This is what I use, and advise my customers.
The discussion is not only about making a 75 ohm RCA, it is about the effects when using RCA in SPDIF interconnects, or in general, how deviating from the 75 characteristic impedance affects the sound.
best
Because (On the clearly false assumption that a 100ps discontinuity matters for spdif), FR4 suffers from very high dielectric loss at microwave frequencies, duroid and the like have much lower losses under these conditions.
They are also expensive and difficult to work.
I still say the whole thing is a total red herring, interpose a VCXO based loop between the biphase clock recovery PLL and the DAC and if the filter is right everything will be fine.
There is also an interesting delay locked loop approach that Wolfson micro describe in an AES paper.
Regards, Dan.
excuse my naivety here as i'm just genuinely curious. does it really matter if you use a 50R BNC as long as the termination resistors (at both ends) are correct? say for instance using the amphenol mini coax/BNC cables, I recently bought some sockets and cables (preterminated, they are only available that way) I bought these for i2s and spdif (all three flavors BNC, RCA and AES) to my ackodac and they are certified for up to 6ghz. I can control the termination at the BNC etc input which is transformer coupled (not the i2s of course).
what do I need to do to make this work? or am I better off wearing the 60 dollar loss for all the cables and PCB mount connectors I have bought and buying the 75R version using the same dimensions, from hirose. I will be using them for connections from RCA and BNC coax, AES (XLR) and i2s from the M2TECH OEM USB ->i2s convertor.
the board i'm connecting to is the PTFE blue teflon version ackodac ESS9012 dac, this version will use these terminations for all 8 x i2s, DXD, PCM, spdif etc I am supplying the parts to be installed on this PCB, so I can send along the amphenol part which fits the same dimension landing pads as the hirose and change the resistors to suit, but if I must I will just grab the hirose tomorrow. no point going t the trouble of using such a system if it itself causes reflections.
thanks for your time
what do I need to do to make this work? or am I better off wearing the 60 dollar loss for all the cables and PCB mount connectors I have bought and buying the 75R version using the same dimensions, from hirose. I will be using them for connections from RCA and BNC coax, AES (XLR) and i2s from the M2TECH OEM USB ->i2s convertor.
the board i'm connecting to is the PTFE blue teflon version ackodac ESS9012 dac, this version will use these terminations for all 8 x i2s, DXD, PCM, spdif etc I am supplying the parts to be installed on this PCB, so I can send along the amphenol part which fits the same dimension landing pads as the hirose and change the resistors to suit, but if I must I will just grab the hirose tomorrow. no point going t the trouble of using such a system if it itself causes reflections.
thanks for your time
Last edited:
We've done that in 1997. Yes, of course that helps, but please consider PLL circuits are jitter attenuators, not eliminators.....
It is not a red herring. I wonder how many among us, taking part of this thread, have done all experiments required to build up the knowledge, let alone are able to measure jitter.
best
Obviously, but once you have the jitter below the converters inherent aperture jitter or at least below the self noise induced jitter of the local oscillator (not that hard), further improvement is pointless.
Easiest way to measure this is build two such clocks a few tens of khz apart, feed them into a mixer (multiplier) and measure the spectrum of the noise sidebands around the difference component.
In fact two clocks locked to the same reference would let you measure right down near DC by using a mixer as a direct conversion receiver.
I would think that anyone who has done microwave EME has probably got a fine appreciation for the importance of low phase noise local oscillators and building synthesisers that successfully attenuate as much reference phase noise as possible (When you are multiplying a 1Mhz reference from GPS or Rubidium up to 10Ghz at the transmitter, and are working with **SMALL** link budgets you tend to appreciate these things).
I did see an interesting suggestion on ProAud that the close in phase noise probably mattered far more then the stuff far removed from the carrier as it tends to cause the Bessel sidebands to fall within the audio band (Thinking of jitter as being phase modulation), this is an interesting assertion which may imply that loop filter bandwidths are generally far wider then would really be good as obviously anything within the loop bandwidth goes straight through a PLL.
I have been playing with a driscoll oscillator with great interest (the thing is quiet, residual noise seems to basically be crystal dependant), but for audio we really need a fundamental mode design (ideally parallel resonant) rather then a third overtone one.
