πŸ’¬ Adjustable Boost Converter with Pass-Through




  • Hardware Contributor

    @NeverDie
    cool ;) i know better but you wouldn't like to solder it i think :)
    about routing, i was referring about notes at : section 5.7 p.16-17


  • Hero Member

    @scalz said:

    @NeverDie
    about routing, i was referring about notes at : section 5.7 p.16-17

    That's interesting. I haven't done a pattern shaped like Figure 5-1 before, but I suppose it's possible to replicate it almost exactly. I guess then, by definition, it would meet their specs.


  • Hardware Contributor

    yep that's what i wanted to explain, not the sch but the routing ;)
    this is a very important section in datasheet. all the datasheet sure, but a bad routing a crppy thing :)

    cool if it can help you, i'm glad


  • Hero Member

    Anyone know if there's a way to extract exact dimensioning information from a PDF file? Seems like there should be. The goal would be to find the exact layout, with dimensions, of Figure 5-1 on page 17 of the datasheet: http://ww1.microchip.com/downloads/en/DeviceDoc/22234B.pdf Since we know from the datasheet the exact width of the chip legs, then I could use the known chip leg width to convert the exact dimensioning information from the PDF into exact absolute dimensions, from which I could create a precise PCB pattern.


  • Hardware Contributor

    following the guidelines should be enough. this is not an rf circuit :) the general guidelines and, when you're not sure there is always this section for each ic. so you can inspire from this. thx to manufacturers ;)
    you can have lot of infos by reading application notes, good practices notes etc...they have lot of resources. very interesting!


  • Hero Member

    If you ask me, chip manufacturers should publish the PCB (gerber files, etc.) for their "typical" application circuit. It would save me from having to re-invent it just so I can take their chip for a test drive. As it stands, you have to buy their costly "evaluation boards"--or else go to the effort of creating your own--to help decide if you might want to possibly use their chip. It should be in their interest to make it as easy as possible for people to test drive their chips. What's the point in getting free sample chips if you have to create a PCB from scratch and wait two weeks for fabrication just to try it out? It makes no sense. Am I missing something?


  • Hardware Contributor

    @NeverDie
    you know, I was kidding a bit ;) you're right, some manufacturers are charging too much!

    I don't want debate (wasting time) but for example, look at the competitive price of nucleo boards compared to arduino boards...You can often find gerbers for RF or lot of other stuff. See (TI, semtech, etc...). so that depends, i can't generalize this.
    Maybe they think that companies and their lab has money to pay for it, and knowledge. And as you have to be a company in electronic field and have a project, to get sample (the theory!), and enough skilled to route it etc. , so they put some recommandation. pure speculation! I think they may be more interested in big batch projects of course, then you can get a cheap sample board!

    That said, here is no need of gerbers for a MCP1640! That's a 5min routing :)

    You should put more effort in reading the datasheet and docs than overhead on finding the exact routing dimension (it's useless for this case) ;)
    With xp, you also may discover that often we need to read more (all the datasheet) than just page1 with typical app circuit which often is not so complete (missing details) but here don't worry for your case this should be ok.

    Enjoy designing :)


  • Hero Member

    The problem for me is that this is suddenly a lot more work than I had planned on. I have about 8 different booster chips I was planning to brute force try to see which worked best in the application. If I have to create stringent patterns for each just in order to try them, it's going to exceed the time I had budgeted for this. So, I'm going to put this on hold.

    I've already ordered the boards that I had posted. I'll try them out after receiving them. Maybe some of them will work well enough to proceed from there, or at least narrow down the list of candidates for which I create more detailed patterns. Anyhow, thanks for the input. At least if they aren't working as expected I'll have some idea as to why. One of the chips has built in MPPC, and so I might just focus on that one, as it really is intended for a solar application.


  • Hardware Contributor

    oki ;)
    i also have designed multiple eval boards..too much stuff to release, but sure i could release some. that shoule be quick job. i have boosters, buck, boost/buck, solar etc.. :smile:

    Here is a pic of my eval board for MCP1640, i did it a while ago when i did my ulpnode..
    0_1479125228025_mcp1640.png


  • Hero Member

    @scalz
    That's good fortune. If you already have an eval board designed for the MCP1640, please do post it. No sense in spinning my wheels to create another one if it's already done.


  • Hardware Contributor

    oki ;)
    hopefully this is quick stuff, because i have an avalanche of stuff to release! and still some non complete release to clean..that's always the same i'm lazy boy, i prefer to play :)


  • Hero Member

    Interestingly, the datasheet for the LTC3105 boost converter with the built-in MPPC does not make any reference to a recommended trace or component layout: http://cds.linear.com/docs/en/datasheet/3105fb.pdf That's in contrast to most of the other boost converters, which do. Since it does not appear to be burdened by layout restrictions, I'll throw together a board for that and post it. Also, if I'm reading it right, it does not appear to require the use of a blocking diode to prevent the current from discharging back through the PV cell under low light conditions, which if true, is also nice.

    That said, I'll change the layout strategy to at least put the capacitors at a shorter distance to the chip. On the boards I had already posted, I had put them on the back so as to keep the board size as small as possible (and thereby fab costs low). So, the board may need to grow a little bigger to keep the caps on the same layer as the chip, but I guess that's probably par for the course.


  • Hero Member

    @NeverDie said:

    If you ask me, chip manufacturers should publish the PCB (gerber files, etc.) for their "typical" application circuit. It would save me from having to re-invent it just so I can take their chip for a test drive. As it stands, you have to buy their costly "evaluation boards"--or else go to the effort of creating your own--to help decide if you might want to possibly use their chip. It should be in their interest to make it as easy as possible for people to test drive their chips. What's the point in getting free sample chips if you have to create a PCB from scratch and wait two weeks for fabrication just to try it out? It makes no sense. Am I missing something?

    After reflecting on this question, I'm guessing that most EE's probably use a spice simulator, or similar CAD tool, for designing and testing their circuits. So, for them, having the physical chip itself isn't so important. Maybe I should look into doing the same.


  • Hero Member

    To expedite things, I just now posted Version 2:

    1. I re-did the routing placement to make it conform with the recommendations, and
    2. I cleaned up the silkscreen to make the component labels more legible.

  • Hero Member

    I just now posted Version 3.

    Changes from Version 2:

    1. Change to using the same type of shielded inductor that my other boost projects are using.
    2. Eliminated unnecessary header pins, and added a ground header pin.
    3. Changed the silkscreen on the MCP1640 land pattern to mitigate against potential for solder bridging.

  • Hero Member

    I've tested the board, and it works.

    Please purchase your PCB's for this project from the following link: https://pcbs.io/share/zyLW7


  • Hero Member

    Here is a photo of the assembled PCB, taken next to a quarter for scale:
    0_1482460523384_mcp1640.jpg


  • Hero Member

    This project is now finished. :smile:


  • Hero Member

    Setting R1=680K gives a voltage output of 2.674v, which is great for charging 2.7v supercaps.



Looks like your connection to MySensors Forum was lost, please wait while we try to reconnect.