Phat DAC 1.8v operation


#1

Hi,

I’ve got a HiKey 960 which uses 1.8v for the logic levels. I know the PCM5102A on the Phat DAC can support 1.8v logic levels but needs to be powered by a 1.8v rail as I understand from the datasheet.

Is there a simple modification I can do to this board to get that support or should I just use logic level shifters instead ?


#2

pHAT DAC uses an onboard AP7333-33SRG 3.3v regulator fed from +5v.

Depending on how sophisticated your available tools are, you could:

  • Replace the AP7333-33SRG with an AP7333-18SRG making pHAT DAC a 5v powered, 1.8v logic device
  • Omit the +5v supply and tap your +1.8v supply onto the output of the regulator.
  • Omit the +5v supply and tap your +1.8v supply onto the 10uF decoupling capacitor

The regulator is the little 3-pin SOT23 device to the right of the C in pHAT DAC. The input (+5v) is the top-middle pin, and the output is the bottom right pin (it goes right into a square via, we normally use square vias to indicate voltage supply so they’re easier to find)

You could also tap your 1.8v source into one of the square vias by carefully scraping away the resist over the top, or into the large decoupling capacitor right next to the lower square via.

If the HiKey 960 pinout is a close match to that of the Pi, and it has a 1.8v source you should be able to cut the +5v trace and use copper enamel wire to route that 1.8v pin on the underside of pHAT DAC to one of the square +3.3v vias.


#3

What about the analog section ? I was hesitant to remove the power regulator as there was no clear mention of running the analog bits on 1.8v, only 3.3v, from the datasheet btw. I was able to probed that out and know where the regulator is on the board and how the rails are generally hooked up.

After probing out the power traces out to the chip and I was thinking about lift the power rail for the digital section by lifting the leg for pin 20 (DVDD) and soldering it on to LDOO via some wire wrapping wire and then hooking it up directly to the 1.8v power rail on the 96boards low speed connector. I believe that’s what’s required to get 1.8v volt operation. Please correct me if I’m wrong as I’m a relative newbie at this kind of stuff. The datasheet wasn’t exactly clear with the language here

I’ll look for those traces you mention and probe them out as well to get a feel of where things run. Thank you

Btw, there’s a lack of audio DAC boards for 96boards systems and the Phat DAC and other TI parts seems to be ideal as they both support 1.8v and 3.3v. I bought one for my RPi3 and it seemed like a good cheap way of getting an audio DAC onto it for experimentation, driver enablement etc

bill


#4

You’re right, AVDD must be 3.3v according to the literature ( I did not consult the datasheet! ). That will make it slightly trickier.

It looks like DVDD, XSMT and CPVDD are all tied together underneath the chip on pHAT DAC.

It seems XSMT can (and should?) be tied to 1.8v if DVDD = 1.8v - https://e2e.ti.com/support/data_converters/audio_converters/f/64/t/521681

But CPVDD also must be 3.3v.

It may be possible to get the tip of a knife right under the top edge of the chip in the middle, to cut the trace underneath, but I wouldn’t bank on it.

It looks like your solution- lifting the leg for DVDD (pin 20, top right) - may be the only way. You would also need to lift XSMT (pin 17) and give it a similar treatment.

It’s worth checking closely if you can get to the trace underneath the chip, since it would be much easier to tap off the capacitor than it would be to lift two IC legs.


#5

I realized that I really need all of those caps to be hooked up to minimize rail noise. I’ll see if I can get right under the CPVDD pin for cutting as there’s a bit of space between the pin and the chip body.

No space at all from the chip and the board. That’s my only hope. Otherwise I have this level shifting board that I just modified to fit for this purpose. That might be easier at this point. I’ll see in a bit


#6

I removed the cap on the top grey line on the left to get better access to the trace underneath the leg of CPVDD and was able to cut it successfully. I used a pretty sharp hobby knife with a pretty pointy blade tip and was able to get underneath it to cut that trace.

I unfortunately expose a bit of the ground plane I believe that’s on top of the rest of the board surround the IC. I don’t expect it to be an issue but I’ll keep and eye on it, cleaned the area a bit to get any remain copper particles out and flux off.

Getting that cap back on is another issue. I haven’t done yet but I have a very small soldering iron that might be able to do the job. I have little experience getting components that small back on their traces. It’ll be an interesting adventure.

Thanks for the picture that really helped btw. Had no idea that there was a trace underneath and that XSMT needed 1.8v as well. I should be able to tap the positive side of the rail caps hooked up to DVDD, LDOO with wire wrap and then jumper them together to a 1.8v source

Thank you again. I’ll let you know how it goes with i2s on the HiKey 960. That’s my next adventure with this project

Also, exposing that trace so that cutting is easier would be a nice revision to the board would be nice btw


#7

Nice work! I totally failed to notice that the capacitor is cutting right across the top of the IC and making it tricky to get a sharp knife under there. Glad you managed to get at it, though!

Getting the capacitor back on will, indeed, be a challenge. They’ll always be much keener to stick to the tip of the iron than the cold PCB. Hot air and solder paste is a saving grace in that respect.

I’ve asked about moving the trace should we ever respin that board. Looks like we should just consider doing a PCM5102A breakout, since there don’t seem to be any readily available ones out there, and certainly none that offer a split AVCC and DVCC.


#8

96boards thing is one of the only groups supported by Linaro and are also the official Google Android development platforms. It’s isn’t a casual hobbyist platform per se but the future is pretty much 1.8v for more powerful platforms like those. Also 96board would likely promote a DAC board as they really lack mezzanine cards for that platform.

Just some ideas to consider here

Having other DAC options with more sophisticated feature sets would be great btw. Going to solder that cap back in a little bit and see how it goes. :)


#9

Whoops, I broken that capacitor. What value is it btw ?


#10

Board data says 0.1uF which sounds about right. It’s a decoupling capacitor for the power supply as far as I can tell. You might find the DAC runs fine without it.


#11

I figured as much. It’ll just maybe generate more noise, but the analog and digital rails are at least separated. I’ll see if I can find a local replacement part first before mail ordering from Mouser/Digikey etc

Next up for my software adventure is GPIO pins and Linux kernel device tree bindings