This is one of those Chinese CNC machines on eBay that’s got all those qualities where you think, maybe its good enough for the price. Of course they never are but that never stops us. $700 all in. CNC 3020 will lead you to them, but really don’t.
They generally are sold as shipped in the USA and actually drop shipped from China. Takes about 7 days usually.
Very well packed, but as usual has that weird smell that everything we’ve bought has from China. So dispose of the bubble wrap as quickly as possible.
Cursory glances the power supply has what sounds like a loose screw rattling around, the machine looks cheaper than the photographs, but then again who expects to get a hamburger from McDonalds that even barely resembles the pictures on the advertising. You know what you’re signing up for when eBaying, Las Vegas rules here. To be fair to the Chinese companies though their advertising pictures are about 9000 times more accurate than the McDonalds ones though. It also looks smaller than it did on eBay, this is why size reference pictures need an apple.
Apparently it uses something called a ‘Parallel Port’ (I keed, I keed), well that won’t last for long.
Our LPKF is a lot nicer, (not as nice the AccurateCNC), but it is going under the chop too.
So it been a while since my last update, my lil’ dog of 16 years had to be put down and I wasn’t feeling it, then family visited from so we stalled abit.
So quick updates for the tl;dr people :-
Head rotation works, but have to add an encoder wheel since we ‘forgot’ about the tool home position inside the head tool pickup…
Y axis is all upgraded, new stepper motor drivers are in 1/2 step is on.
Switched to GRBL and StepDir, instead of, uhh, not stepdir.. Though using the
OpenPNP base firmware revision, i didn’t realise it was different til after porting it.
GUI is almost done, it can place one full board from feeders and trays
Replaced hoses as the old ones burst, which only happens when you say right lets video this.
Trapezoidal for the servos courtesy of Grbl.
Still having some issues with the vacuum sensor, might be bad or low pressure or something else, turned it off for the moment.
So we’re able to pick up and place with the new Grbl firmware and the updated GUI, its all GCODE now. I’ve moved more of the functions into the GCODE vs individual move, drop , vacuum, since the communication is just serial rx/tx and serial is a PITA for this sort of thing, since its all ACK/re-ACK. With an M command its just e.g. M21 pickup part, M22 putdown part and the machine reports yes/no. FTDI’s are decent chips but they will break down comms after long datastreams, I’ve had a lot of problem with them and high speed streaming, as well as overlapped IO support is meh, events don’t work that well so multithreaded serial comms based on events is in theory great, but the practise doesn’t always work well. so its all software handshakery.
We’ve done a lot of changes since the last update.
Grbl’s great, its easy to use and easy to add on too, we were already using the AtMega2560 anyway so lots of space, even with all the stuff I’ve added to it, we’re at maybe 17%of the flash space. I had looked at it when we first started out, but it needed StepDIR and were using – + , our original drivers didn’t support StepDIR. Even luckier I’d put all the stepper controls on the same ports.
Changes for the 2560 are minor, just an interrupt rename and i don’t the sleep mode is working as it ought too, i assumed it was and then wondered how to handle the panel switches but turned out when i added the head movement ACK that it was, I was going to use timer3 to handle switches but then i noticed the main loop was executing constantly, some of the AVR’s have different sleep rules.
One thing i did do though is since the mega ICSP is under the shield and its on the machine, which is 3 metres away from desktop. I left the Arduino serial boot loader on the chip, this way i can just script AVR Dude to load the firmware built in AVR Studio 4 to the mega over serial. so its a lot easier, the downside is that if you press a key too soon the boot loader seems to kick in, i haven’t checked this to be the reason for sure, but it seems like its that. Not a huge deal.
Running a test of the GUI
Pick and placing, it takes a lot to just do this seemingly simple thing.
Head motor, luckily we used a double ended shaft motor in case we needed an encoder, which it turns out we do. Look closely at the top of that shaft, see some thread? yeah we didn’t…. We CAD’d up a mount and fitted the motor.
The motor is 1,000 steps for a full 360 degree rotation, should be able to place the craziest of our boards, we like some gangsta lean in our part placements.
As a side note the magnetic sensor on top of the head, really don’t like the way they’re made or mounted, that wire takes a beating and it’ll fail again so we have to fix that too. You can see all our temporary fixes for the hoses that keep cracking, they’ve lasted 20 years though. The red line is the new one we’re using, which is way more high tech than what was available in those days, it should last longer than the tech is needed.
The rotation head needed a larger power supply, so that is now installed. Its the 24V power supply previously mentioned from all electronics.
