Turing-Welchman Bombe physical build – part 8.

August 19th, 2015

Well, a bit of a setback today. I replaced my main board with a new one that is using latches so the second and third drums hold their position much better between steps. That worked great. Except then when running through a Bombe run nothing would work properly! For some reason the motion of the drums was very uneven and jerky. I spent hours last night trying to figure out why and in the end went to sleep with the problem unsolved. Today, at lunchtime, I decided to take a 5 minute break from work and give it another try. Unfortunately I was rushing and I accidentally plugged my 12 volt battery into the 5 volt rail. No magic smoke escaped from anywhere but my Raspberry Pi got killed! They do contain some voltage protection, including a polyfuse, so it may come right. If not I will have to get another. So far then this project has claimed the life of one Arduino and one Raspberry Pi!

In the meantime I was rethinking the design and am going to try running it in a different way. Previously the Pi told the Arduino when to make the drums step and stop. This is one reason the motion was jerky I think. It relied on me sending the pulses very accurately from the Pi end. I have reworked it so now the Arduino effectively free runs but it sends pulses to the Pi every letter it steps. The Pi can then tell it when to stop. I think this should work much better. I have updated the Arduino software to do this (without the signals from the Pi yet) and it is MUCH better. I have made a little film to show this since I am too lazy to do a big write up today and because my friend Dani, a teacher, keeps telling me off for my chronic abuse of commas.

The other thing I did was order a little USB back plate. I am going to remove one of the connectors and mount it on the Bombe backplate for the Menu USB stick to plug into.

IMG_4304_1 IMG_4305_1 IMG_4306_1



Turing-Welchman Bombe physical build – part 7.

August 16th, 2015

This is turning into one of those projects. The never ending kind!

Lets see, things were going well. During the week I painted the switch housing with crackle black. Being winter, even with a sunny day, it wasn’t warm enough to get the paint to ‘go off’ so I cheated and used a hot air gun. That worked well.

IMG_4270_1 IMG_4272_1 IMG_4273_1

I had to do it in two goes, the inside then the outside. You can see when the paint is just sprayed it is very glossy and smooth. The crackle finish comes as the paint dries. The trick is getting a nice thick, wet coat without the paint running. I prefer to let it dry naturally in the hot sun but the hot air gun works too.


After the paint is dry I like to cure it in the oven. I leave it for a few hours on a low heat, 50-70C, which helps really harden the finish.

I also added the transfers to the drum faces. This took hours. It takes careful positioning and much care. It’s a good thing to do while sitting in front of the tellie. Unfortunately I did this while watching Airport. It was going great until I got to an exciting bit (Petroni giving the 707 full throttle to get it off the runway) and I made a slight mistake. Unfortunately I didn’t notice until I got to ‘W’ and realised I had 4 letters to go but 5 spaces. I’d missed a bloody ‘L’!

So I had to scratch off most of the transfers and start again from ‘K’. That all went fine but then I found I didn’t have enough ‘M’s so one of them has an M made from part of an upside down A and two Is!


I clear coated the discs and again baked them in a low oven.


I also bought a new 12 volt SLA battery and made up a small poser supply board with the boost converter. This bolts onto the main board.

With the drums dry I was able to test things out. Only I hit a snag. I was hoping to just use a friction fit between the shafts of the motors and drums. This was a total failure. There is too much mass in the drums and there was a huge amount of slippage. So I had to rethink the whole way I attached and drive the drums. I came up with this.

I remade the drum shafts with a thicker end piece so I could drill and tap it for a locking screw. If I can find some 3mm grub screws I will use them but normal ones work for now. On the side of the bigger section I soldered on a section of square U tube. That provides a key to give positive location and drive. The key fits into a slot on a washer. The washer fits on the back of the drum.

IMG_4286_1 IMG_4287_1

IMG_4288_1 IMG_4289_1

I tapped the holes in the washer so the four screws that hold the hub in place thread directly into it.

IMG_4290_1 IMG_4292_1

The shaft is permanently attached to the motor shaft. The drums are a slide fit over the shafts. And now the little rabbit clips actually do something. The drums now actually work much the same as the real ones. The slide over the shafts, there is positive location and drive and the clips stop the drums from coming off. It all works really well!

So the front of the Bombe looks great.


