I have used several times the AD-1955 demo board like that from the photo by post #9 about
http://www.diyaudio.com/forums/digital-line-level/135650-ad1955-evaluation-board-schematics.html
and the photo URL
http://www.diyaudio.com/forums/attachment.php?postid=1315495&stamp=1191364151
Unfortunately the AD-1955 demo board is since several years no longer available
This board is much more better than the DAC Yulong DAH1 (DAH-1)
to find about this weblinks:
YULONG DAH1 MARK DAC /Pre amp/headphone amps : hi-end headphone amplifier
Yulong DAC DAH1 Mark headphone amp - Head-Fi: Covering Headphones, Earphones and Portable Audio
and the attached photos
The main disadvantage of Yulong's DAH1 (DAH-1) are the very hot voltage regulators (with the small heatsinks). I have this problem solved by screwing this voltage regulators on the inside of the alu housing (no longer there are this hot spots on the PCB). This work was very unpleasant and time consuming.
I want to have similar evaluation board than that one with AD1955.
Which type could be an adequately substitution (same sonic quality)?
http://www.diyaudio.com/forums/digital-line-level/135650-ad1955-evaluation-board-schematics.html
and the photo URL
http://www.diyaudio.com/forums/attachment.php?postid=1315495&stamp=1191364151
Unfortunately the AD-1955 demo board is since several years no longer available
This board is much more better than the DAC Yulong DAH1 (DAH-1)
to find about this weblinks:
YULONG DAH1 MARK DAC /Pre amp/headphone amps : hi-end headphone amplifier
Yulong DAC DAH1 Mark headphone amp - Head-Fi: Covering Headphones, Earphones and Portable Audio
and the attached photos
The main disadvantage of Yulong's DAH1 (DAH-1) are the very hot voltage regulators (with the small heatsinks). I have this problem solved by screwing this voltage regulators on the inside of the alu housing (no longer there are this hot spots on the PCB). This work was very unpleasant and time consuming.
I want to have similar evaluation board than that one with AD1955.
Which type could be an adequately substitution (same sonic quality)?
Attachments
How about the WM8741 evaluation board?
Wolfson Microelectronics plc: WM8741EV1M, WM8741 Evaluation Board User Documentation
Wolfson Microelectronics plc: WM8741EV1M, WM8741 Evaluation Board User Documentation
AD1955 suggestion
What about this DAC? - it has a replaceable module for both the AD1955 and PCM1794. Price looks to be very reasonable here in China (around $220). At least two of the regulators look to have heatsinks
HA-INFO NG94 MkII
What about this DAC? - it has a replaceable module for both the AD1955 and PCM1794. Price looks to be very reasonable here in China (around $220). At least two of the regulators look to have heatsinks
HA-INFO NG94 MkII
Attachments
Yes, from EVB.
A much more better solution it would be the redesign of an ultimate DAC (what I am doing in the moment), which is still considerably better than this EVB but also much more expensive.
A series of features here will be found, which we find nowhere else - e. g. switchable NOS/4xOS/16xOS, different DACs and switchable different reclocking variations.
The EVB for AD1955 uses for low-pass filter and output stage of the AD797 - for this price really not too bad. The DAH1 from Yulong uses instead AD797 the OPA2134. But I don't know, whether this is the only reason for not so good sound by DAH1.
Therefore I want to find this EVB from Analog devices and that is the reason to start this thread. If there are used devices, I am also interested.
Last edited:
AD1955 output details
In my limited experience with the AD1955, the output filter components have a big influence on the sound quality. The best-looking published filter to my mind is indeed the one that AD shows with the AD797s on the EVB - with the notable exception of the 100pF cap between the complementary DAC output pins. Do you have a schematic for the Yulong output filter? - I might be able to make some suggestions to improve the sound. The AD EVB has one notable omission - analog supply regulators - as designed, it does require a low noise, low impedance external supply.
In my limited experience with the AD1955, the output filter components have a big influence on the sound quality. The best-looking published filter to my mind is indeed the one that AD shows with the AD797s on the EVB - with the notable exception of the 100pF cap between the complementary DAC output pins. Do you have a schematic for the Yulong output filter? - I might be able to make some suggestions to improve the sound. The AD EVB has one notable omission - analog supply regulators - as designed, it does require a low noise, low impedance external supply.
