MSP430

I have added another section so to make it easy to find the topics until this gets split up, I have added this contents...

NEW! SimpliciTI Page
           SimpliciTI using MSPGCC

March 25, 2009

The USI I2C slave device

This is the start of some of my MSP430 adventures. As the question of using the USI port as an I2C slave, often on the popular MSP430f2012 and MSP430f2013, comes up...

 I'll start here.

Now, I may have gone about this all wrong, but it is my experience that the USI port has deficiencies where doing I2C, as a slave, are concerned. I never looked at doing a master as a bit bang version is short and sweet.

Here are links to the code that work for me:

Why this code...

1) There is a flag set for the I2C stop condition but no interrupt is generated. So I check at every SDA rising edge with a GPIO interrupt for the stop condition.

2) Acknowledge bits are done separately as there are only 8 bits of USI shift register. An interrupt is generated when the USICNT goes zero. But this happens when SCL goes high. If there is a low  sitting on SDA for the device ack and a 'USICTL0|= USIOE' is done to do input on SDA for the next byte, there will be a low to high on SDA while SCL is high. This is a start condition and should not be done while the slave is inputting data from some master. So I set up to check for falling edge on SCL in a GPIO interrupt before setting SDA to input.

Since the last time I put this code up I have enhanced it so 'this' slave device will look and feel much like an EEPROM or Dallas I2C devices. 'This' device now receives a control byte and a command byte before going into subsequent I/O.

Note:
The call 'unsigned char i2c_super_command( )' needs to be defined somewhere in your code. It is called after a valid control byte and any command byte is received. The command byte will be in 'usi_command'.

At line 137 there is a LPM_EXIT and this may not suit your needs.

This is written for mspgcc and the likes of interrupts will have to be changed for other compilers.

