for anyone who is following this, or stumbled on this site and wondering what is going on, well you are not alone, I can't work it out sometimes either!
Some intolerable neighbours (3 sets in fact, after a few years of trouble free renting so I am sure it's not me) have forced a house move. This is not a bad thing, I now have quite a bit of space, which means I can set up a decent work area to work on this project.
My homebuilt speakers are almost back together and looking rather smart, just hoping a system of my OHDSP boards can make them sound alright!
These two incidents (speakers and moving) are coindicing very well, as I can now use my speakers with the digital crossover box I threw together. However I am still short of 10Kohm MELF resistors to finish the balanced ADC as I did not get around to placing a farnell order I had put together, but this will not stop me!
Nothing can stop me now!
Still working on this slowly at the moment, other things have had to take priority. Only a temporary set back though.
Have been building up the balanced input ADC boards I received when I found I have run out of 10k MELF0204 resistors, I also do not need to order anything else at the moment, which makes a £3 order from farnell a little pointless (and the shipping would cost more!)
I am working on some new ideas now. Looking to make the "boards" more obvious and user friendly with documentation (and simple things like catchy names, hardware tweaks to interfaces/layouts and so on). Working on a new micro-controller board, packed with features, which could be quite flexible/powerful. Would like to get a better website up as well, as we head into autumn/winter the darker evenings will end up being the time for this.
My Frankenstein clone is alive! Well, okay, I admit that is not true, but the picture on the homepage, and the full picture in the last blog post show a half finished box with some of my boards in partly wired. Well it is now 75% complete, awaiting a populated balanced ADC board (in place of the single ended one), but it works from a SPDIF source. All 8 outputs play music (set as four stereo outputs).
Bit dark to take pictures but just imagine the last picture with a few more boring wires added.
Edit: Oh yes, I am also looking at spinning up an XMOS 200 based DSP/control board, should sit nicely along with the ADAU1452. Would give the option for some DSP/usb streaming without the need for the ADAU1452 board. Looks like there would be some initial limitations (such as the xmos I2S code only supporting 48k in TDM mode, and my rusty programming skills), but the potential for a lot of features could be big.
Just a little update, project is on a go-slow for the next couple of months whilst I deal with some none project related things. Hoping to order PCBs for new revisions sometime in August or September. Might change the DSP board for small vias on the ADAU145x and buy my self a toaster oven and fit it with a controller to do reflow on this just to make life easy.
I have been building up my speaker crossovers, well the first version using hardware I have designed:
Work in progress but the boards work. All 3 DSP boards where I hand soldered the LFCSP chip seem fine. No idea on long term reliability yet but hoping to use the above box to do some testing for this (i.e. just leaving it on a lot).
So I released new board revisions over the past few days, I have done my best initial check on these but I have not yet ordered them to build up. Even if I had ordered them I would still correct any mistakes in new revisions so the files might as well be public.
I am also working on a differential version of the PCM4202 ADC board, I did have a revision 1.0 design but I am not happy with it and so decided to rework this. It should be a more versatile solution than the single ended version and have a better PCB layout.
I also came across Orange Pi SBCs which look very neat with the AllWinner H3 SoC onboard all of them. Early 2015 I had started designing an ADAU1452 Raspberry Pi hat type board, unfinished designs still available at github.com/pwzj/ohdsp.
I stopped working on this as the Pi only support I2S in stereo with no TDM, limiting what I wanted to do with it. The AllWinner H3 SoC datasheet specifies support for all the normal I2S/left and right justified/TDM modes on the audio data lines and so this may work for my original idea. I am very tempted to either spin an ADAU1452 DSP board to interface with this, or make an adapter board to hook into my current DSP design. This all depends on software/driver support for TDM mode, and I need to do a lot more research.
So I finished revision 2.0 DSP board layout, I need to run some final checks over this and make sure it is ok. A tweaked power supply layout with solid ground planes is complete except for some track/polygon overlap errors; not sure why they are errors as the PCB looks fine.
