Replacing the IR LED emitter
Sonagi
despite weekly use of semi-hot water to rinse the catgenie and using
rather powerful detergent (Ajax dishwashing liquid), and running one
full wet cycle every two uses. So I will replace the LED now, and
search the older posts about replacing the LED. The best info I found
is excerpted below.
Patrick said that he used Radioshack "Matched Infrared Emitter and
Phototransistor Detector Model: 276-142"
http://www.radioshack.com/product/index.jsp?productId=2049723
I think this LED worked really well for him. However, there was
another post indicating that this wasn't the case. In any case,
replacing the LED seems somewhat difficult (having to remove motors,
etc), so any pointers would be appreciated.
Thank you.
On Jul 8 2010, 2:44 am, Robert Deliën <Rob...@delien.nl> wrote:
> > I have located someone with a spectrometer that is willing to test the
> >LEDI removed from my catgenie. Unfortunately, his equipment only
> > extends up to 874nm, though being that the 950nmLEDI used as a
> > replacement appeared differently in my camera, the range may be high
> > enough to determine the wavelength of the originalLED. I'm mailing
> > out the oldLEDtomorrow, and I'll hopefully have some results this
> > weekend.
>
> I hope so. 950nm Is in the upper range of IR LEDs, and your newLEDdoes appear weaker on the camera, indicating it is further out of range of your camera. And the fact that your newLEDdoesn't work as well as the original one, suggests a substantial difference in wave length.
>
> On the other hand, 874nm isn't very far into the near-IR range. LEDs are (almost) monochromatic, so if it's out of range, it won't show up on the spectrum analyser at all. Let's keep our fingers crossed. If it doesn't work, I will see if I can call in a favour of an old college at Philips Semiconductors (though I owe her much more favours than the other way around).
On Jun 30 2010, 7:41 pm, patrick <patrickon...@gmail.com> wrote:
> I went by radioshack today and got an IR LED. It came with a receiver
> as well, however I did not install that.
>
> I first put the PUs side by side and looked at both of them with my
> digital camera. I flashed the malfunctioning unit to beta 1 as well.
> The malfunctioning unit's LED was about half as bright as the good
> unit. I then looked at it with my cell phone camera as well. They
> both appeared dimmer through the cell phone camera, again with the
> malfunctioning unit was about half as bright.
>
> I replaced the LED in the malfunctioning unit, which was a real pain.
> The duct that holds the heating element must be removed for good
> access to the LED, and it does not easily come out. Just about
> everything else needs to be unplugged from the motherboard. One of
> the motors, and it's capacitor need to come out as well. When I had
> access, I used a knife and pliers to remove all of the black adhesive
> that I could, and then used a torx driver to push the LED up into the
> housing. The gray wire that splits off to the receiver is attached to
> the cathode, the flat side of the LED. I think the anode wire was
> blue, but don't rely on me, the colors may be different on a 120, or
> older/newer model 60s. I cut the heat-shrink off and desoldered the
> old LED, and soldered and heat-shrank the new LED on. I used hot glue
> to hold it in place. I then spent about 30 minutes reassembling the
> PU.
>
> I powered it up and again used my cameras to look at the LED. With my
> digital camera, the new LED still looked half as bright as the LED on
> my 120. At this point I thought the problem must be elsewhere. I
> then looked again with my cell phone camera, this time the LED
> appeared MUCH brighter, even brighter than the one on my 120. I'm
> guessing the new LED is operating at a different wavelength.
>
> I reflashed to beta 2 and put the cat genie back together. I'm
> testing it now, and had a bit of trouble getting it to a dry state to
> test everything. While I was doing this, I noticed a rapid clicking
> sound coming from the processing unit when it's paused or idle. Is
> this from the sensor muting? Do you think it will damage whatever
> relay is clicking? I never noticed this sound prior to replacing the
> LED.
>
> In the time I've taken to write this, the genie has filled with water
> several times. This is a good sign. A few more cycles will indicate
> whether the new LED has cured the errors.
>
> On Jun 30, 1:34 pm, Robert Deliën <Rob...@delien.nl> wrote:> 40mA is a lot for a modern led. But I think it should work fine at 13.6mA. A video camera is a very valuable tool here. There are more sensitive to IR than photo cameras and it will give you a good comparison between the old and the new led.
>
> > While you're at Radio Shack, get youself a 220 Ohm resistor and a second LED, so you have something to compare, even after the new LED has been mounted. Three AA batteries in series should provide a voltage high enough.