Since we made the shield with the extended pins its easy to add things on to it. And now we’re on StepDir its just one wire to control the head ( Though we make a semi-fatal error at this step in choosing which direction the motor runs )
Setting up a test
After the rotation head was added to the software, i test placed two parts, they’re off, at this point i think its the solder paste or tape missing, so we add some. As you can see though the parts are rotated as they should be. We were uStreaming with the second camera so its all hanging off in space at the moment.
Krs adds some paste to the test pads. She’s the one who wants this thing finished the most as she builds most of our boards
I’m remotely moving the head with the camera view. First home, then moving to the PCB to register the location of the part, then going through the steps of the part placement.
So the mistake in the choice of direction for the head rotation is that the head can unscrew itself and drop down the tool, the sensor doesn’t know the head is down and the tool is now below the PCB holder and auto changer, so the head will gleefully zip around and bang it into whatever it can find, which bends the head, which is not good at all. So we’ll reverse the direction of the motor and use lots of Loctite. This is why the test videos show the parts poorly placed.
Pity these lessons learnt cost us a head each time.. But we move on..
A trial run through all the parts ( using 1206 resistors since they’re like .00004 each) as you can see they placed perfectly.
Ok, maybe not what you’d call perfect but they did place in the right spots, the head down motion was a little too strong and there’s no paste to hold the parts in place so they’ll wander around every time a part is placed. Still the software is working.
The vacuum sensor needs to be set quite accurately or it’ll do things like this :-
The tool is sitting perfectly on the vacuum test pad, the head brought it down to see if was there, blocks the air coming through the tool, but since it wasn’t picked up correctly ( head wrong orientation ) it just left it there.
The head moves to the pad, goes down, the air comes on, there is a settle time, then the machine reads the vacuum sensor, if its not blocked there is no tool, if it is, then there is a tool. That is how it decides if the tool needs to be picked up or put back.
Finally we place a board with paste and all !
So last night(this morning) we finally managed to run a real board. For some reason i decided to go through each of the components and rotate them 90 deg rather than change the rotation in the feeder…. derp… As we did this the vacuum test failed, so you can see me going in change the code ( incorrectly!) and re running, it the bit you might not see is the machine doing a loop of picking up a part… i fix that and we move on.. First run we do a step by step. I noticed a few bugs as we go along so there are some pauses as we fix those.
Stencil pasted a board, you can read about this in our Arduino Mega build blog entry. Its a CNC’d brass sheet.
I taped the board down to stop it moving around, probably not needed.
Some of our feeders need some love and i wasn’t super careful about where i was picking up the parts from so it was a little sloppy, however it was 2AM and i ended up having to leave and come back after one of the alarms at work went off
MMCA used one of his awesome swiss files to ‘fix’ one of the feeders that was catching
Here are some more GUI videos, they’re probably only interesting to those of thats like watching these things over and over..
Annotated some of the steps in the GUI.
One of the feeder pickups i didn’t setup properly, so it was placing them off!
Onto the hotplate, i wanted to see just how bad we could place parts and if they’d fix on the plate, all but one pulled onto the pads, the other worked but was right on the edge of a pad, it’d work fine. All we have to do is take more care when tell it where to pickup
Krs also has a bunch of videos of the machine placing that first board.
Anyway that is mostly it, we’re down to the tuning and bug fix stage so its been pretty fun.
But if you’ve got minerals help out the OpenPNP guys they’re doing good work, our stuff is available to everyone, but its reasonably specific to the Juki though it is GCODE.
Small update: Last night, I started adding the second USB port of the ADK to Grbl as well, its almost finished. I based it on the usb.cpp from ADK but i converted it to C since the c++ generated a larger firmware and there is only one USB port, I want to keep the firmware tight.
BTW If you’re reading this and are on the yahoo zevatech list the moderator who’s apparently building a commercial retrofit for this machine is blocking our posts sharing this open source project. He said he wants credit for a CPU suggestion neither his or our project uses, a fix that wasn’t broken, and a camera solution we didn’t ask for advice on, or use.
So we’re on stage III of the rebuild at Null Space Labs (http://032.la) . What we’re aiming for is :-
Rotation servo in the head
Addition of more checks
Adding more checks was straightforward, added code to determine if the machine picked up a part, with the small compressor we were using it’d start to loose vacuum so if it failed to pick up a part, the delay became longer. This allowed the the compressor to catch up a little, as well as alert the user that something was amiss. Next it checks to see if the part fell off during transport to the board. The machine has a vacuum sensor that knows if the top is being blocked. This lets the machine controller (Arduino) know a part is probably there.
The machine won’t move if the head is down, and the limit switches were put on interrupts. tool changers are tested and put down if they’re up for some reason.