May 14, 2009

The MSP430 as an I2C bit bang master

Revised January 18, 2010
I have only used simple slave devices that do not sink the SCL line so my older code drove this line without a pullup resistor. I have changed the code linked below so that both SCL and SDA are driven as open collectors. There is still no code to wait on a stretched SCL but if you don't use devices that sink SCL, there is no good reason to use it. Also, the delays are now set in the device header with the pin assignments.

~~~
I don't see much of a point in using serial hardware for an I2C master as a bit bang version is short and simple. First is a generic library as SDA and SCL should be hidden at a higher level. This should meet most of the I2C master needs for a straight 8 bit protocol:

Delete the reference to HE_util.h. There is mspgcc code commented out in the header that would be referred to

'for loops are not a good way to create delays, they can disappear with optimization. The actual delay could be set with a more global define and depends on the clock speed/wire length. I have a large delay here for a long I2C line.

Set the port and pins you will use in the header.

Now you can write your specific code for your device(s). Here is an example for reading 32 bytes of an EEPROM:

There is also an example in the header for a temperature device.

July 16, 2009

Setting the DCO with a watch crystal oscillator

There are several routines out there. If you search 'Set_DCO' and 'SetDCO' in Google, you will find them. What may not be obvious is that the routine will change according to the family of devices you are using. If you look at x2xx basic clock, you will find this:

And if you look at x1xx basic clock you will find this:

Toward the lower end of the Set_DCO routine you will find the line:

            cmp.b   #0x87,&BCSCTL1             ; Can RSEL.x be increased?

 In some examples I've seen the compare to 0x0f and 0x07. But XT2OFF is usually set to '1'. So neither of these compares really work. It only seems to work because the routine usually doesn't get there.

 Here is a Set_DCO that I have tested on the x1xx family chip that should work for you:

Set_DCO.s

This is written for MSPGCC so you will need to modify it according to your flavor of assembler. It is written to be called from 'C' with the delta value and saves most of the timer state on the stack.

 
JULY 17, 2009 (rev 8_30_9)

(MSP430F1612, JTAG, watch crystal and LED

 I got a pack of five of these boards on ebay from china. ~$10 in the door. I think the quality is fine, that makes the price right. This was my first quad, second board. This is my first board:

I'll expanded on this entry as soon as possible. But if I can make this work, so can anyone. It only looks daunting if you have not been there. (I say that from personal experience.) Both of the above chips are up and running.

But to start with, you should not try to deal with this job on a pin at a time bases. That may work on a soic device if you are very diligent, but half a mm?!, don't go there. What I found is that there was more than enough tinning that you don't have to add any solder to your work. In fact, you need to sop up what is already there. Other than a bottle of flux and a decent soldering station, you really need an X10 microscope. (If you don't have old eyes like me, get one and keep from getting old eyes!) At that, I don't think you should try this without being able to see each pin as if it were a quarter inch in size, IMHO.

What worked for me was to flux the pins and then use a piece of #22 finely stranded wire laid across the pins as a wick. Work your iron, not too fast, down the row. You will see the solder wick into the #22 as you go. Again, take your time, you are soldering and need to get that iron heat to your work. On the other hand, it is not like normal soldering where you 'add' solder, that is the big difference.

Now, there may be instances where you really do need to add solder, but in the work I have done so far I have found that is a mistake. (Rework would need new solder as too much gets wicked off? A new board and chip seem to have plenty to start, something else.)

Before you solder, get that chip aligned! If the pins are not positioned right on, your chance of failure goes up exponentially. Of course a quad chip is twice the work of an in line as you have two dimensions to work on. You need to clamp the chip down, I did my first ones with forceps and rubber bands. I'll be making a jigging device before I do this again. I used a 4" Bausch and Lomb magnifying glass to set but check with the microscope just to be sure. (twenty years ago I may not have needed the magnifying glass!) BTW, I still have 20/20 three feet and out, but my arms are not that long :)

Just dab two of the opposite corner pins with solder, don't worry about bridging while checking your alignment. When you wick, the bridging will disappear.

When you think the work is done, that is where the microscope comes in. You should see a nice flow from the pin to the board. Bridging will be the challenge, on any pins you see a bridge, just do a little more wicking. If you have moved across the pins with a solid wick and at a reasonable speed, you should see what looks like machine soldering. Don't be stingy with the flux, flux is your friend. When you pass inspection it will all wash off with small dose of flux remover, then soap and water.

Find yourself a hand full of chips and boards at a reasonable price and try your hand at it. You may find, as I did, you will have a working device soldered down the first time out. If you do, you will have no trouble thinking you can do it anytime you need from then on.

I'll take more pictures and talk more when I go there. But after a couple of chips that work when I got done, I would not call this rocket science. Search on youtube, you will find several different philosophies on soldering fine pitch chips. They just make it looks so easy!

Professional SMT Soldering - Washburn Computer Group

OCTOBER 1, 2009

Soldering, One more Time

Here I've taken some photos of my third chip. This went very quickly now that I am more comfortable with this. I'd say start to washing the flux off, this is less than a ten minute job. It is about being comfortable with what you are doing. A little practice will get you there fast.

Before picking up the forceps get that chip aligned under your magnifying glass. Hold the alignment on one side of the chip with your finger then put the clamp on. There are two rubber bands on the clamp. One to the right for a light clamping pressure. The one on the left is just a couple of wraps on the tip and the rest is held out of the way with a piece of wire. This puts pressure from the tip on the center of the chip where the metal part of the clamp would land on the edge of the chip. Now you can pick it up by the forceps and take a good look at your alignment. Yes, a hands free magnifying glass is necessary.

When you are satisfied with the alignment put a drop of solder on the four corners and then you can take the clamp off. To make sure the chip stays seated down I've added four more dabs of solder. You can see that here.

Now it is time to solder it down. Be liberal with that flux. I didn't use anything fancy, just some off the shelf GC liquid flux. You need the flux because you will not be adding solder but instead, removing it. That piece of stranded is #22. I like it better than wick because it is not so intrusive. Just lay the wire along the pins and gently work your iron down the row on top of the wire. You will see lots of solder wick into the wire. sop up as much as you can and take a look under a microscope. If you see any bridging do a little more wicking. When you are satisfied, finish up by touching your iron down the pads and you will have a very professional looking job. You will see an example of 'touching' in the Washburn video. It should look like production work under the microscope.

Here it is before washing the flux off.

The JTAG test, works fine!

Thanks, Dan.

Copyright © 1996 - 2009, Highlands Electronics All Rights Reserved.
Page created: 25 mar 09   rev: 3 feb 10

http://www.lakeweb.net/MSP430/index.html