Any SMD projects?

[therecordingart]

I'd like to jump in with SMD, but don't know where to start. Anyone have a project started?

 

[jdbakker]

This is my work-in-progress: a discrete mic pre project with SMD parts except where it doesn't make sense (big caps, power parts). I'm working on the proto (although my day job hasn't left me much time the past month); hope to have something others can build early in the new year.

If you want something simpler that you can work on Right Now you could do a lot worse than the Dantimax ThatMic mic pre, designed by forum member mcs.

JDB.

 

[JohnRoberts]

I notice a recent discussion of a  cheap SMT digital scope kit. Something like $30 from a chinese source.

get your SMT experience and a small digital scope.

JR

 

[burdij]

How about a PIC project? This is a small section of a larger board I am working on which does 4 channels of midi to CV conversion for a synth project. I wanted to make sure I had all the micro circuitry worked out before stuffing the larger board with the expensive parts.

All the small components are SMD. Here is a view of the board with the SMD parts placed and soldered, poppin' fresh from the oven:

The second shot shows the rest of the parts stuffed. This board is set up for a PIC 18F2520 at 20MHz. There is a midi in and thru connector. One of the white connectors is set up for a connection to a programmer/in-circuit debugger. A header brings out I/O lines to play with.

The last shot shows the board connected to a power supply and the ICD2 in-circuit programmer/debugger (from SparkFun for $106) which is also connected to a desktop running the MPLAB IDE via USB. I have a copy of the MPLAB C18 compiler loaded to do program development. It is a very complete and low cost dev environment. BTW - when you recycle those cellphones, hang on to the wall warts. They are usually compact 5VDC switchers at 3/4-1 Amp. like the one in the picture, perfect for this kind of a project.

 

[JohnRoberts]

Any reason to not use a SMT PIC?

The PIC package is not as tight pin spacing as several SMT parts I use. 

JR

 

[burdij]

No specific reason why an SMD processor couldn't be used. When I did the original layout, I wasn't sure what kind of programming resources I would have available but then I found the cheap ICD2. With the ISPC connector in place, the processor could be firmly anchored to the board.

 

[JohnRoberts]

Come to think of it, for low volume production I have seen socketed PICs  used, so they could be programmed in a socket fixture, then transferred to the production PCB.  Of course to use the debugging capability you need to be connected to the PIC so you'd need to modify a board.

JR

 

[therecordingart]

How about a PIC project? This is a small section of a larger board I am working on which does 4 channels of midi to CV conversion for a synth project. I wanted to make sure I had all the micro circuitry worked out before stuffing the larger board with the expensive parts.

All the small components are SMD. Here is a view of the board with the SMD parts placed and soldered, poppin' fresh from the oven:

The second shot shows the rest of the parts stuffed. This board is set up for a PIC 18F2520 at 20MHz. There is a midi in and thru connector. One of the white connectors is set up for a connection to a programmer/in-circuit debugger. A header brings out I/O lines to play with.

The last shot shows the board connected to a power supply and the ICD2 in-circuit programmer/debugger (from SparkFun for $106) which is also connected to a desktop running the MPLAB IDE via USB. I have a copy of the MPLAB C18 compiler loaded to do program development. It is a very complete and low cost dev environment. BTW - when you recycle those cellphones, hang on to the wall warts. They are usually compact 5VDC switchers at 3/4-1 Amp. like the one in the picture, perfect for this kind of a project.

Nice. This would be cool as well and a CV to midi converter to make my own e-drum brain.

 

[burdij]

The 18F2520 has 10 channels of on-chip 10bit A/D conversion which, combined with an external voltage ref. would be able to do CV to MIDI with fairly good accuracy and stability. That would give about 100 levels per volt or about 8 levels per semitone.

 

[mikep]

Come to think of it, for low volume production I have seen socketed PICs  used, so they could be programmed in a socket fixture, then transferred to the production PCB. 

slightly OT, Ive run into issues pre-programming SMD PICS.  some of the chips' program memory got corrupted in assembly/reflow.  not a big percentage, but enough to make me stop doing it.  anyone else see that? 

mike

 

[JohnRoberts]

I've never used preprogrammed PICs, but I would be suspicious of that explanation. Since you have to apply voltage in a specific way to alter the flash. I would suspect the parts were not properly programmed in the first place. 

Probably not a huge deal to add pads for adding a programming header to rework a small percentage of parts, but I'm not sure I buy that explanation. Did you run the explanation by MicroChip? Was it them who preprogrammed the parts?  :

I like to add a serial number in my code, which I don't think I can easily do, preprogrammed, but I haven't really looked into that.

JR

 

[Andy Peters]

Come to think of it, for low volume production I have seen socketed PICs  used, so they could be programmed in a socket fixture, then transferred to the production PCB. 

slightly OT, Ive run into issues pre-programming SMD PICS.  some of the chips' program memory got corrupted in assembly/reflow.  not a big percentage, but enough to make me stop doing it.  anyone else see that? 

mike

Blame the assembly house for using improper reflow profiles.

-a

 

[JohnRoberts]

Come to think of it, for low volume production I have seen socketed PICs  used, so they could be programmed in a socket fixture, then transferred to the production PCB. 

slightly OT, Ive run into issues pre-programming SMD PICS.  some of the chips' program memory got corrupted in assembly/reflow.  not a big percentage, but enough to make me stop doing it.  anyone else see that? 

mike

Blame the assembly house for using improper reflow profiles.

-a

For accelerated life testing they push the parts past 150'C (that's pretty warm), and not just for a minute.

Retention Bake
Data storage in applicable devices is done by developing a charge on the floating gate structure in the memory cell. Charge loss in this cell structure results in a conversion of zeroes to ones. A 150°C temperature is used to accelerate charge loss in the memory cell and measure the data retention on the EPROM and EEPROM portions of the circuitry. Devices subjected to 175°C are tested at 96 hours and 504 hours. This bake accelerates charge loss in the memory cell, and 96 hours at 175°C is equivalent to greater than 250 years in the field at 55°C. The actual failure rate experienced could be considerably less than that calculated if lower device temperatures occur in the application board.

Here is data for the family I use  http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1413

no fails for 500 hour tests.

=======
OK all good but lead free solder needs peak temps in 250-260'C  range so this testing at 175'C does not prove it will survive, a contract manufacturer running high and long on temp soldering profiles.

FWIW I have experienced a CM who melted my SMT film caps, back in the good old lead solder days, so who knows what a sloppy process might do?

JR