rb2013
Author of The 6922 Tube Review
- Joined
- Apr 12, 2013
- Posts
- 5,930
- Likes
- 507
So reading the Uptone Regen original thread - I see why an external USB bridge can sound superior to a DAC's built in one.
John Swenson the genius Engineer behind the Regen explains on the thread:
http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index10.html
John explains 'Packet Noise'
Alex (SuperDad) from Uptone explains:
John goes into greater detail:
So to summarize the USB receiver chip and processor have periodic high level bursts of activity (when the data packets arrive) that create high frequency noise - that noise bleeds through the isolators in several ways - one way is through the system's PS (like those DC servo laser controllers on CD players) that then modulates with the USB reclocking and even the DAC's clocks (if the USB board shares a PS inside the DAC) and other components (like the isolators thresholds). The second problem is noise riding on the data packet itself generated by the PC's processor and USB MB USB controller (this is why you should have as few other USB devices connected to your Music Server, ie, external speakers, phone chargers, ext HDs, etc..). The more the USB controller has to handle with these devices the more noise that is generated.
Then there is the issue of impedance matching of the USB data output and inputs. Most USB input receivers do not have exact impedance's - so the signal on the cable will create back reflected waves. These modulate with the incoming USB data stream and create issues for the USB bridge or DAC's on board USB receiver and processor - creating jitter. So this may explain why some USB cables sound better then others in your system - they have slight impedance variations that with trail and error you may find one that more closely aligns with your devices USB input receiver. The Regen is made to exact impedance specifications and attaches directly to the back the Bridge or DAC's USB input. But this still does not fix the fact your Bridges or DAC's USB input may not be the exact impedance. So the Regen was designed to absorb those back reflected waves.
The PS and G modulated noise inside DAC does double harm - in addition to the noise modulating with the USB board clocks and processor, it also feeding this noise into the DAC's chips and clocks ground and PS. So thereby polluting now only the USB board but the DAC board as well. Having an external USB Bridge (hopefully plugged into a completely separate PS isolation and filtering device like the Audio Art PB 4x4 Pro to isolate the noise for the rest of the home's power) - can help keep the DAC's PS clean.
What a neat device!
Hey Uptone how about a Regen for the S/PDIF coax input! I'll be first in line
John Swenson the genius Engineer behind the Regen explains on the thread:
http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/index10.html
John explains 'Packet Noise'
This is a term I coined to refer to the power and ground noise a receiver generates when reading a packet of data. For example inUSB(and Ethernet) data comes in packets with a fair amount of space between packets. The receiver chip doesn't do much of anything in-between packets, this doesn't generate much noise on supply planes. But when a packet comes in the receiver goes into high gear processing the data that just came in, this processing generates a lot of highly variable current draw from the board which generates a fair amount of noise on the supply planes.This noise comes in bursts, which is the packet frequency. For exampleUSBhigh speed has packets at 8KHz, which is in the human hearing range. This noise can modulate processes in theDAC(such as the main clock oscillator and theDACconversion to analog) producing subtle distortions which are in the audio range.John S.
Alex (SuperDad) from Uptone explains:
I am sure John will chime in with a proper technical explanation, but allow me to point out a few things--a) The purpose of the REGEN is create aUSBsignal with better signal integrity and impedance than what is at the end of yourUSBcable from your computer or streamer;b) Galvanic isolator chips (some of which are much better than others--all of which generate their ownjitterand should be reclocked afterwards), as used on manyUSB>I2S boards and in the last couple of years in decent DACs, all come AFTER both theUSBPHY and theUSBprocessor;c) The "packet noise", "logic induced modulation", whatever you want to call it--is still happening on theUSBinput board, and despite the presence of isolator chips, the extra voltage spikes, activity, and noise generated will still have an effect on the sound. That's because those galvanic isolators are good at galvanic isolation, but the signal itself gets embedded with tiny stuff (John will have to expound and correct here). Also, I seem to recall that the isolators are good at some frequencies and not others.In addition, to use the JLSounds board as an example (I have one myself, uninstalled as of yet), they have the audio clocks on the board, and as you say, some sharing of PS planes may occur. (And if one does the better thing and feeds a betterDACmaster clock back into the board, it appears that he ran out of isolator channels and the clock would not go through the isolators. He also chose to use the RF transmitter/receiver type of isolator. Not only do those Silicon Labs isolators have about 3.5 times thejitterof the ones John likes, but there may be some RFI emitted.)As a separate example, John when to great lengths in the design of theUSBinput of the recently releases BottleHeadDACto insure isolation, immunity, etc. He used all the right tricks with regards to PHY, processor, multiple isolated regulators, clocking, and the best galvanic isolators and reclock flops--all placed and ordered for optimal results.And guess what?USBcables, computer stuff, and the REGEN all still make a difference!So sorry, when Chord just announced that its newest two DACs--the about-to-ship 2Qute and Hugo TT--have galvanic isolators in theirUSBinput path that make choice ofUSBcables and power supplies irrelevant, forgive me if I don't act surprised if users discover otherwise.