The interrupt based limit switches are nice, anytime a change of the signal is recognised, either in limit or coming out of limit the routine is called and it reads the status of the pin again, and sets the out of home condition, so that machine controller knows the machine should be rehomed before continuing. The code is simple and looks like this in the setup()
attachInterrupt takes the pin number , the function to call and type of signal to look for, HIGH, LOW, and CHANGE, i use CHANGE so that the xLimit1 variable is always set properly, and rather than rely on software tracking I read the state back in.
I’ll probably change the digital read to actually just read the PORTx value directly, you want to minimise time spent in an ISR (interrupt service routine ) as much as possible. 5 is the pin that the XL1 limit switch is connected to on the shield, and homed/xLimit1 are volatile global variables that have the condition, they are set as volatile since they can be changed at unexpected times, either by the hardware directly or the ISR. volatile lets the compiler know not to cache the value or other trickery. This is often the case with things like memory mapped hardware clocks where you read a specific memory location to read a clock, so you’d do something like (clockVal would be set to the specific memory location by the linker or other directive)
const volatile unsigned long clockVal;
We’re not using const for the homed/xLimit1, since we are writing to it via software, a hardware clock changes outside of our code.
Someone knocked one of the feeders slightly off (probably me) as it was being worked on, I didn’t notice and moved the head.
It caught the feeder pin ( bottom left) and bent badly enough that it wouldn’t go back down. So i pulled it apart and used a vice with cut outs and gently squeezed it back into shape, then mmca used his micro lathe to make it straight again. Luckily these are easy to make from scratch.
The servo pulse speeds are more of a pain, the Arduino has some fairly inconsistent timing and some of the functions you’d expect to be fast, aren’t, the delayMicroseconds has a lot of skew, digitalWrite takes a long long time. Once you know about them, they’re easy to deal with i replaced them with bitSet/bitClear and the __builtin_avr_delay_cycles feature in avr-gcc. We captured the output of the servos from the Juki PC with a scope and my logic analyser. That allowed us to see the ramp up and down times to get the head moving at a decent clip. There is still some ripple in the software PWM though, so we’ve converted most of it over to just AVR asm/C and setting the ports individually,the Arduino does what it is meant to do really well and that is be easy to use, but what’s uncommon about it is that it easily lets you drop in code that bypasses or override the built in libraries, in my experience that’s rare in other simplified dev kit software, its typically all or nothing.
Changing the digitalWrite is straightforward. Determine which port the pin you want to set is on and then use bitClear or bitSet instead.
digitalWrite(YCCW,HIGH) becomes #define YCCW_HIGH bitSet(PORTF,3); YCCW is defined as A3, which on the mega is on PORTF, bit 3. digitalWrite LOW is bitClear(PORTF,3) going even further the PORTF can be manipulated directly with PORTF |= (1<<3) to set it, then you can combine them together with PORTF |= (1<<3)|(1<<2) etc to clear PORTF &= ~( (1<<3) | (1<<2 ) ) also there is a _BV() macro that does the (1<<n) shift. so
PORTF |= _BV( 3 )
Using |= and &= should output the closest ASM instructions that you can could in C.
Once we started to investigate the timing from the Arduino that is when I noticed the problems, even with interrupts disabled we were seeing inconsistent results. So I wrote some test code and hooked up the output to the LA and set about trying to make a micro second accurate delay. specs showed a 2uS rise , 5uS pulse and 2uS fall for the initial step pulse, the stepper IO controller is HIGH delay LOW delay HIGH . for the first delay about 8uS works well.
We captured the data with the logic analyser and pulled the data into Excel which is great for this sort of thing.
A quick couple of macros and copy/paste and we’ve got differences in Column F , you can see the length of the pulse in F3 and F4 has the off time length, which gives us our full duty cycle for that pulse.
In H its just the lengths of the off times since the on time is constant. Column I has the differences.
One of the next things excel can do is trend lines, I did this by hand and then arko showed it to me, I’d seen the equation solver before but not this.
First create a line chart with the off times (the part of the duty cycle that’s after the pulse)
Click on the data points in the line to select the line. then use the trendline option
Set the options like so :-
Which gives you this trend line (in black ) and the algorithm for it. I opted to use a table lookup since it has to be super fast for the Arduino.
Skipping back a little after the last blog post we did an 0603 repeatability test, used the layerOnespeaker badge as a base, I exported the centroids from eagle for one side and ran test code. As well as just have the machine draw lines of 0603’s. with a logo-esq list of commands.
The left side has a little wobble, partly pickup, partly the slightly bent head ( we’re lucky that a generous person sent us a new head and feeders you rock Steven ! ) the right side is a similar test but its placing the 0603s on top of each other, that actually worked pretty well a few of them fell over. But given the bent head, no centering or machine vision its working better than we can hand place them. This picture is the version that just places the parts in lines. I also think our stop is a little aggressive in this test, and this is before we really got into the ramping.
