I finally got round to upgrading the illuminator on the up-facing camera, and I'd highly reccommend it as it makes it pretty much immune to ambuient light, and gives a clean image of an 0805 R with a threshold adjust range of 0 through to 60.
I used a ring of white LEDs, as these are way more efficient than red, so can run at 15mA for low heat and long life and still be super-bright.
I pulled the LEDs out of the old ring and fixed the new PCB on it for easy mounting.
I stopped the lens down to minimum, and also reduced the shutter time ( My cam has a rotary switch on the back to set this - the cam I took off an RV1 didn't - may need a neutral density filter if not, or dim the leds down a bit).
I also made a new ring for the fid cam as it fitted inside the hole of the main PCB. I'll stick this over the existing illuminator to avoid disturbing the fid cam position.
I used a homebrew PCB - happy to share gerbers if anyone wants to get proper PCBs made - they are single sided.
Yes please on the gerber. Do you have a part number for the white leds?
PCB files attatched :
gerbers ( single sided, 55x52mm)
ident drawing showing LED polarity ( Leds have a very hard to see notch on the cathode)
top mirror PDF for printing homebrew PCB artwork
DXF of routing outline
LEDs are Osram Duris E3, e.g. https://uk.rs-online.com/web/p/visible-leds/8108196/ or search "LCW JNSH" at your favorite distributor (There are lots of options, mostly different colour temperatures)
Resistors are 1206 100R ( for 12v supply)
Fid cam board needs an external resistor - adjust for brightness, around 100-220R.
I glued a short length of PVC conduit to the diffuser ring to reduce glare from the edge of the LED ring
Very Nice,
I'm up for some PCBs and happy to contribute if anyone is sending them out, otherwise I'll add them onto our next order here.
Alan
BTW are there any problems / gotchas extracting the fixed camera / lighting or is that fine?
Alan
No - it's fairly well aligned by the mount, and the factory cal menu has an easy centring adjustment mode.
Just mark the side so it goes back teh same war round, and ensure it sits square in the mount.
Hi Mike,
>>> I stopped the lens down to minimum, and also reduced the shutter time ( My cam has a rotary switch on the back to set this - the cam I took off an RV1 didn't - may need a neutral density filter if not, or dim the leds down a bit).
On my RV1S it's a Watec WAT-308A with a COSMICAR/Pentax 16mm f1.4 TV lens which doesn't have a shutter setting. The camera, lens and machined mount parts all look nice quality. If you made a note, which camera do you have there in your 4S?
>>> I glued a short length of PVC conduit to the diffuser ring to reduce glare from the edge of the LED ring
Is this mounted above the diffuser pointing towards the part?
Alan
Quote from: alanambrose on December 21, 2017, 02:52:07 PM
Hi Mike,
>>> I stopped the lens down to minimum, and also reduced the shutter time ( My cam has a rotary switch on the back to set this - the cam I took off an RV1 didn't - may need a neutral density filter if not, or dim the leds down a bit).
On my RV1S it's a Watec WAT-308A with a COSMICAR/Pentax 16mm f1.4 TV lens which doesn't have a shutter setting. The camera, lens and machined mount parts all look nice quality. If you made a note, which camera do you have there in your 4S?
Mine's a WAT-508. Without the shutter adjustment you may not be able to stop the lens down far enough & may need to either add an ND filter or increase the LED resistors (or add an external series R). The full brightness may be slight overkill - it is a bit uncomfortable to view!
Quote
>>> I glued a short length of PVC conduit to the diffuser ring to reduce glare from the edge of the LED ring
Is this mounted above the diffuser pointing towards the part?
Alan
No, below, towards the lens, to reduce internal leakage from the LEDs
Inspired by Mike, I've also done a lighting upgrade too and I'm reasonably happy with it. It's slightly different from Mike's in that I chose to keep using 5mm LEDs for the main camera ring and also to interleave red and white LEDs - something about the camera response and the look of the thing. That may be folly. I measure about 8x the ambient light level from each ring and so should be fairly immune to ambient. You might want brighter if you have the luxury of real sunlight. I have not tested the CDFs / re-calibrated the cameras / adjusted the camera response yet. Re the latter, there's one more stop down on my RV1S lens which may do it or I might need a ND filter - which I found you can get OK in the right diameter. I have spare PCBs if anyone wants. Found btw that the reason my fid cam light was flickery was the gnd wire had a break in it where it squeezed passed the camera.
A.
