Last edited:
That is really funny, because I saw that a month ago it showed 17 and I ordered 10 of them too... I don't understand how the stock is 17 still. I also don't know why Mouser has some unknown quantity but my Cirrus rep couldn't get samples. Of course, I probably will only get around to using them months from now...