The ramping is simply having the PWM have a longer off time, pulsing the servo, decreasing the off time, pulse it again and do that for N pulses until the motors are moving fast enough to get to the top speed, that’s what the Excel sheet is for. I’ll go into it in more detail in another post.
Someone from Russian TV decided to film it, Central TV i believe. Unfortunately all the airlines are now tight so i can’t remotely shoot air or disconnect hoses at him anymore.
Adding a camera
This is a Cognex 4100 machine vision camera, machine vision cameras are generally low noise, stable image sources. This one can determine angles, and so on and then send them out via Ethernet or serial. The 4100 has been dropped by Cognex it seems and the software needed to run it is an old version, which has a lot of compatibility errors with Windows 7, its been something I’ve seen before where the menu’s lockup for a long time. But we figured our way around it by using Windows XP inside 7 with its builtin Virtual PC and got it to see the rotation. Think RoboRealm built into a camera.
Inside the camera, TI DSP and a Xilinx FPGA, very nice.
We’re also using some webcams , a couple of them are microscopes, one is from think geek and it’s a wireless HView camera, the other was an eBay special pen microscope usb2. Oddly finding a 12mm drill bit to make a hole for the camera wasn’t so easy.
Adding the camera needs two pins for the servo CW/CCW, three pins for the camera, TX/RX and trigger. The shield has three pins left!, Though I did cheat and use two of the pins as a via since I didn’t’ really want to deal with the LPKF via rivets even though its only a few of them. We do really only need RX though. The trigger just tells the camera to only operate while, or after the trigger is set. The final board will not use the cheats of course, but before I remill a new one we have to
This is our test bed software Pickobear it allows us to test OpenCV (though we have a Cognex camera), the eagle script output CSV is imported into Pickobear and then the machine knows how to place it.
One of the problems with test software is you often leave it in a state where it was testing something specific, and you forgot to remove it , case in point, I had one extra YCW pulse that caused a problem when we were doing a tool change test. After 4 changes it’d refuse to put down the changer, we checked air pressure, offsets, tool head distance and all that, eventually did test code that moved home, then to 0,0, 1000,1000, back to 0,0 and saw it was one Y off each time. A quick check of the firmware and the extra pulse was spotted. But it did allow us to get the numbers needed to exactly place the adjustment for the pickup head.
Next we’re testing out different cameras, I milled out a quick holder from PCB material and the epic blue painters tape. This is a nice’ish USB 2.0 camera that’s setup as a microscope, it came with a nice metal holder and was about $80 from eBay. M4 nuts and bolts seem to be rare around us, never mind the M2.5’s we were looking for locally last week. So zip tie to the rescue temporarily, it does allow a little yaw though but it is ok for testing the software until the bracket arrives.
To make this I measured the hole sizes, the distance apart and picked a distance for the camera, then guessed the size. I then added two drill holes with the hole command, set the size to 4mm for the larger 12mm hole I used the milling layer, draw a circle with the circle command, and set the width (right click circle and choose properties or type change width ) to a very small number, if you use 0.0 as the width eagle will fill in the whole circle and doesn’t cut anything, a width of 0.0 works for milling wires, but not circles. Then milled it out on our AccurateCNC 560H.
Now we fit the new holder, has a nice set screw to hold the camera in place.
Next is adding camera movement and offsetting the board. I did this by adding a mouse click even to the custom control, it then determines the center of the image and a difference calculation, then depending on the direction it needs to go it moves that amount, I still have to calibrate what the relationship of the camera image is to actual axis steps, but its close enough to use right now.
I’m using videoInput with OpenCV and OpenGL to display the data from the cameras, the crosshairs and circles are drawn into the OpenCV image using their primitives.
The rotation head is the next thing we’ll get fully operational. We decided to upgrade the 24V power supply since the little servo can draw more amps than it could safely handle. All electronics has suitable 24V 6.5A deals.
A custom CNC’d bracket was designed by mmca and laid out in solidworks by arko, the motor, pulley etc are all added. This allows us to switch from the simple 90o rotation to full a 360o’s so we can place the part turn 360’s and moonwalk away from it. There is some play in the rotation shaft because of the set screw, so that is being remade so there is no backlash., even though its a tiny amount and the software can correct it. The machine is able to do more than 90o in steps of 90o by simply changing which side the feeder is mounted too, but we want 45o’s for those gangsta leans.
The top bracket/reed switch isn’t mounted or screwed down, that is why its tilted.
New standoffs added and rest of head mounted.
So here is a short video showing the setting of the PCB offset, I home the machine, pick the part I want to use a registration. Then move to the location where it really is and press the OFFSET button., then I move around to other components just to check its working, again the blue lines are OpenCV doing stuff.
