Page 7 of 7 FirstFirst ... 34567
Results 61 to 70 of 70

Thread: Malum IK

  1. #61

    Re: Malum IK

    I have used exactly this circuit to interface between Teensy 3.2 and 3.5 and the DXL bus, yes!

    The RX pin has a 560 Ohm / 1 kOhm resistive divider to translate the 5V of the buffer chip down to the 3.3V accepted by the Teensy.

    Note: The TX pin doesn't have a pulldown here, but the direction does, so in practice, it doesn't matter. The main change I've done since this diagram is increase the R304 value to 10k to draw slightly less current (the time to switch doesn't change perceptibly) and add a 100k pulldown on the microcontroller TX pin, because by general practice, no pin should be undefined on startup.

    Btw, I have a story to tell about this latest board version, and how I had to iterate several times through Macrofab to get it right (at no cost to me! They've been very good about that.) It started with them placing an IC 180 degrees rotated, which caused other failures, which caused rework, which caused other failures of things that previously worked, ... This was exacerbated by not being able to inspect the soldering at the bottom of those QFN packages. In the end we made a new board, using SOIC/TQFP style packages, and that worked fine! From this experiene, I conclude that QFN ("quad flat no lead") is as tricky as BGA to get right.
    But, that has taken a lot of time to get over the finish line. When I finally got a working board on the fourth(?) iteration, I was kind-of low on gas, and it's sitting without a chassis to screw it in to, because of course I need to re-cut the chassis to fit the new board, even though it eventualy ends up integrating more functions and requires fewer external little drivers/boards.

    Also, the AD opamp based video switch seems to work great, although I can't genlock the two cameras to each other, so there is some time to re-sync when I switch.
    Last edited by jwatte; 6 Days Ago at 02:05 PM.

  2. #62

    Re: Malum IK

    Thanks @jwatte,

    I will double check my layout and make sure I did not screw something up in translation, which is highly possible! Or bad soldering...

    As for T3.2/5 you probably don't need to worry about 5v... But in my case T4(or T3.6) than you do...

    Sounds like the microfab ones were sort of a pain! Glad you got one working now.

  3. Re: Malum IK

    That's curious. I've done more than a dozen boards with QFN parts through macrofab and haven't had a single one with the part assembled in an orientation I didn't expect. Plenty of other problems for sure, but none that weren't my own fault.

  4. #64
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,289
    Images
    27
    Rep Power
    276

    Re: Malum IK

    Robotis ran into dead IC problems when using a single two-channel buffer on their earliest CM-900 boards, so they switched to two one-channel buffers at different voltages (5V TX to buss; 3.3V RX to MCU). I've continued using that two buffer solution for a while without issue, but the need for TTL inputs on the 5V TX buffer to guarantee operation is kinda annoying.

    If the newer teensy boards have the same hardware half-duplex stuff that internally disables RX when transmitting, then something like the 74LVC1T45 level-translator is probably going to work well. One side with direction pin is powered by 1~5V, and the other is independently powered by 1~5V. Everything goes tri-state if either of the power supplies browns out.



    I am quite curious how you managed to mess up QFN assembly. Did you use their footprints and stencil patterns? The recommended size and number of vias under the thermal pad? Did they rework previously assembled boards, or fully assemble new boards of the same version from scratch? Did they use decent solder flux during rework? Hot-air gun or oven? The surrounding pads of QFN are easily inspected visually and can be continuity tested for shorts to the thermal pad.

    I've got several boards in the works that use QFN parts (50V/4A stepper motor driver, ATSAM D21/D51, 8 channel DAC, etc) and I was not expecting any serious issue in hand-assembly given past experiences with 0.5mm-pitch VSSOP, DFN, and LGA parts I've already used. I was going to use Seeed for PCB/stencil as initial test then try Macrofab for larger production...
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    bleh

  5. #65

    Re: Malum IK

    I am quite curious how you managed to mess up QFN assembly. Did you use their footprints and stencil patterns?
    They didn't have QFN footprints to match, so I constructed them in KiCad using the recommended dimensions from the data sheets. Also: I use "QFN" to mean all the packages with pads on the bottom but no leads extending out on the side. Some of them were a bit more "polygonal" on the underside -- I used a very small BS-170 analog, and a small 1.8V regulator with a rotated diamond pattern, for example.

    I asked them for comments and they said everything I did was fine and "should have worked."

    My guess is they are still warming up their Mexico-side assembly line, and my board had the unfortunate happenstance of some new employee who probably wasn't going to work out or something.

