Rebuilding the UM1


Some years ago I was given a UM1 that had the logic supply connected to enough voltage to crack the power regulator. Most likely every chip was blown apart.

I had used a similar processor for a video conversion project, which has long sat documented in the video section of my website, an early form of blogging from the late 1990s. I still had a tube of these processors, although the ones I had were 52 pins, where the processor on the um1 had 44 pins.

Since I frequent the surplus shops of the silicon valley, I located most of the chips used in this board. Given supply issues, I have for other clients developed a similar board using a more modern embedded micro. More on this in a later blog. The short version is that I had a UM0 and wanted to keep the same setup codes and configuration on my Caliola.

By connecting a MIDI sniffer to the wire, I was able to detail how the setup system exclusive messages worked. Technically these are supposed to be documented in the manual. The manufacture is not registered so the device setup for these exclusives is the research mark. Which means that this data is open and unprotected, so I could use it.

I noticed that the Octet setup program had a mechanism for uploading new firmware updates. These updates were available on the web. When opened they were found to be in the common format used by the MC68HC11. There was even included a small program called a talker.

Locating the needed chips, the question remained, would it be possible to reprogram the actual UM1 unit from the web updates?

The first step of the process was to remove all the old chips. This was done with a heat gun. The board is slowly heated to 210degrees C. My thermocouple slipped and some of the sockets got a little cooked, the result was a clean board. The front panel was the worse for wear as the switches melted completely.

After the chips were removed, the traces were cleaned up and documented with the Eagle layout program. Oh for the days of Sam’s Photofacts. I wonder if the library still has the shelves of 1950s and 1960s era television schematics. Actually I was able to get schematics for my late 1990s Panasonic, telly. So major manufactures do still provide repair personnel this needed information.

The designer of the UM1 had been dead for some decades now. The successor does provide manuals and firmware files as noted. I took a peek inside these files and located where the device tables go. In order to bring a new system up from scratch, a few additional items such as the serial number are needed. This was easy to get, as it was pasted onto the blown up chip. The firmware for some reason does not contain the device string, which for some reason is always UM1 even on a UM0.

I was a bit concerned that I would have to solder in the new processor. These are surprisingly still available. This package called a PLCC is no longer made. When the EU adopted RoHS this format was discontinued. In the surplus store I was able to find a socket that would solder onto the board, so anyone in the future who needs to fix this can swap the chip out.

Once the new chips were in place it was time to reprogram the unit. I could have connected in through the MIDI interface. The 68HC11 contains a serial boot mode. There is a photograph of this in the attached image gallery. These signals are easy to access on the 74HC02 chip labeled U9. For future reference the pins are Rx on pin 8 and Tx on pin 6. Laurent must have used jbug as the wiring of these pins is straight out of the college textbooks.

The 68HC11D0 contains no internal flash memory. It is easy from looking at the schematic, to see where the memory starts. 0x2000. This is where the jbug11 talker is loaded. The firmware loads in at 0xE000. This is where the SYSEX tables are stored.

The first step is to upload the firmware from the website or setup disks(If available) This can be uploaded by simply renaming the file extension from .frm to .s19. Since the firmware does not overwirte the device information, this will need to be copied from a working UM1 The format is exactly the same as what the UM0/UM1 setup software sends, so this can simply be done by dumping the read firmware sysex

Once this data is written to the first sector of the 0xE000 address, there should be enough code to get the whole unit working. In my case I replaced the MIDI opo-isolator and the U9 74HC02 chip. I was able to connect using the Octet setup program. Just to be safe I re-ran the firmware upgrade.

It should be obvious that a working UMx is needed to get things stared, this blog is about repairing units. There are much better and cheaper ways to make similar devices, which is why my own boards are only functionally equivalent, with no common hardware.

Naturally the diagnostics did not pass as I had not seated in the driver chips. These are often out of stock and I had to get some from other units for testing. Once installed, The diagnostics pass muster and the unit is once more operational apart from a missing chip for the PWM and the melted front panel. The UM1 works fine without the front panel, but there are some unsightly holes where it goes.