Motor Drive Upgrade
The Y motor driver and motor aren’t that great, so we changed it out for this controller
First removing all the old wiring and converting it to the style connectors, it’s a shame to do this as the wiring is just so nice compared to modern equipment. Its not often you see this level of attention anymore.
New controller in place of the old one, it fits pretty well its longer than the old one but luckily there was lots of room, we just have to extend the AC lines a little.
Removing the plastic conduit covers and rerouting the wiring.
Almost there, I’m heading out early tonight and we apparently don’t have crimpers at NSL so rather than botch it, I’ll pick up some tomorrow or bring mine in from home
A new motor was fitted too.
Close up of the belts and cogs etc.
New vexta motor
Machine powered up and homed ok ! sweet..
I’m not sure how detailed or long to make each of these posts, since from reading the comments and emails, its obvious a lot of people skim or don’t read the text and look at the pictures, but I want to make sure we give enough details ( plus the wiki/svn ) so that people can recreate it. We are at the point we can pick and place our boards, surpassing the original software in some ways, a few things are left to be done to make it 100% of the original features sets, mostly to do with feeders. But we’re waaay beyond what it could do in some ways.
Rather than hand build all the badges for our socal security conference layerOne again, http://www.layerone.org/ we’ve gone to a pick and place machine.
Gleep found us a Juki(Zevatech) Placemat 360 (that seems to have been upgraded to a 460 ) pick and place machine. It was sold as ‘working’, the sellers definition was, if I’m completely honest a stretch (outright lie).
This is actually our second pick and place machine, we don’t mention the other one
We’re also interested in acquiring a Zevatech/Juki 460 if you have one for a decent price.
Basically he demo’d everything that didn’t need a compressor, that all worked. Of course everything that needed a compressor as we found out later, didn’t work! Still $1,200 isn’t bad.
I used my supersilent 20a as a temporary compressor, it only has a small < 1 gallon tank, but its actually quiet, we used the 8 gallon compressor at null space which is deafening, so i found a 3 gallon temporary one at harbour freight for cheap in their recent sale. its too small though, so we’ll need shop air at some point. The supersilent was causing the pickup head to fail to work after a few passes, so this caused as a few false starts, the machine needs a solid air supply to function , even in testing.
The existing filter and pressure regulator was a mess, so off to home depot to come back with the best we could find there, which isn’t that great.
This is the old one, remember sold as working. No filter, and all these bits were just lying around inside it.
The machine itself is based on the PC-8801 Z80 4mhz CP/M which I recognised straight away as my old job had me doing game conversions in Japan for the PC-9801.
The whole machine works pretty much on the principal of that if the CNC software said do this, do that, that it executed perfectly. Only limit, head, home and the tool changer have checks.
We fired it up , Krs and Gleep got it picking and placing a few resistors (though they somehow managed to get the tape removal part completely wrong and it was throwing resistors all over the place. Then mmca got it placing QFP parts correctly. The lamp spot system was off, the 90’o rotation was off, the tubes were old and cracking. Compressor filter was non existent and rusted out. We’ve also discovered the whole thing is covered in parts from the previous owners, we’ve scored a few 100 0805s and some IC’s.
The reed sensor was the first thing we found that was broken, a quick trip to eBay and a few days later we had replacements. Luckily Juki is in heavy use, and they use a lot of off the shelf components. Apparently the later 5xx machines do switch to a proprietary drive system.
The reed switch detects if the head is up or down. Its one of the few sensors in this machine. The bend has caused the wiring to break down internally over the years. so the machine gets confused about being up or down, and the software doesn’t cope well with that, it basically needs a full reset afterwards.
The new sensors , $9 from eBay.
I also bought a CPLD based floppy emulator from Poland, it hasn’t arrived yet and we’ll probably be done with the new system before it gets here, and we’ve discovered the speed stays the same but floppy drives won’t last so the SD is still a good replacement.
Placing QFPS (AT90CAN128)
Fashioned a quick tray for the IC placements. We use these great little boxes, also from eBay, for holding SMD components, they double up as handy platforms too.
The feeder is controlled by the head, it moves over the spring loaded pin and pushes it down, this releases air and the notched wheel on the right moves the component reel tape one step, at the same time the protective covering tape is peeled away, allowing the machine to come back and pick the part up. This time, they’re correctly threaded, previously the protective tape was wrapped around the pin in the middle.
Side view of feeders, you can see the reel of components on the left, and the pneumatics underneath. Its important to choose a pick and place with a widely available range of cheap feeders, all too often people buy a cheap pick and place then find out it has none, and it’ll cost $1000’s to get them, if at all.
