Soldering 16-Channel OpenBCI Kit

I just started working with the new 16-channel OpenBCI Kit.  Before you can use the kit, you need to solder the pin headers so that you can connect the 8-channel daisy module to the 8-channel base module.  If you'd like to see my tricks for doing this, feel free to check out the link below.


What tricks have you used to make it easy to solder the headers on?

Chip
«1

Comments

  • Cool just what I needed!
  • edited February 2015
    I feel a bit stupid; I've soldered the daisy module and now a pin (the AGND) is getting in the way of the battery pack's cable lead and it's very difficult to plug / unplug it from the board. Even if my soldering is far from perfect, I don't see how it could *not* prevent a smooth connection.

    Also, at my first attempt the daisy module was not detected by the board. So I got back to my soldering iron, improved a bit my work, and finally I was able to get some signals, testing quickly ECG. However now I'm suspicious regarding my soldering. Is there a way to check that every pin is functioning properly? Apart of course from trying each possible configuration -- which could be a bit tedious with bias / srb1 / srb2 / N pins / P pins

    NB: I hardly touched a soldering iron since high school, this DIY 16-channels version is quite stressful:D
  • wjcroftwjcroft Mount Shasta, CA
    edited February 2015
    Jeremy, had you seen Chip's post on this (above)? I'm sure you've seen the docs section.

    With the SRB* bussed references, you would only need to test SRB2 against each IN*N pin. And then SRB1 against each IN*P pin. (Assuming you use OPENBCI_GUI or other means to change the reference.)

    Usually with soldering pins like this, 'less' is 'more'; very little solder is needed. If you have a situation with too much solder, there is a solder sucking tool likely in your lab to remove some of the excess.

    Another angle is, you could get one of the JST-PH extender cables, thus not need to unplug from the board very often at all.

    http://www.adafruit.com/products/1131


  • Yes, I did read the docs and Chp's post (several times!), but when you're as afraid to break your beloved board as I am, you'd almost need a video :D

    Thanks for the link, I planned to use an extender cable anyway -- maybe with another type of connector as a "jack" looks more solid and practical to me -- it's just that it seemed odd to have such kind of difficulty. It's more the pin going through the board than the soldering that gets in the way.

    I'm afraid you'll learn that I'm better with software than with hardware but... are the SRB* of the main board supposed to be internally linked with the SRB* of the daisy module, or do we need to create some kind of jumper in order to use only one reference for the 16 channels? (With our g.tec equipment this is the case, two extra cables go from ground and reference from one amplifier to the other). Because as for now, a montage using an electrode on the daisy module and a reference on the main produces nothing but noise. :\
  • Hey jfrey,

    Regarding...

    >>>
    I feel a bit stupid; I've soldered the daisy module and now a pin (the AGND) is getting in the way of the battery pack's cable lead and it's very difficult to plug / unplug it from the board. Even if my soldering is far from perfect, I don't see how it could *not* prevent a smooth connection.
    >>>

    I kinda saw this interference as well.  After soldering, I used my wire cutters and trimmed back the offending wire tip / solder to be as flush with the OpenBCI board as possible.  My wire cutters are not very good, so I had to really cut,scrape,cut,scrape,cut repeatedly to get it to be as flush as possible.

    When i was done, the situation was better, but not perfect.  It's at that point that I learned that the OpenBCI board has a power switch on it.  It's a little slide switch on the edge of the board.  "Off" is the center position while "On" is on either end (it doesn't matter which end...they're both "On").  Once I discovered the power switch, I've never felt the need to unplug the battery pack.  For me, that solved the problem of difficulty connecting/disconnecting the battery pack.

    Regarding your question about the SRB pins, my understanding is that the pins are linked between the main board and the Daisy board.  Therefore, if you want a single reference electrode for your whole EEG montage (like most people), you only need to plug it into one of the two boards.  The same is true for the bias connection, you only need to connect your bias electrode to one of the two boards.

