Jump to content

Everything you always wanted to know about the MIMUL motor* (*but were afraid to ask)


Duncan Brown

Recommended Posts

  • Premium Member

The topic name is overselling it at the moment - all I have right now are pictures of the guts.  But as I get farther with schematics, parts identification, troubleshooting, etc. I'll add it here.  I hope it goes without saying that you SHOULD NOT disassemble your motor like I'm doing here.  I am a little skilled at these things and I also have little to lose - the motor was not functioning correctly.  I'm hoping I can get it running properly again, but if not, I was looking at buying another motor anyway.

Duncan

eclair_acl_mimul_001.jpg

eclair_acl_mimul_002.jpg

eclair_acl_mimul_003.jpg

eclair_acl_mimul_004.jpg

eclair_acl_mimul_005.jpg

eclair_acl_mimul_006.jpg

eclair_acl_mimul_007.jpg

eclair_acl_mimul_008.jpg

eclair_acl_mimul_009.jpg

eclair_acl_mimul_010.jpg

  • Like 1
  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

If you can't get it working, it would be possible for me to build completely new electronics for that motor using only the original Thomson motor itself but scrapping all the original circuit boards and replacing them with my new boards.  I would recommend a update which has a small oled display and graphical user interface with freely selectable crystal speed, footage counters etc.  Price range is something around 1000 usd and takes something around 3 to 4 months to build. 

The limitations are that the electronics built into the camera's base are bypassed so built in exposure meters would not work. Attaching the original handgrip would be possible by making an adapter cable which allows connecting it directly to the motor. Power is connected directly to the motor as well.

It would be much more advanced than the original motor and one can select any crystal speed from the full range the motor can handle, not just couple of preset speeds like on the original one.

Link to comment
Share on other sites

  • Premium Member

Next, I will have to say that the idea of not even bothering to troubleshooting the circuitry, and just replacing it, is appealing. 

I historically fix things back to their original state, rather than improving them.  As an example, I'm building up a series of 1970s pinball machine designs that were thought to be lost to history, and I am doing it with the original technology - relays, stepper units, etc, rather than taking the easy route and throwing a computer controller in each one. 

But this positively stone-age digital technology (double sided boards, through-hole components, TTL logic chips, giant crystal) could largely be replaced with a board the size of a postage stamp, possibly with less effort than trying to understand and repair what's currently there.  The only hard part might be the motor driver circuitry, but that's already there and presumably in OK shape, given the behavior of the motor.  But I will make a first stab at documentation, troubleshooting, and repair.

The motor currently only has two speeds (very slow and very fast) depending on the position of the dial, neither one of which is synced properly (the light stays on, and my phone tachometer app shows them not to be precisely on any expected speed.)  My first thought was that the switch wipers on the circuit boards were crudded up...only to open it up and find a sealed, expensive switch that's probably good for millions of operations... so that's probably not going to be the problem.

Then I thought the problem must be with the circuit board connectors.  For repairability reasons I am very thankful they connectorized the boards, but over decades of lmicrovibrations, connector contact points can fail - usually fixed with nothing more than unplugging and replugging them.  But then I see they used gold-plated tooled pin and socket connectors.  It's still possible a connector failed, but far less likely with those.

In other words, this beast was very well made, sparing no expense, and designed with harsh environments and decades of use in mind.  Well done, Eclair.

Duncan

  • Like 1
Link to comment
Share on other sites

6 minutes ago, Duncan Brown said:

In other words, this beast was very well made, sparing no expense, and designed with harsh environments and decades of use in mind.  Well done, Eclair.

I guess Thomson-CSF -- a defense contractor! -- wasn't a bad choice ?

Link to comment
Share on other sites

  • Premium Member
5 minutes ago, Heikki Repo said:

I guess Thomson-CSF -- a defense contractor! -- wasn't a bad choice ?

Do we know for sure they built the entire motor?  Their name is on the actual motor component, but I suspect anyone could purchase motors from them.  There is no question this was built using "mil-spec" parts though.  I'm still trying to figure out what the Texas Instruments 84xxx series of chips is.  I'm sure they are equivalent in function to the same numbered 74xxx series parts, and their official mil-spec versions were the 54xxx series.  Might have to go dig my old TI data books out of a box.