I got close to finishing some changes on the ADAU1966 DAC board and then ripped up the jumper/spi header/power supply sections as I am just not happy with the layout. I also thought switching to a 4 layer board would help. As I am aiming to use this as the main analogue output source for several projects the extra cost of a 4 layer board is justified to me.
I also invested in a new toy:
It's a Siglent SDS1202X - initially I ordered the SDS1102X but changed the order. As it is only 1GSa/s shared between 2 channels the full 200MHz bandwidth is limited when using both channels at once. As I would like to be able to probe up to 50MHz square waves the extra bandwidth is useful even if it is only really accessible with only 1 channel on.
It's a nice piece of kit, I have only played with it a couple of times so far. The features it has are unbelievable for the price, even some of the megabuck scopes I have played with have less features than this by default! I had been hoping to redeem the free decode option currently advertised on the Siglent website but heard nothing from my email yet. You do get decode free for 30 uses anyway, although I am not yet sure if things like SPI decode are any use on the 2 channel oscilloscope, unless the SPI clock can be used on the trigger input.
Just a minor update. One of the issues I faced with building the boards was the lack of clear parts lists/parts types. Initially this is because I was not totally sure on what parts would be, how much they would all cost and so forth. I just designed what I wanted based on what I could get rather than making it a cost driven design.
Now the limit is because of the lack of information I have put into the schematic designs, and that the fields I need are missing from each component. Adding them is a slow/time consuming process. I am also trying to get KiCad to output a nice CSV BOM file automatically. Presently there is misalignment between the heading of a column and the contents of the column, but I am not sure why as it should work.
Someone kindly pointed out the eeschema library file was missing from the git repository. I did notice a few weeks back and had committed the file. I obviously did not push the update to the master branch though!.
Which brings me to board revisions 1.1/2,0. The ADC board revision 1.1 is almost ready to release with correct power pin assignments, I am also working on a newer revisions of the DSP boards with a master SPI connection. There is a new revision of the DAC board with the mating slave SPI connector. This means the ADAU1452 should control the ADAU1966 for faster sample rates without the need for a microcontroller.
I am also revising the PSU and buffer board. This is mostly component spacing and silk screen tweaks. On all boards I have decided to switch to some lower cost regulators in digital supplies. The TPS7A45xx series are quite expensive but I will still use these in the analogue supplies due to their low noise across the audio band..
I have also been playing with RightMark Audio Analyser. There seems to be a problem with Intermodulation distortion+noise readings using the SPDIF input on the DSP. THD at lower frequencies is high on the SPDIF input as well. Otherwise the measurements seem to fall in the noise floor of my Xonar U7 soundcard. Some of the noise issues are from low level 50Hz and multiples (a lot actually, all the way up to 500Hz). This is a very crude test so take all results with a pinch of salt, but it is interesting anyway.
Results shown below. First result is the Xonar U7 analogue output connected to the Xonar U7 line input. Second is connecting the Xonar U7 analogue output to the DSP platform ADC analogue input, and the DAC analogue output to the Xonar U7 analogue input. Third is the Xonar U7 SPDIF output connected to the DSP board directly, and the DAC analogue output to the Xonar U7 analogue input
Finally soldered an EEPROM (well a socket actually) on one of the DSP boards and it self boots just fine.
Here is a bad phone picture showing the DAC buffer board (single ended outputs) and the PCM4202 ADC board wired into my test chassis. Boards are stacked so the DSP and DAC are underneath.
So, the reason the PCM4202 sounded so different to the digital input is because.... I have wired left to right and right to left (or LRCLK may be inverted). Also I think the phono cable I was using was faulty as well.
Having replaced the cable, swapped the channels in SigmaStudio (which is nice!), and set the levels between the two to almost the same, the analogue and digital inputs sound pretty much similar to me. If anything I would say the digital input sounds a little more clinical. I guess that is partly down to the sound card I am using to generate the analogue audio from my laptop in the first place (a Xonar U7).
An electronics engineer and a long term electronics hobbyist. I like tinkering with stuff and making things.