    Finally, just to clarify your statement " a montage using an electrode on the daisy module and a reference on the main produces nothing but noise", are you plugging electrodes into any of the channels (IN1N-IN16N) in addition to the reference SRB2 reference electrodes?   Unless you change the configuration of the hardware through the GUI, the default setup is looking only at the INxN inputs vs SRB2.  All the other pins are ignored.

    Chip
  • I'm glad I'm not the only one with this plug problem :D You're right, with the switch it's not that often that we need to disconnect the battery pack, even though it's more practical when storing or travelling (I've already took it for a 14,000 km trip!).

    Indeed I tested my electrodes with one into the SRB2 of the main board and one into a N pin. It's working for example between SRB2 of main board and 1N, and between SRB2 of the *daisy* board and 9N, but not with SB2 main + 9N or SRB2 daisy + 1N. Any idea of the faulty soldering?

    The BIAS "link" *seems* to work (a bit harder to tell when it's improving signals when I barely hold electrodes in my hands :D). I do see a "BIAS" label for a shared pin between main board and daisy (next to the battery-pack-cumbersome AGND), but as for a "SRB"...?? 
  • wjcroftwjcroft Mount Shasta, CA
    Jeremy,

    I'm guessing that Joel @biomurph will comment here about the linking of the SRB's between boards.

    re: the AGND pin length. Another approach could be: use your solder sucker tool to remove most of the solder from the header socket in that location; then with your diagonal cutters, cut off flush with board. Then apply a bit more solder. OR: remove socket with sucker ENTIRELY from board. Trim pin length then with full access to pin. Then resolder. It may be a good idea to update our assembly tutorial to include trimming the pin before soldering.

    re: JST-PH extender plugs / socket. These connectors are very reliable once clicked into place. Quite sturdy and capable of thousands of operations. Wide acceptance in industry. http://www.jst.com/

    William
  • Sorry, I will not try to cut part of my board -- even if it's just pins -- right now. Plus there's some chance I'd cut myself. Especially because there's some chance I'd cut myself :D

    I did not find a soldering more ugly than the others on by board and I did not see a clear duplication of the SRB pin in the schematics -- but I'm not used to read them -- so while waiting for an educated response, and because I don't want to make things worse by randomly re-soldering pins, I created a "Y" cable (on male pin to plug the reference, two females pins to plug to both SRB2). It seems to work, at least with ECG signals. By tomorrow I will try EEG signals, I'll keep you updated if the workaround holds.
  • biomurphbiomurph Brooklyn, NY
    edited February 2015
    @jfrey The SRB pins are not connected automatically. You could solder pass-through wires, or use a Y connector, like you describe.

    When the SRB2 pin is internally connected to a P input, that channel's P pin is still connected through the muxer as well.
    When the SRB1 pin is internally connected to an N input, that channel's N pin is disconnected from the amp by the muxer.

    The BIAS_INV connection between the main board and daisy module makes connection between every channel that has 'include in bias' selected from the GUI. That pin is not the electrode BIAS output! The BIAS output is available on both pins labeled BIAS.

    And yeah, the damn GND connection pokes through the bottom of the board and interferes with battery plugging... sorry about that!
    I use a pair of flush-cutters to trim the solder and pin down to board level. Here's an example of a flush-cutter

  • Perfect, exactly the details I needed, thanks! I'm glad it's working the intended way :))

    As for the GND pin... well, we can take that, and it'll make an excuse to do an OpenBCI v4 ;)
  • wjcroftwjcroft Mount Shasta, CA
    Could we mod the docs page to suggest that the AGND pin be trimmed BEFORE being soldered? So that it is flush with board bottom surface. That can be done with a regular diagonal cutter, which is in everyone's tool kit.  :-)

  • biomurphbiomurph Brooklyn, NY
    great idea william!
  • Great idea which never made it into the docs :( Pitty
  • edited March 2016
    So, are the following two actions the "unofficial" recommendations of this thread?

