What's new
What's new

temporary inspection camera for machine monitoring

jpf44

Aluminum
Joined
Apr 1, 2005
Location
Scranton
Greetings,

we have an assembly machine at a customer site that is doing some weird things every 60000 parts or so - we would like to setup a camera to record to flash disk or something so we can monitor. flexable head type that will record 8 hours if anything is available - any recommendations that's not to expensive.
 
not really at a customer sight - probably no. some of these things move quickly I need decent quality video in a small area.
 
That's how I'm leaning - just got a quote for a rental on a high speed industrial camera $2,500 a week for rental. Go Pro it's probably going to be
 
I agree on the Go Pro but would like to here more about the problem. I am amazed and intrigued at the thought of catching an intermittent problem every 60000 parts or so. Is it possible ?

Make Chips Boys !

Ron
 
System throws an error when something goes awry this is time stamped in the log I'm hoping to correlate time stamps to see if we can catch something - we just cannot put a finger on what's going on and would like to get some live action of the mechanisms when they are feeding/assembling the parts. Application challenging - small delicate aluminum/brass parts, couple of grams in weight, weird geometries, etc. etc. while not really "high speed" end-end cycle is about 4 seconds the shuttles, lift pins, inspection stations etc. run at milliseconds you can't really see what's happening 2 much going on all over the place. I have about 3000000 thru the system and had a problem about 3 months ago and that was traced back immediately to out of spec parts feed into the machine (pulled a handful out of the hopper they were all bad) - I suspect the same this time however cannot do 100% inspection on thousands of parts (I've spot checked hundreds all in spec. as best as I can measure onsite) also parts get 100% destroyed when problem occurs so...
 
I agree on the Go Pro but would like to here more about the problem. I am amazed and intrigued at the thought of catching an intermittent problem every 60000 parts or so. Is it possible ?

Make Chips Boys !

Ron

Yes, I have a current issue where a part is oriented in a machine via vision system. I may get 1 bad part every 3 - 8 months, or I may get 4 in one week and then none for another 4 months. Through the use of these security cameras, we have determined it is not an issue with the vision system programming, nor is it an operator induced issue (intentionally causing a bad part or doing something against operating procedures). It is PLC program or a physical issue (bad connection, broken wire, electrical noise, etc). We added another monitor to the computer on the machine and wrote a temporary "viewing" rung in the ladder to display all of the important flags / inputs to the sequence and have a security camera watching that now to see which flag is giving us grief. This thing runs 60,000 parts a week, the security cameras are the best method of catching this type of one-off, random event.
 
IMHO video is probaly the wrong way to go, may be far better just setting up a digital cam with external power with a remote triggered shutter, more than a few can be set-up to take bursts of 10 frames in quick succession Sink the file number to the part count and its easy to then go back and see whats happened unlike trilling through hours of video.
 
IMHO video is probaly the wrong way to go, may be far better just setting up a digital cam with external power with a remote triggered shutter, more than a few can be set-up to take bursts of 10 frames in quick succession Sink the file number to the part count and its easy to then go back and see whats happened unlike trilling through hours of video.

In my case, the next piece of equipment inspects for the problem the mis-orientation causes, so it will reject it and the operators write down the time of the rejection. I can then go back an pull the previous 30 minutes or so and see what happened. I believe in the OP's case, the machine stops/jams and logs it, so would be easy to go back and pull that footage.

In my case, we made more progress in that we saw "what" happened in the plc logic this time, we just don't yet know "why" or "how" it happened. However, we were able to put in code to deter the "what" from occurring again. Time will tell and we have some tests for figuring out the "why/how" question, but at least for now the customer is safer.
 
A lot of standalone IP wireless cameras fit the bill as they have on board SD card slots. I've had decent luck with the Foscam cameras.

GoPros capture much better footage but have limited battery life. You'll need some sort of external battery for it run for 8 hours.
 
We think we have figured out part the why now..... it has to do with packet timing on the ethernet, we basically got 3 signals from the camera at the plc within 140 ms. ----- With only 1 camera trigger. This cannot happen as the camera issues 1 packet of info for every trigger. The reason appears to be network lag. What we saw was apparently three past triggers worth of data. We've put things in place to keep this from causing the issues it was while we continue to monitor with the 1080P cameras to see if we have cured it with some other setting changes related to the ethernet communications. If we continue to see this lagged comm, the most likely culprit is the 10 year old Netgear ethernet switch and that will be the next thing to look at.
 
We think we have figured out part the why now..... it has to do with packet timing on the ethernet, we basically got 3 signals from the camera at the plc within 140 ms. ----- With only 1 camera trigger. This cannot happen as the camera issues 1 packet of info for every trigger. The reason appears to be network lag. What we saw was apparently three past triggers worth of data. We've put things in place to keep this from causing the issues it was while we continue to monitor with the 1080P cameras to see if we have cured it with some other setting changes related to the ethernet communications. If we continue to see this lagged comm, the most likely culprit is the 10 year old Netgear ethernet switch and that will be the next thing to look at.

Relatively cheap to preemptively upgrade, but would be good to KNOW that it is the culprit.
 
A lot of standalone IP wireless cameras fit the bill as they have on board SD card slots. I've had decent luck with the Foscam cameras.

GoPros capture much better footage but have limited battery life. You'll need some sort of external battery for it run for 8 hours.

Go Pro's are overrated IMHO SJcam is near enough the same features for 1/5th the price yet still the same mounting and underwater potential. I have a Sjcam M10+ a older model admittedly, but have used its high speed modes to capture - slow down mechanical stuff before. It can be externally powered too, std micro USB so dead easy to do. Hence there’s only the memory card limit as to recording length, but most the Gopro's and clones all allow - have a dash cam mode were it just keeps rewriting to the memory card, hence a large card and you can capture 10's of hours and if nothing happens - you leave it powered it will just keep looping and over writing the card.

Only key with any camera set-up is half decent lighting especially if your going for faster frame rates, oh and as i found out the hard way trying to watch a single turn clutch set-up that kept performing 1.5 turns some LED torches in some modes are just flashing fast and when your as unlucky as me the dark spells can do really well at sinking with the frame rates just as the action happens.
 
Relatively cheap to preemptively upgrade, but would be good to KNOW that it is the culprit.

Yes, we don't use Netgear (or any standard networking routers/switches) anymore for Ethernet/IP. We switched to dedicated Ethernet/IP switches specifically designed for machine tools / automation some years ago. Learned our lesson on that one when IT decreed that we had to install managed Cisco network switches for all Ethernet/IP applications.

That somewhat worked until IT decided they needed to "manage" the switches, then they shut down half the plant for a day and a half until they could get all of the changes "undone". Of course.....nobody at the plant knew they had messed with the switches so it took most of a day to figure out why nothing would run. And that was the end of the "managed switch" decree for any equipment/automation.

I don't want to change it until we see the problem again. I want to know that it wasn't the ethernet settings we changed first. In an intermittent case like this I want to change 1 thing at a time and test for results. Yes, it's slower than tossing a parts cannon at it.....but you fix the root cause when you do it this way. Band-aides never work long.
 








 
Back
Top