(http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2016.45.53.bmp) (http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2016.46.04.bmp) (http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2017.07.51.bmp)
(http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2016.00.22.bmp) (http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2016.14.54.bmp) (http://anagram.net/nuts/Versatronics/Cameras/2018-02-05%2017.14.43.bmp)
Bigger images here:
http://anagram.net/nuts/Versatronics/Cameras/ (http://anagram.net/nuts/Versatronics/Cameras/)
Have you made a programmable current driver for your boards? Looks very tidy.
I already bought leds Mike used but not gotten around to making the boards yet.
What camera is on your RV1S?
>>> Have you made a programmable current driver for your boards? Looks very tidy.
An LT3030 - the '12V' supplies looked a bit iffy.
>>> I already bought leds Mike used but not gotten around to making the boards yet.
I think either approach will be a big improvement. It could be that Mike's use of SMT white LEDs for the main ring will be better. If I ever re-spin I might make suitable for white LEDs as well and make the current adjustment range a bit more accommodating.
>>> What camera is on your RV1S?
Watec WAT-308A with a COSMICAR/Pentax 16mm f1.4 TV lens. From memory, it goes to F16 and takes 27 x 0.5 filters - although the LED ring screws neatly into the filter threads. Mine was set at F8 which (with 8x increase of LED light output from 6EV to 9EV) implies that if necessary I would need to stop down to F16 and add a ND4 filter. There's a space about 42dx3.5 under the diffusers to fit an ND filter if necessary. Without actually trying it yet, I'm guessing that the adjustment can be made in software though. Also, the camera itself has 18dB AGC.
When I've bought my RV4s (many years ago) it was in a bad shape (although Groves advertised it as a full working order ;) ), mostly blind, loosing steps and many other problems, so there are many mods and changes were made during theses years.
I've read that it's vision is reliable only with Plexiglas covers. But it's difficult to adjust machine and feeders with all this shields and metal crate. Also my working space at that time was quite small, so I wanted to use it as a table top machine, which I can remove and store compactly. So I decided to get rid of all metal structures, leaving only MDF table, machine itself and feeders. So the first thing I've decided to modify - vision.
1. For fixed camera I've just glued Rebel 1W white diodes with thermal conductive glue directly to the aluminium ring. They are powered by switch mode constant current driver. They are so bright that camera can work now under direct sun light. To increase contrast I've just inserted a paper tube blacked with black marker. Also I do not use diffuser as it creates unnecessary light spills. Camera was dimmed with fully closed iris and shortest time.
2. For On-the-fly vision I've removed red LEDs and installed 3+3 Rebel 1W LEDs, again with switch mode constant current driver. The camera has no adjustable Iris, but fixed. So I've just cut new iris from a piece of food aluminium foil, blacked it with permanent marker and made a tiny hole in the center with a needle. Then unscrew lens tube and swapped the irises.
Also I've made a light protection around the arm and head from foamed plastic. This stops any light to get inside the on-the-fly camera compartment. (I'm leaving in Italy and we have a very bright sun :) )
3. Fid camera: standard LEDs were substituted with modern high brightness versions. On the photo you can see red light even at day light.
Also I've removed main power transformers an installed 3 switch mode PSU: separated X axis, Y axis and A- +Z-axis. And separate PSU for Arm electronics supply. Machine now works much smoother and never loosing steps.
And just to make easier to identify feeders and loaded components, I've printed labels, laminated them in plastic and glued to the feeder. Now every line is clearly marked and you can write this nonpermanent marker which component is where.
Nice I like the labels a lot. Also interesting to hear the diffusers are not necessarily needed. Do you remember how many stops headroom you had on the lens? You left the shutter the same? I guess we should get some help from the AGC also. (BTW I couldn't get the RV4S WAT-508 spec from watec but I got the RV1S WAT-308A spec here:
http://anagram.net/nuts/Versatronics/Cameras/WAT-308A.PDF (http://anagram.net/nuts/Versatronics/Cameras/WAT-308A.PDF) )
For my machine I just measured:
ambient lighting - 4 2/3 EV (iso 100)
no diffusers - 12 2/3 EV (i.e. 256x 8 stops over ambient)
1 diffuser - 9 EV (i.e. 12x 3.7 stops reduction)
2 diffusers - 8 2/3 EV (i.e. minor reduction, leaving 16x / 4 stops over ambient)
... which suggests you can get a long way to sorting out the lighting by just removing the diffusers. Also, two layers of diffuser don't seem to give any advantage (or surprisingly disadvantage) over one.
A useful reminder of the value of replacing the PSUs.
Alan
I forgot to mention another label on another side of feeder table, which identifies lines when you insert tape.
