Robots, et al.

TI App Note: AN-2020 Thermal Design By Insight, Not Hindsight is a really good app note to help calculate thermal impedances and estimating heat sinking required on a circuit board.

Page 4 (table 1) contains an error in its calculation of the thermal impedance of a 12-mil via.  The problem is the calculation uses the finished plating hole thickness as the outer radius, since this is the finished thickness it should be the inner radius.  This changes the via thermal impedance from 261 C/W to 239 C/W.  Equation (3) should be:

Just thought I’d tell the world, I hate errors in app notes.

I’ve been doing more embedded coding lately and need music that’s up-beat, with no vocals.  My latest pick is the album LP4 I can listen to this album for 8-hours straight on repeat – AMAZING.

I especially like the random clips they add to the songs like a German clip before ‘drugs’ or the clip at the end of ‘Party with Children’:

I’ve been thinkin’ what to do with my future, I could be a mud doctor checkin’ out the earth underneath.

I’ve been a fan of the relativly cheap power supplies sold at MPJA.  In the past I’ve been used to the larger supplies that will trip relays when over-currenting the supply and that is usually enough to get my attention.   The newer solid state supplies are much lighter and don’t have any relays in them.  I have not yet blown up any circuits, but prememptivly I decided to add a constant-current buzzer.

Overall, these are nice supplies.  We got a few at work and other then the fan being loud on one of them we have no complaints.

I knew that the signal to show whether these supplies were in contant-voltage or contstant-current mode was being sent to the low-voltage LCD display so I started probing around in the connector between the power board and the LCD board.

The white connector on the bottom of the board connects to the actual high-power regulator board.  I discovered that pin 4 (or two if you count from the other side), carries a signal that is 12V when in constant-voltage mode, and 0V when in constant-current mode.

The plan:  use buzzer, turn it on when constant current mode is enabled.  First I tried just connecting the buzzer between the signal line and 12V.  This didn’t work and the LCD got very cranky – must be this line is not as low-impedance as I hoped, time for plan B:

I needed a way to drive an active-low signal without having to drive 74xx logic or anything.  The solution: P-type Mosfet.  When the gate is at 12V (Vgs = 0V), the FET is off, and therefore the buzzer is off.  When the gate is low, Vgs = 12V, the P-FET turns on and the buzzer is on.  The great thing about using MOSFETs instead of BJTs is that there is no gate current when the circuit is not switching (I don’t anticipate this thing jumping in and out of constant-current mode, that means I need to turn the currrent limit up).

I wired in a TO-92 leaded PMOS, connecting R2 to the gate, C3 (12V) to the source, and connected ground on the buzzer to the ground on the backlight for the LCD.  This seems to be working great and doesn’t seem to be too loud once it’s in the enclosure.

The guys at MakerHaus and I are getting together to make an Open Spectrometer.  We just kicked off our Kickstarter and I’ll be posting the status of the hardware on here and the Open Spectrometer website.

So far I have roughed out code on the F28069 and I should have a block diagram up soon.  The code is being hosted on GitHub.

Long-term plan is to use a slightly more user-friendly architecture, I just got in my LM4F232 Evaluation Kit, and hopefully alpha prototypes will be a daughtercard for that dev kit.

Over the past month or so I’ve been devising a quadcopter.  What I’ve come up with is pretty much the same as every other vanilla-quadcopter.  I decided that I don’t want to use ESCs to control the motors – I’m going to do sensorless field-oriented control (FOC) using a F28069 controlStick.

I’m hoping that using FOC control is going to improve the efficiency of the motors.  I’m not entirely sure about this, but I doubt efficiency is the main concern of the cost-down RC speed controllers.

The main reason I’m doing this is because it’s a awesome engineering challenge.  Who doesn’t want to do 4-motor sensorless FOC control with an an awesome DSP.

So far the progress I’ve made on this is to layout a 3-phase H-bridge board with low-side current sense.  I submitted it to batchPCB and the boards came back.  So far I am able to power them and not draw any current (when the gate drives are disabled).  I’m hoping to get some preliminary firmware going to get at least one motor spinning in the next week or two.

Front side of board

Backside of the board

This is really awesome Finite Element Method Magnetics – I’ll have to check this out soon.

This weekend I attended the Open Source Hardware Summit.  It was great getting to meet all the people who’s blogs I read and hardware I look at (and sometimes buy).  One talk that especially inspired me was Amanda Wozniak’s talk on Open Sourcing the Engineering Process, this encouraged me to finally start a blog and document all the engineering I do for fun.

It was great to have lunch with Jeff from Mighty Ohm, and seeing a talk on manufacturing with Ian from Dangerous Prototypes and Eric from Seeed Studio.

I also was inspired by Geoffrey Barrows talk to order one of the optical flow sensors from Centeye.  I’ll be sure to post when I get them in.