Looking for feedback on brewing shield with touchscreen

@peekay123
I tried the IL9341 straight from the spark Web IDE.
Worked like a charm! Only had to change the arguments in the constructor to

Adafruit_ILI9341 tft = Adafruit_ILI9341(D4, D5, 0);

I set the reset pin to 0, as it is connected to the spark reset pin. The pin is not used in the library, so it doesn’t matter anyway.

I will send a board and LCD your way tomorrow.

I can probably get the LCD’s without pins. So I can use

and

1 Like

Small update:
Got SWD debugging working, see my other thread. Display is working fine. Still have to test the touch screen functionality.

The 5V linear regulator works, but it generates quite a bit of heat.
The board + spark + display draws 250mA. So the LDO has to dissipate (12V-5V)*250mA = 1.75W
This might become problematic when the board is put into an enclosure.

I might have to reconsider the choice to use an LDO and go for a switching power supply instead.

If you do go with a switcher, take care to minimize noise/ripple on the output rail.

If the switcher produces too much ripple (e.g. poor layout, high ESR caps, etc) you can push the on-core LDO into oscillation. Key symptoms of that problem are very high LDO power dissipation (aka, it gets very hot) and/or erratic behaviour of the core.

The core doesn’t need an elaborate or difficult switcher, but it pays to make sure it is at least decent.

@AndyW, I was thinking something along these lines:

http://webench.ti.com/webench5/power/webench5.cgi?VinMin=11.00&VinMax=13.00&O1V=5&O1I=0.8&base_pn=LM2675&AppType=None&op_TA=30&lang_chosen=en_US&optfactor=3

LM2675MX-5.0, with low ESR 47u or 68u output capacitor and 47uH or 68uH coil.
This chip gives a much higher efficiency than the LM2595 (90% vs 82%).

WeBench gives an 8mV output ripple at 260kHz.

Total BOM count is 7, so it is still pretty small. Only 4 more than the LDO.

Looks good.

Pay attention to the layout with the lm2675.

@AndyW
I designed the new version, with the switching power supply. I hope you can look at it and critique my layout.

Other changes:

  • Pull down resistors on ACT1-ACT3
  • Pull up resistor on BUZZ (to 5V, pin is not 5V tolerant, but with 10k resistor).
  • Input cap on ADC changed to 10n
  • 10n input cap on touch screen interrupt pin (this is the trace closest to the coil)
  • 6p terminal block instead of 3x2p, cheaper

And of course I had to move everything to make room.

New schematic:

PCB:

http://gerblook.org/pcb/MXHrnx8ztU74CmPrLH6JgP#front

1 Like

Hi Elco,

Just a couple comments:

  • C4’s return is quite long… see if you can get that ground connected more directly to the Core’s GND.
  • As a general practice, the planes under L1 should be split in the middle with a gap, vs. shifted to one side like you have it. Or better yet, remove the ground plane under the inductor completely. Unless the inductor is a shielded type, then it doesn’t matter much, but I’d still gap it center it just because I might not be able to buy a shielded type of inductor down the road. The ground plane on the bottom under the inductor is not serving a purpose for heat or much for current either, so you could clear it away there as well. Do some research on this as well, this is kind of stuff I do because I’ve run into issues in the past during EMI testing and SMPS tuning. YMMV :slight_smile:

Thanks for the feedback @BDub.

  • C4 return path is GND pin on header --> bottom plane --> GND pin on spark. It’s not that long? I can place it under the spark core though and it will be shorter.
  • The plane under the inductor is 5V, not GND. The datasheet had the plane split like this in their reference layout. The inductor is a shielded one: http://katalog.we-online.com/pbs/datasheet/7447714680.pdf
  • The plane on the bottom is all GND

Ahh, ok… didn’t see that GND pin on the 3 pin connector. That’s not so bad, but ideally you want both legs of the cap to be between the the input connections and connections to the spark core.

If you have a reference design, definitely go for that first. It’s shielded anyway, so no worries… just a more expensive part. Reference designs are a good place to start, but also be cautious as they are not always going to give you what you need.

Don’t be afraid to stitch your GND planes from top to bottom a bit more… it’s basically free.

love this idea.
i have a lot of touchscreens and spark cores lying around. would love to help where i can.

Hi Korneel,

Are you Dutch by any chance? Your name sounds Dutch…

If you want to help write the code for this board, I can send you a prototype.

Elco

Hi Elco;

yes i am dutch :smile:
would love to help you, i am great in copy pasting code and troubleshooting stuff, not so much in writing blank code.
still happy to help.

I soldered the new prototype and the things I have tested so far seem to work :smile:

Switching power supply tested at 89% efficiency at 400mA (12V to 5V).

3 Likes

Nicely done!

Not only does it look great…I think you have done an awesome job in the short time that you have been working on this project…Just simply very well done…Wish I have the electronic and programming skills to help push it down the road to completion just a bit faster as I’m very eager to try it out…

1 Like

Glad I didn’t place my order for the BrewPi shield tonight before looking at this thread.
Is it too late to start looking at using the Photon instead of the Core?

PS. Can I preorder the BrewPi Spark Shield?

Thank you

It is too early to use the Photon instead of the core :smile:
I plan to have these in stock in 3 weeks. When the Photon is ready, it will replace the core.

Pre-orders will be possible after the official announcement.

this is awesome! very excited to pre-order

1 Like

excuseme but ther is the modbus library in the library list? i don’t found it.

Sorry for the lack of updates guys. I have a big one here: http://www.brewpi.com/introducing-brewpi-spark/

Because it comes with enclosure and all, I decided to call it the BrewPi Spark, instead of BrewPi Spark Shield.
I hope that’s okay, because every unit will have a spark core inside I thought it was appropriate.