Feeder with pneumatic assembly
The expansion board
This is the board inside the machine, it is a couple of 8255s which are the defacto standard for PC parallel IO, almost every PC has had one or more of these, they’ve since moved into the ASIC’s but the principle is the same. It memory maps each of the input/outputs of the machine so that host PC can see them. I pulled off the floppy image, copied the files to my PC and reverse engineered the controller code with IDA.
I found an IMG of the floppy online, this was MFM encoded . So i converted that to a raw binary file, and then used cpmtools to copy the files from it. I was hoping to find some of the saved files so we could reverse the format and write a quick tool to do the placement. Once the files were copied off i tried a few of the different PC-8801 emulators, M88 etc, but had no joy in getting it running. So finally I just pulled apart the CP/MCOM files in IDA and see what we could find.
The teaching process is tedious, so reversing the format would have been worthwhile.
This board takes the IO from the PC, buffers 74LS240 it and uses power darlingtons FT5723M to switch the 24V signals for the pneumatics. As well as read the various sensors and the + / – for the motors. The motors and stepper drivers are off the shelf, but very nice, we even have newer versions of the motors and controllers at NSL.I’ve removed the bottom connector to make it easier to take pictures.
The grey cable that has been added later is the automatic tool changer, this is soldered directly in the spare connections , 5V and 24V VDC. The 5V powers the small adapter board in the ATC and the 24V is for the pneumatic switches.
The remaining signals are multiplexed IO that are demuxed by a 74Ls138 on the ATC board, which deviates from the way the rest of the board works as the rest are all controlled by the darlingtons directly.
Each function of the machine is basically <control> – <buffer> — <pc> – <memory map>
So if you want the head to go down, you flip a bit in the PC’s memory. Its all digital IO, nothing fancy at all. The only extra part is the 5V TTL to 24VDC for the pneumatic switches.
Stepper drivers and power supplies.
The stepper drivers are on the bottom, the other one is to the right under the tray. the two power supplies are just visible at the top right, one is a 5V the other a 24V. The power filter is in the lower left.
Stepper motor driver
Since the machine was in bad need of service, we stripped it down, here the XY belts are visible. The top side has the the driver motor and the bottom side gets its power from a rod under the bed on the right side, so both belts are moved in unison. The ATC is in the top right and the frame in the middle is what is left of the PCB holder.
Tearing it down.
mmca stripped the head down. here it is removed from the gantry. mainly because there is a piece of string visible , and we can’t figure out what its for.
Shims, we don’t think these are factory shims.
The strange piece of string inside the head… What could it be for?
Bottom view of the tool pickup and the 90’o rotation.
These 4 arms are moved towards the part and clamp it gently, this straightens the part for placement, it can also rotate the part by 90’o ( which sucks for us because i always like to put parts at 45’o)
The laser, focused lamp (this machine continues to surprise us ) which is used to position the head in teaching mode.
We’re removing the lamp and replacing it with machine vision, so some measurements are taken.
The hoses are removed and marked with a letter , the corresponding connector is also marked with the same letter.
This is how the previous owners repaired the 90’o rotation arms….. so that explains the string. this was removed and repaired correctly. The 90’o does just that, it rotates a part by 90’o that’s all this machine can do, so we’re going to change that to it can do arbitrary rotations.
This hose had cracked, a few others did too. I found a few temporary replacements at the auto parts store 4mm ID, 8mm OD fuel priming line. The plan is to replace all the hose.
Stripped machine screw in the head. Replace from grainger, M3x8mm 0.4mm thread 5.5mm head size.
Spent some time measuring all the screws and what not. The machine is old enough that it came from proper manuals with circuit diagrams.
We’re replacing the IO board, the plan is to throw in a TI Stellaris ARM lm3s9b96 chip instead, (TI were good enough to send us a bunch a while ago, thanks TI!)
This board is a dumb board, it just marshals the I/O and does the switching of the 24VDC with darlington’s.
Here we’re removing and verifying the connector sizes and function (the manual had some errors) so its good to do that. It also gives us good insight into what’s going on.
Checking how the machine works with my trusty fluke.
I threw together the connector layout in eagle and printed it out to verify it, early revision.
Measure the hole size and distance. Our board is exactly the same size so its a drop in replacement, we’ll just lose the two larger connectors and change it to USB.
Here we were figuring out how the ATC worked, at first it was though to drive it directly , but there weren’t enough wires. So its 24V, 5V and control signals, the small interface board at the front is a 74LS138 decoder/demultiplexer with a few buffers and more darlington drivers , it switches the 24V on and off based on the 4 control signals coming in.
Automatic Tool Changer
The tool wanted is lifted up when the machine wants to change it, on the right are the pneumatic switches that are controlled by a 24V signal.