    They did turn at least one chip upside down, which might very well have fried other chips by leak-through. I think their re-work was all hot air, although their original reflow soldering is IR double-sided.

    The main problem for me was that it's impossible to inspect the solder under the parts once it's done. I can measure continuity, and for a number of components, there was continuity between pins that should be there: gate+drain for the BS-170, for example, and after the second time it came back, +5V and GND were somehow connected, which I couldn't figure out where it would happen, because 17 different components were strung between those two rails ...

    There were some other problems, too, for example a 560R resistor turned into a 560K resistor. Again, most of that was probably attributed to the employee having the bad day, and then follow-on problems from the re-work (plus perhaps original problems that I couldn't diagnose initially, that then remained after the first re-work.)

    When the final revision came back, having shorted out +5V to GND (which previously had worked fine) they gave me store credit and I re-spun the board with SOIC chips, and it worked swimmingly. Except for this one Soberton supposedly reflow-compatible piezo buzzer, which I couldn't make go in any of the three separate boards, so I'm giving up on that particular part. I don't blame them for that one. I'll solder a connector on at the pads for that part, for an external unit instead. There was no SOIC version of the AKM motor driver I used for the "feeder" motor, so I swapped that out for a TI 8-SOIC DRV 88xx and that worked fine. There was also no QFP/SOIC version of the dual-rail buck controller, so I kept that one, but that one had worked well every time.

    The first board had a bunch of design problems that were mine, though, so it's not all on Macrofab. I had forgotten to have pull-downs/ups for a bunch of pins that might be "undefined" on start-up. (I added 100K pull-downs to all of them in version 2.) I had a factor 10 too small capacitor in the compensation circuit for my 5V buck regulator (I re-worked that with a small ceramic capacitor while debugging that version of the board.) I had a Teensy pin coupled directly to an input for the power control IC, and that ended up draining the turn-on pin to 1.1V, even though the Teensy isn't documented as having a power-off pull-down on its GPIOs. (I'm using a 3.6 on this board, btw.) However, that board still had very hot spots (85C +) which I now believe were also symptoms of bad solders under parts; given that that was the first iteration of the board, I thought the problems were all on me, and I paid for the next board.

    So, full story:
    Board v1: severe hot spots, but also a number of design bugs. Wrote this one off.
    Board v2: hot spots from chips mounted wrong. a few bad components. presumed solder bridges under QFN/DFN-like parts. Re-worked several times, but ended up showing +5V shorted to ground. They credited this board back to me.
    Board v3: SOIC and similar vanilla parts. Worked first time, once I got it in my hands!
    Last edited by jwatte; 5 Days Ago at 04:20 PM.

  6. #66

    Re: Malum IK

    Attachment 7632

    Not sure this is all that useful. The board is fairly crammed, so it's hard to tell from the CAD. Maybe I'll remember to take a photo when I get home.

  7. #67
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,289
    Images
    27
    Rep Power
    276

    Re: Malum IK

    More crammed than this 33x50mm thing?

    Click image for larger version. 

Name:	tb67s249-v6.8.2.png 
Views:	8 
Size:	156.6 KB 
ID:	7640

    ATSAM D21 for control with 32k crystal, primary TB67S249 stepper driver, secondary DRV8871 brushed DC driver (for solenoid-based shutter), two opto/limit sensor inputs each with monostable pulse shaping outputs, (22) 50V/10uF 1206 ceramic caps, (2) 63V 12.5mm diameter electrolytic caps, linear regulator, and a few other passives.
    Last edited by tician; 4 Days Ago at 02:22 PM.
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    bleh

  8. #68

    Re: Malum IK

    [quote]More crammed than this 33x50mm thing?[/COLOR]

    Probably not on average, but the dual-rail buck converter area, and the video-switch-and-general-inverter area are both quite full!

    Interesting (to me) learning: I originally used SMD connectors that could be easily populated in the design, but those use more board area, so I actually saved space by switching to pin headers.
    For mass production, SMD connectors are cheaper, but for Macrofab, they have pin headers in their "house parts" so I don't pay for the soldering of all those through-holes when I stick with those ...

  9. #69

    Re: Malum IK

    Maybe it is just me, but I can not get to either of your (jwatte and tician) attachments...

    Gives me an error about invalid_attachment...

  10. #70
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,289
    Images
    27
    Rep Power
    276

    Re: Malum IK

    I was never able to access jwatte's and mine is no longer showing the thumbnail or even properly linking the image. Might be because of "invalid token"/timeout -> re-open in new tab -> restore comment -> post.
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    bleh

Thread Information

Users Browsing this Thread

There are currently 3 users browsing this thread. (0 members and 3 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •