Monday, June 18, 2012

UV exposure Unit

Up til now I have been making my own circuit boards at home with the toner transfer method, and while I have had some success it seems very hit and miss, and doesn't get amazing resolution. The toner is porous, doesn't always adhere well, can't be used for very thin traces, and is susceptible to melting and spreading out while being ironed on. I decided to upgrade my PCB making process to one using photo lithography. I bought some dry film photoresist sheets on eBay, but in order to have consistent results an ultraviolet exposure box is necessary. I designed, built and programmed this box, and am very happy with how it turned out, as well as the results I am now able to achieve with it. I used almost 500 UV wide angle LEDs emitting at 395 nm as the light source, and a handful of red LEDs mixed in to allow me to line up the masks and get things set up without prematurely exposing the resist. The power supply and timer/controller is in the box on the right hand side, as well as a lockout magnetic switch to only turn on the UV when it detects the box is closed. The indicator shows when the UV panels are powered on. The time may be easily entered using the keypad on top, and is displayed with four large seven-segment LED displays above the keypad. Buttons A through D allow commonly used times to be stored and recalled as presets using the microcontroller's EEPROM. If the lid is opened while the UV is on, the controller turns it off, turns on the red lights, and pauses the timer until the lid is closed again.
I've added the CAD files for my original design Here, and I have updated the schematic with the changes I mentioned in the comments, available Here. There's more than one new design for the display board, both use only 3 I/O pins and should drive the display with even brightness. The first one uses four 74hc595 shift registers, and four ULN2003 darlington arrays. If the high current 595 shift register variant is used, the darlington arrays can be omitted. The second version multiplexes the LED displays together, using only two 595 shift registers and one darlington array, and four BC558 PNP BJTs. This design saves quite a few parts, but may be slightly harder to write code for, and may be slightly dimmer due to the multiplexing. Either design should be better than what I have used in my old design, and the control board should also be compatible with the Sparkfun serial seven segment display, if you do not want to bother with building your own.

9 comments:

  1. Any Plans or schematics? Nice job really interested in this UV Box

    ReplyDelete
    Replies
    1. Thanks! I can post the Eagle Cad design files for the controller board, but there are a lot of things I would change about this device if I built another. Firstly, the LED display uses a BCD to 7 segment decoder to supply current to the displays and some NPN BJTs to multiplex the displays together. In theory this works well but in practice the LEDS do not receive enough current, so the number 1 is brightest and the number eight is dimmest, and neither are near the max brightness that the display can handle. The decoder chip has a maximum supply current that is too low. either a different decoder chip or a completely different control scheme is necessary for it to work properly.

      Delete
    2. Also, the button pad for entering the desired exposure time is less than perfect as well. I tried to save some I/O by using 4 ADC pins to read the button matrix, instead of the typical 8 digital I/O pins. Each row has a different size pull up resistor, and each column a pull down, such that each button pressed creates voltage divider and a different analog voltage on the column, which can be read with the ADC and the button pressed can be determined. In practice this has been pretty unreliable and sometimes will register a button press adjacent to the button which was actually pressed. If I did it again I would connect the rows to a voltage divider chain, and then place very weak pull downs on the column pins, that way nearly no current will flow through the button switch when it is depressed, and it should work much more reliably.

      Delete
    3. The light arrays themselves have been working great, and the device works well enough that I have not really bothered to fix the issues I just mentioned, but I wouldn't follow the design here exactly. If you used the four of the high current variant of the 595 shift register to drive four common anode LED displays, you could run the display with only 3 I/O pins and it would work at full brightness, without having to multiplex them. That would probably free up enough I/O to monitor the keypad matrix using digital pins, avoiding the problems I had here with the analog method I chose.

      Delete
    4. Cool. The Eagle files would be awesome. I have been searching the Internet for some different designs and I came to a conclusion that yours looks the best! Its clean and professional looking. Did you design the power supply for the unit or did you scrap one off an old printer or something else? How are the PCBs held up? Is there glass or something in there?

      Delete
    5. Thanks! I bought the PSU on ebay, a 12v 5A supply, I think it was only about $15. Any other supply less than 20v and with high enough current rating should work though. Just design your LED array with more LEDs in series and appropriate current limiting resistors. Yeah the bed of the unit is a piece of glass halfway in between the two LED panels, I use another smaller sheet of glass set on top of the PCB to hold the mask against it flat. I don't think I have a picture here with the glass, but you can see the lip it rests on inside the box in the first picture. I'll post the files later, I can update the schematic with the changes I described above also, if you like.

      Delete
    6. Sounds good. Anything will help. Thanks

      Delete
    7. Alright, I updated with my old design files, as well as some new ones incorporating the changes I described. Good luck with your project, let me know if you have any questions!

      Delete
  2. Hello the link you included for the design/files doesn't work . How can I get them, please?

    ReplyDelete