We’re using Power MOSFETs to control the 24V instead, a 6 pin ROHM US6K1DKR in a TUMT6 package ( time to create a new device in eagle again !) I ordered 100 from Digikey yesterday and should have the board layout finished today. Then we can mill out a test PCB and see how it works. (parts arrived a few minutes ago!)
You might be amazed, I was , about just how simple this machine is, you could run the whole thing from a set of on/off switches, albeit very slowly. But that is great for us though as it makes it very easy to replace the PC software.
The next big thing is going to be testing the new power MOSFET and building the new PCB. The chips will be here today have arrived.
So new eagle package
Cut out a few to test.
Apparently I goofed on the measurements, I did change it around a tad after the first revision. Teeny part.
Soldered it anyway
So the next step is adding cameras etc.
mmca explaining the new part to be cnc’d out for the camera
mockup of the mount
The head has to be recalibrated so the bottom of the tool is 62.5mm from the table, with a .1mm accuracy, so we as usual went overboard and used grade B gauge blocks.
69.5mm to .00005 inches accurate.
Gauge blocks are fascinating, they stick together like magnets if you put them together by making surer there is no air between them, but if you just stick them together they won’t. Super flat. these aren’t grade a or better, but they’re nice. mmca has the coolest stuff.
Starting to rebuild it
Machine vision tests
This is work in progress, testing RoboRealm/OpenCV and teaching it components, it works well!
Using a panda board a HP HD Webcam for testing the vision.
Playing around with layouts for a quick test tool. two grey areas are for the cameras.
Well that is it so far, my Motorola Atrix decided that the fingerprint reader would become burning hot to the touch. So I pulled it apart and removed it, but somehow managed to make it do a full hard reset (or a docwho76 as we call it ) and it deleted a bunch of my pictures. google+ had failed to sync them. But we’ll keep documenting the project,
charliex, tom, Charles Bynaker, and 8 others are discussing. Toggle Comments
Yesterday I was rewriting code for Null Space Labs ‘Cylon III’, a blinky toy thing we make and use for people to have soldering practice, it spawned from the EMSL Cylon and then through a few iterations from a simple 9 LED thing to a charlieplexed circle.
Cylon II (with an IR LED to mimic a tv-b-gone)
We use the ATTiny2313 which a lot of people complain has no legs ( the MU package ), I think its great but it can be a challenge especially with the milled versions of the PCB to fix and replace the chip.
Anyway I was playing around with different clock speeds in AVR Studio with the Atmel Dragon. The dragon is a terrible piece of hardware when it doesn’t work right, which is often. It has a design flaw that causes a couple of chips to blow up regularly, so regularly in fact we keep a stock of the chips at Null Space Labs. My dragon has been repaired three times so far.
Then there is the software itself, the way it interacts with the USB has it disconnect and reconnecting after using the AVR Studio programming interface, sometimes it’ll hang and you’ll hear the USB connect/disconnect cycle until you unplug it ( which increases the chance of the chip blowing up issue ) and exit AVR Studio 4, which is really annoying especially during debugging.
So I select the <4Mhz option and AVR Studio warns me that I won’t be able to program the chip if i set it this low, I say no and put it back on the internal 4Mhz. However I fail to take into account that CKDIV8 is set ( clock divide by 8). At this speed the Dragon is having problems reading and flashing the chip, even at 51khz which is the lowest speed. It’ll read the chip id once, or read fuses once, then require me to exit the programming software, and reload it.
So i think, ok reset the fuses since it’ll let me do one operation. But in AVR Studio you have to read the fuses then set them, you can’t set them by hand then program, this uses my one attempt before the Dragon fails to communicate. It reads the fuses, but then fails to set them.. I try every variant, unplug the dragon, nope, this fails since it looses the USB setup or whatever AVR studio is doing, reset the target board, disconnect the target board. Nope, the only thing that works is exit the programming interface in AVR studio, let it reset the Dragon and back in again, at which point it reads the fuses again and then I can’t do anything with it, since it no longer communicates with the chip.
Anyway at this point I’m thinking its time just to replace the chip, but then I remember about our AVR programmer which is a smaller , simplified version of the USBAsp, we call it the Kraken.
We only use the ICSP mode and have our own SIL 6 pin programming header. The pogo pin version is much nicer than my hack.
Krs pulls one out of the back and builds it in less than 20 minutes, it’s programmed and ready to go. I setup AVRDUDE and read the chip id, that works, read it again, still works.. So i try just to set the LFUSE so that the clock is set back to 8Mhz, and bingo it works straight away.
USBASP/Kraken 1, Atmel AVR Dragon 0
Atmel sells the Dragon for $49 http://store.atmel.com/PartDetail.aspx?q=p:10500053 it comes with no case (which doesn’t help the chip destroying problem either ). I personally think they should rev it, or sell it with a box,or at least a warning. Using it with a USB Hub will help alleviate the chip killing problem.
