Today we were surprised to see our sensor boards and digikey parts arrive today. As
the box cutter inched its way through the packaging, the smell of fresh solder filled the air. The initial reaction to
the revealed sensor boards were "Wow, that's small." This was coming from Jared and I. Ilya was kind of baffled by our reaction since
we had designed the boards ourselves. So we started to analyze the boards and the first things that was clearly wrong was the silk screen layer.
All of the reference designators were unreadable. This was unfortunate but thankfully it wouldn't affect the performance of our boards.
And so we analyzed some more. This time something larger came up. Somewhere along the line the wire holes for our CSAs were lost in layout.
This poses a large problem since our glider application is largely dependent on current sensing. As for the IC's, there was one problem
regarding footprints and two problems regarding the parts we had ordered. The first problem was that the temperature sensor's footprint was
to large. Once again it was one of the first footprints that we had made that caused a problem. In terms of packaging, the 8-SOIC is
NOT the same as the TSSOP-8. Also digikey decided not to send us the light sensor, the only I2C sensor we could easily solder which still fit
on the pad. Go Figure.
Sensors comma Website
As of now the Sensors Board has been sent out to be manufactured, and the parts have been ordered from Digikey.
Since Ilya chose express 2 day manufacture, we should have both the parts and the boards by next week so we can
get to soldering them together. With the sensors on hand, we can test the sensors, generate sensor specific code
and attributes, and complete the model of the system for our presentations on the 19th of August. We definately
wont have the slaves and master in by the 19th, but we can at least have the Sensors. That way we will have some
meaningful data to collect and show with the two Dev Boards at the Presentation.
Also the Website is live. Woo! We've added some basic information about the project on the project page and some
images should be up shortly. I guess if you are reading this, you can see for yourself.
Again, Again, Again
We had thought we had finished the slave and sensors for production a week ago. It took many revisions but we felt we had finished all the hardware side of the project. Thus we began investigating codeing and reviewing the work that our CS counterparts had done while we waited for the boards to be sent out for manufacturing. It wanst until Monday of this week that we found that ther were problems with the masterboard we had sent out many weeks ago. Advanced Circuits informed us that two of the chips do not actually fit their footprints. The Microcontroller's pitch, the distance between the centers of each pin, was 2 mills too short. On a chip with 16 pins per side, the small error in pitch compounds into large errors down the line. This footprint had to be fixed. The other chip was the RS232 converter chip. The mistake there was we indicated the wrong package from Digikey, resulting in Advanced Circuits trying to fit absolutly the wrong package onto the footprint. Since they had already bought the parts, we also changed the footprint of the RS232 converter. With these changes we were actually able to decrease the size of the master board, but in the long run we lost time and wasted some money on the now faulty master boards. To make up for lost time and money we compiled the new master and 3 slaves onto one board so that there would be less overhead with labor and we would recieve the masters and the slaves at the same time.
Satisfying Size Constraints...Again
This week has been extremely busy. There has been a large amount of revamping of these sensor boards, mostly due to the size constraints of the slave board but also because we need to be able to mount each of these sensor boards on objects other than the slave board. Consequently there are now holes that require a .25 inch diameter circle of space. The connectors had to be shoved apart, increasing the length to about .9 inches. It wasn't a very large increase in size but it still had some impact on the final design. We are still hopeing to fit 8 sensor connections on the sensor boards, the only requirement being 2 of these connections must rely on there off-board mounting capabilities. Despite all of this, every sensor board has been laid out and we are now officially ready to draw up the slave. Oh, and one other setback. Each sensor board layout must be converged into one large layout. Hopefully there is some sort of import feature or we'll have to say goodbye to our free afternoons.
Satisfying Size Constraints
Yesterday we had a rude awakening regarding the expectations of the size of slave board. After placing the components in Layout the dimensions were around 2.4 by 2.1 inches. This did not include any additional overhang (approx.5 in) from attatching the sensor packs (2.9 x 2.6 yikes!). It all came back to the spawn of a design, the bidirectional current sensing board. The board was too large, or too long. The width was reasonable, however the length (1 inch) was not. The amount of overhead space each sensor took up above the board needed to be decreased. So the original design of this bidirectional current sensor involved 2 current sense amplifiers, 1 op amp used as an adder, and 1 differential amplifier used as a subtractor. As it turns out, the differential amplifier wasn't necessary at all. We turned a single op amp into the adder and subtracter acquiring the same result. We managed to shave off .2 inches from the length of the board which is quite significant since our intended application will not be so forgiving in the dimensional department.
What's all this then?
Another day of waiting for the master boards to come in. Since the sensor boards we have made all fit in the current size, Alex has begun routing the Slave board. The width between the connectors depends on the sensor boards and can have a large effect on where things can be placed and routed on the slave board. Once all of the sensor board layouts are completed, and we can not think of any other sensors to make, we will manufature them in one fell swoop; all of the boards will be printed on one dual layer plater and then cut out. This, of course, means they must be the same size to facilitate separating the individuals.
That was Easy
The size of each sensor board has been finalized,
or so we hope. This has reduced the slave board in size
to about 2 x 2.5 inches. There will be some overhang with each sensor pack, but
that was to be expected since there are eight sensors on each slave board that are
1 x .5 inches in size. The BMA250 has been replaced today with a lower power accelerometer made
by VTI Technologies. It should also be easier to mount because of the package. The only drawback is
the price, where it costs about $9 per sensor as opposed to $5 per sensor. The temperature sensing board has been completed today. It could have been made into a single layer board but since the bidirectional current sensing board
had to be two layers it has been made into the standard (size and layers). The website is simple and on its
way to completion. Hopefully we can get some media posted ASAP.
Case of the Mondays
Jared is currently completing the Bi-directional Current Sense Amplifier sensor board. Earlier today We had made a major change to the setup. Instead of using a dual opamp package we went with a single op amp mixed with a 1V/V gain differential amplifier. This removed four 0805 resistors from the board allowing for reduction in size. The dimensions were originally 1.25 inches by about .65 inches. The board is now 1inch by .5 inches and the connectors are also .3 inches closer. Since the connectors are now closer in proximity the slave board can also reduce in size