With the issues of non-buffering of feeder commands, does a triple feed actually ever work ?
Not sure if the part in question will need multiple imaging or not.
Yes, I think so. Provided the part is imaged in one shot, it should index after lifting the part out of the tape, indexes again (if I remember correctly) just before it puts the part down, and the third index is after the part is placed.
Something like a 6.3mm diameter electrolitic can be visioned in one shot and comes in tape requiring 3 indexes. Gets a bit painful if it rejects the part, which often seems to happen with parts having two long thin leads like this (also HC49 SM crystals etc).
I don't think its possible to set number of indexes to 4 though. While it accepts 4 as an entered value, it either always reverts back to 3 when you go to check the figure again, or it just does 3 despite being set to 4 - can't remember which.
All i can say is there must be some wierd software problems or machine faults!!
Multi indexing does work. I have parts that require up to 5 indexes and have never had any problems even when feeders are next to the fixed camera.
The machine picks the part; indexes once; then visions the part; takes the part to its place position then stops at its settle position and cycles througth the rest of the required indexes to bring the next part to its pick position.
I am a little baffled as to why so many machineds have this fault!!!! I have 3 RV4s and they all index fine without a problem, maybe i,m just lucky!
AFAIUI the problems started when they introduced the flying camera, where it visioned before the first index completed and sent the second index too quickly. Obvously anything that needs more than 2 indexes will be too big for OTF vision.
I think l have seen some occasional double feed failures when both the feeder and the placement positions were very near the fixed camera, but they sorted themselves out with retries
I know Phonoplug added proper buffering to his replacement feeder boards but wasn't clear under what circumstances it was needed.
Where are you defining the number of indexes required?
If you use a standard defined feeder with an index of 2 and alter the index of one lane to 3 in RVplace, RVplace will use that index of 3, but it will continually forget it both during the job and whenever you close the program. The solution (though I have never bothered) is to modify or clone the feeder giving the lane(s) you require a defined index of 3, no doubt if this lane is very close to where the parts are fitted on the board and simple to image (e.g a Case C electrolytic) one might run into the buffering issue sometimes unless you tweak the place parameters to be just slow enough.
In order for multi indexing to be successful you have to create the feeder in RvSetup and define all the parameters there.
You must also ensure the number of retries are at least the same as the number of indexes, i set it at one more, multi index will not work if the retries are not set correctley.
Any modifications done to the feeders in RvPlace will not be saved, you must set up the feeders in RvSetup.
OK just to show I'm not going completely mad...
As I'm building some boards today that use one of these parts I took a bit of video to show what it does.
http://youtu.be/JZE2nilsi7s
The number of indexes, if not clear in the video, is set to 2. There is no index as soon as the part is picked, which there would normally be, and only 1 index sent after the part is placed. You can hear the feeder indexing 3 times in quick succession - this is because I have programmed the board in the feeder to advance 3 indexes for each index command received.
What version software is installed? the latest is 3.1
Are you using a newer style feeder or if it is an old style has it been modified for onboard 12v?
In the vid the feeder is miles away from the camera so no problems of short travel!
The feeder MUST be set up for multi indexing within the RVSetup programme, if done within RVPlace it will NOT work, the settings will NOT be saved.! also the retries MUST be the same or higher than the number of indexes.
Aaha, 3.1 is not the latest version! I have 3.5.
This could be the biggest clue yet though. Is there any chance you could zip up all the files you have in the 'rv' folder (don't need any of the sub-directories) and email them to me? Don't bother including 17MB 'tutorial.exe' if its also there!
I'm running a job that will take a couple more days yet, so if you can send me them pretty soon I'll go through them to try and find which is the offending file causing the issue and report back.
I'll PM you my email address.
Cheers!
The top issue is version 3.5 which is what i run on. The 3.1 i quoted is what the RVPlace and setup report it as in the help menu.
Sorry for the confusion.
In my early days of having my 1st machine i also had problems with multi index, the only fix i have implemented is to set up a dedicated feeder in RV Setup -- this is the only time it works and i have never had a problem since!
I know it seems simple but if your feeder is not configured within RVSetup then it just doesn't work!!
