This Blog will document projects I have in progress or completed. You can search for my handle on Flickr, Facebook, YouTube, or Picasa for photos. I will link some of those here.
This Blog will document projects I have in progress or completed. You can search for my handle on Flickr, Facebook, YouTube, or Picasa for photos. I will link some of those here.
I am a big fan of Roy Underhill’s “The Woodwright’s Shop”. Last fall was the 32nd season, and he’s still wearing the same hat. The second episode was titled “The Eleven Grooved Box”, a project he uses in his woodworking school. You can see it here. I was attracted to this project because he uses Stanley 45 combination planes to make all eleven grooves. I have a Stanley 45 and have been looking for an appropriate project so I am trying to duplicate what Roy does.
I won’t go over all the steps to make the box here, you can watch the half hour video for that, and you should if you want to understand the rest of this page. But I will pass along some things I learned, and show how I made those #$%@! spline grooves. Each corner of the box has to have two matching grooves plowed for splines, without these the box would be very weak. You can see these in the corners of the lid in the photo. Cutting those spline grooves with an old Stanley 45 might be easy for Roy but for everyone else it’s a pain. A millisecond of inattention and the sides of the groove are ripped up. So like any self respecting woodworker, I made a jig.
My grooving jig for the Eleven Grooved Box
After almost giving up on this project, I sat down and analyzed what is happening. When plowing the grooves, you have to hold the plane perfectly perpendicular to the 45 degree mitered surface. The fence on the Stanley plane rides on the reference surface. But the skate is captured in the plowed groove! If you let the plane roll to the right, the fence lifts off the reference surface a bit and not much happens. But if you let the plane roll to the LEFT, the fence digs into the reference surface and pulls the blade to the left. The result is a horribly shredded edge on the left side of the groove. In the video, Roy has an Iron Arm and holds that Stanley perfectly aligned through the whole operation. My arm is made of rubber so I had to make a jig to get the plane to behave.
V1.0 – My first jig attempt was a piece of 2×6 cut off at one end at 45 degrees, with a stop block attached. The stop block helps control tear out at the end of the cut and makes it easier to initially align the work piece with the plane. It did not help with the left side shredding problem, in fact made it worse.
V2.0 – Added a second 45 degree cut at the end of the 2×6, creating a 90 degree angle at the tip. That spaced the work piece farther away from the fence. I reasoned that the longer roll radius would pull the blade over less if I let the plane drift off axis. It helped a little but still not satisfactory.
V3.0 The third revision adds a reference surface for the BOTTOM of the plane fence.
The fence is now constrained by hard surfaces in the down and right directions, and in the up and left directions by my left hand. It can only move back and forth like it’s supposed to. The whole thing gets clamped in a bench vise for the plowing operation, just like in the video.
The bottom reference surface is constructed at the rear of the jig by screwing on a 3/4 piece of scrap cut off at 45 degrees at the top. That 45 degree surface is 90 degrees from the fence reference surface of the jig. The two screws are in elongated slots so the added piece can be adjusted up or down, which in turn moves the small block of hard wood up and down the fence reference surface. I made a one time tweak so the plane is level and aligned with the work piece bevel at the beginning of the cut and tightened the screws.
There is a small block of hard wood on which the plane fence actually rides, that sits loosely on top of the added piece.
The hardwood block has screws inserted in each end so it won’t slide off the jig when the plane is working.
I start the cut with one light pass at the end so I can see where the plane is plowing, then knife down the grain like Roy does. With that, and the v3.0 jig I’m getting perfect spline grooves in the mitered surface.
Stanley 45 tips for the Eleven Grooved Box
1. The eighth inch wide inch cutter needs to be as sharp as possible. I use one of the cheap “Eclipse Style” honing guides. I had to file the rounded jaw slightly to get it to grip the small cutter firmly. Use a simple wooden stop gauge to set the cutter for a 35 degree sharpen angle. Problem is, the narrow cutter can’t keep the gauge from wobbling during the honing process. The local hardware store had nylon bushings exactly the same diameter as the guide roller, with an ID the same as the guide screw shaft. I pulled the knob off the screw shaft and hacksawed a screwdriver slot across the end. With the knob removed, I can put a nylon bushing on each side of the guide and they act as outriggers to keep the whole thing true to the stone.