And what is the best opamp therefore?
I have overlooked your advice. This could be have similar sonic behavior - thank you thereforeMarlowe post #2: said:
Exactly this I also note. In the attachement you will find the schematic of Yulong's DAH1, unfortunately without the front/display unitabraxalito post #8: said:
Last edited:
all of them.
http://www.diyaudio.com/forums/digi...-opamp-i-v-conversion-dac-25.html#post1947424
there is no best, there is suitable and not suitable for the selected schematic
Yulong DAC enhancements
OK, thanks. It seems to be incomplete in other ways besides the display (which doesn't matter to me). For example, there are no wires taking data into the AD1955
The output filter seems to be directly lifted from the AD application but the AD797 has been substituted for a BB one with considerably lower bandwidth. I'd suggest deleting the 100pF caps from the DAC output to ground, but I'm not sure what the outcome will be for the sound - so try it and see. They can't be doing any good - DAC designers generally work to achieve the lowest output capacitance so having external caps looks odd to me. Perhaps they're to soak up glitches from the DAC, but they tend to screw up the following opamp in the process. Another bizarre feature is the lack of feedback around the output discretes on the headphone buffer - the feedback loop has been closed around the opamp but not its output driver. Really weird.
The power supplies are only made from standard 3-terminal regs, so a move to generously decoupled adjustables will improve ripple and noise (assuming this hasn't been compromised by grounding problems). Adding extra RC filters before the regs may further reduce ripple. Notice that the DAC offset currents are fed in direct from the +15V rail through two 2k2 resistors. If these resistors aren't closely matched (so the CMRR of the following filter can provide some help) the +15V power supply noise is being fed in unfiltered straight to the DAC output. This is an oversight copied directly from the AD EVB (albeit with a value change to track the change in rails from 12V - 15V).
Other improvements are bound to be possible to the layout...
OK, thanks. It seems to be incomplete in other ways besides the display (which doesn't matter to me). For example, there are no wires taking data into the AD1955
The output filter seems to be directly lifted from the AD application but the AD797 has been substituted for a BB one with considerably lower bandwidth. I'd suggest deleting the 100pF caps from the DAC output to ground, but I'm not sure what the outcome will be for the sound - so try it and see. They can't be doing any good - DAC designers generally work to achieve the lowest output capacitance so having external caps looks odd to me. Perhaps they're to soak up glitches from the DAC, but they tend to screw up the following opamp in the process. Another bizarre feature is the lack of feedback around the output discretes on the headphone buffer - the feedback loop has been closed around the opamp but not its output driver. Really weird.
The power supplies are only made from standard 3-terminal regs, so a move to generously decoupled adjustables will improve ripple and noise (assuming this hasn't been compromised by grounding problems). Adding extra RC filters before the regs may further reduce ripple. Notice that the DAC offset currents are fed in direct from the +15V rail through two 2k2 resistors. If these resistors aren't closely matched (so the CMRR of the following filter can provide some help) the +15V power supply noise is being fed in unfiltered straight to the DAC output. This is an oversight copied directly from the AD EVB (albeit with a value change to track the change in rails from 12V - 15V).
Other improvements are bound to be possible to the layout...
I/V conversion opamp
Interesting thread (in between the distractions from Carlos) - thanks for the reference. I don't find I agree that the AD797 is amongst the worst for this job, but I do take your point about not provoking non-linearity in the input stage. It does fall within the feedback loop for an inverting connection ISTM. I have tried to SPICE the AD797 in this application but for some reason the simulation won't converge with the AD-supplied macromodel. What I find strange is that AD clearly does have far better amps (cheaper ones too!) for this job, yet they specify this one. Perhaps it helps get better noise figures for the datasheet?
Does anyone know if the Berkeley Audio Alpha DAC uses the AD797 in its output filter (it does seem to have the AD1955)? I'm asking because I've read plenty of rave reviews for it online.
Interesting thread (in between the distractions from Carlos) - thanks for the reference. I don't find I agree that the AD797 is amongst the worst for this job, but I do take your point about not provoking non-linearity in the input stage. It does fall within the feedback loop for an inverting connection ISTM. I have tried to SPICE the AD797 in this application but for some reason the simulation won't converge with the AD-supplied macromodel. What I find strange is that AD clearly does have far better amps (cheaper ones too!) for this job, yet they specify this one. Perhaps it helps get better noise figures for the datasheet?
