802.15.4 mesh connect board assembly

[CcanDo]

We need help finding an assembly solution. The RF module includes a Uart. The present design is REV 4A, hardware, REV 32, C code.

The RX side is found to perform. The TX side is a different part number and has a 20% failure ratio. The Board assembly has been attempted in house and by 4 seperate CM. The highest failure ratio, from one CM was 60%. The definition of "failure" is; code will not download from the compiler/desk-top to the module.

The assembly has included precautions for ESD, pick and place all components except the hand soldered RF module. The ESD hand solder station is set at 720 F.

There are no known drop in sockets available from the RF module manufacturer.

Perhaps the forum will respond and offer, a 20 % scrap factor is typical. I hope not.

 

[zappedagain]

Ohh... lots of questions. No, 80% yield is not acceptable.

Can you remove the Tx from the assembly, install it in a known good assembly, and program it there?

Is it a programming error (could be power supply issues; noise issues; grounding issues)? Or are the Tx modules dead (probably back to power supply issues).

ESD can kill the transmitter, but the digital side (programming) will probably live through most events in an assembly house (assuming they have ESD controls in place).

I'm making some assumptions here on what you mean by assembly. What is getting assembled?

Is you soldering iron at the hand solder station an ESD-safe iron? Is the workstation ESD safe? Have you tested it to be sure?

You state the revision of your module, but not the manufacturer or model number. What is it?

Z

 

[IRstuff]

So, have you done no failure analysis? What exactly is failing? How many parts are there out on this module? What is this code being downloaded to within the RF module?

You're giving absurdly scant information for this type of problem, and it's not the sort of thing that can be diagnosed remotely. Why haven't your engineers been tasked to figure out the problem?

TTFN
faq731-376

 

[CcanDo]

Since posting the question, the simplest solution to answer your questions and more of our own, locate a programming fixture that will hold the module.

The module has an embedded 8051 PIC, C code WILL BE loaded on the 8051, before being soldered onto the board. (presently, protocol has the module soldered to the board, then down-loaded through a header on the circuit board)

Subject to the module being corrupted, prior to removal from an ESD free bag, in an ESD free area and programmed with known safe-gaurds, that part of yield will then be known.

No, there has not been a forensic anaylsis done, though it has been discussed. However, signatures don't disclose failure or corruption. Signatures have only been compared on a flying probe test and systems analyzer.

The module is ; California Eastern Labratories, CEL, ZICM2410P2-2 w/QFN-48 case. The module board dimension creates need for a special fixture socket.

http://www.ironwood.com

or possibly plastron are capable of producing that fixture. A good machinist is also a possible. Or, perhaps this forum board has such a fixture or other suggestions.

Thanks for your support and am looking forward to your suggestions.

 

[zappedagain]

That module is designed to be soldered to the board. Operation in a socket could be very degraded due to the inductance of the socket leads, especially the ones marked GND-RF.

I've typically seen newbie engineers damage modules like these by powering them up before connecting the GND connections. the part gets powered up and finds a GND path through a signal pin and damages the part. Some fail immediately, others fail in the field. Verify that every pin (besides the NCs) is connected. You can't assume that all of these pins with the same name are connected on the board.

Z

 

[MikeHalloran]

Zapped brought up a good point.
Make sure your incoming inspection is not powering up the modules.



Mike Halloran
Pembroke Pines, FL, USA

 

[CcanDo]

Thank's Guys,

Down-load protocol goes from the PC to the CEL dongle, to a conductor ribbon, w/a female header on the end. For security, the ribbon conductors were loto flipped on the header. The daughter board schematic was configured to match the flipped header. Care was taken in assembling the shielded ribbon conductors with air gap between the conductors.

Potentially, could we have corrupted the dongle and created a field within the flip ?

 

[zappedagain]

loto flipped???

Does your programmer/technician know about the flip, so (s)he isn't flipping it again to correct the problem?

'created a field within the flip' - strong enough for an ESD event? Not likely.

Z

 

[CcanDo]

Yes, the tech is installing the headers correctly. I know that was a reach.

The latest, those that would not load, will load IF you power the board w/ 9 VDC battery (3.0 VDC after the VR). Then the board will work, but only sparadicly.

As you may know, the user instructions insist the program is loaded without power to the board. The boards that work as designed are loaded without power.

 

[zappedagain]

The instructions are probably figuring that the programmer is power the IC being programmed and nothing else. If that part isn't isolated from your power net then the programmer is attempting to power the entire circuit and failing to provide enough current.

When you add the battery you reduce the load on the programmer, but may be introducing power supply sequencing issues.

Try installing a Schottky diode for power to the CEL module. When only the programmer is installed, this part will get reverse biased so the programmer doesn't have to program the rest of the circuit. Your circuit will have to be able to handle the 0.3V voltage drop when the normal power supply is there and forward biases the diode.

Z

 

[CcanDo]

Greg ? or, close ?