The Quarky Engineer

Since my GTR is getting close to completion, I need to get the sensor wire harnesses done. I used some rope and string to figure out the lengths and routing, then used that as a guide for each harness. Each sensor harness is twisted, and then the sensor harnesses are combined and twisted for each main harness.

Normally I would have combined the +5V and GND wires from each sensor in a given harness, but with so many sensors I will be connecting to a number of different capture devices. Each of those devices has their own +5V supply, and in the case of the Motec ECU there are 3 independent +5V busses.

I still need to add the larger connectors on the ECU side, but I’ll do that after a test fit.

I took a crack at hand soldering those 0805s, 0603s, SOICs, and QFPs. It took me three tries on the second QFP to get it right. Fortunately the hot air rework tool makes removing them easy. In the end I got both QFPs down and tested every pin. The 0805s and 0603s were indeed pretty easy.. easier than I thought they would be.

Next up I’ll do another of the same board with solder paste and the air station, then another with a skillet, then repeat all three again. The mantis makes it really easy to see your screwups! (Also Matthew R. Wilson pointed out this practice board doesn’t have solder mask between the pins on the QFP, which makes it a tad harder)

It is interesting to look at how the auto-routing performs in Diptrace. Attached are the following:

(1) My 2 layer hand routing, with as much ground plane fill on the back side as I could do. I tried to do as much of the routing as possible on the topside so the ground fill would cover as many things as possible.

(2) The auto-routed 2 layer which doesn’t have any planes. The autorouter uses a pretty simple layer direction method, with more traces on the lower layer than the upper one.

(3) The auto-routed 4 layer with a ground plane fill on one inner layer, and a power fill (both 3.3V and 5V, with the 3.3V parts on the upper part) on the second inner plane. After adding the fills I let the auto-router do the rest.

For doing fine PCB work I picked up a Vision Optics Mantis. (3D microscope). It is a really nifty device with a single large ‘viewport’ instead of eyepieces. It gives you a very stereoscopic view that you can pan and tilt around.

Optically it appears like large objects at a reasonable distance, so I can wear my regular long range vision glasses and things are in perfect focus. Neat technology. (I don’t know Paul Yaw, you might just have to get one).

Ok hand soldering 0805s I can see, but those 0603s are tiny! I clearly need something with stereoscopic magnification.

Well, it worked. Sort of. I did a second pass moving the components around and I was able to hand route in two layers, with most of the second layer being a ground pour. I still need to do some trace cleanup, but DRC passes so tolerances are ok. Certainly it is good enough for a prototype.

This was perhaps the most interesting drink I have tasted this year. A port from 1875. It is hard to imagine drinking from a bottle made in not the previous century, but the one before that.

It was actually pretty good. A bit sweet, and the port from 1970 was actually better, but the history of a bottle this old!

Happy Thanksgiving everyone!

Doing a little spring cleaning today and I found this old shirt. I bought it at the Rush concert in Indianapolis back in 1986, when I was my 15 years old 120lb self. I picked a random t-shirt this morning to wear, and funny enough it was the R30 Rush concert tour. Needless to say I don’t think the one from 1986 it fits anymore.

Paul Yaw reminded me today how hard it is to make a list of favorite musicians…. but most influential to my own appreciation of music is a shorter list from my younger years. (And thanks to Thomas Talavage for being a partner in enjoying Rush concerts in the college years)

It’s like Tetris, but more fun!

So I’m trying to decide if I should ditch the Schottky diodes (1N5817s) connecting to ground in the analog input stage. This is the pre-stage to the ADCs, and the design is for analog input from 0-5V sensors (automotive). 4 Op-amps for 2 inputs. The second op-amp is a sallen-key filter, which then goes to the ADC input (which is a 0-2.5V ADC), thus the divide-by-2 between stages. The two diodes at the input, one to ground and one to +5v, just provides protection from over voltage and reverse voltage on the analog inputs.

The overvoltage protection is a good idea, as it is reasonable to think someone might accidentally hook up something that is going to Vbat (>12v). Hooking up to something with negative voltage is not very common.. and the most probable cause would be by also having the main power connection backwards. In that case you would have no power on the 5V rail (since there is a diode on the power input), but you could potentially get -5Vs relative to ground on an input pin. That would damage the op-amp even with the limited current input… so yea.. I guess I should keep them there… or just not wire it up backwards. 😉