I may replace the face plates with some lasercut and engraved ones. I went and found out how much that would cost (less than $100NZ) if I can provide them with a suitable file. But I wanted to make sure the Bombe actually works before I do that.

Good thing too. It doesn’t bloody work! I made another mistake. The issue is how I drive the steppers. Because of the mass of the drums jiggling about I am using a holding current to keep them in position accurately. When driving steppers you make them move by switching voltages through several coils. Each step is caused by the voltage changing in a particular sequence through the coils to make the motor move. Between steps you can either keep the last voltage there or turn all the voltages off.

If you turn them off the motor is free to move. If you hold the voltage the motor locks into position. For what I am trying to do I want it to hold. My code does do this. But because of the way I share the drive signals to the second and third drum only the first drum was being held. It steps very accurately but the other two would not. I used an analogue switch to switch in the drive signals when the second and third drum need to move. But when they are not moving there is no holding current since the output doesn’t latch.

So I have redesigned the electronics. Instead of 4066 switches I am going to use 74HC373 octal latches. I think with these I can still switch the output when I want (using the latch enable) but when they aren’t changing they will hold the same voltage on the motor driver inputs so they will hold their position.

I hope.

So the back of the machine is looking rather sad.


I pulled out the main board and will modify it to take the new latches. There should be just enough space. I just need to get some desoldering braid so I can remove the old components then clean things up to rework it.

The other thing I have done is ditch the USB hub and am connecting the menu thumbdrive direct to the Pi. I am getting a small USB panel socket and will mount that to the main board so the USB thumb drive a can be plugged in easily.

Turing-Welchman Bombe physical build – part 6.

August 11th, 2015

Right, pressing on here since I have cars I should be building so need this thing done! The other night I started making the covers. Well, the top and bottom ones. I will need to get more steel for the sides and back. On the top of the housing I brazed on a metal strap to which the handle can attach rather than just bolting through the flimsy sheet steel top.


The top and base are simple steel squares drilled for screws that go into tapped holes in the frame. One little rant here. My cordless drill is pretty good except for one silly feature. It has an LED light on it but the light is totally useless! Whatever you are drilling is always in the shadow of the chuck! I think that’s what happens when you let marketing people come up with features, not engineers!

IMG_4232_1 IMG_4236_1


To mount the castors I positioned them inside the base so I was sure they cleared the frame and marked where to drill the holes. I only had some 3/16ths bolts to attach them at that point and they were a little big. Another problem is the base of the fixed wheels isn’t flat! So the bolts sat at weird angles. I sorted that out today.

IMG_4237_1 IMG_4238_1

That was about all I got done last night. It was cold then all of a sudden it got very noisy in the garage as it started hailing! It was like someone dropping ball bearings on the roof. I came in soon afterwards!


Today, after an interesting meeting at work, I decided to go out for a drive for some fresh (and chilly) air. It was a nice day so I took the MG out and went and bought some new nuts and bolts for the wheels as well as hinges and a catch for the base. I also got the handle. When I got home I found my new slotted screws had arrived from RS. Earlier my boost converter had also arrived.

Tonight I mounted the wheels properly. The screws I got were too long.


It wouldn’t hurt anything leaving them like that but it looks rubbish so I cut them all down. I have a trick for doing this. I take a steel plate of the approximate length I want for the screws and drill and tap a hole thought it to take the screws. If the plate isn’t quite thick enough you can adjust the length with washers under the screw head. Tighten the screw through the plate then chop off the extra thread with a hacksaw. Then sand the cut off part, while still in the plate, on the linisher.

IMG_4247_1 IMG_4252_1

IMG_4249_1 IMG_4250_1

The screws will all end up the same length and when you remove it from the plate it also cleans up the cut end of the thread nicely. Things look much nicer without extra, unnecessary threads everywhere.

IMG_4253_1 IMG_4254_1

Since my correct slotted cheese head screws arrived I did the final mounting of the motors and the main board to the front plate. First I had to  open up the holes the motor shafts fitted through. This is because the motors have a stepped boss in the middle of them. I needed to open the hole up so this boss would fit into the plate and the motor mounting face was hard up against the face plate. I opened up the holed with a hand reamer then finished them off with a sanding drum in the Dremel.