Duncan

Link to comment
Share on other sites

10 minutes ago, Duncan Brown said:

Do we know for sure they built the entire motor?

Good question. The lack of schematics for them would point that way. After all, there are surprisingly comprehensive schematics for the early ACL, even if not for the light meter. But nobody admits to having seen anything on the motor electronics.

Link to comment
Share on other sites

  • Premium Member
6 minutes ago, Heikki Repo said:

Good question. The lack of schematics for them would point that way. After all, there are surprisingly comprehensive schematics for the early ACL, even if not for the light meter. But nobody admits to having seen anything on the motor electronics.

State secrets.  Need-to-know basis.  If I recreate them, I will probably be whisked off to The Village.  (Be Seeing You!)

Duncan

  • Haha 1
Link to comment
Share on other sites

  • Premium Member

I suspected the 1536000 Hz crystal was chosen because it divided evenly for all their speeds.  Sure enough!  I did the math so you don't have to:

8fps = divide by 192000

12fps = divide by 128000

24fps = divide by 64000

25fps = divide by 61440

50fps = divide by 30720

75fps = divide by 20480

Duncan

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member
28 minutes ago, Phil Rhodes said:

It's better than that.

1536000 / (24000/1001) = 64064.

It even works for 23.976!

It's an easter egg!  Too bad there is no secret button code to access it...

Duncan

 

Link to comment
Share on other sites

  • Premium Member

normally this type of systems can't manage with only one pulse per frame from the encoder so probably it is more complicated than that math.  Additionally it needs to be dividable from the crystal using the 70's technology without making it too large, I think they did not have programmable counters available yet so one needs to divide using binary + decade + ring counters which easily makes multi-speed system less than ideal in size (too many ic's needed) . It is possible that they managed this by using complicated multi stage rotary switch setup which manages the intermediate frequencies so that it makes possible to generate the reference frequencies with fewer ic's used.

generating the reference frequencies accurately is the easiest part of the system though, both back then and today

Link to comment
Share on other sites

  • Premium Member

when trying to repair the system, the easiest way is to follow up the signal paths with a oscilloscope until you find out where it breaks. you need to find out which wires bring the motor encoder signal and the speed reference signal to the phase frequency detector and then test with 2 channel oscilloscope where the signal changes unexpectedly. It may be a pfd issue too but you want to start from the beginning of the signal path (from the motor encoder and from the crystal frequency generation board) to see the first break in the signal path

Link to comment
Share on other sites

  • Premium Member

A little scoping around on that top board with all the SN84XXX series chips gives the expected results - the SN84L93 chips act just like a 7493 counter chip, so I still don't know what the 84 series is, but at least all my 74xxx data sheets will work.  And those counter chips give more and more divided versions of the 1.536MHz source frequency.  Interestingly there are some where the duty cycle is wacky and the frequency is not divided precisely; that may be part of the root cause of the motor running a bit slow and/or the fact that it will only run at 2 different speeds.  I'll know more once I trace out the complete circuit, but this gives me an idea of where it might be going wrong.

Luckily, this frequency-dividing circuitry just runs freely when the circuits are powered up, regardless of whether the drive part of the motor has been told to be on or not, so I can work on this without the motor flailing the whole time.

Duncan

Link to comment
Share on other sites

  • Premium Member

Some additional pictures I've taken along the way.  One of the transistors that are mounted facing the board pulled up a bit so you can read the part number.  And a bunch of the switch to document how all the fiddly little wires attach.

 

eclair_acl_mimul_054.jpg

eclair_acl_mimul_055.jpg

eclair_acl_mimul_056.jpg

eclair_acl_mimul_057.jpg

eclair_acl_mimul_058.jpg

eclair_acl_mimul_059.jpg

eclair_acl_mimul_060.jpg

eclair_acl_mimul_061.jpg

Link to comment
Share on other sites

  • Premium Member

Progress report:  I've been slowly working through stuff, mostly on the top TTL logic board so far, making some progress.  That's some convoluted logic there, that's for sure!  I'd never run across a "3-2-2-3 AND-OR-INVERT" gate (7454) before.  So far I've only broken off two wires (one of which I'm 100% sure I know where it goes, the other one I'll have to go back in my pictures and hope I can figure it out) and blown up one transistor, letting the magic smoke out (it's on the backplane board, and I hope when I remove it I'll be able to read a part number on it), which is pretty much how these things go, so all is well.  But it's why I like having a spare in hand before I do this!