>
> > On Jun 30, 2010, at 19:17, patrick wrote:
>
> > > Radioshack carries a couple of IR LEDs. I'm not sure if they're
> > > suitable though. One is rated with a voltage drop of 2V, and it's
> > > rated at 40ma. With 5V and a 220ohm resistor, the current should be
> > > limited to 13.6ma. I have to wonder if it will be bright enough. Any
> > > thoughts?
>
> > > On Jun 30, 11:26 am, Robert Deliën <Rob...@delien.nl> wrote:
> > >>> As far as the LED goes, is it on a separate module? If it is, I'll
> > >>> look at my 120 and my 60 with a camera, and then try to swap the LEDs
> > >>> between the units and see what happens.
>
> > >> The LED is in the bottom of the processing unit, one of the two components directly in contact with the light guide. Both are 'glued' in place with some kind of tar like putty. Once removed, you can mount a replacement using hotmelt glue.
> > >> Use a camera to figure out which one is the LED because they may be swapped on individual units.
>
> > >> We're very curious about your results!
Robert Deliën
That component seems like a good fit. It should be a regular 5mm IR LED. I you have any choice, you should select a wavelength of 950nm and an as narrow as possible beam angle. Replacing it is a bit tricky, but very feasible.
First you need to take off the rear shell, as you probably did before.
Next, you unplug all connectors from the PCB, except for the keyboard flexifoil connector.
Now you can remove the front shell from the bottom part.
Removing the hopper motor now isn't that difficult anymore, but if you're skilled in soldering you shouldn't need to: I didn't.
The water sensor LED and Detector are connected to the controller board with 3 wires: Red, blue and grey. The grey wire is split into 2, leading to both LED and Detector. In my case, the blue wire goes to the LED, but I recommend you to verify that it's the same for your box: When the connector is plugged into the controller board, the wire next to the keyboard flexifoil is the one going to the LED, blue in my case.
Alternatively, or additionally, you can identify and mark the LED using a digital camera,before disassembling the processing unit.
The LED is held in place by some kind of bitumen. You can easily knock it out though, using a big allan key as a driver. Put the alan key on the LED, from the exterior-side of the bottom part, and gently knock it with a small hammer. The LED should be knocked inside now. Scratch away the bitumen from both the LED and the hole it came out. Cut the heatshrink open lengthwise, remove it and desolder the old LED. This is tricky because the wires are short and space is limited. You could also knock out the Detector, so you can do this outside the box, but I chose not to, because I didn't know if that would damage is too badly.
Put on the new heatshrink on the wires, solder the new LED and shrink the heatshrink. Put back the LED in the hole and secure it with hot-melt glue (from a glue gun). After is has solidified, you can put back the front shell and connect all wires again. This is a good moment to load GenieDiag and see if it works. If it does, load back CatGenie, put back the rear shell and you're ready to go!
Cheers,
Robert.
Sonagi
Watt 220 ohm resistor to test the led. Did I buy tithe right resistor?
Thanks.
Robert Deliën
(Sent from a mobile device; Please excuse grammar, typos and brevity)
Sonagi
very bright. However, now I have a much serious problem. After a
couple of full cycle, the bowl stopped rotating. I checked, and it
seems the bowl motor is jammed. It was weird because I did not remove
the motor during the LED replacement. I can force it to move with my
finger (very hard push), then it might move for a few seconds, then
will stop. Once it stops, the bowl motor makes a loud grrrrring sound
and some vibration. Then it seems to heat up really hot. I know Robert
tried to get spare motors from China but they won't deal with small
orders. Do you have their contact information by any chance? I have a
Chinese speaking friend and maybe he can maybe able to order.
This would be my third major problem with this catgenie (1. bowl
wouldn't rotate due to gear mismatch 2. dying LED leading to false
positive 3. broken bowl motor---which might be caused by too much
spacers (four in my case) to fix the first problem).
I guess I am back to my trusty Semiautomatic rolling litter box with
flushable litter. I guess I will be on the lookup for cheap used units
on Craigslist.
Thanks.
On Apr 10, 2:07 am, Robert Deliën <Rob...@delien.nl> wrote:
> Yes, that seems to be the right resistor: I will adjust the current to about 15mA and that's what I'd recommend for an IR LED too. Good luck!
>
> (Sent from a mobile device; Please excuse grammar, typos and brevity)
>
Robert Deliën
> very bright.
