Well, I've gotten to the design point where real money starts to be involved. I'm going to need licenses from the HDCP folk, along with any other audio formats the device will handle. That just leads into having some initial mock-ups built. That'll also cost money.
How much money is the question. HDCP membership is going to be around $1500 a year. I figure another $5000 will cover the consulting, and fabrication of the first prototype device.
Found Noritake Itron who makes VFDs (Vacuum Florescent Displays). Not only will they build custom displays, and mount them on PCBs with all the needed drive components, they can also incorporate push buttons for the user interface. I may be able to get them to build the full UI board.
Have the basic ideas for the power supplies complete. There will be a 115/230 volt, 50/60 Hz transformer which will output around 9 volts AC. Both the main board, and DAC board will have their own power supply sections. Rectifying and regulating the power to the needed levels.
Most of the main board, and all of the DAC board will be behind relays. The DAC board being dual regulated, with a series regulator, for most of it, and then shunt mode regs for the 3.3 volt reference signals. The shunt regs draw power all the time, especially when the devices behind them are shut down. So there needs to be a way to cut them off from the transformer completely. Thus the relay. The main board will also have a relay cutting down on draw when the system is asleep. Only enough power to run the UI board, and it's IR receiver will be in use. The UI board will be able to close the relay on the main board to wake it up, and then the main board closes the DAC relay when it comes time to initialize it.
All power supply sections will be linear rather than switched mode. That'll cut down on the electrical noise considerable (both internally, and also fed back out onto the power line). Linear supplies are not as efficient as switched mode, so this processor probably won't qualify for Energy Star status, but that's fine. One doesn't buy a super car for fuel economy; if I may be allowed an automotive analogy.
Looking at power supply designs, based upon the requirements of the various boards. The DAC board needs various voltage levels, and an inverted polarity for some parts. The main board is a little more simple in it's needs, plus no need for extremely accurate reference voltages as are required when entering the analog world.
I think I can get everything done with a single transformer with an output somewhere between 7 to 9 volts. I'm used to the requirements of power amplifiers, looking at logic level electronics everything is to a much smaller scale. That makes things both easier and harder.
Wow, high quality clocks, like Crystek's CCHD-950 are really expensive. Sticking 16 of them on a board will add over $400 to the price. That's no good. So I'll have to find a clock distribution circuit which can send the clock signal to each DAC without adding much noise or jitter.
Things are slowing down a little bit. Mostly because the first batch of research work is done, and now the actual designing has to happen. It's easy enough to look up a part, or read some datasheets in my down time, but I need dedicated hours to do any serious designing. So I'll be limited to mostly weekends.
Otherwise, the Sil Image sheets are great. Though their chips have so many interconnects, that I'm probably going to have to locate them on the main board rather than on a sub-board. One less board might actually make things cheaper.
Silicon Image really came through. After receiving my NDA they delivered no just two datasheets, but nine huge PDFs. I've got some reading to do.
I have selected which part will be used for the 1-to-8 I2C-bus switch. I lucked out, the biggest switch NXP makes is the PCA9548A, and it is exactly the size I need.
Nu Horizons replied as quickly as the previous distributor I contacted. I have to say, I'm impressed. I guess I've been working in the wrong field too long, software companies always take several days to get back, and then they don't have answers. These hardware guys are on the ball.
Anyway, in order to get the datasheets on Silicon Image's HDMI parts I need to complete an NDA. That's fine, I've done that before for other bits of information. Oh, also I can't buy parts until I acquire licenses from the HDCP and HDMI people. Though, that was expected.
I was trying some PCB layouts yesterday and today. Was using the PCB123 software which the one manufacturer provides free of charge. The biggest problem with it is that has such a small library of parts, and doesn't import any normal formats. So I spend most of my time drawing footprints, rather than laying out the board. I didn't even get to any complicated components yet.
Trying something else now. DipTrace had good reviews, and I see why. After spending many hours trying to make PCB123 do what I wanted, I found it so easy to work with DipTrace. I still have to create a few parts from scratch, but even that's a painless task. Will have to see how it handles silk screening on the bottom side of the board.
I requested datasheets on Silicon Image's SiI9233A HDMI receiver, and SiI9334 HDMI transmitter. We'll see Monday if this reseller is as responsive as the previous one I contacted. They appear to be a much larger company, so I'm not getting my hopes up.
Similar to my I2C problem, I also have to get I2S sent around the board to the various DACs, even getting it from the DSP board to the DAC board could be a problem.
Enter LVDS. I do love my differential signaling. I figure I'll encode all eight, stereo I2S channels channels as they leave the DSP chip with a M-LVDS transmitter (the M is for multipoint). Since each I2S channel has three signal wires, I'll need 24 transmitter inputs, which will deliver the output over 48 lines. I'll route the now differential signal off of the DSP board, and on to the DAC board. There I'll send one set of signals (six lanes) to each pair of DACs. Right next to every DAC will be a M-LVDS receiver. That works out to two receivers terminating each M-LVDS bus. Since I2S carries stereo data, refer to my previous post about setting the pair of DACs to look at either the left or right signal.
It's just an implementation detail, but I still need to consider how to distribute the three data lines for each I2S channel over the transmitters, since they have inputs in multiples of four. Going with three, eight-input transmitters probably isn't a good idea. Putting, muxed data, channel clock, and bit clocks on separate chips is probably a really bad idea, due to skew between parts. Even letting the one, odd wire be isolated from its companions would more than likely end up with errors. No, the best idea is probably to waste one transmit input out of each four, so there are no odd ones out. Maybe I'll find something else to send over to the DAC board with differential signaling, or just save the traces.
The DACs are where the volume control will be done. They sit on an I2C bus and receive the commands from there (along with much of their setup routine). The problem is the DACs only can listen on one of two possible addresses (set by grounding a pin). So having 16 of them poses a problem.
Enter the, NXP I2C hub, that'll let me create multiple, separate buses behind the hub, and be able to address each channel individually. I will still use the address selection in the DAC, so I only have to create 8 buses, instead of 16.
ESS's reseller, Shaw Electronics, came though quickly with the datasheet for the ESS ES9018/ES9012 DAC. Excellent news, when operating in mono mode, the input can come from PCM channel 1 or 2. So all I need to do is duplicate the an I2S signal, and send the copies to two separate DACs. Then I can program one DAC to look at the left data, and the other to look at the right. I also learned that the stereo-only part, can also operate in mono mode, with the same specifications as the full eight-channel part. As I never plan to use it that way, I might be able to save a couple bucks by going with the 9012.
I plan to operate the Sabre32 DAC in current output mode. So I need a I/V stage. The one design I've seen, uses a pair of TI's OPA1632, which are fully differential. On the other hand, ESS's Maximizing DAC Performance for Every Budget, says the best performing part they tested was Analog Device's AD797. I can use a pair of them, and still get a balanced, differential output.
I contacted a local reseller of ESS Technology parts. I'm looking for the datasheets on the ES9018 Sabre32 Reference DAC, and the ES9102 ADC.
What I'm curious about, for the DAC at least, is the format of the I2S signal when running it in mono mode. If I have to separate out the left and right channels from a stereo stream, that'll require more parts. It'd be nice if I could just split the stream as is, and send copies to two different DACs. Then tell the DAC to either use the first or second stream. Hard to guess without the datasheet, and ESS doesn't have them for download from their site (that I can see).
I figured I'd like to have a log of the development of the projects created under the Yooso banner. This way I can have a reference to look back on, and others can follow along if interest builds. Creating the blog was easy, now to start on the rest of the work.