John goes into greater detail:
Originally Posted by bibo01
Interesting explanation.
In the presence of a galvanically isolated USB interface like JLSound where clocks are on the isolated side and there are two separate ps (one for xmos chip and one for oscillators), can that "fair amount of noise on the supply planes" still be relevant? If yes, how?
The two separate PS share ground.
If we measure noise on both PS, what kind of measured value should we aim for in order to leave DAC unaffected?
Very good question. The isolation helps but is not nearly the panacea many people think. Lets travel through the system and look at both the power and signal and what happens to them as we go through the system.So lets start with aUSBreceiver with bursts of high frequency noise on both the power and ground planes. This PG (power/ground) noise will modulate the data being sent to the isolators. It will slightly increasejitterand the amplitude of the pulses will vary with the noise.This noisy power also goes into the driver side of the isolator. The signal going across the barrier (light, EM waves, magnetic field etc) gets modulated by this PG noise as well. The PG noise also changes the threshold of the input receivers, addingjitterto the signal.On the other side of the barrier we have a couple things happening, the varying signal level, caused by the PG noise in the driver, also causes the receiver current to change, even with no signal applied. Thus the receiver causes PG noise on the "clean side" directly related to the PG noise on the "dirty side" It is definitely attenuated, but not by nearly as much as most people expect. Then we also have traditional logic noise caused by the fact the output is a normal logic signal, every time the output changes it creates noise on the PG planes on the clean side. Thejitteron the signal created by the PG noise on the dirty side is still there PLUsjitterintroduced by the isolation scheme. Thisjitterchanges the spectrum of the logic noise on the PG planes on the clean side.So then we feed the signal through a reclocking flop, which is supposed to get rid of all thatjitteron the input. Well it helps, but no reclocking flops are completely effective. The PG noise at the flop still causesjitterto show up on it's output, PG noise changes the threshold where the flop detects the "switch" of the clock, thus increasingjitteron the output.The result of this chain is that PG noise on the "dirty side" can still make it through to the "clean" side. It IS attenuated, but not completely gone.Cascading such stages can theoretically help, but in order for that to work the reclocking clock has to get fed back through the isolators which significantly degrades the clock so it turns out cascading doesn't help much. (a two stage cascade does make things better, but not by a huge amount)On the issue of PS supply noise THAT is a whole story in itself which needs to get tackled separately.John S.
So to summarize the USB receiver chip and processor have periodic high level bursts of activity (when the data packets arrive) that create high frequency noise - that noise bleeds through the isolators in several ways - one way is through the system's PS (like those DC servo laser controllers on CD players) that then modulates with the USB reclocking and even the DAC's clocks (if the USB board shares a PS inside the DAC) and other components (like the isolators thresholds). The second problem is noise riding on the data packet itself generated by the PC's processor and USB MB USB controller (this is why you should have as few other USB devices connected to your Music Server, ie, external speakers, phone chargers, ext HDs, etc..). The more the USB controller has to handle with these devices the more noise that is generated.
Then there is the issue of impedance matching of the USB data output and inputs. Most USB input receivers do not have exact impedance's - so the signal on the cable will create back reflected waves. These modulate with the incoming USB data stream and create issues for the USB bridge or DAC's on board USB receiver and processor - creating jitter. So this may explain why some USB cables sound better then others in your system - they have slight impedance variations that with trail and error you may find one that more closely aligns with your devices USB input receiver. The Regen is made to exact impedance specifications and attaches directly to the back the Bridge or DAC's USB input. But this still does not fix the fact your Bridges or DAC's USB input may not be the exact impedance. So the Regen was designed to absorb those back reflected waves.
The PS and G modulated noise inside DAC does double harm - in addition to the noise modulating with the USB board clocks and processor, it also feeding this noise into the DAC's chips and clocks ground and PS. So thereby polluting now only the USB board but the DAC board as well. Having an external USB Bridge (hopefully plugged into a completely separate PS isolation and filtering device like the Audio Art PB 4x4 Pro to isolate the noise for the rest of the home's power) - can help keep the DAC's PS clean.
What a neat device!
Hey Uptone how about a Regen for the S/PDIF coax input! I'll be first in line