Repairing the front panel did not have quite the same success. I socketed these chips, which means the display plate no longer fits. The surplus hex switches are also much larger and do not have the same pinout, so give the wrong midi channel. There is also quite a bit of flicker on the LeDs. Most of these units are set once anyway, so the front panel is a nice extra, but not needed.

It is interesting to compare my “clean room” copy to the original. There is no hardware in common. I do parse some of the same setup code which is what makes my unit a clone.

The repair on this is also quite expensive, The obsolete processor sells for 10s of dollars, when in stock. [Currently at this I have some similar E2 processors for sale on eBay]. Atmel processors sell for a few dollars each, These contain many times the storage space and do not need the external logic. The replacement parts on the UM1 were around 35 dollars not counting the driver chips. It probably would be cheaper to buy a new unit, if such were available.

Over the next few months, I expect to sell some of my compatible prototypes online. I much prefer consulting and building devices to order. I also am willing to take on repair of the octet systems as well as any Broadmore power rolls that may still exist.

RepRat main controler

I mistyped the name Reprap in the firmware code. Happy accident, I now have a name for my Repstap/Mendle project. This fits in well with the ratsnest of wires on the SDK500 mega168 project lashup.

As I do not care for the bloat of Arduino, I am writing my own firmware in assembly. Much of this I borrowed from one of the MIDI compact flash projects. This way I can use my already small footprint FAT12/16/32 code.

SD uses a serial interface for comms, so much of the low level code for initializing the card and reading blocks needs to be re-written. Using an AVR freaks tutorial, for butterfly SD card connectivity is a good way to test the code.

The supplier makerbot had a few remaining v1.2 controllers. This enabled me to not make a board from scratch. Since the SD card socket is in place I swiched over from the rats nest to the v1.2 board.

Weirdstuff, a surplus store in the silicon valley has some SD/MMC adapter cards for about 20 cents [USD] at this price I bought a couple dozen. With a heatgun and a thermocouple poping the connector off is a simple process. Provided one does not melt the plastic.

The makerbot/reprap “motherboard” controler v1.2 board has the pads and holes in the right place for these sockets. The shield on the cheap sockets is a bit smaller. Removing solder mask in these places allows for the MMC card to be attached to the board. Image 1 in the gallery below details where the solder mask was removed.

The board can be assembled either for use with an ATX supply or with a power brick. The 3.3 volt regulator I got from jameco has a different footprint than the recommended one. It is not a good idea to power a 3.3v regulator off of a 5v regulator. There is not enough headroom for the regulator to work. Looking at the ATX connector, it seemed the most efficient way to get power would be to tap the 12v line and add a standard 5v regulator. It worked out nicely to make a little bridge using the leads on the filter caps. This also gives some points where the bench supply can inject power.

The schematic indicates that the 12V is sent through the RJ45 connectors to the external controller boards. I like to recycle/reuse stuff, so I got some block of RJ45 from weirdstuff, which I de-soldered. Some of these I melted the tabs as I did not have any heatsheld. I mounted one of these extra connectors in case I want to inject power through this connector.

It is a bit annoying that the power switch is a relay trip. This means the AVR must always be powered. I still need to make some usb/rs232 boards for the FTDI chips I have about. These use an odd pitch footprint, which requires at least a breakout board. Ironically, I want to get this stuff working so I can drill the boards. I guess that is the rep part of the system. The rapid part is what my (C)NC is not.

The gallery below shows some of the steps in bringing these boards online. I moved the ex at90s8515 code from the mega88/mega168 to the mega644. The mega168 to mega644 path is fairly straightforward. Some of the interrupt vectors are different. Mostly due to the second optional USART.

Now that I have a more stable less of a ratsnest prototyping setup, I expect to make more changes to the firmware. I am still considering using i2c to drive the steppers. The current layout consumes most of the pins. This makes adding multiple heads/axis a bit awkward. There are also a shortage of pins which can be used for button panels and readout displays. Of course these can be implemented through the serial coms, which would involve removing the existing code from the system.

3D printing for the rest of us, The reprap.org project

Recently I have been active posting to the reprap forums. There I started a thread relating to my observations after reading the forums almost daily for a year. This thread can be read on those forums.

I have a rather macabre and dry sense of humor. Much like the mummys here in the hall of mystery.

