Saturday, July 30, 2011

More Custom PCBs Arrived

I submitted another order with the wonderful Dorkbot PCB prototyping service a few weeks ago, and promptly got it back the usual 14 days after ordering it.  From top to bottom:
  • The second spin of the SerialCouple thermocouple ADC board.  Given a thermocouple connected to the right side of the board, this board will convert it into ascii values and feed it back over the standard 6 pin FTDI serial connector on the left at 9.6k or 115k.  This board corrects a few problems with the first revision; namely the MISO/MOSI pins between the AVR and MAX31855 being reversed (which was correctable in software, but sloppy), and the board missing all of my custom silkscreen, which I had very carefully placed on the wrong layer in Eagle.
  • The board on the right is a relatively simple breakout board from the GoodFET connector to a five pogo-pin bed of nails targeting the IM-ME.  I did integrate a forward diode to drop the power supply from 3.3V to the 2.5V the IM-ME normally runs on, but I'm not sure if that's even required.
  • The final board on the bottom is the next step in my IR communications experiments.  It has an MSP430G2452 connected to a 38kHz IR receiver, a constant current IR LED driver (borrowed from the DP USB IR toy), and all of PORT1 brought out to an edge connector meant to be plugged into a breadboard or as a daughter board for other projects.  The 4 position DIP switch onboard is meant for being an address set, such that several of these boards could be deployed as addressable general purpose 8 bit I/O boards.  They could then be used to build a small inter-room telemetry network for controlling lights or turning on computers without WoL, or anything else that could be sent over 1200 baud or less.  I will certainly be posting about this more once I have a few of them talking to each other, but I wouldn't expect that to happen with my current summer schedule.
 And would you look at all that beautiful silk screen on the back; with my name and everything! 
    My summer internship has been keeping me as busy as ever, and I see no foreseeable chance to get back to my electronics workbench in Davis before school starts, so I'm afraid I'm going to be spending most of the summer doing little more than pining over these unpopulated boards.  The next Electronics Flea Market is in two weeks, so anyone in the area should come out and say hi, either in the parking lot at De Anza or at the breakfast at Bobbi's afterwards (around 9:30am).

    Saturday, July 23, 2011

    New Computer History Exhibit

    For those that don't live or visit the Silicon Valley, the Computer History Museum is an amazingly well put-together museum in Mountain View, cataloging the various stages and advances in computing in not only the most recent fifty years, but even earlier computing processes before digital computing.

    Previously, their exhibit consisted primarily of one large warehouse room filled with miscellaneous computer equipment, roughly sorted in chronological order from slide rules, to WW2-era tube calculators, up to the end of the mainframe and beginning of minicomputer eras.  It was one of those rooms that lay-men would spend half an hour walking through with the guided tour, but us nerds would need to make the pilgrimage a second day because 10am-5pm just wasn't enough hours to drool over things like the innards of a Cray-1 and one of Google's first rack servers.

    Last December (2010), the CHM closed the main archive exhibit and announced that they were working on a new and improved exhibit to be opened in January.  Of course, I had to go back to school in January, but my Dad and I were sitting around this morning, having little important to do, and decided that we should finally go experience this new exhibit.

    Now don't get me wrong, it is a technically much better exhibit.  They actually don't use the original warehouse space at all, but used the other half of the building, to the right of the lobby.  They do an amazing good job of separating out the different ages and categories of computers and technologies, and interpreting them, giving the lay-men an idea of where what they're looking at fits in in the computing history timeline.

    As someone who already pretty much knows that core memory is and what punch cards look like, it is a much less interesting museum.  They have maybe 15-20% of the raw hardware displayed that they previously had.  We walked through the exhibit for an hour and a half, getting lost three or four times in their Ikea-like maze of walls, and read all of the nice little interpretive signs under the iconic pieces of historical hardware, and... that was about it...

    We had fun playing "count the number of pieces of hardware on display we can pull out of our junk boxes," but as someone coming already aware of the general progression of computing history, the entire experience ended up feeling very... cursory.  I walked out the end thinking "That was a nice introduction, now where is the other half of the new exhibit?"

    They're doing  a very good job, the new exhibit is a much more appropriate lay-man's introduction, and at $12 the museum is still a steal, but as a serious hardware nerd, it's hard to not look back at what they had previously and be a little disappointed.

