ESS has quietly rolled the ES9022 over to the ES9023. While the product brief link on ESS's DAC page is broken I contacted Bryan at Shaw Electronics and got the datasheet. While ESS no longer requires an NDA to obtain datasheets one still does need to contact the distributor to get them so I suppose I probably shouldn't get into too much detail on an open forum. Suffice to say the ES9023 looks VERY similar to the ES9022.
I am somewhat fascinated by these chips and am contemplating doing a build with them since I think they're pretty cool. In particular, I find myself wondering
1) What exactly the difference between the ES9023 and ES9022 is.
2) What actual minimum load impedance is. Both datasheets say 5k but parts of the 9023 datasheet can be interpreted to indicate 3k or even 700 ohm. And 5k doesn't make much sense since in most applications you'd have to buffer the 9022/9023 with an op amp, rather defeating the purpose of having an op amp integrated into the DAC. I've found remarks in various threads indicating folks are driving headphones directly with the 9022. No measurements, though, and the 'phones impedance typically isn't stated---this is of the clearer threads, calling out a 70 ohm impedance. Personally, if the parts are happy with a 1k load then I'd be happy too.
3) How THD varies as a function of output level. -94dB typ at 0dBFS is pretty decent but most DACs exhibit flat THD with falling output level to some corner level with a linear increase below that (figure 24 in the PCM1795 datasheet is a good example). I'm curious if this is the case for the 9022/9023 and what the corner level might be.
4) How sensitive the integrated op amp is to output loading. Unlike most op amps there's no stable capacitive load spec and, curiously, application diagram calls out a first order "RC" lowpass on the output with 0 ohm resistors and 4.7nF caps rather. Quite different from the usual 2+kOhm source impedance second or third order filters called out for most DACs. Or the 22-100 ohm resistors used to prevent an interconnect driver op amp from oscillating. I like the low source impedance, but the filter pole with a 0R is high enough I'm kinda curious why the output caps are there at all.
5) What the best point to tap ground is if one wants to approximate a balanced output. I'd normally expect this would be AGND as connecting to Vreg would increase the amount of interconnect noise coupled into the DAC's reference tank. However, the pin description for Vreg can be interpreted to mean it's the preferred connection point.
Anyone happen to have data that would help answer any of these questions? Just trying to knock out as many variables as I can in the design phase rather than doing multiple prototypes.
I am somewhat fascinated by these chips and am contemplating doing a build with them since I think they're pretty cool. In particular, I find myself wondering
1) What exactly the difference between the ES9023 and ES9022 is.
2) What actual minimum load impedance is. Both datasheets say 5k but parts of the 9023 datasheet can be interpreted to indicate 3k or even 700 ohm. And 5k doesn't make much sense since in most applications you'd have to buffer the 9022/9023 with an op amp, rather defeating the purpose of having an op amp integrated into the DAC. I've found remarks in various threads indicating folks are driving headphones directly with the 9022. No measurements, though, and the 'phones impedance typically isn't stated---this is of the clearer threads, calling out a 70 ohm impedance. Personally, if the parts are happy with a 1k load then I'd be happy too.
3) How THD varies as a function of output level. -94dB typ at 0dBFS is pretty decent but most DACs exhibit flat THD with falling output level to some corner level with a linear increase below that (figure 24 in the PCM1795 datasheet is a good example). I'm curious if this is the case for the 9022/9023 and what the corner level might be.
4) How sensitive the integrated op amp is to output loading. Unlike most op amps there's no stable capacitive load spec and, curiously, application diagram calls out a first order "RC" lowpass on the output with 0 ohm resistors and 4.7nF caps rather. Quite different from the usual 2+kOhm source impedance second or third order filters called out for most DACs. Or the 22-100 ohm resistors used to prevent an interconnect driver op amp from oscillating. I like the low source impedance, but the filter pole with a 0R is high enough I'm kinda curious why the output caps are there at all.
5) What the best point to tap ground is if one wants to approximate a balanced output. I'd normally expect this would be AGND as connecting to Vreg would increase the amount of interconnect noise coupled into the DAC's reference tank. However, the pin description for Vreg can be interpreted to mean it's the preferred connection point.
Anyone happen to have data that would help answer any of these questions? Just trying to knock out as many variables as I can in the design phase rather than doing multiple prototypes.
Last edited:
Everypne who has a datasheet fpr the ES9022 has signed an NDA, so I doubt if anyone will discuss with you here openly. Most of the questions you post you can get an answer from ESS via the distributor. They answered all my questions regarding the 9022. And you do not need a buffer for the output. They recommend you connect a capacitor directly at the output for HF filtering, so it is stable for capacitive load to the recommended value, and probably beyond. If you ask them the right questions, you can even figure out the output impedance.
So I can only suggest you contact them.
I knew of the 9023 as well, but they withdrew the webpage. And I also wonder what the differences are to the 9022.
Patrick
So I can only suggest you contact them.
I knew of the 9023 as well, but they withdrew the webpage. And I also wonder what the differences are to the 9022.
Patrick
Yes it will work pretty well W/O the buffer...
But for the better sound performance
good buffer is recommended,
(just try it and You will hear the diff.)
that could be simple like AD797 op,
or JFET circuit...
Employ another device to cope with reactive loads
not the main device? (and avoid condenser which should be from high values
min 10uF to fit the let say 10K r of next stage...)
cheers
But for the better sound performance
good buffer is recommended,
(just try it and You will hear the diff.)
that could be simple like AD797 op,
or JFET circuit...