Aha, so that confirms you old LED was indeed weak and possibly the cause for your water sensor problems.
> However, now I have a much serious problem. After a
> couple of full cycle, the bowl stopped rotating. I checked, and it
> seems the bowl motor is jammed. It was weird because I did not remove
> the motor during the LED replacement. I can force it to move with my
> finger (very hard push), then it might move for a few seconds, then
> will stop. Once it stops, the bowl motor makes a loud grrrrring sound
> and some vibration.
First thing you need to check is if you bowl isn't turning too heavy. On the box I had the problem, I could hardly turn the bowl by hand. Heavy friction on the bowl will cause your bowl motor to wear out quicker.
Second think you need to check is if you can turn the bowl motor manually. You can try this by turning the cog driving the bowl at the bottom of the processing unit by had. If your motor is worn out, it turns lightly in at least one direction.
Finally you can check if the processing unit fits the base properly, though that is not likely to be the cause.
> Then it seems to heat up really hot.
How do you know? This probably means that your bowl has heavy friction, as the motor has to work hard.
> I know Robert
> tried to get spare motors from China but they won't deal with small
> orders. Do you have their contact information by any chance?
Well, not wanting to deal with small orders was just my assumption, as they never responded to my email. But perhaps it's just a language problem. In China it's very common to do business over IM, perhaps that will work better.
This is their website:
http://bt-motor.cn/
You can find their contact information there.
> I have a
> Chinese speaking friend and maybe he can maybe able to order.
I'm in for 3..5 pieces of each bowl motor, scooper moter and drain motor, depending on the price.
> This would be my third major problem with this catgenie (1. bowl
> wouldn't rotate due to gear mismatch 2. dying LED leading to false
> positive 3. broken bowl motor---which might be caused by too much
> spacers (four in my case) to fix the first problem).
They're not built to last, that's for sure. Very sad.
> I guess I am back to my trusty Semiautomatic rolling litter box with
> flushable litter. I guess I will be on the lookup for cheap used units
> on Craigslist.
Or visit the LB-Central forum. People are offering boxes there regularly. A CG60 motor will do fine and you'll only need the processing unit to be shipped (and a pair of long rubber gloves, I'd recommend).
Dylan Farrell
This message uses recycled electrons
Robert Deliën
I have also asked for 110V spare motors and I'll post here when I know more.
Sonagi
and reducing the amount of litter after forcing the motor gear to move
with a screw driver. I hope it will last for a few more months until I
find spare parts.
However, now I have another problem. After replacing the LED, I am no
longer getting the false positive error anymore, but now am getting
the water filling timeout error. I made sure that faucet is fully open
and water pressure is high. My water valve seems to work okay, and
drain motor also works well. So the catgenie will fill the water
(close to the top), stop the water, then will blink the first pattern
indicating water filling timeout. So I have used the drain mode and
tried again, and same error. The bowl is full of water (more than the
non-error state) and the bowl is also rotating. I might let the
catgenie rest for a few hours before trying again. My guess right now
is that the new LED is too powerful and somehow is penetrating through
the water and the infrared sensor is still sensing some light, so the
catgenius is thinking there is no water. The rare false water sensor
error. I might have to replace the infrared sensor with the one that
came with the emitter, or try putting some tape over the infrared
sensor to make it a bit dull. Also this might have been because I am
running full cycle on a clean machine (no poop), and the water sensor
seems to work better with dirty water. Any insight would be
appreciated.
---your previous post where I found some info---
>The LittleLanguage has only two error patterns for the error led. One being 0x01 and the other being 0x05. They spell out 00000001 and 00000101 binary. The pace frequency is 1 byte per second, so for the first pattern the LED is on >for 1/8th of a second and off for 7/8th of a second. For the seconde pattern, LED is on for 1/8th of a second, off for 1/8th of a second, on again for 1/8th of a second and off for 5/8th of a second.
>The patter you're descriping would be 0x55, but that one isn't active in software yet, so I'll have to assume it's the first pattern you're seeing.
>The first patter indicates a filling timeout. And it's usually not associated with a water sensor problem. It's just a safety measure that limits the filling time to prevent floods. If we open de water valve and the water sensor still doesn't see >any water after 2 minutes and 15 seconds, we have to assume something went wrong. That could be be a rare false negative of the water sensor, but since your box realy doesn't fill up, it's probably something else: A closed faucet, >low water pressure, a faulty water valve, relay, relay driver or a software bug.