I've disassembled fixed camera to check settings. Iris =1/16, Shutter speed = 1/5000
A had several times failures because of DC socket on the camera, so I've removed original DC power and Video
connectors and installed mini-DIN socket and lockable plug. No problems with fixed camera any more.
Well have some green boards made from Mike's gerber. I have spares if anyone wants a set.
I have an RV1s and wondering if I should do the upgrade or not. Sometimes if things are not broken its not good to try and fix them. But I wonder if it will just improve reliability of component detection generally. The camera does not have any adjustment on it so I believe I will need to control the brightness of the leds to get the best level.
Should or?
Trev
If you never find you need to tweak vision settings then there's little point.
Dimming the LEDs only gets you so far - the /main point is to increase the light level such that any change in ambient light level is negligible compared to the illuminator. Even if the camera has no adjustments, the camera lens probably does have an aperture adjustment.
If this is already fairly closed you may need to add a filter to reduce the light level into the camera.
The simple method I used was to cut out the old LEDs, drop the wire from the 12V 12 LED ring below through a hole on the aluminium LED support ring and fix the PCB in place on top of the aluminium ring with double sided tape or your preferred fixing method.
Solder the wires to the old supply wires and cable tie to one of the camera supports, keeping things tidy. Finish up by placing a 15mm copper end feed straight coupler on top the the camera lens to stop light getting to the camera, I didn't fix or glue it as it sits nicely.
I also used a black marker pen on the inside edge of the translucent white ring, but don't know if the penmanship made a difference as there's no lens flare. Cost under 3 quid, takes about half an hour.
https://www.ebay.co.uk/itm/2x-White-40mm-12V-1210-3528-9-SMD-LED-Car-Angel-Eyes-Headlight-Halo-Ring-Light/183572284970?hash=item2abdc2c62a:g:IAwAAOSwwnVcdSNT (https://www.ebay.co.uk/itm/2x-White-40mm-12V-1210-3528-9-SMD-LED-Car-Angel-Eyes-Headlight-Halo-Ring-Light/183572284970?hash=item2abdc2c62a:g:IAwAAOSwwnVcdSNT)
https://www.ebay.co.uk/itm/Plumbing-End-Feed-Copper-Fitting-Straight-Coupler-8mm-up-to-67mm/192517422892?var=492606537586&hash=item2cd2eebf2c:m:mmMolovfPLzEMZuU-s0WLmw (https://www.ebay.co.uk/itm/Plumbing-End-Feed-Copper-Fitting-Straight-Coupler-8mm-up-to-67mm/192517422892?var=492606537586&hash=item2cd2eebf2c:m:mmMolovfPLzEMZuU-s0WLmw)
Cheers
Mike
Just to add my 2p worth. I recently also replaced the dim and dingy component camera illumination ring. After replacing with that in the pictures below I had to change the setting on the rotary dip switch at the bottom of the camera from 0 to 2, and that brought the threshold adjustment figures for most CDFs down to around 0 where previously most were edging to -35 with the old illumination (or 100+ with the new illumination but the camera still set to 0). The key thing though is that its very insensitive to ambient light now so I no longer have to darken the room when using the machine and hope the sun doesn't shine through the windows!
ahem apologies for the huge photos!
My camera definitely has no shutter control so it sounds like I need an external resistor to set the led brightness to a point that does not saturate the lens. Anyone have a recommendation for using an ND filter over an external resistor?
I will make Mike's board up and test it out. I have noticed that over time some components seem to find a placement offset. I locked their x and y position in the cdf file but they still seem to drift. I think this is due to changes in image detection due to light levels. I literally saw placement perfect to start with and then it just moved over time but only for some components.
I will do the fiducial camera lights as well to help with auto correct. That always seems to struggle.
Trev
Quote from: trev on March 11, 2019, 09:26:50 PM
My camera definitely has no shutter control so it sounds like I need an external resistor to set the led brightness to a point that does not saturate the lens. Anyone have a recommendation for using an ND filter over an external resistor?
I will make Mike's board up and test it out. I have noticed that over time some components seem to find a placement offset. I locked their x and y position in the cdf file but they still seem to drift. I think this is due to changes in image detection due to light levels. I literally saw placement perfect to start with and then it just moved over time but only for some components.
I will do the fiducial camera lights as well to help with auto correct. That always seems to struggle.
Trev
The whole point is to increase the light level so that ambient chages are a smaller proportion of the light the camera sees, so although reducing the illuminator to the point where it no longer saturates the camera may give some improvement, the man objective is to reduce the camera's sensitivity to the point where ambient light is insignificant compared to the illuminator.