To be really sharp, you have to flatten and polish the back side of the cutter as well. This is complicated by Stanley having made the 45 cutters slightly curved. You can use the Charlesworth ruler trick but you will need a thicker than usual ruler because of the curve. I found it good enough to just free hand polish the back by putting a lot of finger pressure on the tip.
2. Use a good ruler and measure the distance between the fence and the skate at front and back. Mine is typically wider at the rear, which causes the skate to bind in the groove. Loosen the fence rear lock screw and push it around until the measurements are the same.
3. Wax (Paraffin from a candle) the face of the fence, the bottom of the fence, and the bottom of the skate.
4. Don’t overtighten the cutter lock bolt. It doesn’t take much to hold the eighth inch cutter in place.
5. Use an eighth inch drill bit to set the depth stop. When the groove is finished, lay the drill bit in the trench and if it sticks up above the beveled surface, back off the depth stop and cut a little more.
Glue up tips for the Eleven Grooved Box
1. Glueups have to be rehearsed. Make sure you can get the box assembled before the glue starts to grab.
2. I’m using Titebond III which has a little bit longer open time than Titebond II. I don’t have a Roy Underhill style glue pot.
3. Use an acid brush with the bristles cut off to about 3/8 inch to apply glue. Avoid applying a lot to the inside edge as squeeze out is difficult to remove there.
5. Apply glue to all the miters and grooves then wait a minute for that to soak into the end grain. Then apply another coat of glue and insert the splines. An easy way to apply glue to the splines is to lay a sheet of foil or waxed paper on the bench, make a puddle of glue, and roll the spline around in that.
5. Don’t forget to insert the top and bottom panels. DAMHIKT.
I have a drill battery that for some reason will not work in the factory charger. (possibly because I rebuilt it myself) So I need a way to kludge on a laptop power supply to do a trickle charge. Not having a socket for the business end of the battery, I have to attach wires somehow. I tried a standard clothes pin, will stretch to the one inch spacing if you bend the spring. It is not satisfactory. So I made a wider clothes pin by gluing two together in series. The glued blades are sawn off and the tips shaped a bit. Works great.
I can make a triple too. A standard clothes pin will open less than a half inch. A double will open an inch, a triple an inch and a half. Those are the spade tips I used to make the actual contact with the battery terminals.
Adafruit’s LiquidCrystal backpack library doesn’t use the Arduino SPI library to send data to the 74HC595 chip. They just shiftOut bits directly, which works fine if the only device is the backpack. However, the fazjaxton CAN library does call the SPI library. This caused major problems when switching back and forth between CAN and LCD because the clock rates were way different.
I decided to investigate the LCD 74HC595 SPI implementation by Juan Hernandez from the Arduino playground which uses the Arduino SPI. It took quite a bit of time to get that library working because Hernandez uses a different pinout connection from the ’595 chip, and did not have the LCD backlight control. Since I already have the PC board cut for the Adafruit pinout, I had to figure out how to rearrange the Hernandez code. Once I finally got his HelloWorld-SPI demo working though, it was simple to add the backlight bit, and I could modify the head end sketch to match. I changed the LCD code to use the same SPI clock rate as the CAN and they play together very well now.
This is the backpacked head end with new code displaying light level and temperature from remote station 31.
The completed printed circuit board needs to be stuffed with components.
Populating a PC board is best done by placing the shortest components first and the tallest components last.
So first to go on are the Z wires. These connect traces on the bottom to traces on the top. They are necessary because this homemade board does not have plated through holes.
The next higher parts are the resistors.
Next capacitors, chips, and the crystal. Some of the leads have to be soldered on the top and on the bottom of the board. Again, this is because I do not have plated through holes.