Is there anyone out there who can explain to me why a knob with 5 marked positions (75/50/N/12/8) is attached to a switch that has 6 positions?  I mean I guess it's way easier to buy a 6 position switch than a 5 position switch, but...   As near as I can tell, electrically, it's basically a position under "8" that does the same thing as "8".  Can anyone with one of these motors (that works!) verify that's how it behaves? 

Geeky details of what I've figured out in the next post, so you can skip that if it's going to make your eyes glaze over.

Duncan

Link to comment
Share on other sites

  • Premium Member

I've been working at this from a few directions at once, switching when I get frustrated or bored with one and moving on to another.  

The switch is made by EBE but uses an industry standard naming scheme.  MX 2/4x6     MX is the size (17mm), 2 is the number of wafers (ganged switches stacked one on top of another), 4 is the number of poles (electrically separate switches, though they all move together) and 6 is the number of detent positions.  The switches have 12 possible detent positions, so by mechanically limiting it to 6, you can have two switches operating together around 180 degrees each of the wafer.  Then with two wafers, that's where you get the 4 switches.  There is a "common wire" ring for each wafer that the switch connects to each contact in turn as you spin the switch.  The common ring has two breaks in it, to make two 180 degree common rings for each wafer.  That's a standard off the shelf part.  In this case, they wanted to use the same common wire (+5V) for two of the sets of switches, so they bridged those gaps again with solder.

The upper switch (nearest the knob) has one part that sends the 5V into 4 places in that convoluted logic, to select the speed 50/25/N/12 - apparently the default behavior of the circuit with no 5V signal being applied to any of the gates, is to select the 8 speed.  Since the last two positions have no wire attached, this is where I get  the notion that the position below 8 is just 8 again.

The other upper switch sends the 5V one place for 50/25 and another place for all the others.

One of the lower switches does the same thing.  This is clearly involved in the fact that my motor went a fast speed at either 50/25 or a slow speed at all other positions.  It's like there is a different way the circuitry controls the motor for the upper two speeds.  The lower switches use two different common wires, and send their selected signals off to more motor-y things, not the logic things, so I haven't chased that all down yet.  I will note that there are a LOT of wires going into that motor case, so that's going to be a whole other level of complexity to tackle at some point.

There are 3 7493 4-bit binary counter chips.  The first one simply takes the input frequency from the crystal and divides it down.  I think the net result of that is a 48KHz signal feeding most of the rest of the logic.  The other two work together to divide down the frequencies even more, with various divisions fed into this convoluted logic in different ways, but the main clock input to those comes from the little daughterboard attached to this logic board, so that frequency changes with the positions of the switch.  That daughterboard is labeled "oscillateur à rampe variable" which google tells me means "variable ramp oscillator" so that all makes sense.  Unfortunately, the two chips on that board have no markings on top, whether on purpose or not I can't tell.  So that will be interesting to figure out schematically, but I can certainly investigate it with an oscilloscope (if I get it all running again!)

One thing that makes it hard to do all this is the fact that everything is plastered with a "conformal coating" - clear goo that hardens and seals all the exposed metal pins and leads and traces away from the elements.  Great for longevity, not so great for connecting meter or scope probes to parts.  The secret is to do it from the back, where all pointy leads and pins are easily reached through the thin and easily broken coating.  But it makes for a lot of flipping boards over and back, which is where wires start breaking.

There are two components on the logic board that look like wirewound resistors and are in series with the logic signals, which makes no sense unless they are delay lines - an analog method of making signals have the desired timing relationship.  That would be weird, but this whole design is a little weird.  I cannot measure any conductivity through them.  I sure hope that doesn't mean the voltage from my multimeter torched them?  That would also be weird and make no sense that 5V would be fine but 9V would be fatal...but in order to do a delay line you need a ton of very tiny wire, so anything is possible I suppose.  More investigation needed on those.

Duncan

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...