Employ another device to cope with reactive loads
not the main device? (and avoid condenser which should be from high values
min 10uF to fit the let say 10K r of next stage...)
cheers
According to my experience (please see the ES9022 thread) there is no nned to use a buffer unless you have difficult load to drive. (The 9022 has an output impedance of a couple of hundred ohms.) Even in that case, 2 matched pairs of 2SK372V will be a much better choice than any opamps at unity gain.
As always a matter of taste.
Patrick
As always a matter of taste.
Patrick
Depends on when the datasheet was acquired; I obtained my 9022 datasheet post-NDA.
Them = Ismosys? I didn't get much from the US distributor.
Good to know, as is the output impedance. I take it a couple hundred pF in parallel with 1k wouldn't be a difficult load (e.g. a couple meters of mic cable with a receiver op amp on the other end)? Also, were you referring to this thread? I searched before starting this one and didn't find much, either on DIYA or otherwise (though I find search engines profile me as an English language user and return mainly English hits even when the query isn't language restricted), but would be pretty interested in the schematic for your 9022 build.
I've gotten datasheets for several ESS DACs directly from Shaw Electronics (US disti) without the need for an NDA, so as far as I'm concerned, there's no gag order. I'd still recommend going through the official channels to get the datasheet, though. Just because...
As far as output impedance goes - that might be a matter of THD rather than current drive capability. I'm guessing that you can only get the promised performance with a very light load (kOhm). At heavier loads, the THD might rise. That's common for op-amps and many quality op-amps tout the ability to drive, say, 600 ohm loads with a low THD.
Driving high capacitances (nF) makes me uneasy unless the manufacturer specifies a maximum. I'd stick with no more than their recommended value for sure. The phase margin slope vs cap load tends to be rather steep. At least that's the case for common op-amps designed for around 60 degrees of phase margin. That's also something that needs to be taken into consideration when driving longer cables (meters).
~Tom
Look at the Nuforce UDAC2 thread on head-fi to get an idea of what headphones this DAC will run. AFAIK, the UDAC2 runs directly to the headphone out socket of the UDAC2. Recently had it at an audio meeting
- into AKG701s (62ohms?) it didn't sound good
- into HD600 (300R) it sounded bloody fantastic
- into HD681 (32R) it sounded fantastic
I'm not sure that the lower impedance causes it problems, more likely the current requirements needed - the AKGs being a difficult drive
- into AKG701s (62ohms?) it didn't sound good
- into HD600 (300R) it sounded bloody fantastic
- into HD681 (32R) it sounded fantastic
I'm not sure that the lower impedance causes it problems, more likely the current requirements needed - the AKGs being a difficult drive
I assume this was intended to read "anything lower". In my case the answer is I work pretty much exclusively with balanced interconnects and the optimum tradeoff between input impedance and noise on the receiving side of the interconnect is usually around 2-3kOhm, modulo whatever interconnect and RF filter impedance might sit between the differential receiver and the DAC. One may not have the luxury of controlling the other end of the interconnect. And increasing the receiver's input impedance can be an issue even in full DIY situations. For example, low noise op amp, low efficiency hi-fi drivers, probably no problem. Current feedback op amp, 105dB horns, problems---both from the increased noise being audible in the high efficiency speaker and the degradation in current feedback performance.
Hmm, no measurements there either. Suppressed by the NDA, I presume. Looks like jkeny got you the datasheet.
Agree. I'm hoping Dustin might provide some guidance
A review of ESS's DAC lineup is interesting (numbers are for 8 channel mode on the 8 channel parts):
- start with an unbuffered voltage output and shared supplies
- 9023 - -94dB THD, 112dB DNR
- buffer the voltage output, separate analog and digital supplies
- 9006 - -102dB THD, 120dB DNR
- switch to 64 pin package and add more supply pins
- 9008/9012/9018 - -108dB THD, 120dB DNR
- go back to 48 pin package but change to current mode out
- 9016 - -110dB THD, 124dB DNR
- keep current mode out and switch to 64 pin package again
- 9008 - -118dB THD, 128dB DNR
- do some die layout improvements over the 9008?
- 9012/9018 - -120dB THD, 129dB DNR
Obviously, if one wants performance one can throw money at it and do a real nice job with a 9018 board. But where's the challenge in that?
Besides, there's a big thread on the 9018 already.
Sure, but as I remarked above I'm more interested in highest price/performance. For example, the 9006 datasheet implies the part hits -102dB THD in differential voltage mode and ESS's eval board app note says it does -104dB in differential current mode---yes, this contradicts the hypothesis above, but I hadn't noticed the app note had 9006 data as well as the 9008 when I wrote the previous post. If that's accurate one could quite reasonably say the extra 2dB isn't worth the second op amp in an application where cost is a consideration. As another example, the 9006 unbalanced current out performance spread in between the 49720 (-106dB) and 4560 (-105dB) is 1dB but the 49720 costs six times as much.
The obvious way of answering such questions for the 9023 is to do an eval board with a bunch of different options and measure them all. But at -94dB THD the 9023's already just about at the limitations of my measurement gear. Anyone got a spare AP2?
The obvious way of answering such questions for the 9023 is to do an eval board with a bunch of different options and measure them all. But at -94dB THD the 9023's already just about at the limitations of my measurement gear. Anyone got a spare AP2?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.