If the camera doesn't have the shutter time control, first thing would be to stop the lens down as far as it can go, then look at adding an ND filter
>>> If the camera doesn't have the shutter time control, first thing would be to stop the lens down as far as it can go, then look at adding an ND filter.
On my RV1S, I just replaced the LED ring without changing the camera settings or using an ND filter etc. I leave the interior lights on when it's placing and it seems fine. I think the camera AGC sorts it all out OK.
Alan
Well I have it working of sorts but did not have a lot of time to fiddle with it. Mike's board is extremely bright!
I have ordered an ND4 filter so will experiement a bit when that comes.
There seems to be a few different board designs now so everyone is going to have different settings which is a shame.
Thanks for all the comments though, the forum certainly is keeping these machines alive.
I just got round to upgrading the lighting on the RV4 flying camera - the old one was starting to get really picky on ambient light levels as the LEDs degraded. It got to the point where a difference of 5 either way on the threshold made the difference between recognising or not.
All I can say is WOW what a difference - now it just works all the time, whether or not it's sunny outside, or whether the LED tape I put inside the frame is on or off.
The original light is a pair of white acrylic blocks with three 3mm red LEDs down holes in the back. I didn't have any 3mm whites to hand so made a small board with a strip of SMD white LEDs and a 1mm white acrylic diffuser. This is a bit more uniform than the original, but don't know if this is much of a factor.
I then put a neutral-density filter (grey tint acrylic) over the camera to desensitise it.
The factory settings mode allows the camera to be viewed, with thresholding on or off. I set the LED current to get a clean thresholded image image with a threshold value of 220
Thanks, that's really good to know.
Trev
Also - if you take off the mirror assembly, after replacing it, do a camera centre adjust in factory setting (Put tool 5 on the nozzle by hand), as any slight change in mirror position makes a noticeable difference to accuracy.
I realised this when I noticed there was a placement offset between parts placed using the fixed and flying cameras.
Hi Mike, do you have any images of your new OTF camera illumination mod?
Bit hard to get pics in-situ.
Black tape around is same as on original
Appreciated, thanks.
Mike, how is the OTF camera performing with the white led upgrade?
Quote from: trev on November 21, 2022, 11:46:50 PM
Mike, how is the OTF camera performing with the white led upgrade?
Huge improvement - almost never need to tweak thresholds now, and pretty much unaffected by all but direct sunlight.
Mike, is the wiring to the LEDs 12v? I am thinking of tracking a board with some white me LEDs on.
Yes, 12V - it does dip a bit when the vacuum pump is on, but that's not an issue as it's always on when you need vision, so consistent.
I have a switched mode power supply so don't think I will see the dip. I found that initially the OTF camera would work but as the day progressed things would just get more and more unreliable. Maybe sunlight. But it was just a shame. Looking froward to implementing this now as the RV4s running with OTF can actually run at a decent pace.
Managed the white led OTF upgrade. Seems to be doing the trick. Took a bit of calibration and a placement offset adjustment in the cdf file but placed great after that. Real speed improvement to be able to use OTF reliably.
Thanks Mike for the initial info.
Quote from: trev on July 05, 2023, 03:13:54 PMManaged the white led OTF upgrade. Seems to be doing the trick. Took a bit of calibration and a placement offset adjustment in the cdf file but placed great after that. Real speed improvement to be able to use OTF reliably.
Assuming you removed the mirror, did you do the camera centre adjust as mentioned above?
Changing illuminator shouldn't need a placement offset adjust, only the brightness threshold.
Hi Mike
When I put the mirror back on, I did focus, centre and ratio? in calibration. I do not believe I had the offset previously but talking to a friend who used to run RVs he said he used to have one cdf for OTF and one for Fixed because he always had an offset he had to put in.
Are you saying that if the calibration is performed correctly then the OTF calibration will take care of any offset?
Just going on first principles, once calibrated, both cams should be aligned to "true" centres and no offset should be needed in the CDF.
It could be there are some oddities in the software ( or maybe a bent nozzle shaft), but I've never had any need to do any manual offsetting, apart from the occasional pick offset on asymmetrical components
But the OTF camera doesn't do a rotation or movement. The camera is fixed to the arm so it can't correct any error and check it's fixed. It seems to take one reading any done
Not sure what you mean by "doesn't do rotation or movement" that - both cameras measure offset and rotation to provide a correction for the placement position.
After doing the camera centre adjust, this should calibrate the camera positions relative to the nozzle, and placement should then be the same regardless of camera used.
There may be some threshold differences between cameras, but not placement offsets.
As mentioned above, after doing the upgrade I noticed some offsets between fixed and OTF, but these went away when I did the centre adjust - don't recall if I only did just the OTF, or both.