What do you make of the delay locked loop approach used by Wolfson?
Regards, Dan.
I wonder how many jitter problems are caused by PLL hunting. I was using the old MC4044 in RF synthesizers back in the 1980s and the D type phase detector, common in many ICs these days, had a horrible dead zone leading to jitter on the output
Hi Dan,
Thanks for the feedback
Our first PLL had its' corner at about 3 Hz. Our current mixed mode PLL (Grimmaudio CC1) is at 30 mHz (miliHz) if I recall correctly.
Most DIYaudio people do not have microwave experience, and even if they had, jitter requirements for audio differ quite a lot from those for communication systems.
I have no clue abot the Wolfson approach, do you have a direct link ?
best
here is the AES conference paper that wolfson produced:
http://wolfsonmicro.com/uploads/documents/en/A high performance SPDIF receiver_Oct 2006.pdf
As ever with such papers, it is half marketing fluff, and the basic technique could trivially be implemented with a small PLD and a VCXO downstream of the line receiver if desired.
I am not convinced that jitter on an RF clock for audio is meaningfully different from phase noise on a local oscillator in a comms system, the issues are very similar (and even the resulting bessel sidebands are the same).
I always liked the trick of switching in a progressively narrower loop bandwidth once the system comes into lock, and 0.03hz sounds like a perfectly reasonable loop time constant once everything is locked up, a few hz is almost certainly far too high.
Convincing audio types that you need to think RF when working on this stuff can be an uphill struggle (Solid ground planes, power transfer, short connections, small loop areas, yadda yadda....).
Regards, Dan.
http://wolfsonmicro.com/uploads/documents/en/A high performance SPDIF receiver_Oct 2006.pdf
As ever with such papers, it is half marketing fluff, and the basic technique could trivially be implemented with a small PLD and a VCXO downstream of the line receiver if desired.
I am not convinced that jitter on an RF clock for audio is meaningfully different from phase noise on a local oscillator in a comms system, the issues are very similar (and even the resulting bessel sidebands are the same).
I always liked the trick of switching in a progressively narrower loop bandwidth once the system comes into lock, and 0.03hz sounds like a perfectly reasonable loop time constant once everything is locked up, a few hz is almost certainly far too high.
Convincing audio types that you need to think RF when working on this stuff can be an uphill struggle (Solid ground planes, power transfer, short connections, small loop areas, yadda yadda....).
Regards, Dan.
indeed, well though I got no answer here, I have decided to stick with 50R (RF standard) for exactly that reason and this decision was confirmed by the designer. internal i2s and spdif connections should be 50R. 75R,110R on the cable and connector from outside, then from the transformer on the receiver onwards, or otherwise internal i2s connections from usb convertors should be 50R; internal connectors, resistors and cabling should be specified as such. that makes more sense to me as well, how can 75R possibly be maintained right to the dac input pins anyway? and i2s is closer to RF frequencies than anything audio
more important to properly couple S/PDIF to receiver chip
Will it do any good if i mod a DAC by using, for instance a Canare RJ-R 75 ohm RCA female receptacle, and connect it using a 75 ohm coaxial cable directly to the receiver chip ?
However doing so will there be any impedance discontinuity at both ends of the internal coaxial cable solder joints ?
Will this mod make it better or worse ?
Canare Corp.: 75 ohm RCA Connectors: RCA Receptacles
Canare RJ-R 75 ohm RCA female receptacle
Will it do any good if i mod a DAC by using, for instance a Canare RJ-R 75 ohm RCA female receptacle, and connect it using a 75 ohm coaxial cable directly to the receiver chip ?
However doing so will there be any impedance discontinuity at both ends of the internal coaxial cable solder joints ?
Will this mod make it better or worse ?
Canare Corp.: 75 ohm RCA Connectors: RCA Receptacles
Canare RJ-R 75 ohm RCA female receptacle
quad shield cable RG6 cable
What brand and model of quad shield cable RG6 cable did you use ?
What brand and model of quad shield cable RG6 cable did you use ?
- Status
- Not open for further replies.