QFN soldering MAX31850K troubleshooting...

Hi,

Finally got myself doing some QFN soldering on a MAX31850K breakout board from OSH Park.

I thought things went well, but, obviously, I could not get the thing going fully operationnal :o

I'm trying to see if I have bad connection on the board somewhere, but it's pretty tiny... I've checked the basics, such as voltage and grounds but I don't know how to make sure all the parts are ok... Is there a way to do that? (without taking everything apart)

Since the MAX31850 chip itself is giving out an address (I2C device), I take that as a good sign, but does it mean that the soldering on that part is all good? Could I have that thing fried and still have it give out a valid address?

Thanks for your input!

D

If it's returning an I2C address then you're doing something wrong. It's Maxim's own 1-wire system.

I prefer the MAX31855 as it's more like SPI (although not entirely.)

Thanks for pointing that out.

But, if the chip gives back its 64-bit serial code, is there a good chance that I don't have to investigate the chip connection?

The connection is obviously OK if it's responding to the address request. Do you get any data from the temperature request? Even if you have no thermocouple connected, it should give you the error bit that says "No thermocouple connected!"

You can also test the other error bits by grounding the thermocouple or connecting it to the supply voltage.

The internal thermometer should always be working. If that is not responding then the chip is as good as dead.

It's returning NAN. I'm not very good at C but looking at the Dallas Temperature library code, it seems to be struggling to "temperature LSB"

That happens with or without thermocouple connected. I'll investigate connection downstream of the chip leading to the thermocouple.

Does the MAX31850 chip detects open thermocouple?

Well, I might have an explanation for the wierd behavior I'm observing.

If anybody can confirm that'd be great...

I reflow soldered on a temperature controlled stir plate, and I have to assume that I magnetized the whole board and components...

::slight_smile:

Can I recover from this?

DuzMontana:
...but looking at the Dallas Temperature library code, it seems to be struggling to "temperature LSB" ...
Does the MAX31850 chip detects open thermocouple?

What library are you using? I don't think the data format for MAX31850 is the same as for DS18B20. For example the first has 14 bit temperature value, the latter 12 bit. The ship detects both open and shorted thermocouple and that informtion is part of cold junction temperature data.
If I remember correctly Adafruit sells the board with the MAX31850 chip and may have a library you need.

Thanks Kallek,

I'm using the Adafruit library. I'm pretty convinced that the board I got is the exact same as the Adafruit's. I was hoping to do a smart move by learning to solder these QFN parts, but that ends up just being a total waste of money. That's what happens when one cheaps out... >:(

I just ordered the board from Adafruit to go on with my life.

I'm still puzzled thought if I magnetized my board!! There are some ferrite parts on there that aligned with the stir plate magnet...

DuzMontana:
I'm still puzzled thought if I magnetized my board!! There are some ferrite parts on there that aligned with the stir plate magnet...

The FB1, FB2 and C1 are there to reduce the noise of the thermocouple signal. I do not think you can permanently magnetize ferrite and even if you could you should still be able to read the temperature.
It seems Adafruit library has only minor modifications compared to the original Dallas temperature library. They are trying to detect the fault...

case MAX31850MODEL:
      if (scratchPad[0] & 0x1) {
 return NAN;
      } else {
 return (float)rawTemperature * 0.0625;
      }
      break;

... but you can tell more exactly what type of the thermocople fault there is by checking the bytes 2 and 3, thats where DS18B20 has alarm data.