Killingbeans
Major Contributor
True, but both I/V stages failing completely out of the blue? Sounds unlikely to me.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Review and Measurements of Holo Audio May --- Probably the best discrete R2R DAC
- Thread starter WolfX-700
- Start date
Many thanks indeed. I'll pass it along to the repairer.
FWIW both balanced and unbalanced outputs outputted simultaneously, constantly - so weren't selectable..
Just to say - I'd only be connecting a Melco N1dX music server to the May. There's no volume control. Magma have also said the impedences of the May and my pre amp are fine to work together; but I'm still concerned about 5.8v being so much higher than the 2v I found to be optimal previously. Tambaqui has a volume control - at twice the price..
Can I just check with you folks - would I be able to use the May in same way as I used the Mirus i.e. Melco N1dX server straight into May then going into pre amp, with control via the Melco app? Or would the May require insertion of a streamer? Thanks for your patience.
Extreme_Boky
Active Member
- Joined
- Jun 9, 2020
- Messages
- 155
- Likes
- 102
Is that the new headphone amp? Do you have better angle photos / higher res?
He's only posted 2 pics but the release is coming very soon.
Extreme_Boky
Active Member
- Joined
- Jun 9, 2020
- Messages
- 155
- Likes
- 102
This is what I found... it shows the design approach quite well. Looks good.
John,
Mastering Engineer and D90SE user here.
Apologies for the 4 year later bump. Can you elaborate on the above point?
No EQ needed. Use its built in linear phase or minimum phase sharp roll off digital filter any you’ll get perfect flat frequency response
Topping D90SE
AdamFrandsen
Active Member
Does anyone know if the FPGA in this provides dynamic error correction? The reason why I ask is because of resistor drift over time - my logic tells me an integrated chip like the TI dac11001B would be preferable to a discreet design, since it can be easily replaced... I am thinking a dynamic system would involve switches connected to each resistor, multiplexers, a precise current source, a high resolution adc to measure current drop, and the fpga of course… quite complicated, one could even add a temperature sensor that makes the system check itself if any deviations re detected or set it to perform a check at each startup for example… and if one resistor fails and you need to replace it you would need to reprogram the fpga with its current deviation from the target value - or find one exactly matching, seems like an impossible task
Last edited:
Killingbeans
Major Contributor
Maybe just make sure that only the bare necessity of current runs through the resistors at any time?
As far as I can tell, resistor drift is largely dependent on how much the resistor gets "cooked" during its lifetime. If it never experiences much more than ambient temperature, the drift would happen at a rate that not even your great-great-grandchildren needed to worry about?
As far as I can tell, resistor drift is largely dependent on how much the resistor gets "cooked" during its lifetime. If it never experiences much more than ambient temperature, the drift would happen at a rate that not even your great-great-grandchildren needed to worry about?
Last edited:
AdamFrandsen
Active Member
To achieve true 24bit I believe you need a resistor tolerance better than LSB of 0.00000596% (1/2^24) - VPG resistors that are the most precise are at 0.001% tolerance, that amounts to 16.61bits. The VPG resistors drift is max 0.002% per 2000 hours, so it already drifted too far for a 16bit dac in 2000 hours, no?
Last edited:
Most "R2R" DAC designs are not fully binary; the MSBs are normally unary (equally-weighted) to reduce matching requirements. to something reasonable. For example, if you use 4 unary bits, you reduce matching requirements by a factor of 2^4 or 16. Some audio designs use quite a few unary bits to reduce matching (tracking, etc.) requirements. I think I mention that in one of my DAC articles but am not sure. For audio you can get away with quite a few unary cells; for RF applications (my prior life), it is a trade for bandwidth, since additional cells add area and parasitic capacitance and such.
The impact of errors from resistor drift depend upon whether they are all drifting together or randomly. If they all drift the same then it is only a gain error. Reaching 16+ bit performance in a segmented R2R-type DAC is very difficult, and maintaining it over time and environmental variables (e.g. temperature, supply noise, etc.) just makes it harder. One of the things I ran into on an IC design was that the metal to the resistors was corrupting the performance due to the difference in temperature coefficient (tempco, very high for metal) and resistivity if the metal to different resistors. Matching that was painful because I had to watch not only the resistance itself but also the current density to each cell.
The impact of errors from resistor drift depend upon whether they are all drifting together or randomly. If they all drift the same then it is only a gain error. Reaching 16+ bit performance in a segmented R2R-type DAC is very difficult, and maintaining it over time and environmental variables (e.g. temperature, supply noise, etc.) just makes it harder. One of the things I ran into on an IC design was that the metal to the resistors was corrupting the performance due to the difference in temperature coefficient (tempco, very high for metal) and resistivity if the metal to different resistors. Matching that was painful because I had to watch not only the resistance itself but also the current density to each cell.
Killingbeans
Major Contributor
That max drift is specified at what constant temperature?
And if they all drift randomly, it raises the noise floor?
Considering how people rave about the "clarity" of vinyl playback, despite of its 11bits(?) best case dynamic range, I really wouldn't worry about resistor drift in discrete R2R DACs. IMO, this whole resurgence of R2R is based on nothing but romantic notions anyways. So, even if the drift turns out to become an actual audible issue in the future, the hobby community is more or less guaranteed to rebrand it as a desirable feature
EDIT: Corrected a brainfart typo
Last edited:
Yes, generally.
Records have achieved 50~60 dB dynamic range IIRC, and the best tapes push 80 dB though 40 and 65 dB respectively are more in line with what I recall for most examples.Considering how people rave about the "clarity" of vinyl playback, despite of its 11dB(?) best case dynamic range, I really wouldn't worry about resistor drift in discrete R2R DACs. IMO, this whole resurgence of R2R is based on nothing but romantic notions anyways. So, even if the drift turns out to become an actual audible issue in the future, the hobby community is more or less guaranteed to rebrand it as a desirable feature
I assume the only way for digital to sound good to some audiophiles is to use a NOS filter-less R2R DAC with tube buffers feeding a SET amplifier. A recent thread elsewhere complained that all good-measuring equipment sounded "cold, lifeless, sterile". I was reminded of when I did nothing except to raise the noise floor of a component using a bit of colored noise (rolled off the high end). People often preferred the noisier version, but complained about the higher noise floor. (This was in the analog days, no DAC involved.) I decided then that many people associate defects with some sort of premier quality, e.g. preferring hand-wired point-to-point as a sign of craftsmanship not found in perfectly wave-soldered boards. Of course, there were (are?) all sorts of articles on how the boards themselves degraded the sound, despite evidence that the examples were related to poor layout and not the PCB material itself...
Killingbeans
Major Contributor
Similar threads
