Messages

Isn't that annoying when it happens?  RS are possibly the worst at it, we once used to have a SOT25 part from them that we used 400+ of at a time and they shipped them in bags of 1 from cut  tape, some fool had done that to their entire stock. In fact even now nigh on impossible to get a strip/reel of FT232 unless you buy them off FTDI directly.

While Mike seems to have you sorted if anyone ever needs 10x10 trays I always have lots, also empty reels and more limited numbers of  20x20 14x14 5x5 and 3x3 trays.

I've never used craigslist myself was surprised to notice pick and place machine craigslist listings come up in google results. Currently there are apparent listinga for a Europlacer (or 3 or 4..) one of those nasty little Maddeltechs and a Phillips machine (with an incorrect YOM). A quick browse would shows a considerable amount of kit on there, especially if you like lathes & CNC. Much of it seems to be posted under "London" even when it isn't really.

You can still get industrial ISA PCs, it's the consumer level ones that are harder to get hold of and indeed now command a premium on places like eBay. There are a few niche sites online that sell some of this stuff too even those weird 5.25" quantum drives if you remember those. Getting Win95/98 machines onto a network is getting to be something of a pig however, if you come across an truly ancient PC that hasn't been configured in a while that way you run into all sorts of nonsense with drivers(if you can find them) installation disks (for the edition of 95 you are running) and then things like uptodate SAMBA installations or windows shares won't talk to them anyway. Try and resort to something oldschool like FTP and instead you'll find all those programs require some specific version of trumpet winsock and DLL hell ensues. The best solution is to get it as far as being able to read USB drives or get something equally retro and exotic like  a superfloppy or ZIP.

I would strongly suspect that too many significant changes happened between Win98 and 2000/XP for it to be feasible to get it working on those platforms

Another slightly sneaky option would be to dual boot into a lightweight Linux installation that is capable of talking to modern networks, USB sticks etc

I almost certainly wondered about rack and pinions wrt the RV4 when we still had one, but this is the first closeup I've seen of an RV1 and behold, yours is plastic, on an RV4 it is metal and longer, IIRC I'd say the motor was different too. I don't think anyone came up with a part number and I wasn't keen on dismantling ours at the time to experiment with whatever one might find online.

One would assume its being sold by Aub like the two posts below this one?

It's been 5 years since I moved on from one of these machines, but I always used RVGerber as that is how the training worked and in turn that is how data came to me. For RVGerber you primarily need the silkscreen layer and the paste layer, RVgerber attempts a crude form of OCR to match identifiers in your BOM against labels in the silkscreen, it then has a fairly good guess at which pads might relate to that device, or you can highlight the correct ones with the mouse. Double mouse click moves to the next device, right click dumps the part at the selected pads mid point, space rotates it 90 degrees.

So the workflow goes something like  Import Gerber->Import BOM->Scan Idents->Locate the parts->Check it all. It's a fairly slow job not helped by the rendering speed - scrolling is basically impossible. So using Mikes tool is never a bad idea it can turn a 2 hour job into a 10 minute one.

As a prior step it is probably a good idea to make sure you have predefined any "CDFs" required, these are the names and sizes/models of the various component packages you are going to use. Good practice would be to define a set of rules for part names and package names and then stick to them otherwise you can end up with the same part named slightly differently or with a different package name in different products and this actually gets quite annoying.

RVCad I'm pretty sure never worked so you can ignore it.

The training manual that comes with the machine talks you through a job start to finish, it is actually pretty good and I believe there is a copy on here somewhere...

If you have nicely defined fiducials the machine is good at recognizing, Auto-Fid correct actually works quite well, if they aren't so good however it can do some interesting things, size, colour, contrast and lighting all have an effect. Just using the crosshairs on the fid camera is not exactly a huge drawback as long as the camera calibration is OK, lining the nozzle up instead is trickier but dodges the calibration issue.

You can manually tweak placement of individual components inside RVPlace if they are not quite where you expect them to be but this tweaking is not persistent, however it is a useful thing to be able to do if for whatever reason your machine has a weird offset in a particular part of the place area.

Well I doubt it is in any way custom. No idea which one it is but I think possibly it comes with the feeders with gears rather then a springy clip + pusher nozzle.

Somewhat surprising to me but apparently you can now buy those little Neoden in the UK, from Blakell Europlacer of all places. I don't see them on the website but apparently £10-12k gets you one without the import hassle + support and training.

It is a lot to gamble for sure, being Kickstart averse myself I can't begin to guess what others make of such a sum. They really should have had something a little more ready before coming to Kickstarter, especially on the software front to reassure people. Such an effort is far from trivial and the idea they can knock up a new platform over the next 12 months (or less according to the text) seems ridiculously ambitious. Like so many of these projects I always worry when a bunch of people only one of whom has more than 10 years experience in anything really think they can solve the industries problems just like that. The target seems low too, it seems to me its fairer to your backers to find some VC or a loan on the back of some strong pre-orders & interest, the risk here is instead spread between a handful of people who presumably are not wallowing in cash (or they'd just buy something known and 2nd hand).

