ESS Sabre dacs can be operated as current out or as voltage out. Distortion is *much* lower using I/V. OPA1612 is presently the preferred opamp, it is used by ESS on newer eval boards. (AKM also uses OPA1612 for AK4499 I/V.)
Please see the somewhat dated (but still relevant) documents at the ESS downloads page: ESS Technology :: Downloads
Particularly the following: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Please see the somewhat dated (but still relevant) documents at the ESS downloads page: ESS Technology :: Downloads
Particularly the following: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Last edited:
I have a 1704 diy dac with opa861 on iv.
Is this the best option or is ad844 staked better??
Only listened to 1704 with opa861.
I have also a 1702 dac but with the classical op amp..not tryed the opa861 there.
I'm using pcm61 dac in nos mode with ad844 as i/v and buffer.
I paralleleled two pcm61 dacs per channel and reduced ad844 pin 5 resistor from 1.2k to 820ohm, I was expecting increase in output but it reduced.
As I read if I parallel dac chips current multiplies, if current multiplies output will increase. While using single dac chip with 1.2k resistor output was higher. With paralleled two dac chips (with reduced iv resistor 820ohm) it reduced also.
Isn't it should be higher than single chip with 1.2k pin5 resistor?
I omitted pin 4-9-10-11-14-15 while paralleling pcm61..
Any idea where I made a mistake? Or my logic is not correct?
I paralleleled two pcm61 dacs per channel and reduced ad844 pin 5 resistor from 1.2k to 820ohm, I was expecting increase in output but it reduced.
As I read if I parallel dac chips current multiplies, if current multiplies output will increase. While using single dac chip with 1.2k resistor output was higher. With paralleled two dac chips (with reduced iv resistor 820ohm) it reduced also.
Isn't it should be higher than single chip with 1.2k pin5 resistor?
I omitted pin 4-9-10-11-14-15 while paralleling pcm61..
Any idea where I made a mistake? Or my logic is not correct?
Well your output should increase if you re-insert the 1.2k at pin 5. Removing the second DAC and just adding a second 1.2k in series to ground will give you the doubled output I assume you were looking for by adding a second DAC. Open loop, voltage at the TZ pin (pin 5) should be roughly equal to Current In * RTZ. So by lowering this resistor, you are working against what I assume to be your goals of increased output.
Even still, if the DAC outputs are identical, and the resistors are spot on values, and the only significant load at pin 5 is the resistor, then you should not be seeing a reduction in gain. But the PCM61 datasheet gives a +-30% range for both output swing and output impedance in current mode. Resistors have tolerances, parasitics, and potential non-linearities. Additionally, any DAC output capacitance or phase mismatch between the two DAC outputs could have an effect (I wasn't able to find information on these vis a vis the PCM61).
On the AD844 side, it has a 65R worst case input impedance which would exacerbate any difference in output impedances between the two units compared to the typical 50R input impedance. Also, the heavier load may depress output levels at the high impedance node at pin 5, especially without NFB. From the data sheet, the standard closed loop configuration has the devices at pin 5 operating into a minimum load of about 200,000 (40,000 current gain in output stage * 50R load), with an output impedance of about 3 megaohms, I haven't read through most of this thread to see what people are loading this node down with, so I could be way off base on that point--performance at pin 5 might be better open loop as impedance drops, but in that case increases in output by any means will reduce performance.
So there are a number of possible explanations, although an actual reduction in output when adding the second device does seem to require either your DACs landing on opposite ends of the +-30% window or quite a few deviations from the typical specs across the board. I attached an LTSpice .asc I made to play around with some of these values. At 1k, the AD844 parasitics have pretty much zero effect, but I did find this paper with a useful CFA macromodel and numbers specific to the AD844.
A few questions that might help solve the mystery:
What output levels are you seeing before and after adding DAC 2?
What tolerance are your resistors, and can you measure their actual values?
Are you able to measure the level at the AD844 input in both configurations, and possibly with just the second DAC connected as well?
What output level do you see if you reinstall the 1.2k resistor in place of the 820 in the parallel DAC configuration? This one will tell most of the story, but the others might provide additional details.
Or, if you're not all that interested in solving the mystery and just want to double your output, re-install the 1.2k or remove the second DAC and load pin 5 with 2.4k.
AVCC_L and AVCC_R are shorted together in the output stage (I/V) board for the dual ES9038Q2M dac. Also, it was designed to use OPA1622 on an 8-pin DIP adapter for the differential summing stage. However, his DIP adapters leave OPA1622 ground pin floating which then causes increased distortion. On the good side, the price is very reasonable. Seems of lot of Iancanada users are not using the I/V board, they find resistor I/V and transformers to give better sound quality. Sorry to say it but in my book that means voltage output mode sounds better than the I/V board.
I don't remember,
the second board with the traffo was for I/V with a resistor aftr for the I/V or the board was just for the voltage outputt pins of the ESS and the traffo used as an OT or for XLR purpose ?
A verroboard may suffice but sure the inexpensive boards fro Iancanada with a correct layout is perhaps no hassle good enough ? I don't know... mines are still in a box, projects are stacking on each others !
the second board with the traffo was for I/V with a resistor aftr for the I/V or the board was just for the voltage outputt pins of the ESS and the traffo used as an OT or for XLR purpose ?
A verroboard may suffice but sure the inexpensive boards fro Iancanada with a correct layout is perhaps no hassle good enough ? I don't know... mines are still in a box, projects are stacking on each others !
The transformer board is used instead of the opamp I/V board, so, yeah, voltage out is what it looks like.
Regarding a veroboard type solution, it can work pretty well if copper foil is used to make a ground plane. At the request of some of the guys in the 'ES9038Q2M board' thread, I wrote up a though-hole-component design with assembly steps and pics for a diy output stage board. Its still available if someone wants to build one more or less from scratch. Fair amount of work though. Probably easier to do with SMD components and surfboards mounted on veroboard: SURFBOARDS 9000 SERIES FOR I.C.s Surface Mount Breadboards and
Adapters
Last edited:
ah, ok... I did'nt remember Ian was using also passive I/V with a traffo and a resistor with this board, just remember it was a Lundhal. Already used myself little lundhals long time ago (20 years perhaps) in cd payers but not for i/v purpose curent increase a the outputt of a curent dac chip, just for filtering before the RCAs/XLRs...
AD846
The AD846 was stable in an open loop configuration- I used it that way on a number of occasions.
The very high transimpedance was generated by successive cancellation of the base currents of the transistors in the top and bottom current mirrors in the signal current path, leading to an extremely high output impedance.
I presented an ISSCC paper on the design.
We didn't really have a market for the concept, but it seemed like a neat idea at the time.
This really is a blast from the past.
The AD846 was stable in an open loop configuration- I used it that way on a number of occasions.
The very high transimpedance was generated by successive cancellation of the base currents of the transistors in the top and bottom current mirrors in the signal current path, leading to an extremely high output impedance.
I presented an ISSCC paper on the design.
We didn't really have a market for the concept, but it seemed like a neat idea at the time.
I remember an oscillation of the AD846 (I think) first silicon. Lew Counts (may he RIP
) burnt his finger on the part I was debugging. It took some effort to determine what was happening. It was a pity that it didn't achieve much commercial success as it was a neat part to design.The AD841, AD843 and especially the AD847 seemed to do better.
Sorry for the trip down memory lane...
Replacements for AD844/AD846
check out post #5 under
Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?
check out post #5 under
Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?