Robert Deliën
> and reducing the amount of litter after forcing the motor gear to move
> with a screw driver. I hope it will last for a few more months until I
> find spare parts.
That's some good news, at least. I'm still trying to get those motors, but no news yet though.
> However, now I have another problem. After replacing the LED, I am no
> longer getting the false positive error anymore, but now am getting
> the water filling timeout error.
That almost sounds as if your new LED is too bright, making the sensor detect false negatives.
> I made sure that faucet is fully open
> and water pressure is high. My water valve seems to work okay, and
> drain motor also works well. So the catgenie will fill the water
> (close to the top), stop the water, then will blink the first pattern
> indicating water filling timeout.
So if I understand correctly, the bowl fills up but far above the usual level, until it finally stopped because of a filling timeout?
> So I have used the drain mode and
> tried again, and same error. The bowl is full of water (more than the
> non-error state) and the bowl is also rotating. I might let the
> catgenie rest for a few hours before trying again.
I don't think that will help. And even if it does, it may still be unreliable.
> My guess right now
> is that the new LED is too powerful and somehow is penetrating through
> the water and the infrared sensor is still sensing some light, so the
> catgenius is thinking there is no water.
I totally agree. The refraction index of the light guide changes when submerged. That causes the injected IR light to break out, but not all of it. If the remaining part is still strong enough to trigger the sensor, it will conclude the absence of water.
> The rare false water sensor
> error. I might have to replace the infrared sensor with the one that
> came with the emitter, or try putting some tape over the infrared
> sensor to make it a bit dull.
That's rare indeed. I'd recommend to partially cover the LED end of the light guide with black tape. That will at least confirm or our theory.
> Also this might have been because I am
> running full cycle on a clean machine (no poop), and the water sensor
> seems to work better with dirty water. Any insight would be
> appreciated.
No, I don't think that will be of much influence. And even if it does, your box should work reliably even if it's completely clean.
Sonagi
> That's rare indeed. I'd recommend to partially cover the LED end of the light guide with black tape. That will at least confirm or our theory.
>
sense the water.
Using electrical black tape to cover about 50% of the led has allowed
me to run ONE full cycle. So I guess my LED is too bright. :( So the
LED can't be too dim or too bright. LOL The optical sensor is not
reliable at all! There must be some other way to detect other it more
reliably.
I will run the full wash a few more times to make sure that false
negative error is gone. I guess I could replace the LED with something
dimmer (but which one?), but I am too stressed for it now. Maybe using
the matching infrared detector (that was bought with the emitter)
would prevent this problem as it might be less sensitive than the old
detector. Did anyone encounter similar problem after putting in a new
LED?
I really need to get a spared processing unit. I hope the black tape
hack will allow me to run the machine without any problem for a few
months. Thanks for the tip.
Dylan Farrell
This message uses recycled electrons
Sonagi
black tape from the bottom of the LED (from the underside), opened the
catgenie and put some black tape around the LED from the back. Though
I don't how LED leaking light from the back will reach the infrared
detecter as this is still encased in black substance/glue. But I tried
it anyways, and I had one successful run with your method. We will see
how long it will last. Otherwise, back to Robert's method.
While doing this, I noticed that I ripped up the silicone tubing for
the detergent dispenser. No wonder there was no detergent residue on
the toilet. However, luckily, I didn't see any leakage to the
sensitive IC circuits/parts from the broken tubing. I patched it up
very crudely using hard plastic tubing, zip tie, and crazy glue. Do
you know where I can get a similar soft silicone tubing that will work
with the catgenie dispenser?
Thanks.
On Apr 14, 5:31 pm, Dylan Farrell <halvsb...@gmail.com> wrote:
> I had a similar experience for the first dozen times or so after I changed out my led. I thought that it might have been a problem with the water valve closing. I think the real problem was that after replacing, I didn't cover the BACK of the IR LED, and there was some stray IR light reflecting out the back. I put a nice thick layer of black electrical tape around the LED, and haven't seen the problem since. It seems to have fixed the issue for me, but I never thought about it being an issue of too strong out the front.
>
> Sonagi, Did you glue the IR Emitter (LED) back in after you swapped it? did you use something clear, or black? I'm curious if my LED just needed to burn in slightly, or what, as something similar happened briefly to me. That's it, I'm going to make up a capacitance water sensor that works with the cat genius code... just as soon as I have a little free time. Maybe even this afternoon.