Don't understand your comment about setting up a dedicated feeder in RV Setup.
1) You have to have different feeder definitions for feeders with different numbers of lanes.
2) When allocating feeders, you can't add the same feeder twice (say you need to 10x8mm feeders in the job), so you would have to create another with a slightly different name.
I have to say I don't really understand the difference between 'Save' and 'Make unit' in the 'Feeder unit create/modify' function of RV Setup. When I was taught how to use the machine it was said that you 'Load' a suitable feeder as a starting point, make any changes required to number of lanes, pitch etc, 'Save' as your new type, then 'Make unit', but I don't know what the point of the last two actions is when it seems like you could do this with just 'Save', surely?
It would still be really useful if you could email me your application files. I will test drive these to see if it makes any difference and let everyone know.
The 'save' function is to save a basic copy of the feeder just created so it is easily available for future use.
The 'make unit' saves a copy of the feeder created into the feeder allocation ready for selection within RvSetup for inclusion in the job file.
I know its seems a funny way of doing things but we are talking about 1980's software, lol!
The feeder creation and setup is a little long winded and could have been made a lot simpler for setting them up but hey we stuck with it, lol! You end up with hundreds of different feeders for different jobs and a good system of naming them is required to keep on top of them.
I will zip up my files and email them you later today.
In my experience time spent on getting the feeder creation and allocation correct is time well spent in preventing retries and pick errors giving a smoother uninterupted place.
Just A Thought!
Check the feed cables from the control module to the feeders.
There is a mixture of power (240v) and signal going thro these, they are plugged in and out of the feeders regularly and the pins wear and tarnish.
Check these cables out for worn and damaged pins, this may be a possible cause of feeders missing command signals and miss feeding or not feeding at all.
Have been having a play around with this multi indexing and think i have a handle on it now!!
The majority of components that i use are far too large to be successfully imaged and placed with the on the fly camera so i use the fixed camera for most.
There is not a problem imaging and multi index feeding even with the feeder next to the fixed camera and no matter how fast i have things this always works for me.
The software pauses at the point of 'settle before place' and indexes as many times as i set it (I have even had it index 10 times!!)
I have put the same component (a large TO265 power transistor) to be visioned by the on the fly camera without changing anything else and it also works fine, indexing 10 times again!
Smaller components such as a SOT89 i have put in a feeder next to the fixed camera, had it multi index and place, there were no problems, machine paused at the point of 'settle before place', multi indexed and then placed it.
I then changed nothing but put it to vision on the fly and all hell broke loose and multi index went out the window.
I then removed the 'merge moves' and hey presto it multi indexed fine. So i set everything to place as fast as it can and without the 'merge moves' it placed and indexed without any problems. There is a small trade off for placement speed but is still faster than using the fixed camera.
I have now tried various feeder positions in relation to the components place position and the only time i begin to see problems is when the feeder and place positions are very close to each other, approx 6 inches. (I think we are finding the limitations of the machine!!) To overcome miss indexing in this situation remove 'merge moves' or use a slower place speed, i have found that just putting the rotation or Z axis on medium or slow is enough to allow multi index.
But i have observed that the feeders MUST be configured within RvSetup.
In the first screen in RvPlace you can access the 'Edit feeders' but any changes you make within that screen are NOT saved, they are lost as soon as you close that screen.
In the second screen in RvPlace you can access 'edit feeders' Changes made in this screen are saved ONLY until you exit that screen and go into any menu on the first screen, wierd i know but thats what happens!
So to use multi index successfully the feeder must be configured fully within RvSetup then everything is saved and available in RvPlace.
I recommend that a slow speed is used to start with so that all parameters can be configured correctly , then speed up the place to a point were it is just successful, with carefull optimisation of feeder positions and place speed settings multi index does work with little compromise of placement times.
Good work, that would explain why I didn't even encounter the issue for years, well that and the fact we place a very boring and quite static range of packages.
Hurrah!!!
Many thanks to SkyWalker for sending me his files. I've now had chance to investigate.