But why it can’t flash the ATTiny2313 when the USBAsp/Kraken can just seems like a software problem. Now if only we had access to DebugWire I could do without the Dragon altogether.
Still after you’ve seen a few of them with holes burnt through the chip and sometimes even into the PCB, you’ll know why Atmel calls it the Dragon. It’s funny how the people who make the chips can’t get their programmer to work well, when one guy in germany can.
I’d like to say no Dragons were harmed during this process, but one of our members ( who coincidentally named the Kraken project) blew his up while I was working on this problem ).
Since this blog was written around 8/22.. We’ve blown up another three dragons..
I’d been meaning to build a USB isolator for a long time, especially since people tend to rely on and my laptop/pc for programming and testing of the new boards they’ve built. Regular hub’s don’t offer much protection and the USB chips will shut down if they detect something, so an opto-isolated USB adapter is a good thing.
I found a USB isolator from circuits@home, had BatchPCB make a few of the PCB’s ( Not recommended as they are too expensive and their customer support is non existent, try Silver Circuits instead )
After krs built and I tested the USB Isolator it seemed to work pretty well, so the next step is to adapt it for an NSL project. In this case I went back to the chip itself and decided to start from scratch. We’re using the Analog Devices ADuM4160 USB Isolator.
Step 1 – Find Chip, datasheet and Design
Find the datasheet, and another other points the OEM has, Analog are generally really good at this.
So this is probably one of the most frustrating aspects of design, sourcing the chips. We’ve decided to look for the ADuM4160, the ADuM5000 and the Atmel AT43301 chip… Uh oh its my nemesis, Atmel… Almost every design I’ve ever done that’s relied on Atmel has either met out of stock for months, or discontinued.. Ok so off to findchips.com to do a quick survey of the land.
Ok good, a spread of stock over different suppliers with a decent quantity for a project of this size. Prices as usual vary widely. Newark often come in the lowest, $9.78 vs $12.23 at digikey and oddly $19.41 at element14. Even odder given Farnell/Newark/element14 are basically the same place.
Also I notice there is a eval board, so we’ll note this part number and look up the datasheet for that too, it’ll more than likely give us the schematic to work from, as well as layout tips. The chip datasheet covers local layout aspects for the PCB for the chip itself, but the evaluation board gives you it all usually.
Ok the ADum5000 ( which we may not use in the hub since a hub potentially requires more power )
It is in stock, but a low count/spread if I were doing a commercial production run I’d look into it further. Again Newark has beaten Digikey pricing (Newark also ship everything in separate white envelopes inside the bag with markings on the outside vs digikeys plastic baggies )
On to the Atmel….
What a shocker! no stock… no spread. Ok so time to head over to Atmel, and see what’s up with the chip. Ok –AC version is non ROHS, so it is gone. Its a few years old so its not surprising its been EOL’d
AT43301-AC AT43301 5/20/2008 11/20/2008 No replacement, same for the -au… Well that tales the biscuit, obsoleted, no replacement. checking eBay turns up no results, and its too difficult to deal with the part finders on Google to source a chip. So we’ll have to start looking for a different chip. So far my Atmel experience is staying true.
Step 3 – Find a replacement chip or design.
I’ve had mostly good luck with TI , I like the Stellaris series they bought, though the chip i used wasn’t in mass production last time i checked. Poking around we get the TUSB2046B its available in LQFP32 too
I used a BusPirate V3 board to test out these rivets. They are effectively thru-plating for PCB’s, normally you need to do messy multistage chemical stuff to do it at home. This isn’t as good a result as a chemical plate but it makes up for it in ease of use. It’s not the cheapest option, which is usually just using a bit of wire in the hole and hand soldered in, but that can cause problems with VIA’s under flush mounted components.
Basically you put a rivet in the hole, flip it over and hit with an automatic center punch.
Here are some of the results so far. I didn’t spend as much time as i ought to have aligning the backside of the board, so the drills are slightly off.
Bottom side, they take up less space on the side that’s being compressed.
I used a needle to insert them one a time, a wire strung with a bunch might work.
A few more added.
Takes a little practice to get them smooth, making the automatic center punch ‘snap’ is key. They’re almost flush when that happens. (circuit in the back is Jack Gassett’s butterfly (the logic analyser update for the sump) http://blog.gadgetfactory.net/ )
some closeups with the HiView wireless usb microscope, because of the way it works you see black rings where there is really copper if its an angle.
This one went funky, needs some after attention, I didn’t put the center of the punch in the rivet !
trying out different pressure settings on the punch