>
> Dylan Farrell
Robert Deliën
> negative error is gone. I guess I could replace the LED with something
> dimmer (but which one?), but I am too stressed for it now. Maybe using
> the matching infrared detector (that was bought with the emitter)
> would prevent this problem as it might be less sensitive than the old
> detector. Did anyone encounter similar problem after putting in a new
> LED?
The easiest and most reliable solution to this problem is to put an extra resistor in series with the LED. Just cut the blue(?) wire and solder a resistor in between. The trick is into finding the right resistance.
For that you could use a variable resistor (potentiometer) first. Put the PU on the base, fill it with water using GenieDiag and close the tap. Turning the potmeter should make the valve click at some point, allowing you to tune for the right resistance. Don't turn it too sharply, because there's an averaging delay in software. LEDs are not linear, so I recommend you to get a potmeter of 500 Ohm and one of 2 kOhm in case 500 is still too bright.
When increasing the resistance, you should be able to find two points:
1. First a transition point from water detection to false negative detection.
2. Second a transition point from no water detection to false positive detection.
To find the first point there should be water in the box. To find the second there shouldn't be any water in the box. These point should be very widely apart: Perhaps you won't even be able to find the second one with a 2kOhm potmeter.
Once you have found both points, go back the the first one, increase the resistance to stay well clear of this point. Remove the potmeter, measure it's resistance, round it up to the nearest E12 value and get a fixed resistor with that value.
I'm rather curious about the resistance of the potmeter at the second point too. Could you please post these values?
CAUTION: Picking a resistance too close to the false negative edge may cause brief false negative detections while filling. This will reset the filling safety timeout, allowing longer false negative detections to flood your box. Still not very likely, but I have to tell you.
Robert Deliën
My LEDs is covered with white hot-melt glue. But my sensor is still covered in the original black goo. I think stray light from the back is too few to trigger your sensor, but better rule it out.
> Sonagi, Did you glue the IR Emitter (LED) back in after you swapped it? did you use something clear, or black? I'm curious if my LED just needed to burn in slightly, or what, as something similar happened briefly to me.
LEDs shouldn't need to burn in. If you had this in the beginning, you're close to the false negative edge.
> That's it, I'm going to make up a capacitance water sensor that works with the cat genius code... just as soon as I have a little free time. Maybe even this afternoon.
Write me down for three pieces! ;-)
Robert Deliën
> the detergent dispenser. No wonder there was no detergent residue on
> the toilet. However, luckily, I didn't see any leakage to the
> sensitive IC circuits/parts from the broken tubing. I patched it up
> very crudely using hard plastic tubing, zip tie, and crazy glue. Do
> you know where I can get a similar soft silicone tubing that will work
> with the catgenie dispenser?
The silicon tubing is easy to obtain: Any store selling RC model planes/boats/cars will carry it as fuel tubing. They usually have two types: One for petrol and one for methanol. The latter is silicon tubing and should be fine.
Don't try to glue it, that won't work. Just cut out the damaged part and connect the ends back together using a small hollow rod (by lack of a better word).
Sonagi
using black tape internally doesn't seem to do the trick after one
successful full wash. So I am back to taping up the LED from the
bottom of the case.
I found the tubing and the joinder rod from an Amazon seller
Amazon.com.. they cost about $9 including shipping. (Straight Tubing
Couplers,3/32" - Robart Manufacturing, Silicone Fuel Tubing Standard
3' - Great Planes, both Best Service Stores). My crudely patched
detergent tube seems to work, but I will replace it to prevent
possible future leaking.
As far as the potentiometer is concerned, the closest item seems to be
this: Potenthttp://www.radioshack.com/product/index.jsp?productId=2062300
. 1K-Ohm Horizontal-Style Trimmer
Model: 271-280
I also need to buy a cheap multimeter (there one for $7 on Amazon) and
also learn to use it.
What is the best size for the heat shrink tubing? Also can I use a
hair blow dryer to shrink it as I don't have a heat gun. Otherwise, I
will just use glue gun and black tape to insulate it.
Meanwhile, would it be possible to use one or more of 220 ohm
resistors I bought earlier to test the LED to reduce power to the LED?
I have like five of these, and so I can solder them serially and
hopefully that will dim the LED sufficiently to prevent the false
positive error.
Thank you as always.