Does anyone know if the Berkeley Audio Alpha DAC uses the AD797 in its output filter (it does seem to have the AD1955)? I'm asking because I've read plenty of rave reviews for it online.
Nevertheless uses "Analog Devices" the AD797 in their EVB. I ask by "Analog Devices" for the reason in those days. If I recall right, I get follow justification:
"The output requirements between PCM1702/1704 so as BB successors and AD1955 are completly different - the R2R network by Burr Brown's multibit DAC's needs for I/U converter in all cases OP-Amps with FET inputs".
This means for me, that the AD797 isn't suited here because its input error current (BjT's instead jFET's at the input stage).
Unfortunately I cannot verify, because I haven't exactly internal schematic of AD1955.
In the meantime there are several new operational amplifier with jFET input available like the successor of OPA627/637 - the ADA4627. Perhaps this could be a better replacement for AD797
http://www.diyaudio.com/forums/soli...-1-outperform-burr-browns-opa-627-opa627.html
Last edited:
and so what? for example TI uses more funny schematic in their EVB, all that they(AD, TI etc) need is providing parameters on static signals - THD and noise, thats all, about the sound(dynamic signals) no one thinks, enough is once to listen and then forget this "beautiful" sound
DF-16x S/H-2nd order 6bit DS modulator-DWA(Scrambler)-continious time current output (dual RTZ) DAC (MOSFETswitches) - (BG ref-opamp/mosfets based ref current generator)
Last edited:
DAC waveform reconstruction
OK, let me clarify a little. You'd said on the thread you referred to that:
and I was responding to this by accepting that there are indeed aspects of the input stage that don't appear to be suppressed by negative feedback when operating in non-inverting topologies. However it does seem to me that with an inverting stage, the input stage distortion is suppressed by the feedback. Admittedly there isn't much loop gain around at the high frequencies associated with the AD1955 output transitions. Any clearer now?
Yes, indeed - glad you noticed that one, it still has me scratching my head - using a 2.2nF capacitor across the feedback resistor and an NE5534. I'm not totally fluent in digital audio theory but this looks to me to violate the requirement for a zero-order hold stage and therefore must generate significant inter-symbol interference. I've been looking long and hard for online mathematical treatments for what happens in the case of ISI in the reconstruction stage, but so far have drawn a blank. Clearly the measured figures for the PCM1792 et al are exemplary, but there's something wrong in using what's effectively an integrator to reconstruct the waveform. Do you have any references which explain what happens?
OK, let me clarify a little. You'd said on the thread you referred to that:
and I was responding to this by accepting that there are indeed aspects of the input stage that don't appear to be suppressed by negative feedback when operating in non-inverting topologies. However it does seem to me that with an inverting stage, the input stage distortion is suppressed by the feedback. Admittedly there isn't much loop gain around at the high frequencies associated with the AD1955 output transitions. Any clearer now?
See also a datasheet output stage for 1794\92, it a super funny scheme
Yes, indeed
- glad you noticed that one, it still has me scratching my head - using a 2.2nF capacitor across the feedback resistor and an NE5534. I'm not totally fluent in digital audio theory but this looks to me to violate the requirement for a zero-order hold stage and therefore must generate significant inter-symbol interference. I've been looking long and hard for online mathematical treatments for what happens in the case of ISI in the reconstruction stage, but so far have drawn a blank. Clearly the measured figures for the PCM1792 et al are exemplary, but there's something wrong in using what's effectively an integrator to reconstruct the waveform. Do you have any references which explain what happens?
This seems only to thicken the plot - the requirement (if such there be) for a JFET output opamp seems to have completely got lost at TI/BB seeing as they're specifying a 5534 - which is a bipolar input device! So does AD know something that TI/BB does not?
From memory, the 'requirement' for a FET input opamp relates to maintaining monotonicity when the output resistance is code-dependent, as is normally the case with CMOS DACs. I very much doubt that anyone can make a 24bit DAC monotonic (TI doesn't mention this aspect of the PCM1704's performance) so perhaps this is not so relevant in audio.
I found this treatment by Bob Adams very interesting - it doesn't mention the AD1955 specifically, but is indicative (amongst other things) of the general principles embodied in the part:
Bob Adams AD presentation
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.