Reprap is a bit of a mystery. Something real that has come out of what is a dream and a bit of a lie. There is still much promise to be had with the current technology. Most of the old timers must think me a bit loopy as I tend to rant against software versification. That only the newest hardware and software are worth using.

For almost 20 years from 1982 to 1998 I made my living directly off either selling or working on the design QA team of apple products. Steve Jobs was my boss’s boss. We did what Steve wanted. If not, well lets just say Apple no longer designs laser printers. At least my products were a success. The short of it is I use Apple products. I do not download revisions because I am told to. I download what I need when I need it. Because it is a better product.

This is what damaged apple in the late 1990s. I did not agree with it then and do not agree with it now. If it is not broken, there is no need to fix it for the sake the repair bill. Which is usually padded anyway.

If this is how you feel, then stop reading now and return to the facebook fad crowd, because this sheep is an Arcadian sheep, who believes in recycle and re-use. If it is not obsolete, then why bother with it, it will become obsolete soon enough.

So what made Apple great, and why I like so much the work of the Disney brothers is that they produced something new. Reprap is like this a new way of looking at how one can make things.

Unlike the lie of most dreams, reprap has some truth that makes it quite likely to succeed. There are a lot of issues still to be had. One of these issues is speed. This has been a problem with my existing CNC equipment. It is too slow and the design path takes forever. I am lucky if I can scan in a part and make a 3d model in less than 16 to 18 hours.

Then comes the issue of making the part printable. There are several divergent paths for the reprap. The core team made a java applet. This does not work on the mac. Since the hardware was expensive and built to last, there is no one to test the code, so the code sits somewhere in cvs hell. The problem is that most modern programmers depend on the corporate need to upgrade early and often. Little thought is given to simplifying the code to run on the lowest common denominator.

So instead of bothering to fix the mess, a new branch went off using python. So we are in the jungle anyway. One adapts. This is an experiment anyway, with just the right amount of anarchy.

The real gating force is that my machine tools are slow. So the best place to start is with the bottleneck of the 1989 drazzi motor controller. I need to make some rollframe parts for a client, so what better than to upgrade the system and mill these parts. At the same time look into how the system can be improved.

The shop is not really a place for the latest in computer hardware, So the goal is more for an NC controller than a CNC, where the build files can be read off of a smart card. I already have the electronics for reading compact flash card MIDI, so making these changes should be quick. We will see. a days project turned into a week, and is more likely to take a month.

The real question is making the motors move. In practice we are just drawing a picture. In 1773, Swiss engineers could do this with cams, so it must be a trivial process. The Swiss even used a Z axis on the pen or pencil so the final pictures look hand drawn. The doll that draws with a pen even knows when to shake the ink from the tip. The artist doll with a pencil blows the dust away.

Of course the french kings that these toys were made for were beheaded, But that is only poor solution to the problems of management. The new managers, were quick to realize that scaled up these simple toys could be used to rivet battleships automatically. This was also a simple solution to the Irish problem. Since automation did away with labor.

As I have noted, my late friend John Grass sold me the machine to prove a point. That one needed to understand the basics first. I was the only one who could make the drazzi work. It only would run one program and draw a picture of a running horse. The reprap seems to have the same issue, it currently is designed to keep the unemployed labor occupied dreaming of change, and not of what is needed.

Make a mechanical computer

Through my computer history and horology contacts, along with some targeted searching on the net, I have collected a good deal of the designs for Babbage’s Analytical engine. The next step would be to visit the archives where the papers are stored. Doran Swade estimated that it would take his team some millions of dollars to do much of what I have done.

Of course our methods are different. Doran, is working to make it just like it would have existed in the 1850s or 1860s. Doran has good arguments as to why a mechanical computer, designed in the 1830s is not likely to have been made before about 1853. His goal is to answer the question, could it have been done then? His approach is to match the metallurgy, surface finish and make a 19th century machine in the 21st century.

My approach is more pragmatic. As far as Alan Bromly or Tim Robinson were able to learn, Babbage did not have the concept of an instruction set. This makes any modern re-construction suspect. The other issue is that the Analytical Engine was huge, about the size of a steamboat or locomotive.