    1) purchase the JST extension cord so you don't have to unplug from the main board socket -- and for those who install the main board in the Cortex board holder, have the cord coming out of the holder opening below the board perhaps with the battery pack hanging off the cord (maybe zip-tied to the headset)?

    2) the connection from the ear clip electrodes (one for BIAS and one for SRB) *both* need to be connected to those respective inputs on *both* boards?? Because above it mentions there is no direct shared SRB pin and the shared BIAS_INV pin is not the same thing as BIAS. So if I want the ear clip electrodes to be the SRB and BIAS on both boards, it would seem they have to be connected to both boards?
  • wjcroftwjcroft Mount Shasta, CA
    @leka0024, hi.

    (1) yeah, I mentioned that tip on another thread, it also applies if you use a lithium cell behind the mainboard, and want to recharge it without disassembling. In that case two (shortened) JST extensions allow the mainboard to be unplugged from the lithium; then the lithium connected to the charger.

    (2) only the SRB2 pins need the Y cable. The Bias pins do not need a Y, since the 2 Bias pins already have the same signal.

    William

  • Hello @wjcroft , thanks for the reply!

    1 - good idea! That's a better solution than having it hang out on the cord.

    2 - Why are the two BIAS the same? I confirmed with a meter that the BIAS pins and the BIAS_INV thru-hole are not the same node. So wouldn't you want the signal/voltage from (one of) the ear clip electrodes to be attached to the BIAS pins on both boards??

    (a short explanation or a link to another thread is fine!)

    Thanks
  • @wjcroft

    I'm going to assume the comment

    "The BIAS_INV connection between the main board and daisy module makes
    connection between every channel that has 'include in bias' selected
    from the GUI. That pin is not the electrode BIAS output! The BIAS output
    is available on both pins labeled BIAS."

    means that that somehow the micro and/or A2D chip are in fact connecting the two BIAS nodes on the two boards together, under the proper circumstances.
  • wjcroftwjcroft Mount Shasta, CA
    The ADS1299 chips share the BIAS_INV amongst themselves as a consensus determination of what Bias signal to produce. BIAS_INV is a 'bus' shared between the ADS1299 chips. It is not the Bias output. Each ADS1299 can independently generate the proper Bias on it's own. In a daisy configuration like this, only one of the Bias outputs needs to be used.

    However if you look at the mainboard and daisy schematics, both ADS1299 have their Bias outputs wired to the respective headers. So only one of these 4 Bias pins needs to be used. See the TI ADS1299 pdf spec sheet for more info on the Bias drive circuitry.

    William

  • edited March 2016
    Thank you @wjcroft!

    So for the case of using two ADS1299's in daisy config, in order to detect 16 channels of EEG data ... why isn't the situation the same for the SRB2 pins on the two boards??

    And since it is the case that a Y adapter is needed ... why wouldn't they have made another thru-hole in the main board so that the node could be shared directly with the daughter board??
  • wjcroftwjcroft Mount Shasta, CA
    I think Joel was looking for maximum flexibility for way the input channels are configured. Yes, it would have been possible to avoid the Y cable if the SRBs were bridged automatically. If you look at the TI data sheet there are a huge range of possibilities for configuring the SRB busses. Some labs may want to do unusual configurations.

  • biomurphbiomurph Brooklyn, NY
    That's exactly right, @wjcroft
    @leka0024, if you are going to use the tech for one setup perpetually, then I would suggest doing the little bit of soldering to bridge any connections that you want to keep.
  • edited March 2016
    @biomurph, understood. We purchased the 16ch kit so we could detect 16 channels of EEG signals on a subject, and we're not going to make the poor guy wear THREE earclips!

    So I just clipped one of the wires that had the pin socket already crimped (from the kit) and soldered the other end from the main board SRB2 pin, where it sticks out the bottom side.


