Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Michael Kelly, Apr 30, 2016.
Sorry I misunderstood the question, There is no external 5V input option.
First of all Michael, well done!
I am especially happy that you left us several options for data output. A universal board.
A few questions of curiosity.
Please explain further about how the board is powered. I see a 5V jumper. Is that for powering the Pi from the board? I also see what looks like a 12-48V barrel connector. What does that do? Some HDMI audio pinouts show 5V from the HDMI cable. Are you powering the board through HDMI?
I see a couple of lovely Murata transformers. What data streams are isolated?
All good questions. Let’s take them in order.
The 5V jumper, when installed, enables 5V to the HDMI connector. This not for powering the board.
There is a separate jumper labeled PI_5V to allow the PI2AES to power the pi.
The barrel connector brings in the raw 12V to 48V input voltage. This feeds both the expansion/I2S header and the on-board [email protected] switching supply.
The SPDIF optical is naturally isolated since it’s optical. The RCA output is isolated via a Murata transformer, as is the balanced AES via the XLR connector.
Hi Michael, I can not wait.
one question - if I desolder the RCA, can I connect it to a BNC socket to slightly (psychologically) improve signal, especially feed downstream DAC better due to better 75 ohm impedance matching in the signal pass? I am planning to make a new home-brew aluminum box similar to what I did for your 502DACs. Thanks.
The RCA is built into the same part as the OPTO, so I don't think it would work. Sorry, but there was no room for BNC, RCA and OPTO. And the only combo was RCA and OPTO.
BTW, we use an off the shelf 3 foot RCA cable originally used for Composite Video, so it is 75 ohm. You've seen the results in the test plots. I suggest spending your efforts on getting a good DAC to attach downstream.
Fair warning, I'm going a little techno geek on this one. Not that I know or understand that much, just curious.
The article at the link suggests that the recommendation for 1.5 meter digital cables relates to rise times around 25 ns. Posts that I found in another forum seemed to find measured rise times for commercial products in the 18 to 25 ns range.
I noticed that Michael measured excellent jitter using a cable with RCAs and shorter than 1.5m. It likely that Michael's boards and measurement rig are proper 75 Ohm termination and, therefore, don't get data transient reflections, but we can't guarantee that our audio equipment is as good.
I admit to being confused about what "rise time" means in a digital audio signal. Looking at the WM8804 datasheet show the master clock having 25 ns for a complete cycle time. I don't get how the rise time discussed in the article can be right with a 25 ns cycle time.
One interesting thing in the article is the idea that shorter rise times can make the transmission line insensitive to different cables. It suggests that 25 ns rise times are used to meet FCC tests which would not apply to DIY. If Michaeal measured rise time on his boards, assuming we can define the right way to measure rise time, I am guessing that his rise time will be much less than 25 ms. If true, it would make cable length and possibly cable construction not that important.
Sorry. Ending ramble. Turning off geek and returning to human.
Rise time is the amount of time it takes for an electrical signal to go from its lowest level to its highest. For ARS COAX that is approximately 0 V to 1 V. It is a figure of merit that reflects how strong the driver is. Measuring at the far end it will also take into account the amount of energy required to charge the inherent capacitance of the cable. While this capacitance level seems infinitesimally low to us, it does require a finite amount of energy to overcome. The longer the cable the more energy required.
What this all means in practical terms is that in order to recover at the clock and data at the far end we must be able to detect these transitions from low to high (rise time) and from high to low (fall time). If they are too long, the receiving circuitry has difficulty with properly decoding the signal.
reflections represent the fact that when we send an electrical impulse down a wire when it reaches the other end, unless it is a perfect impedance match, some of that energy will reflect back, this will then be joined with new edge and create what we refer to as ringing.The greater the mismatch, the greater the reflection, the greater the ringing.
However, having said all this, it is not like analog where every increase of any of these undesirable factors creates a corresponding degrading of the audio. It is digital, so as long as none of these factors are sufficient to prevent proper detection and reconstruction of the data, then they actually are relevant. Of course, every receiver has its own capabilities. But they are not so sensitive that a signal that is received well at 1 m by one device will not be received by another device. They are generally quite capable of handling this.
Sorry for the droning on, but this is an area where I find that there is in a enormous amount of uncertainty due to the fact that as humans how we assume they way things work, but that is not how computer circuits actually work. Make no mistake, a digital audio link is a computer circuit.
Thank you Michael. Not droning at all. A very clear explanation. As you know, there is a lot of misinformation out there, hence my earlier confusion.
This just shows how good your digital boards are and how much we can learn from you.
I’m glad i could help. And besides, being an engineer generally means being insufferable know what all!
PI2AES Preliminary Data Sheet and HW Reference Manual
This is a real Swiss army knife in digital playback !
Michael, in your HW reference Manual, you noted that the co-ax is 50 ohm. I thought BNC impenitence would be 75 ohm? Would a traditional linear voltage supply better than the on-board switch voltage converter?
Thank for the info, I fixed that error. I also noted that the output is via a CLIFF OTJ6 and the connector is RCA, not BNC.
I don't know if a linear supply would be better. I personally doubt it. But, I can't say with certainty that it would not.
Could we bypass the onboard switch DC-DC voltage converter? I didn't see a jumper for it. I know I have been microscopic picky but I do feel linear regulated 5+ supplies (one for Pi3, another one for 502DAC hat) improve the sound. By the way, USB battery phone charger was not better because the charger internally uses switch DC-DC voltage converter to get the 5V output from 3.3V batteries.
But you haven't even heard the sound yet!
In any case, no it is not practical nor was it designed to allow that. You can still separate the PI power.
Very nice indeed!
I agree with the quote, we should listen first. Do I understand that the 5 VDC primarily powers the HDMI pins and the Pi? Further, that the 5V HDMI pins are only required if the device on the other end needs power?
Yes and yes. The HDMI has a jumper for 5V so you only need to use it if the other ends needs power. Useful for adapter boards that convert I2S LVDS to I2S parallel.
Is there going to be a case available for Pi2AES?
Yes, I am doing the design now. Same concept as the 502 DAC case.
Make the toslink cut out bigger.
How much bigger?
Separate names with a comma.