Bromly left some fascinating articles written in the early computing and horology journals. These observations combined with a simple paper training computer called Cardiac, produced by Bell labs is a great entry point into making a 3 digit decimal computer using Babbage’s concepts.

By scanning in much of the publicly available information, I have made some 3D models of the adding mechinisims, These look perfect for making on a repap 3d printer. This is where I was a year ago when I discovered the world of three dee printing.

So over the course of the next few blogs, I will take a look at the reprap process, and just how practical will it be to repstrap a Mendel.

Welcome to the Egyptian Hall of Mystery

Welcome to the Egyptian Hall of Mysteries…

The yellow Drazzi …

Many years ago my late friend and mentor John Grass sold me a desktop CNC machine. John was a machinist of the old school, and could not make the machine work. I think he wanted to prove a point, that having something was not the same as wanting something.

Here is an image from the sales brochure:

What the machine looked like when new.

The machine came complete with all the manuals, and a Dos program called RobotMart. Many times I searched the net for more information on this. Several years after I acquired the machine in I was able to upgrade the Dos program and purchase a few more collects and accessories. I set the machine to making roll transports for Piano roll scanning. I also used it to create some parts for a mechanical doll. More recently I modified some hobby gears for use in a difference engine model.

At the time I lived in a 2 bedroom apartment, I leased the machine in exchange for shop space at a friends company. This worked out well. I was also able to make an EDM sinker machine and a watch pinion cutter. The the dot com era came to an end. My friend moved his tools into a Silicon Valley Garage, where I continued to use it for making roll-scanner parts.

We nicknamed the machine “Yellow Drazzi” as the eastern European name rhymed with an alien species on the Babylon-5 show, popular at the time. Where there were Green and Purple Drazis.

My work took me to Kentucky were I worked for piano movers designing a floppy disk MIDI playback system for pianos. This company failed after a few months and I was left with the floppy disk player to manufacture on behalf of one of the partners. This poor machine sat idle.

My workshops are in a barn like structure in my parents back yard. This is a potting shed, not water tight. I did not want the machine rusting, so left it in my friends unheated garage, Which also gave me access to the Bridgeport mill and atlas lathe. When it came time to make the feet for my Caliola Pipes, having to travel 60 miles to use it I moved the machine to my parents garage.

Over the years I have owned this machine, I have wanted to speed up the axis, which move around 5 to 10 inches per minute. When the machine sat idle, the fans developed a rattle, which was annoying. A few weeks ago I took the controller apart. Replacing the fans with new ones that glow with blue LEDs, these fans were half the price of the black ones.

While I had the machine apart I drew a schematic of the controller. The electronics were made in 1989. The stepper chips were manufactured through 2008. Searches of the network, I found better programs such as KCam and Mach3 to run the machine. The controller was compatible with these programs. It is like getting a whole new machine to play with.

For the last year I have been reading the Reprap forums and blogs reprap.org. Having worked with printers and scanners in the 1990s I have collected quite a few rods and steppers. Having spent so much time consuming information from these forums and blogs, It is time to start sharing my own progress under the name of sheep. Since I am simply just another sheep in the fold.

The first plan was to use the Yellow Drazzi as a repstrap. This may be a bit slow, as the old stepper controllers are resistive limited and max out around 10 inches/minute. A desire for speed improvements, has lead me to completing the reprap (now Mendel version) stepper electronics. While this will not run the coffee can steppers, I should be able to mill the Mendel parts on the Drazzi.

The ultimate goal here is to be able to replicate, not only a Jaquet-Droze style keyboard player, but the connected pieces, such as Babbages analytical computer, The mechanical calculators from Swilgue’s astronomical clocks, and the Antikeythera device. This while at the same time looking for work/contracts, speaking at conventions and working out a new act for the Dickens Christmas fair. So there may be gaps here as these project progress. No promises on updates, bit there is a lot to share and much more to come…

Bird in a box.

Somewhere I seem to have lost almost a year. Should you find it, please return it. Hopefully I can maintain a better blogging schedule. There are some exiting new things happening.

Last spring I came close to making a small 20 note organ in 1 week. Everything was completed except the pipework. Shortly after I started making and voicing the pipes, a family of birds moved in and made a nest in a box of pallets, which are for my w146 band organ.

