Hacking the HP45

[dragonator]

It may very well be possible to limit the width of the board, but to the width of the HP45 itself might be very difficult. Currently I am at a 40x75mm controller. The width of the HP45 is 19mm. The 26 pin header is already around 36mm wide, so I will have to put that sideways on the back if I want to even get close. Less pins is not an option. I think I can realistically reduce the width to 30mm, maybe 25mm. I will give it my best try, less width is added to the wish list.

As an interesting side note. If full color is what you want, you can also install an HP23. It is on my list when Oasis is done. It is basically a wider version of the HP45, with 3 colors. It only has around 100 nozzles for each color, so the swatch size is smaller, but it has 3 colors.

[Wonko]

If you want to get the pin count down, you could use a CPLD chip like the Altera 3000 to create an SPI interface to control all the pins. SPI can be connected to many CPUs via hardware SPI interfaces or with the Arduino Due even via DMA, reaching 1Mbits per seconds easily, and 8Mbits with some precautions.

Those CPLDs are small FPGAs for under 2 Euros in single quantities. A programmer clone cost under 20, and the Altera software is free.

[dragonator]

I like the idea for future versions, but we are quickly reaching outside of my skill set. While I still want to learn more about FPGA's and CPLD's in the future, for this project I want to keep electronics on the edge of my current skills.

Current pin count to the head controller is 23, excluding power lines. I want to run the ground and primitive power through a separate cable because flat cable is rated at 1A, and this circuit peaks at 5+ amps. 14 are for primitives, 2 for primitive drivers, 2 for nozzle condition circuitry, 2 for address 4017 or shift register (I can probably run the addresses without the TLC drivers), 2 for the 10x and TSR resistor and 1 for future endstop.

There are still options, but thus far I see too much complexity (for me) in moving the primitives to serial. The way I want to handle my data buffer requires me to have the primitives parallel. I plan on using a 4Mbit parallel ram chip to buffer my full sweep. I really do not want to mess with 1Mbit data to the printhead and reading said 1Mbit from the SD card at the same time.

[Philipp]

Hello,
I'm not quite sure I understand your plugs and cables correctly. In my view, it is extremely useful for the cable guide that the plug for the ribbon cable is vertical, otherwise it will be difficult with the movement.

For the print heads are two possibilities, either colored, or to widen swatch. Both are interesting. I know the color print head, it is even narrower swatches, which is bad for the printing performance and the nozzle wear.

It's all very complex.
Best wishes
Philipp

[dragonator]

The challenge has more to do with the PCB layout than the flatcable orientation. A flatcable can be bent in such a way that it is in the direction of movement. I am still trying it, but I give no promises. A lot of stuff needs to be on the controller.

[dragonator]

While optimizing the amount of chips required, I ran into a minor wall. I am somewhat at a loss right now. Here is a quick recap.

First I looked at the MIC5891 Wonko suggested.

http://ww1.microchip.com/downloads/en/D ... ic5891.pdf

I now understand it, but it has 2 issues. 1: it cannot handle 3.3V logic. Faster microcontrollers are 3.3V so this is a downside, but not a dealbreaker. I can use level shifters. 2: it is slow. Datasheet suggests a turn-on delay of 2us (manageable) and a turn-off delay of 10us (slightly less so). This is rather unfortunate, because I like the chip. Is this only the output enable delay or is this also for shifting between pins? If it is faster at shifting I will simply not touch the output enable and I will order the chips this weekend.

Then I looked at the 6810, but it is no longer in production and my sources do not have them in stock.

It left me thinking. Addresses do not need much in term of current. I have used HEF4017 and HCF4081 chips before. These things are rated at 15V max, so I can run them straight to 12V. The only thing I now need to do is shift the 5V (and future 3.3V) to 12V before I enter the IC's. How hard can it be?

Apparently quite hard because everything I have made so far does not work. The 4017's have magically stopped doing anything, not to mention the 5V to 12V level shifters I made (1 npn and 1 pnp transistor) have a huge amount of turn-off time (over 60us).

I can probably get it to work if I give it a few more hours, but this has been quite a setback. If anyone has a better idea than 4017's ran at 12V for address control, I am all ears. Ideally a solution needs to be controllable by 2 to 3 3.3V logic pins, provide 12V and take less then 2us to shift from address to address. Also feedback on the MIC5891 question above is appreciated, because I am still very interested.

[Philipp]

Hello,
what's with the A6812 20bit, or the A6818 32bit. It works with 3.3V logic voltage.
I have found no information about the production status.
Here is a distributor:
http://www.ic-chips.com/de/product-deta ... 24757.html

What is the end stop used for?

[dragonator]

The A6812 is a beautiful chip but it is discontinued and hard to get. It is also fairly slow. I did some looking around, and found several replacements for it:

http://www.mouser.com/catalog/645/usd/377.pdf

The most interesting chip I think is the HV9308. It is sadly 5V logic and the latch is required to operate it (meaning 3 pin operation) but other than that it is perfect. 12V output, 32 outputs, and 500ns switching time. It is harder to come by, but not out of production:

http://ww1.microchip.com/downloads/en/D ... 072213.pdf

Slower but more available is the HV518. It also has 32 12V outputs, but takes in the 1-2us range to switch. Slow, but fast enough.

http://ww1.microchip.com/downloads/en/D ... 072413.pdf

I will look further into this topic. I might order a few different chips for testing to see if they can handle the speed.

@phillip: The endstop is for use in 3D printers. I plan on using this controller mainly for 3D printers. I had pins left and with cabling running to the head already, it is simple to add an endstop. For the printer I plan to build with it there is also a motor on the head gantry, so I can drop the endstop pin if I have to.

[Philipp]

Hello,
I would only point out the problems that cable should be bent as seldom as possible. I would try the end switch not to be mounted on moving parts. However, it may be useful to define the starting point of swatches by a backdrop and a switch.
Regards
Philipp

[dragonator]

I agree that cables should be bent as little as possible, but it is not something that should not happen at all. How much it bends depends on how the circuit fits, not the other way around. I did try it vertically and it does work quite well, so the cable is for now vertical.

The endstop will always be mounted on a moving part with this type of gantry. There is no real other way. The head on this printer will be driven (for now) by a stepper motor, so I need no indexing for starting point. Also any starting point would need a 0.1mm accuracy.

I designed a new circuit. This one has 2 TLC59213A's, 3 MIC5891's, 1 LM311 comparator and a SIRA12DP-T1-GE3 N channel mosfet. The current board is 25mm by 75mm. It is about as narrow as I can make it.

The left side is the address side and sits closest to the HP45. It has the 3 MIC chips on one side of the board, and nothing else. I might add a few level shifters from 3.3 to 5V on this side as well, to control the MIC chips. Only 5 pins connect the bottom to the top board.

The right side is the top board, furthest from the printhead. Mounted on the right is the 26 pin header and to the top the high current primitive driving side. The comparator and mosfet are mounted on either side of the board in the middle. The TLC's are to the bottom. A few resistors and capacitors are also mounted here. The top side has surface mount components on both sides to save the space.

I will order the parts for this board (plus a few parts for some future experiments) this week. If the circuit works on a breadboard, I will finish and order the PCB's.

HP45 breakout board Oasis test 2.png (158.78 KiB) Viewed 15898 times