Now I attach the CAN connector, the contrast pot, and the header that mates with the LCD. These are the tallest parts
Finally, some rainbow wire to connect the backpack to the Arduino. Also the backlight switch transistor at this time because it took a while to find one in my junk box.
Finally solder it onto an LCD display and see if it works.
Remote light level and temperature on the display
The project needs quite a bit of work on the software to be effective. The two SPI functions seem to interfere with each other. Suspect a library conflic.
In this part I’m showing a method for transferring the laser printouts to double sided PC board material.
Double sided printed circuits must have the top and bottom masks perfectly aligned. I begin by using a sharp scribe to pierce the exact center of each corner hole on both the bottom and top layer printouts.
Next I insert one of these tiny nails (3/4″ #18 gauge) into each corner of the bottom mask. I could use sewing pins but nails have flat heads which keeps them standing up during the next steps.
The next photo shows the bottom layer printout with alignment nails inserted into the mask’s corner holes. If I was making a board without convenient holes, I would add holes to the Eagle drawing outside of the board area and just trim them off after completing alignment.
Now I take the top layer mask and carefully slip it face down onto the alignment nails, smooth it out and tape the top edge. The tape maintains the mask alignment from here on.
Next I removed the four nails. In this photo you can see how the two masks align face to face. Holding the aligned sheets up to a strong light should show any problems.
I’ve scrubbed the blank PC board material with fine steel wool and wiped it down with lacquer thinner. It cannot have dirt or fingerprints. Now that my masks are aligned I can position the blank over the bottom printout. I use bits of Kapton tape, though the board should stay put without tape if the paper is not moved before applying the Iron. I cut the blank about a quarter inch oversize on all sides to make this step easier, the excess will be trimmed off after etching.
It’s time to heat up the Iron. I use the highest setting, which I think is still a little cool. I let it sit on the mask/board/mask sandwich for two minutes to make sure the copper is hot enough to melt the toner. There’s a slab of MDF underneath the paper so my workbench doesn’t get cooked.
I remove the Iron and immediately use a small rubber roller (brayer) to press the mask firmly onto the copper. Not sure this step is necessary but I think it helps even out the transfer when the Iron face is not perfectly flat, or the Iron temperature may not be even across the face. When the board has cooled a bit, turn it over, repeat the heating and rolling process on the other side.
When the board has cooled enough that you can handle it, trim away most of the paper but leave a little bit around each edge. Be careful at this stage, pulling the paper off will ruin the toner transfer.
Next the paper has to be carefully removed from the board in such a way that all the toner is left on the copper. I put the sandwich into a tub of warm water with a small amount of dish soap. In a half hour or so, the paper will begin to disintegrate. Just rub the paper gently with fingers, center towards the outside, and it sloughs off easily. If a bit is stubborn, give it more soak time. You must remove all the paper from the areas that will be etched. Paper left on the masked areas is OK.
This is the transferred board dried and ready for inspection. The toner is somewhat delicate so you can’t touch it. I use a visor magnifier and check all traces for complete coverage. Repairs can be made with a fine point Sharpie pen though I believe a typewriter correction pen would work better. Sharpie ink is too thin.
I painted the quarter inch of excess copper around the board edges with typewriter correction fluid and it stood up well to the etch bath. Also have to watch for smudges in the mask. My old laser printer has a tendency to drop toner in random places, I have to scrape these off with the scribe.
Finally into the etch bath. I’m using the aerated cupric chloride method as learned from Jim Williams. I have a good size aquarium pump feeding four air stones at the bottom of a 4 inch square etch tank. There’s about a pint and a half of Muriatic acid in there. I pull the board out every minute to check progress. In this photo it’s about 2/3 done on one side. At that point I turned the board over so the other face got the bubbles. Took about 15 minutes to do both sides.
It can be difficult to see when all the copper is removed. You have to inspect both sides carefully. Here I placed the board fresh from etching onto a florescent light so I can see through it. The alignment is looking pretty good as evidenced by light through the pad holes, and I’m not seeing any shadows where the copper was removed.