It "Looks the part" and they are right not to chase speed, the claimed 30 micron accuracy is up there with a proper machine and if a plastic feeder really works that is pretty impressive. Placement area looks vaguely sensible to. It seem to be getting a little stick online from opensourcers etc "why didn't you use openPNP?" etc and while there may be some justification in that its not like anyone else has used that stuff to make a decent commercial product, the Liteplacer is almost comical, tho for some it may be useful..

https://www.kickstarter.com/projects/chipsetter/chipsetter-one-a-desktop-pick-and-place-machine/posts/1692469

Not my listing..
http://www.ebay.co.uk/itm/Dima-HP-110-Pick-and-place-machine-with-150-feeders-/272241108706?hash=item3f62d5a6e2:g:-sQAAOSwl-FXNLKW

We looked at this model as part of our research when buying a machine, they seem to be pretty robust well built machines. Dima still exists (part of Nordson) so there should even be support available, Dima was represented by KaiserTech and then etek-europe and currently APP Electronics.
Feeders are very compact as the reels sit in the trough underneath looks like they started around €600 each (8 and 12mm) new about 3 years ago.

If it helps, back in my RV4 days I setup all our PCB's using RV Gerber, it isn't all that bad but a typical PCB would take me about 2 hours to put in and then check over, partly because the graphics rendering speed and scrolling are so very slow. However RV Gerber is very fussy about Gerber formats and will come up with a PCB completely out of scale if you output with a different resolution (IIRC it wants 3.2).
Altium outputs a whole bunch of extra columns in its pick and place file there don't seem to be options to customise it, you can simply select CSV or text. I'm pretty sure this means you will need to tweak Mikes import utility code to handle the CSV file properly. It expect

"RefDes","PatternName","Type","Value","Layer","LocationX","LocationY","Rotation"

"C1","SMALL1206","SMALL1206","2.2u 50V","Top","2859.3","2691.9","270.0"

Proteus looks like

"J1","CONN-D9F","D-09-F-S",BOT,0,-764.63,-976.252

Protel looks like

"Designator","Footprint","Mid X","Mid Y","Ref X","Ref Y","Pad X","Pad Y","Layer","Rotation","Comment"
""
"R4","1005[0402]","69.882mil","156.496mil","69.882mil","156.496mil","44.291mil","156.496mil","T","360.00","SMR 2K2 0402"

As you can see, in Protel we have been using the comment field to store our internal part numbers -handy to avoid defining the same part multiple times slightly differently

As with any pick and place machine I would suggest it is a good idea to establish any new 'packages' a design introduces and create those 'CDFs' before you start anything else.
Once you have defined a component don't go and tweak its Pin1 definition at a later date, it will mess with all your existing defined products.

The RV4 is a contrary little swine, components are defined as they are orientated in the tape which means they do not and indeed often cannot follow standard conventions such as https://blogs.mentor.com/tom-hausherr/blog/tag/ipc-standards/ this means you may need multiple component definitions for the same package simply because the rotation changes between tape and tube (hence warning above).

Don't forget something like 0805 is a footprint not a full package definition. For the best results you will then have variants of this for the different heights, these might be picofard (thin, typically pale), 100nf+ (pretty thick usually brown), std nf values (somewhere in between) etc. Radically differing heights using the same package may not image consistently and thicker parts will of course get rammed into the board quite hard if the defined height is much smaller.

Dense population of  electrolytics can cause a problem in any process, particularly with lead free where many parts reflow quite close to the limit. However lead free solder melts around 217C so one would assume you could successfully use a Galden fluid with a boiling point around 230? Lots of things will easily be getting to that temperature in forced air convection if the board has other "massive" devices to heat or a ceramic PCB, is the electrolyte boiling or is it something more basic like it getting hot too quickly? A quick google does reveal several manufacturers excluding VPS as a process for electrolytic capacitors, Panasonic seems OK with it with provisos so this is something I'm going to bear in mind when going through with our next Oven purchase.

Also possibly of note a few parts specifically exclude IR or at least limit it to one pass, typically LEDs but I have also seen the same on thermistors.

Every 300 parts is a awful lot, this symptom sounds somewhat similar to what preceded us changing the Arm control board and the one in the PC (DSP Carrier?) with phonoplug's ones. In that instance I was fairly convinced it was losing XY position before dropping Z (it is a while a go now but I see to recall a distinct change in sound right before it went batshit). The new boards made the problem go away completely but it was nowhere near as frequent as every 300 parts more like 2-20K which for us was bad enough given it is quite capable of destroying a PCB depending on where it drops.
The replacement boards were a complete stab in the dark as we couldn't find any dodgy control lines or loose / faulty sensors, but it seems unlikely the same thing would go wrong twice assuming this is still that machine.

As I understand it, how well this performs will be dependent on how well thermally balanced your PCB is and if this oven has got the initial warming stages right. Some of the reasons Vapor phase hasn't made the inroads its advantages suggest it should are because not getting these two things right can result in way more tombstoning that you would see in a good convection oven. I'm told major Vapor phase manufacturers have the ramp/preheat stage licked these days, however when you scale up the process they rapidly get very expensive to buy.
This does look like a handy device to have, if you use sensitive or expensive components this has to be better than those hot drawer things on eBay.