IMG_4256_1 IMG_4257_1 IMG_4258_1

The motors were mounted with the correct screws and washers and they are held in with Loctite. I also mounted the main board to the back of the faceplate and wired in the motors. I temporarily reassembled things with the correct screws to make sure it all fits. I also powered up the board to make sure it still worked (which it did).

IMG_4259_1 IMG_4260_1 IMG_4261_1

Pint glass for scale!

I still need to mount the USB hub and work out the battery. I tried a 12 volt 4Ah SLA battery which will fit. The one I have says it can’t be tilted (which seems odd) so I will get another that I can mount on it’s side in the base. I’ll order than from RS. I’ll make a clamp to hold it down to the base. The USB hub I might mount with some double sided foam tape which should be more than enough to hold it. The little boost converter will mount to the main board.

IMG_4263_1 IMG_4264_1

Tomorrow I will make the housing for the LCD that goes on the left hand side to represent the relay unit. Here is how it looks now with the (unfinished) drums in place.


Whiskey glass for scale!

Turing-Welchman Bombe physical build – part 5.

August 9th, 2015

Right, this weekend I had so much to do. Clean the house, cut and stack firewood, work on the car. I did none of that. I worked on this thing instead!

Earlier in the week I painted the three drums. Three coats of gold then three of clear over that. They came out great!


I also went and got more acetylene on Friday so I could braze up the frame. The plan was to spot MIG it all then braze the joints. As it happens I was out of MIG shielding gas too (so another trip to BOC this week then)! So I just gas tacked everything then brazed it. I started with the top and bottom squares then I clamped everything down to my steel plate to do the sides and was very careful to keep everything square.

IMG_4161_1 IMG_4171_1

IMG_4172_1 IMG_4179_1

The brazing went very well and after a quick sand down I started on the various cross pieces things would mount to. I started with the front cross pieces which the brown Tufnol would attach to to make up the front. The drums are mounted on a sheet of Tufnol set back into the frame. Smaller pieces fill in the game. Some angle iron made a convenient mounting place for these small pieces. I drilled and tapped them so that the pieces can be simple screwed into place. The main plate is attached to two long, vertical bars set the correct distance back into the frame.

IMG_4203_1 IMG_4180_1

IMG_4181_1 IMG_4182_1

AFter all the brazing was done I sanded, wire brushed the frame. To get into some of the inside corners I used a spot sandblaster. It’s very important to remove all the flux left from the brazing. It turns into a solid, glass like substance that can be hard to remove. Sand went everywhere! The front pieces attach like below. All the mounting screws are temporary. I have some slot head ones coming which will be much nicer than the cross head ones.


The pieces on the face have beveled edges so they fit nicely into the frame. It’s very hard to see how all this is done on the real Bombe so I am sort of making things up as I go now. Since I am not making the entire thing though I don’t have to be (indeed I can’t be) exact. I can just be Bombe like!

The main front plate is a sheet of Tufnol drilled to take the steppers, the pointers and the main circuit board.

IMG_4189_1 IMG_4191_1 IMG_4188_1 IMG_4192_1

With everything temporarily mounted I was able to see how the drums (only loosely assembled) would look. I think that’s getting there! You can just see the face plate is mounted behind the vertical bars it bolts to. This is so when I make the side covers I can fold the front edge right around and bring it right back to the face plate. The bar there gives it something to rest against.

The main board attaches to the back of the face plate. I machined up some 50mm long brass standoffs to hold it up past the steppers. The space between the motors and the circuit board also provides a handy place to hide all the motor wires.


Next I made the little switch unit. On the real Bombe this seems to have two little paddle type switches. I am just using push buttons. To mount them I cut a piece of oak (same as I used on my Orwell computer case). After cutting to shape I drilled two cutouts underneath it so there was room to poke the switched through and do up the locking nut. I kept the block thicker than needed while I drilled then later planed and sanded it down to the required thickness.

IMG_4205_1 IMG_4207_1

IMG_4209_1 IMG_4210_1

The curved cover that goes over the switches I made from steel. That was a simple job but I needed something round to bend the metal over. I looked for cans or piped or anything the right size to bend the steel around. In the end I found that the rear axle of a 1929 Riley was exactly right, so I used that! Vintage cars are good for more than just banging my shins on after all!


