An Interem Solution to the Pneumatic Valves Problems

This week I’ve been occupied with (finally) putting away all the display mats I bought last year at the Anderson and Skyline matrix auctions. They have been cluttering up the shop since last summer with their variety of packaging. I finally took some time to make more dividers so I could file the mats in the storage boxes I use. Making the dividers involved some dreary repetitive work at the metal shear, then stacking the strips and cutting notches in them on the metal-cutting band saw. Most of these mats are now either in their boxes or in my own surplus section. Unfortunately when making the dividers I miscounted so I have to make a few more before I can finish, but at least now I have the procedure down pat.

The problem will then be reduced to finding a place to store all the boxes because the cabinet I was using does not have enough room. I might have to just make my own cabinet. Later on I can go through the boxes one at a time, cast everything and make proofs, and verify how my holdings match the font contents listed in the specimen books.

Also, preparatory to making some display matrix blanks, I’ve cleaned and lubricated my mill. I will also have to check some of its adjustments before I proceed, to ensure I get flat mats. Jim Rimmer used a Ludlow Supersurfacer to finish his mat blanks but that might have been a case of everything looking like a nail when all you have is a hammer.

Tonight I revisited the problem of my pneumatic solenoid valves. I need four valve units that work properly in Normally Closed (N/C) mode and give good airflow when energized. My valve collection turned out to have 12 Normally Open (N/O) valves of which 6 are electrically different (in what manner, I’m not sure yet) and 6 N/C valves. I determined that I could modify the internal springs to convert an N/O valve to an N/C one, and also that good or bad airflow seemed to be a property of the shutter in the valve. I was stalled on the problem of modifying the shutters to improve the air flow.

Tonight, though, I decided to avoid wasting time trying that. I only need 4 working valves. I have 6 valves that work in N/C mode so I don’t have to fool with springs; all I had to do was collect 32 shutters that gave good airflow and swap them around so they are all in 4 valve bodies. I tested all the shutters on the 6 N/O valves and noted which gave good airflow. I disassembled four of the N/C valves and the 6 donor valves and put the good shutters into the N/C valves. Swapping the shutters only took about half an hour (and would have taken less had I not jostled one of the disassembled valves and spilled all the shutters on the floor).

The result is four valves that give reasonable air flow in all eight ports. There are still noticeable differences in air flow so once I have a prototype assembled and connected to the caster I may find that some of the ports still need better air flow. There are still some donor valves available though, and I can raid those if necessary.

Tagged with:
3 comments on “An Interem Solution to the Pneumatic Valves Problems
  1. Christophe Slychan says:

    Nice to read that you’ve got a functional set now. I’m keeping an eye on your blog.

    As for electrical differences, it’s stated in the identification code in two places: at 6th position (right after the “BX758”), and at the end.

    The former can be:
    8 – 8 coils without diodes
    D – 8 coils with flyback diodes with common ground (you control the valves with positive voltage, relative to common). That means common is at GND and you apply +24VDC to inputs.
    G – 8 coils with flyback diodes with common VCC (you control the valves with negative voltage, relative to common). That means, common is at +24VDC and you short inputs to GND.

    These diodes are for shorting the voltage generated in the coil when you turn the valves off and the magnetic flux is decreasing (self-induction electromotive force).

    And the second parameter (at the end of the valve code):
    12 – 12V DC ± 10%
    24 – 24V DC ± 10%
    JJ, XX, KK – special models that require a controller board named PRB and UDB, supplied by Matrix.

    The coils are connected in pairs and (at least in some models) wired to two little round sockets with plug-in cables, and hidden under a black plastic cover. In both cable groups you’ve got 6 wires. Blacks are common (2 per socket, i.e. 4 per valve block).
    The color to output assignment is in the datasheet.

    Now, the question is if Bill Welliver’s interface can work with your valves…
    If you have BX758 D – then the interface MUST supply positive 24VDC (GND and outputs).
    For BX758 G – it must supply negative 24VDC (+24V and outputs)
    For BX758 8, it will work no matter if it supplies positive or negative voltage. The interface must have its own flyback diodes though. They can be discrete components (like 1N4148) or parts of an integrated circuit (like ULN2803).

    I could tell you some more if I had the interface’s schematic, or at least some good closeup photos of the PCB.

    I’ll also slowly develop a production version of my controller; it’ll fit onto a Raspberry Pi and feature a complete interface for connecting the valves and photocell to Raspberry’s GPIO. The prototype is already built, tested and working, but a bit large (the circuit is built on a perfboard three times as large as the Raspberry). My current device is in two enclosures, one contains the Raspberry, interface and a power supply, and the other contains valves & photocell (and is attached to the caster). They’re connected with a leftover 20-pair 37-pin D-SUB interface cable for some unknown device. But it’s too unwieldy for production-grade controllers.
    See the prototype (and the rest of our type foundry) at
    I plan to put the Raspberry and interface together with the valves, and the only external component would be a 24VDC power supply. The controller would then be connected to the paper tower, 24VDC and air supply, and LAN (via Ethernet cable) or WLAN (via USB WiFi interface).
    John Cornelisse says that he’d find quite a few interested parties around Europe.

  2. kpmartin says:

    The third set of valves seems to have some coding that does not match the data sheets I have (I forget exactly what did not match, but I think I mentioned it in the article where I discovered I had three types of valve).

    My intent is to make a unit that it ideally software-compatible with Bill’s interface, so you could plug mine into a Mac and use his host software. I actually don’t know how his interface appears to the computer but my first guess would be that it appears as a (USB) serial port. As such the same host software could talk transparently over Bluetooth SPP profile devices. Conversing over WiFi would require using something like the Telnet protocol.

    I will have a PCB mounted on the back of the valves and manifold, and it will be rather sparsely populated: the 4 driver chips for the valves, 31 resistors and 31 LED’s to show what’s being driven, a voltage regulator, and a processor chip that can drive 31 digital outputs, one digital input, and a USB port. I might consider Bluetooth and/or WiFi but the complexity of those make using a bare processor impractical and you’re back to needing a full OS on board. Maybe some sort of USB dongles exist to accomplish this…

    I had thought of using a Raspberry Pi or Arduino or something like that, but by the time I do the work required to break out all the digital I/O I might as well just use the right processor for the job directly on my board (assming I don’t want the WiFi and Bluetooth). I have access to equipment for soldering SMD devices (and friends with the expertise to keep me on the right path) so it should all go together pretty slick.

    For now I have a breadboard with the driver chips and shift registers stolen from the hardware that came with the valves, and I’ll be driving this with my PC parallel port on an old Windows 2000 laptop. You don’t need new hardware for proof-of-concept!

  3. Christophe Slychan says:

    Thanks for the info,
    I personally use two MCP23017 expander chips (I2C to 16 I/O ports) with four ULN2803 chips (8-channel Darlington driver). As for the photocell input, I just use one of Raspberry’s GPIO pins.

    I also don’t know much about the IOWarrior module Bill’s interface is based on… might be worth looking into.

    And as for the software, I’m writing a pair of programs: one for typesetting and the other for casting. The casting part is on the Raspberry itself. It has a console user interface and I connect to the RPi via SSH over the network, then run the program. In the future, I’ll write a web UI to make it more user-friendly.

Leave a Reply

Your email address will not be published. Required fields are marked *