It was difficult getting photographs of the birds. With some effort I managed to snap this picture. The birds were bold and seemed to enjoy watching me voice the pipes. Later when I took the nest apart after the young one left, I discovered that they were using the shavings from the pipes, as I made the pipes! Clever birds.

Bird who moved in

Bird in a box

Finally a I got around to downloading the photographs of the completed pipes. Here is the image.

It has taken me some months to get these items uploaded. I was called back in to complete the Theater organ I was installing. This project took until October of 2009!.

The 20 note organ was a success. I was able to take it to Several events and rallies. Below is a photograph from 2009 sutter creek.

I was also able to take the little 20 note organ to Participate in the 2009 Dickens Christmas Fair. Since the photographs were taken I added a nice top. I also made a new costume. Part of the reason I have been behind on my blogging. Of course, when one is so busy with a production like Dickens (there were 1000 performers this year) I have no photographs of me entertaining the folk entering the cow palace.

If someone did get a photograph, I would like to see it. Of course I have plenty of photographs from when I was in the Dark Garden window. That seems to be the nature of the show.

Everything at once.

Not sure if there is a point in counting the days anymore. Even without pipes the organ made it’s public debut on the weekend. Saturday It was shown at the MBSI Ardenwood meeting. Where there was a nice table on the side porch to display the work in progress.

I forgot the camera, however several others did manage to take some photographs. These may be sent to the MMD for publication.

Several friends stopped by Ardenwood, Including Russ Doering, who gave me some pipe wood. I think the scrap from these pipes were used to make the pump and regulators.

Sunday was an eventful day. I wanted to be in three places at the same time. Two of the events were close together, So the organ first went to the AMICA meeting, Hosted by Bob Gonzales in Marin. I then stopped by for the last hour at the Shoenstein open House in Benicia.

My experiment to locate a time machine sometime in the future, seems to have failed, So I was not able to attend the second day at Ardenwood. Even so I was fortunate to attend two event happening at the same time.

It was not my intention to take the organ into the Shoenstien factory, as they were showing off a rather nice large organ, and a much smaller one. Mentioning I had the organ in the car with me, I was invited by the owner Jack Bethards to bring the organ inside and show it. Jack is good friends with Russ. Probably the wood pipes I made the wind pumps and reservoir passed through Shoenstien at some time.

This makes the little one week wonder as I have been calling it a success in that many of the people who indirectly helped collaborate on it, got to see the results first hand.

While showing the instrument to a visitor at Ardenwood, I noted that the wood used was made from found scraps of wood. I was real pleased when they remarked that the organ was “green.” as this was the point of the project.

After the Shoenstien Open house, a few friends of Harold, who made the crankshaft and roll mechanism, also in the same week, had a nice dinner to wind down from a Hectic weekend.

It was also nice to see some of the same people, both days at different events. This means I was not the only one multiplexing everything at once.

Aldo Laus at Mell Wright’s excellent busker site estimated that it should take 54 working hours to make the 650 or so parts at 5 minutes per part. This is way over optimistic. I spent about 16 hours a day for 8 days which works out to 128 hours. A lot more than the typical normal 40 hours allotted for work. Aldo estimates, that there are about 290 or so parts not counting the pipes. Many of these parts took more than an hour to layout measure, check and measure again, before cutting.

I think I was wise to start with the bellows, as having built and recovered a number of singing bird bellows, I find the valves on these items to be tricky. The valves do need time to settle in.

I did not make the roll mechanism as this was done by Harold the project instigator. I suspect Harold spent more than 5 minutes on the parts he made.

This week will be spent making and voicing the pipes. I am not aware of any public events in the near future, Probably just as well as I will not have such an artificial deadline to deal with. I will admit that I had fun with this project. One of my former co-workers at Shoenstein, asked me if it was a Joke. I think they were correct in a way. That there was humor intended in setting out to do the nearly impossible task of making a quality product in a week.

Not that the final product was a joke, but the desire to do it in a week, was a bit tongue in cheek at first. Then developed into something serious.

Day 7, everything but the pipes.

Technically, the parts for this organ were made in a week give or take a few days.