I then soldered on a simple steel back. The back is right up against the face plate so not really needed but on the real Bombe it looks like there is a flat metal plate just behind the keys under the cover. You can see this is painted crackle black. So I put a back on mine so when the whole cover is painted crackle black you will see the back of the switch cover is the same.

IMG_4213_1 IMG_4214_1

The cover attaches to the block with four small wood screws on the sides and the whole thing bolts to the frame.


Now we’re getting an idea what this thing will look like finished.


I next need to make the covers.  They will screw to the frame. I am not looking forward to drilling and tapping all those holes! The bottom will have the castors on it. The top will have a handle so I can move the thing. I was originally going to power it from a wall type plug in supply. The main board has a rectifier and a regulator so any voltage above around 15 volts, AC or DC will work. But then I thought well, the whole thing can be portable. I am adding a carry handle after all (which the real Bombe definitely didn’t have!). So I have ordered a small boost converter and can run the whole thing off a small SLA battery that fits inside the case. The whole thing will be completely standalone then. It’s not efficient stepping the voltage up then down again but for this that won’t matter. A run only takes 20 minutes and it draws well less than 1 amp of current so even a small SLA will be able to power it for many hours. I also still need to do the lettering on the drums.

Turing-Welchman Bombe physical build – part 4.

August 6th, 2015

Tonight I machined up the drum hub shafts from some brass rod. These are very simple. A beveled end and a groove for the rabbit clips to slip into on one end and a hole bored in the other that slides over the stepper motor shafts. The rod is a tight press fit into the white plastic part that everything on the hub bolts to.  All the torque of the motor drive goes through this. As there is no load on the motors it should spin the drums easily.

IMG_4147_1 IMG_4157_1 IMG_4158_1

The real Bombe has a fairly substantial steel frame. For my Bombe I decided to make a slightly lighter weight, but still steel, frame. Since it was a fairly quiet day and there was a break in the weather I went for a drive at lunchtime to get the steel. I thought I had some left over from building the Austin 7 body frame but apparently I had used it all! I went and bought some lengths of 16mm square steel tubing as well as some other pieces of angle iron.

Last night I had drawn up a plan of how I will make the frame.  It’s quite simple, just a rectangular box with some places to mounts various plates and components. The outer covers will be steel. Since I am just doing the indicator drums I can’t exactly copy the entire housing so I am just going for a Bombe type look.

I am starting with a simple rectangular box frame. The top and bottom are simple steel squares with a vertical joining them in each corner. I first cut the top and bottom pieces. The corners are cut to 45 degrees so I can weld it all together square. I made the first carefully to the right size then used that to mark up all the others. I have to hand cut and file everything carefully.

IMG_4139_1 IMG_4143_1

IMG_4140_1 IMG_4141_1

Then I end up with 8 pieces all the same length. The little holes in the middle are because I am going to gas braze the frame together. The hole lets heated air out when you’re getting everything hot with the welding torch. Not sure that’re really necessary but I have always done it! The little holes are handy later for painting since you can bung wires into them to hang the thing up.

IMG_4150_1 IMG_4154_1 IMG_4155_1

The verticals are simply cut with square ends. Everything will be MIG tacked into place so it’s all square, then I will braze it.  I do need to go and  get a new bottle of acetylene and some brazing rod first. Everything is done on a flat steel plate so it is all level. I need to tack everything just to check it’s all going to fit. It will be a tight fit but this is meant to be a desktop type model so it can’t be too big.

When I was out I popped into Supercheap Auto and bought some gold paint and some clear coat to cover it. I will paint the drums with this. This is just a test on a spare sweet tin I had. I want to see how resilient the paint is.


I also received the other day the little castors which will go on the bottom of the Bombe just like on the original. There are two fixed and two swivel wheels on it.


I also got some dry transfers for the lettering finally. This stuff is now very hard to find. None of the stationary or art shops have it anymore. I don’t think Letraset has been around for a long time. But you can still get similar transfers from modelling places. Woodland Scenics will be familiar to anyone who does model railways. They do dry transfer sheets for doing lettering on things suck as trains and buildings. I ordered some from a model shop online here and had it the next day.


Since they cater to all scales there should be a size of lettering that will be appropriate for my drums (around 1/8 inch is about right). I am going to do a test first as I want to see if I can clear coat over the transfers to make them more robust.

Next Page »