WS6318 - weird ON_OFF pin behavior and no communication

Hi there,

I’m the HW engineer of gsm device with WS6318 module. It’s connected with STM32 with 5 line communication. The other pins like WISMO_READY, ON_OFF, etc are connected to STM32 too. The main domain of power is Li-ion battery and it’s nearly directly connected to module VBAT. The STM32 has its own LDO (from Li-Ion to 2.75V) and the maximum voltage at STM32 part is that voltage. There is no direct connection to VBAT on STM32 side (so there is no change that STM32 could supply from one of its pins the VBAT voltage).

We have outsourced the manufacturing of board and mounting of components. That firm had noticed some serious problems after doing X-ray scan of finished boards. The inside of WS6318 looked different at each part mounted. There was lack of few BGA balls in the inside of WS chip. That was the first spin with that company, first board looked good and second and third looked weird (as I said - moved or lack of balls under BGA component inside WS module). The two boards with problems were taken for module exchange and the first one was given to me for the hw and sw tests.
The second spin in that company, after buying few WS6318 chips, was a failure too. They send me 2 mounted boards and both of them had short-circuit between VBAT and GND. They removed WS6318 modules from boards and the short-circuits disappeared. After tests, as they said to me, there was some short-circuits between power and other pins on module (just like internal short-circuit in module). They ordered few other chips and right now they are waiting for them (delivery time is huge, they will receive modules in January).

I don’t know if they used too high temperature in soldering process or there was some problems with ordered module before the soldering process. Right now we are very pissed :imp: , because we had lost few months and can’t even predict when we receive good boards.

At the end I would like to describe that first board, from first montage spin, which didn’t have any short-circuit on the first look. After some time we have developed software driver for WS6318 module. First thing - no communication. After some other tests I’ve noticed that there is something wrong with the power on of the module.

Pin connection:
WISMO_RESET is HI-Z on stm32 side
SIM_PRES is connected to 2.75V
ON_OFF is Hi-Z or Low controlled from STM32
WISMO_READY is pull-downed with 100k and connected to STM32 input
TP5 is connected with 10k to ground.
RTS/CTS/RX/TD/DTR are connected to STM32, but before power-on they are Hi-Z

The before power-on state is WISMO_RESET have high value (from internal pull-up in WS)
We are reading WISMO_READY low

After we put Low on ON_OFF pin, we notice that power consumption increases by about 110-120 mA, but WISMO_READY is still low, and no communication by RX/TX. After remove of low signal on ON_OFF pin, the current consumption falls by 110-120mA immediately! This is one thing - something is wrong with ON_OFF pin. The documentation states that it should turn on module, and then Hi_Z on ON_OFF should not turn module off, but low signal after proper power on for >5.5s should do that. Second thing - power consumption, as I remember it should be about 70 mA after power on.

The other weird and very wrong thing is that, if we put low on ON_OFF pin we can measure on RX or TX pin about 3.8-3.9V (Li-Ion voltage)! That pins should be in 2.8V logic! After making Hi-Z on ON_OFF pin we can see about 0 or 2.7-2.8 V (I don’t remember it right now) on Rx/TX pins (the VBAT voltage disappeared).

Do you think that it’s effect of wrong soldering temperature profile or broken module? VBAT voltage on communication logic pins are very weird…

To summarize:

  1. What is the main cause of that weird behaviour of module with ON_OFF pin control? Some internal shirt-circuit?
  2. Did you hear about that kind of problems with soldering that module like lack of BGA balls inside the module?
  3. Is it really so hard to buy that module with delivery time counted in days, not weeks/months?

For question 2 : you need to make sure the factory applies the Customer Process Guidelines (available on the Sierra Dev ZOne) word by word : reflow profile, stencil specs. When Sierra releases an LGA module, a second reflow test (equivalent to the assembly at your EMS) is done on several thousands of units.
Besides, all modules are tested on production line before shipment, and they all should look the same. If not, then it is most likely due to the second reflow (as anticipated above)
If after checking that your EMS did strictlly apply the process guidelines, you can send the X-Ray picture to your Sierra Wireless distributor for analysis by them or direclty by Sierra.

Question 3 : Sierra distributors usually hold a stock of modules for fast delivery. Check list of distributors :

Question 1 : have your EMS x-rayed your board ? 3.7V on UART looks like more an internal short, but I can’t be 100%, assertive. Anyway, from my point of view, you should make sure that you have 100% reliable assembly processe before investigating this issue as it may be linked.