    *In my opinion* - it would still be better to have the SRB(2) pin shared directly in the board stack, perhaps running to a switch where you could enable or disable that link, as needed.
  • wjcroftwjcroft Mount Shasta, CA
    re: three earclips

    In QEEG and neurofeedback, a common montage is "linked ears", where the left and right earclips are connected via a Y-cable, then plugged into the single reference socket. In this montage, the usual place for Ground (Bias) is somewhere on the midline, For example AFz on the Electro-Cap. The midline location can possibly offer better noise cancellation (inverse mains injection) to reach both sides of the head.

    In the case of the daisy, it would be a DOUBLE Y, with the two references first connected via another Y cable, then that goes to the 2nd Y out to the ears. Or obviously the cable could be more like a bus.

    William

  • edited March 2016
    @wjcroft

    Interesting idea! Thanks for sharing that. I just assumed that using A1/A2 (either or) as reference and bias was the standard way. But I like the idea of using a midline electrode as the bias, to be physically closer to the 15 electrode signals we're recording & analyzing on the Mark3 headset (F, C/T, P - 7,3,Z,4,8). I guess then, we'd probably want to use FpZ or OZ as the bias? Or perhaps into a Y, again?

    Do you suspect increased performance for using both A1 & A2 into a Y, for the reference? Or could we get away with just one earclip on either location?
  • wjcroftwjcroft Mount Shasta, CA
    I have never seen the Bias / Ground applied to both sides with a Y. Midline is pretty standard. As I mentioned the Electro-caps put this at AFz, (on the 10-10 system) midway between Fpz and Fz. But any midline position can work.

    re: increased performance [for linked ears]. It gives a bit more centralized (non-cortex) location for the reference. It does effect the signal amplitude slightly. This is why the QEEG people use it. But also many neurofeedback protocols with multi-channels recommend linked ears, especially if they have channels on both left and right sides of the head.

    But for simple applications, there is no problem with just doing one ear as reference and the other as bias.

  • @wjcroft

    Thanks again for all the info in this thread!

    The Mark3 headset only has midline locations at FpZ and OZ, beyond the FZ,CZ,PZ locations that we're using for signals. So it would have to be one of those. Hence I was just wondering if tying those two together and then feeding that to the board for the bias, instead of just one midline location, might boost performance in the same sense that tying the ears together and then feeding that to the board for reference, instead of just one ear, apparently does.

    If you don't know or if it hasn't really been done, fair enough.
  • wjcroftwjcroft Mount Shasta, CA
    For the reference leads it's not so much a 'boost' as a more balanced perspective regarding left right channels.

    For the Bias wiring you suggest, yes that should work, though I'm unclear it would be a measurable advantage. I've not seen any other montages using two ground connections on the head. But should be fine.

    With the Mark IV, other midlines should be available, e.g. AFz, FCz, etc.

  • hey guys, 

    I just got down to soldering the pins for the Daisy Module and I have tried it out using OpenBCI Python. It didn't get detected so I suspected bad soldering work. However, I do have one question before I get crazy before trying to re-solder. I am aware that the Daisy Module had undergone some re-works and I have read the page about it. I am not sure if my hardware model is the re-worked version or not. Perhaps someone could take a look?


    I just need some clarifications before I move on, because I am not sure if this version of the module has been reworked yet or not. 

    Regards,
    Samuel
  • wjcroftwjcroft Mount Shasta, CA
    @samuelpo94, hi.

    I merged your post into this existing thread on Daisy soldering. Take a look at some of the earlier posts. Joel @biomurph may comment here, but I believe the "reworked" Daisy modules were in the December 2014 timeframe. And they were all tested before being shipped. Later Daisy's I don't believe need any rework. Your board does not have the rework, hence was not one of the early Daisy's.



    Some of your solder blobs look a bit unusual, such as the AVDD. If you follow Chip's tutorial (first post above), you may be able to correct that. Do you have a solder-sucker? Using a soldering iron with a sucker, you can undo and redo pins that may not be making proper contact.

    William

Sign In or Register to comment.