Robert Deliën
> Amazon.com.. they cost about $9 including shipping. (Straight Tubing
> Couplers,3/32" - Robart Manufacturing, Silicone Fuel Tubing Standard
> 3' - Great Planes, both Best Service Stores). My crudely patched
> detergent tube seems to work, but I will replace it to prevent
> possible future leaking.
Good. $9 Sounds a bit steep to me, but better too expensive than not for sale at all ;-)
> As far as the potentiometer is concerned, the closest item seems to be
> this: Potenthttp://www.radioshack.com/product/index.jsp?productId=2062300
> . 1K-Ohm Horizontal-Style Trimmer
> Model: 271-280
That one will do. Just use the middle pin and one of the other two.
Don't you have a local electronics store? A Fry's, or a small local owned store?
> I also need to buy a cheap multimeter (there one for $7 on Amazon) and
> also learn to use it.
Make sure you get a digital one. That will be easier to learn.
> What is the best size for the heat shrink tubing? Also can I use a
> hair blow dryer to shrink it as I don't have a heat gun. Otherwise, I
> will just use glue gun and black tape to insulate it.
I usually use lighter. Just stick it in the blue bottom part of the flame, wave the flame up and down the length and turn the tubing, so it will not overhead. Practice a bit before you go 'for real'. When in doubt, you can gently rub it with the shaft of your soldering iron too, but the lighter works much better.
I'm not comfortable with US wire gauges, so I cannot recommend you any size here.
> Meanwhile, would it be possible to use one or more of 220 ohm
> resistors I bought earlier to test the LED to reduce power to the LED?
> I have like five of these, and so I can solder them serially and
> hopefully that will dim the LED sufficiently to prevent the false
> positive error.
Yes, that will most definitely work.
Just put all 5 of them in series and put this series in series with the LED. By shorting more and more resistors, you can vary the total resistance. Once you know how many you need, you can remove the bridged resistors. You can even consider that a permanent solution.
Sonagi
> Yes, that will most definitely work.
> Just put all 5 of them in series and put this series in series with the LED. By shorting more and more resistors, you can vary the total resistance. Once you know how many you need, you can remove the bridged resistors. You can even consider that a permanent solution.
resistors (220 ohm each), the false positive is reduced (like one
false positive during the wet cycle) and the catgenie seems to recover
from the error eventually after detecting water after initially giving
out false positive error. I am kinda puzzled. When I first put the new
LED, I ran successfully two cycles without any errors, but now, I am
getting false positively consistently.
I guess I need to add 1K ohm potentialmeter to my 1100 ohm resistors
(220 X 5 ohms). So I hope something less than 2200 ohm resistors would
provide any dampening to the LED. Is my case just a fluke with really
bright LED? When I looked at the new LED with my phone's camera after
putting in five resistors, I could barely see the blue light, but
somehow this is enough to trigger the infrared sensor. BTW, this is my
new LED, http://www.radioshack.com/product/index.jsp?productId=2049723
(rated 2V 40mA). Is there a more compatible LED I can buy (without the
need of these resistors)?
Thanks.
Robert Deliën
> resistors (220 ohm each), the false positive is reduced (like one
> false positive during the wet cycle) and the catgenie seems to recover
> from the error eventually after detecting water after initially giving
> out false positive error.
As long as the false positive exists, CatGenius will not recover. But it is gone, if only for half a second, the washing program will continue an it will no longer care about the water sensor status.
> I am kinda puzzled. When I first put the new
> LED, I ran successfully two cycles without any errors, but now, I am
> getting false positively consistently.
That is weird indeed. I cannot think of an explanation for that.
> I guess I need to add 1K ohm potentialmeter to my 1100 ohm resistors
> (220 X 5 ohms). So I hope something less than 2200 ohm resistors would
> provide any dampening to the LED. Is my case just a fluke with really
> bright LED?
Possible, where the fluke is a different LED than we expected. Patrick, another group member, used the exact LED from the exact same Radioshack set and it worked for him. The problem with Radioshack is that they are not a controlled source: They buy component from brokers to put in these sets. Whatever they can get their hands on for a good price is what they put in these sets. So there may be a difference there. Check out the reviews on the article.
But it may just as well be something else, like a faulty light detector. Though I think finding a replacement for that is a long shot. The one in the set may work, but if it doesn't, there's no way back as it is hard to salvage the old detector from the black goo.