There seem to be two versions of 'place.dll' around. One is about 1.7MB, the other 268k. It seems that the 268k version has been supplied separately to the main install package at some point, I guess at a later date, and has a newer file date than the 1.7MB version. HOWEVER. Indexing of parts that will be visioned in more than one shot does not work with this dll:
- With parts in static locations (ie trays), it will always pick from the same pocket regardless of it already having taken a part from there.
- Parts in vib feeders work ok, but it keeps retrying for ever when the tube runs empty, regardless of 'retries' setting.
- For parts in index lane feeders, the first index (when the part is picked) doesn't happen, so you have to set number of indexes to one more than what you need. On top of this, setting number of indexes to greater than 3 (so you actually get 2) doesn't work either, no matter where you set the number of indexes (defining the feeder, 'edit feeders' etc).
Not sure what the supposedly newer version of place.dll was supposed to improve. Perhaps in moving over to the one that works I might notice some other difference, however, being a lot larger, my suspicion is that the 268k version should be heading to the recycle bin quite rapidly never to be given the time of day again.
So thanks to all those that have helped with clues and info over various different threads I have commented in about this issue! Finally one real pain in the 4R5E seems to be fixed! I have attached both files to this post.
SkyWalker - in a previous post here:
Quote from: SkyWalker on April 05, 2011, 04:04:11 PM
Place.dll on all three of my machines is 268K dated 12/09/01
you say you have the bad one on your machines, though in the files you sent me was the good dll.
Also
Gopher, in the same thread a few posts up you mentioned having the 268k version. I recommend changing it!
Quote from: Gopher on April 04, 2011, 12:32:17 PM
Place.dll here is 268KB with a date of 12/09/01 ....
Incidentally having a copy of all the files from SkyWalker, I ran them and mine all through a binary compare (WinMerge). 5 were different:
- DeIsL1.isu - apparently an uninstall file so not surprised they are different
- Gerber.exe - yours is older though much larger (4.2MB Vs 644k). Mine reports to be V2.8 when run while yours, V2.5. Will attach V2.8 to this post if you want to give it a go.
- Framegrab.dll - yours is a bit newer and marginally larger
- place.dll - already mentioned
- Pxc2_95.dll - yours makes my RVplace crash as soon as it tries to use a camera. I suspect you have a different video grab card to me as this is a dll supplied by the card manufacturer.
Wouldn't let me attach gerber.exe to last post - over size limit - so here it is.
Nice work there phonoplug ;D Another step forward in our collective knowledge - well done sir.
Will get round to looking closely at all my machines and tidy up the software and all its versions by the end of the week.
Ok, so the first problem I have found in using the larger size dll (that does multiple indexing correctly) is that it indexes the first component in a job too soon.
Has anyone ever noticed how the pick dwell is longer than normal for the first part on a board, each time you begin to place each board? Well with place.dll that is 1.7MB, its indexing the part before the head has retracted, during this abnormal extended 'pick dwell'. This gives mixed results varying from picking nothing, to picking two parts from both adjacent pockets (eg 2x 1210's which are very close together and picked with a larger tool), to picking the part from the pocket post-index, meaning one part is waisted and the next pick from that lane has an empty pocket.
Both place.dll versions 'feature' this extended dwell time for the first part on the board, but the 268k version indexes the lane immediately as the head retracts so it works ok.
I'm tempted to add a little inline circuit to just delay the data enough so I can remain using the larger (better) dll...
Seems to be working OK - I was nervous about using this part as it was a combination of triple feed and potentially problematic imaging but placing fine.
Noticed that it seems to pause to do the second feed before placing :
http://www.youtube.com/watch?v=YARcM0TVq44&feature=youtu.be
Thats exactely how it should work, the machine pauses at the point of 'settle before place' and the second and any subsequent indexes are done before the part is then placed on the board.
Quote from: phonoplug on August 14, 2012, 04:38:29 PM
Ok, so the first problem I have found in using the larger size dll (that does multiple indexing correctly) is that it indexes the first component in a job too soon.
Now discovered this only happens if you have auto fid correction ticked. If you manually do the fids at the start of each panel, or even run through it manually, but using the 'auto' button for each one, it indexes the first part for the job correctly.
Another bug on the list...