EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: itdontgo on July 15, 2013, 06:38:01 am
-
We've had a few boards back with a short in the GSM module; we're talking say 1 in 100. We've built about 1000 of these boards and I have about 10 down the side of my desk with latched up output amps.
At first it was assumed that this was caused by a surge making the voltage regulator overshoot or from customer feedback it sounds like a few of them connected it to low voltage AC which eventually blew the module. When we looked at the scope though the power supply appears to perform well with very little ripple.
However I come across one unit which worked fine - until you called the PCB and the board started using a lot of power. The output power amp chip (TriQuint TQM6M4068) latches. The voltage on the board fails and this chip heats up - only releasing when the power is cycled. I'm now surmising that this is what happened to the other boards - the difference being their operator didn't switch them off and just let the modules cook!
http://www.triquint.com/products/d/DOC-A-00000735 (http://www.triquint.com/products/d/DOC-A-00000735)
The fact I have a board which can recreate the fault - if it is the same fault - on demand is excellent!
First thing was to connect a Li-Ion battery directly to the module and RF ground to rule out power supply issues. The fault still occurred. Beyond this I'm assuming the fault lies within the Quectel module. Every other input to the module goes through a 3v regulator. All I can think of now is to get the module off and put it back in the reflow oven to make sure the ground connection is properly soldered. I've done this before with a CCTV camera board that only worked when you twisted it.
I'm all out of ideas! Can you tell?
-
For anyone designing with the Quectel M95 GSM module - I recommend you keep your voltage to about 3.7-3.8v. Looking back at the older boards which never had a problem they ran at 3.8v. The Quectel datasheet reference design was 4.2v so we changed the design to suit. The faulty unit latches up at this voltage. I don't see an overshoot in the power supply but it could be too fast for the scope I'm using (200MHz though :-// ). At 3.8v the faulty unit is fine.
Strange as the PA chip should go up to 4.5v:
TQM6M4068.pdf Datasheet (http://www.adventcontrols.co.uk/TQM6M4068.pdf)
-
Oh and the reason it didn't work with the lithium battery... carelessness with the soldering iron shorted the audio output which locked up the baseband chip! Doh!
-
The power amp on every GSM output module i've looked at is more of an injection locked oscillator than an amplifier. it's very very common for these to oscillate unless a lot of care is taken with the grounding, shielding, ramp up and down, input and output matching etc.