Making the pipes was saved for the last task. Debugging the pumps, took more effort than expected. The valve design chosen was the alternative one. The valve leather I had left over from another project has been stored rolled up. This needs to lie flat for awhile. When the pump sticks were connected, the valves did not fully seat and the back pressure blew out the side of one of the feeders.

At least the check valve flaps are easily accessible. Holes were drilled where the corner tacks pass through the material. This allows the leather to lie relaxed on the valve seat.

another issue was that the releaf valve was not centered on the hole. Once these issues were corrected the system pumps a steady 4 inches of wind.

dcp_2967

By the time the pumps were working, It was too late to cut the wood for the pipe sides. The wood has been re-sawed into thin sheets. These need to be cut into rectangles, starting with the largest pipes in the group.

dcp_2969

As for the player unit, My friend Harold, who got me started on this project, Is making some of the needed items, such as the spool holder, and idler clutch assembly. These will be ready in the morning. So I plan to attend the MBSI event and at least show what can be done in a week.

If there is time, some pipes may be made for the events on Sunday. The nice thing about this project is that even if it could not be done in a week, It will get done, and be a nice addition to the collection.

Day 6, too close to call.

At the end of a long day. Progress is not quite where I would have liked it to have been. This may still happen. I have 24 hours left, which accounting for a few hours of sleep probably gives 17 or 18 hours in which to get something playing.

ideally I would have the bellows pumping wind so the pipes can be made and voiced quickly. Instead, I was able to get the roll player section completed first. Mostly due to help from my friend Harold.

I decided before it got too dark, to saw up some more of the local wood. The band saw blade was dull, while cutting through a piece of the pear wood, the blade broke. I am pushing this saw to the limit. re-sawing the thick slabs of cherry is hard on this tool. Hopefully I can get a new blade in the morning.

Without a working band saw, I am unable to finish the pump sticks. Fortunately I had planned to do some of the gasketing after 9pm. Even though this is a semi-rural area, the sound does carry at night. Although as I was performing the leather work, I could hear a train which was a mile away quite clearly.

dcp_2964 This is an interesting way of making a gasket. I am not sure if this is the best place for this method, but it seems worth a try. This is how Wurlitzer gaskets the regulators to the wind trunks.

First a strip of felt is glued down the center of the frame. Then gasket leather is glued on ether side of the felt. It is important not to get glue on the top of the felt. The only packing leather I have is thick. I think what I used is some extra pouch leather I got for making Wurlitzer valves. Once we get pump sticks and the wind line drilled, then perhaps It will be seen if this works.

dcp_2965 Finally we show the progress. All the main parts are in place and mounted to the sides. I did get a sheet of MDF for the pipe rack. Not sure when I will have time to cut it. I may just mount the pipes onto thin strips of board (I have a lot of that material.)

It will be interesting to see if this can be made playable in time for the event Saturday. Much of the detail, like the sanding and finishing of the cherry sides will not happen. The most critical part that needs to be fitted is the supply spool holder. This got delayed as a 10mm drill was needed. Not sure where the time to make this will happen.

Probably allowing for a week+ 2 days of preparation for this project was a bit optimistic. On the other hand, without the self imposed deadline, It is easy to defer many of the tasks. The important thing is the challenge of what can be done.

It should also be noted, that while the goal was to do the project in a week, the project has had the help of several people. Which is the focus on why this project has been selected, so it can be shared with whomever is interested.

Of course, the deadline of the 16th of May 2009 is artificial. Undoubtedly there will be improvements. Music will need to be acquired, spools made. Thee knob of the handle turned. The case varnished and decorated.

My late friend John Grass, used to say, the first part to make was the hardest one. So with this project I have at least started something.

Day 5, build an organ in a week

All the major parts are completed.

No photographs as I forgot to take the camera out to the workshop.

Harold made the spools and the drive wheel for the take up.

The plastic tubing was acquired.

To save time the top of the pressure box will be a single sheet of plastic.

Optimistically I wanted the bellows and pressure box connected by the end of the day. This did not happen. A few bearing blocks need to be made along with the pump sticks. Having everything else ready to assemble should make for an interesting day tomorrow. It will probably be another long day.