> When I looked at the new LED with my phone's camera after
> putting in five resistors, I could barely see the blue light, but
> somehow this is enough to trigger the infrared sensor. BTW, this is my
> new LED, http://www.radioshack.com/product/index.jsp?productId=2049723
> (rated 2V 40mA). Is there a more compatible LED I can buy (without the
> need of these resistors)?
There is, but the trick is to find the proper one. There are no type numbers on LEDs and Digikey lists numerous similarly-looking IR emitters.
The new LED appearing very faint on your phone is unexpected. Perhaps it's a clue, perhaps it's just a fluke, as camera's are not measurement instruments.
I didn't realize the LED can be too bright too. I can get you the one I use, but our postal service doesn't have a good record or delivering my parcels...
Cheers,
Robert.
Richard W.
Ive been watching this scenario unfold over the past couple of days. I also
have a unit with a bad IRLED. I went and took a close look at the IR set
from Radioshack that your working with and I think I found a possible
culprit to your problem. That particular photo sensor is a photodiode, not a
phototransistor. So you are getting at least some current through it all the
time. It is not a pure on/off sensor. I know radioshack describes it as a
transistor, but look at the reviews, I believe it may be mislabeled. This
may be the reason for the false positives.
Cheers,
Robert.=
Sonagi
emitter, and not the IR sensor. Were you able to find a suitable
replacement IRLED that works relatively well without the false
positives?
Sonagi
attaching five 220 ohm resistors.
> There is, but the trick is to find the proper one. There are no type numbers on LEDs and Digikey lists numerous similarly-looking IR emitters.
> The new LED appearing very faint on your phone is unexpected. Perhaps it's a clue, perhaps it's just a fluke, as camera's are not measurement instruments.
>
> I didn't realize the LED can be too bright too. I can get you the one I use, but our postal service doesn't have a good record or delivering my parcels...
try adding another 1K ohm resistor (in addition to 1100 ohm resistors)
and maybe even try the infrared salvaged from a touchless soap
dispenser. http://www.amazon.com/Lysol-No-Touch-System-Dispenser-Refill/dp/B0035HIGGA
(I bought a few when they were on sale for $5/each to just play
around. It has a motor, four AA batteries, IR LED and sensor). I
burned the IR led already by pushing pure 6 volt through it, but I
read one of your old posts at the other forum, you said: "1.18 volt
across the LED and across it's series resistor 3.00 volt. The series
resistor has a resistance of 220 ohm, inducing a current of
3/220=0.014 amp), so that shouldn't burn out the LED i hope.
Thanks.
Richard W.
All figured out in one post!
No, I havnt had the time to mess with it. I know it's bad, but it's on a spare unit.
Sent from my iPhone
Sonagi
replacement saga. So I tried various resistor combinations (from 0 to
2.2K ohm) and here are the results.
Crude method without using catgenius diagnose firmware:
1. 0 ohm with no resistor: 2 successful full wash then false negative
water error indicating that the LED was too bright and it resulted in
not detecting water.
2. 1100 ohm: almost a dozen successful full wash then false negative
water. LED was still too bright.
3. 2.2K ~ 2K ohm: No more false negative water LED error, and I was
able to run the machine until yesterday. Then I started getting false
positive water LED error indicating the LED was too dim.
------------------
More refined method using catgenius diagnose firmware:
So after attempts, I finally tried Robert's method using the Catgenius
Diagnose method to trigger false positive LED error using
potentiometer. I used 1K resistor AND 1K potentiometer, so I had a
range of 1K to 2K to fiddle. I knew that 1K was still bright, but 2K
was too dim. However, changing 2K to 1K slowly didn't result in false
negative water where LED becomes too bright. Also, changing 1K to 2K
when the bowl was without water didn't result in false positive water
where LED becomes too dim. If I was able to reduce the ohm further (by
getting 2K ohm potentiometer which Radioshack didn't have at the
time), I might have triggered the false negative water but this
information wouldn't be too useful as I knew 1100 ohm could result in
false negative water error already. I figured lowering the ohm all the
way to 1K would bring back the false negative water error, but that
didn't happen this time.