    Friday, July 8, 2011

    K Thermocouple Amplifier

     For various projects, my friends or I have needed to amplify a standard K type thermocouple to the 10mV/C ideally suited for computer and microcontroller based analog-to-digital converts.  This amplification is critical, because the thermocouple by itself only puts out 41 μV/C, which is TINY.  Additionally, thermocouples operate by measuring the temperature difference between two points, so you need to know the temperature on the "cold" end of the thermocouple as well.  Traditionally, this was done by submerging the cold end in a bucket of ice water, but that's a lot of work and becomes messy, so the preferred method is to use some sort of electronic compensator such as Linear's LT1025.  I have previously built the reference application from the LT1025 datasheet, but now that I've finally taught myself how to use Eagle, I decided reimplementing it on a printed circuit board might be useful in the future.
    This board is designed using entirely through-hole components, meaning it is very easy to solder together your own copy, to amplify two type K thermocouples to 10mV/C to be fed into any sort of ADC
    • The left connector is six screws which connect to your power supply and analog-to-digital converter.  From top to bottom these are:
      • V+: Something like 5V.  As long as the power supply provides a clean voltage larger than the output you need, the exact numbers don't explicitly matter.  The maximum voltage between V+ and V- is a function of the dual op-amp you use.
      • GND
      • V-: This can be shorted to ground through SJ1 if you don't need to measure temperatures below 10C, otherwise apply something like -5V.
      • Tcold: This is a 10mV/C signal of the cold junction reference on the circuit board.  Useful if you also want to measure the ambient temperature.  If you expect the ambient temperature to be below 0C, you will need to install a 10k resistor in R5, which is otherwise unneeded.
      • CH1: 10mV/C signal of the first thermocouple to be fed into a volt meter or ADC.
      • CH2: 10mV/C signal of the second thermocouple, also to be somehow measured.
    • The right connector is for the two K type thermocouples.  The polarity is important!  Make sure you get the positive wires in the positive terminals, which are marked in the silk screen.  I expect the thermocouples to be terminated with bare wire, so if they come ending in HPC connectors or otherwise, you just need to get a screw driver and disassemble those.
    The Bill of Materials can be very easily sourced from any major online distributor. Digikey is my personal favorite:
    • 7x 0.1μF capacitors (in 0.1" packages) (Digikey) C1-C7
    • 5x 2 position, 0.2" spaced wire-to-board terminal blocks (or any other way to get four and six positions) (Digikey) J1, J2, J3, CH1, CH2
    • 2x 255k 1% resistor (Digikey) R1, R2
    • 2x 1k 1% resistor (Digikey) R3, R4
    • 2x 100 ohm multiturn trimpot (Digikey) R6, R7
    • 2x 8DIP IC socket (Digikey)
    • 1x LT1025 DIP (Digikey) U1
    • 1x LTC1051 or any other dual low-offset op-amp. (Digikey) U2
    • Optionally 1x 10k resistor - R5

    I used ITead Studio's prototype PCB service, which is a very interesting service.  $14 out the door got me 10 copies of my sub-5cm x5cm board, and two random other boards of other person's design.  They're decent boards...  There is definitely some problems with drilling accuracy, but it doesn't look like any of the holes are too problematically off-center.  If you're planning on using this service for a board design, I suggest pushing all your masks out a few more mil than you really need for through-hole.
    Part of the deal is that for 10 cents, ITead will make two more copies of your board, and swap them with other orders, which is very fun.  I managed to track down where one of the two random swapped boards came from; it's a stepper motor driver for a CNC mill.  The other board is a little baffling; it seems to be some sort of surface mount breakout / some sort of LED bar / random stuff.  Pretty baffling...

    Unfortunately, I don't currently have access to my electronics work bench, so I can't assemble one of these boards yet to see if it actually works.I'm certainly not going to need ten copies of it, so if you want to be adventurous and try testing it for me, I'll mail you one of the boards to play with.

    PCBs in stock: 5

    If you would like to make any changes, or use this design in a larger project, or just have boards fabricated yourself, the entire design is licensed CC BY-SA, noting that the original schematic is heavily based on Linear's datasheet application note.  Download the Eagle/Gerber files here.

    Sunday, July 3, 2011

    Bus Pirate - Free PCB Build

    The wonderful thing about Dangerous Prototypes (other than them making some really cool stuff), is that they give away free PCBs throughout the week.  I managed to snag a free PCB code a few weeks ago and built a Logic Shrimp. DP was kind enough to send me ANOTHER PCB code for that, so I ordered everything I needed to build a Bus Pirate.

    First thing first; THE SILK SCREEN FOR THE CONNECTOR DOESN'T MATCH SEEED STUDIO'S CABLES.  The key is on the wrong side, so my cables (when they get here) are going to be folding back over the board.  Rats...

    Other than that, soldering it together was really quite easy and straight forward.  The resistor networks were about the only new package I had to deal with, and were easier than I expected for leadless packages.  I also used 0603 LEDs, where it calls for 0805, since that's all I happened to have in stock.

    Personal note: I just started working at PhaseSpace this week, which means I'm going to be very busy for the next three months.  My goal for the summer is to get first spin prototypes of all the board designs I have in the pipeline, but will probably not be able to really test / develop them further until school starts again when I have all kinds of free time again.