All that is left now is to clean off the toner with Lacquer thinner, trim it to final size with my big garage sale tin snips, and drill holes with a #64 bit in a Dremel tool. This composite photo shows the top and bottom after drilling. You can see the pad alignment around the drilled holes is very good.
In my previous posts here, I constructed Controller Area Network shields using single sided PC boards. They work well, but now I want to combine the MCP2515 CAN driver circuit with a 74HC595 serial-parallel converter so that an LCD display and CAN network can share the SPI buss. This is more complex than the CAN shield alone so a double sided PC board is in order. The two circuits I’m referencing are both available in Eagle board files. After about a week of learning Eagle and dozens of Auto route passes, the following design emerged.
I’m using a laser printer toner transfer method to lay out PC boards. This board is the same physical size as the common 16×2 LCDs, 3 1/8″ x 1 3/8″. Ironing on the toner spreads it a little so I need a design that has plenty of clearance. Since there are no plated through holes, I need to minimize vias, and force wiring to the bottom layer if it is not possible to solder the component on both top and bottom. It amazes me how nudging a component a couple of millimeters can drastically change Eagle’s routing.
There are many web pages describing how to etch PC boards using laser printer toner as resist material. There is an active forum on Yahoo with lots of information. Results reported vary a lot depending on the printer and paper used. All toner formulas are not equal.
Laser toner is ground up plastic with black material mixed in. Letters are formed by electrostatically attracting toner particles to the paper, then melting it in over a heated fuser roller. It is possible to re-melt the toner and transfer some of it to the copper surface of a blank PC board. The paper that works best is high quality photo grade stock, but thin. This type of paper has a high clay content to make the surface glossy, and ink that hits the clay drys immediately and won’t smear. Slick magazines like these qualities, so many people use magazine paper to print their PC board patterns.
My laser printer is an old Lexmark 4039 10+. It has an extra dark setting that makes toner transfer easier, and has a rear single sheet paper feeder and front exit slot. This gives a very straight paper path so the printed copy doesn’t curl up. Here is a photo of the two prints needed to make my LCD backpack. The paper is torn from an old Digikey microcontroller catalog. Note: always cut the torn edge off with scissors before running through the printer else it will catch inside and jam. Don’t Ask Me How I Know This.
It’s difficult to visualize how the top and bottom of the design (top photo) is going to work when applied to the PC board (Don’t Ask Me How I Know This either). Eagle has a “mirror” button in the print requester. You use this when printing the top layer, but not when printing the bottom layer. If you did it right the two prints will match up when placed face to face. Why mirror the top and not the bottom? Remember that both images are mirrored when they are transfered to the PC board – so the top gets mirrored twice resulting in a non mirrored resist, and the bottom gets mirrored once resulting in a reversed image. Eagle knows this. Any text Eagle puts on the bottom layer is automatically backwards on your screen but prints correctly on the PC board.
Adafruit recently announced an LCD “backpack” that has decoders for SPI and I2C communications between an Arduino and an LCD display. As is their policy, they open sourced their modified LCD library and made the schematic available. The SPI portion of the device uses only a single 74HC595 eight bit serial to parallel chip, which I happened to have a few of. So I decided to try sharing the SPI buss between the MCP2515 and a liquid crystal display. This means I can add a display to one of the nodes by using only one additional pin on the Arduino. With the help of Adafruit’s example code (and a very similar entry in the Arduino playground) I have a 16×2 LCD breadboarded using pin 9 as slave select for the ’595 chip.
In this photo there are only five wires connecting the Nano to the CAN chip and the LCD.
Adafruit’s library can turn the LCD backlight on and off also, so you can blink the display to draw attention. This particular LCD doesn’t have a backlight though. Also I commented out all the I2C stuff in the library and saved 1k of memory.
Next step: cut a PC board with the CAN chips *and* a 74HC595.