SMD Heatermeter on a RPI B+? Yes, it can be done......


 
v6.2

I made a few small updates to the v6.1 design:

  • Switched to tented vias
  • Broke out the remaining 4 RJ45 pins to solder pads
OSH Park project: https://oshpark.com/shared_projects/6VXPWWjp
Eagle files: https://www.dropbox.com/s/h53mtvkdkjf6k93/HM50SMD_C_V6.2.zip?dl=0

08f79ce136aa31a07f579f4834f452f8.png


eb261f3382e0de4d23f3031cb0152d81.png
 
Looks good but you haven't fixed the stupid f*ckup that I've made.:(

The 0.1uF capacitor just below IC2 should be much closer to IC4, at this position it does nothing! That's what you get when you let the software do all the work.
 
Those soldermask files have all of the vias that wouldnt be used for a solderpaste stencil

EDIT: Just opened it up in eagle and exported the tcream and bcream layers with dimensions. Hopefully that will get it.
 
Last edited:
Peter F said:
Looks good but you haven't fixed the stupid f*ckup that I've made.:(

The 0.1uF capacitor just below IC2 should be much closer to IC4, at this position it does nothing! That's what you get when you let the software do all the work.
Click to expand...

Hmm, I didn't realize that was an issue. That'll be a bit harder to move. My OSH Park order already went to the fab so no fixing it for this order at least. You been seeing issues on the shift register with that cap so far away?
 
Stafford Brunk said:
Hmm, I didn't realize that was an issue. That'll be a bit harder to move. My OSH Park order already went to the fab so no fixing it for this order at least. You been seeing issues on the shift register with that cap so far away?
Click to expand...

I haven't seen any issues, but it's bad practice and I will rectify it in my next revision. But don't worry, your boards will work just fine.
 
If you're going to break out the RJ45 1,2,7, and 8 pins...I would also suggest putting an 0805 resistor pad (with a cut-able trace between the pads) beside the 100K resistor that goes to the thermocouple connector. This would allow you to easily cut the trace to interrupt the circuit for a switch that is used for accessories such as the RDTC board.
Stafford Brunk said:
I made a few small updates to the v6.1 design:

  • Switched to tented vias
  • Broke out the remaining 4 RJ45 pins to solder pads
OSH Park project: https://oshpark.com/shared_projects/6VXPWWjp
Eagle files: https://www.dropbox.com/s/h53mtvkdkjf6k93/HM50SMD_C_V6.2.zip?dl=0

08f79ce136aa31a07f579f4834f452f8.png


eb261f3382e0de4d23f3031cb0152d81.png
Click to expand...
 
Haha, looks like I should have posted this before I placed my order.

I'd actually like to route those extra pins directly instead of via solder pads. There's also that capacitor that needs to be moved and that trace switch. Unfortunately, I'm pretty sure I'll have to re-route the board which will possibly re-introduce the possibility for noise (which is a problem since my scope is currently out of action. need to get a new one).

Have we thought about going with a 4-layer board? I understand the costs thing but the extra two layers may give us additional routing flexibility + noise reduction. I'm eventually planning to see if I can replace the Raspberry Pi with an Intel Edison so may see if I can go 4-layer to shrink the board down some more.
 
Stafford Brunk said:
Have we thought about going with a 4-layer board?
Click to expand...
With the regular HeaterMeter board, the size of all the connectors dictates the size of the board and the overall device. You need one side just for all the probes, then the RPi connector is massive (although technically you don't need all the pins), and finally there's the width needed by the LCD. You can save space by moving the connectors to the same side as the LCD starts, but then they run into the Pi's connectors. To get around this you can flip the whole thing 180 degrees but it isn't going to save you even 10mm. Believe me, I've flipped flopped stacked and laid out this board 100 different ways in mockups and the Pi always gets in the way!

The CHIP is a little better because it is like 2/3 the size of a Pi and doesn't have a bunch of giant connectors on it.
 
Yup it is plugged into my computer right now! It works by itself but the wifi is a little on the dodgy side, there's still a lot of software to write before it works as a HeaterMeter host.
 
Interesting... IDK where I am in the pecking order, but should be getting a couple of those at some point as well. I'm looking forward to seeing what it can do.....
 
I wasn't really thinking of using 4-layer for size reduction but more for noise/EMI control. Note that I haven't done any work on actually trying to reroute the board and look at noise so take this with a grain of salt.

Bryan Mayland said:
The CHIP is a little better because it is like 2/3 the size of a Pi and doesn't have a bunch of giant connectors on it.
Click to expand...

That's actually why I want to mess with an Edison (thing is about as big as a postage stamp). It isn't nearly as cheap as the CHIP though. I should be able to use an Edison -> RPi conversion board to get the software side working and then try and clean up the hardware.
 
Although the noise levels of the of the V6.1 board are within the limits and don't cause any problems anymore, I still feel that they could be lower. One of the main causes for noise on the TC signal still is HF noise on the 3.3V power caused by the PWM of the fan output.

I did some testing to see if I could reduce this noise and found that an inductor in the 12V PWR line helps a lot;

Untitled.png


I'm not sure what the best value for this inductor is, 47uH works well but I have ordered a few different inductors from Mouser to see what the best value is.
 
Oh that's certainly interesting putting it there. The v4.3 board has a space on it to put an inductor between the 3V3 and the analog side (as recommended by the Atmel documentation) but I found it actually created more noise to have it there so it will be removed in the final version. I'll have to try one in that spot.
 
It's just like the toroids I use on my RC planes to isolate noise from the 12V lines going to the ESC/Motor and the 12V lines for the RF transmitters/receivers. In the rare instance that this doesn't work, I use two separate batteries to create independent circuits altogether. What about two power adapter ports? :P
 
The weather was nice and I had some brined pork belly in the fridge waiting to be turned into bacon, so the ideal opportunity to play with my latest prototype!

HM%20Bacon-1.jpg


HM%20Bacon-2.jpg


Results,

HM modifications : back to the drawing board (ie. no improvement)

Bacon : Not tasted yet but the smell was divine!:D
 
You must log in or register to reply here.

 

Back
Top