Instead, I filled the bowl with water (using genius diagnosis
firmware), then started the drain pump. This allowed me observe when
the water stops being detected. Then I restarted the water to see when
the water would be detected (indicated by the red LED being turned on
and water being stopped). So I was filling the bowl with water (with
low water pressure) and draining at the same time to see water sensor
behavior. I tried a lower ohm, then the bowl seems to be filled
slightly more (1cm above the bowl litter mark line). I tried a higher
ohm, then the bowl seems to be filled slightly less (0.5 cm above the
bowl litter mark line). By doing this several times, while fiddling
with the potentiometer, I chose the ohm value (total 1770 ohm) that
seemed to let the sensor detect the water the quickest while avoiding
the false positive water led that plagued me just shortly before.
The above inconsistency in triggering water sensors errors seems to
indicate that either my new LED is fluctuating in brightness (due to
LED's low quality, or power fluctuation within Catgenie, or ambient
temperature is affecting the LED brightness, or water sensor is
progressively getting dirtier). My aim is to avoid cleaning the water
sensor for at least two months without getting either false positive
or negative error. So as my water sensor becomes slightly dirtier with
usage, I will try decreasing the ohm by turning the potentiometer to
make the false positive error disappear without triggering false
negative error.
I used an extension cable to put the potentionmeter above the
processing unit and put black tape to put it there so I can easily
adjust the ohm. High moisture might degrade it but it will probably
last a couple of years before having to be replaced. Right now, I
noticed that the water pump seems to stutter (open then briefly close,
then open fully again) so I am worried this might be the next part to
fail.
Anyways, that's it for now. If you got any ideas or tips, please do
share.
Robert Deliën
> replacement saga. So I tried various resistor combinations (from 0 to
> 2.2K ohm) and here are the results.
>
> Crude method without using catgenius diagnose firmware:
>
> 1. 0 ohm with no resistor: 2 successful full wash then false negative
> water error indicating that the LED was too bright and it resulted in
> not detecting water.
> 2. 1100 ohm: almost a dozen successful full wash then false negative
> water. LED was still too bright.
> 3. 2.2K ~ 2K ohm: No more false negative water LED error, and I was
> able to run the machine until yesterday. Then I started getting false
> positive water LED error indicating the LED was too dim.
That's weird: The brightness really isn't that critical. The brightness isn't linear to the series resistor, so you may have reached a point where a small difference in resistance has a big impact on the brightness, but still it's not what I expected.
I have found a component bag with 3 IR LEDs in it. I suspect these are the remainders of my own replacement LED. I think they are all similar and I think I bought 4, to serve my 3 processing units and still have a spare one. I can send you one, if you like. But it will have to wait until next week, as I am on a holiday this week.
> More refined method using catgenius diagnose firmware:
You need to know that GenieDiag probably still uses the old water sensor algorithm: You probably noticed that the relay doesn't tick with this firmware. Unfortunately Google has deleted the release notes, so I cannot check it from here.
The new algo doesn't pulse the LED. Instead the LED is always on. This may cause the LED or the photo diode to run warm.
> So after attempts, I finally tried Robert's method using the Catgenius
> Diagnose method to trigger false positive LED error using
> potentiometer. I used 1K resistor AND 1K potentiometer, so I had a
> range of 1K to 2K to fiddle. I knew that 1K was still bright, but 2K
> was too dim. However, changing 2K to 1K slowly didn't result in false
> negative water where LED becomes too bright. Also, changing 1K to 2K
> when the bowl was without water didn't result in false positive water
> where LED becomes too dim. If I was able to reduce the ohm further (by
> getting 2K ohm potentiometer which Radioshack didn't have at the
> time), I might have triggered the false negative water but this
> information wouldn't be too useful as I knew 1100 ohm could result in
> false negative water error already. I figured lowering the ohm all the
> way to 1K would bring back the false negative water error, but that
> didn't happen this time.
GenieDiag may not be completely comparable to the situation with CatGenius. When I get back, I will rebuild GenieDiag with the new water sensor algo, so it will be fully comparably.
> Instead, I filled the bowl with water (using genius diagnosis
> firmware), then started the drain pump. This allowed me observe when
> the water stops being detected. Then I restarted the water to see when
> the water would be detected (indicated by the red LED being turned on
> and water being stopped). So I was filling the bowl with water (with
> low water pressure) and draining at the same time to see water sensor
> behavior. I tried a lower ohm, then the bowl seems to be filled
> slightly more (1cm above the bowl litter mark line). I tried a higher
> ohm, then the bowl seems to be filled slightly less (0.5 cm above the
> bowl litter mark line). By doing this several times, while fiddling
> with the potentiometer, I chose the ohm value (total 1770 ohm) that
> seemed to let the sensor detect the water the quickest while avoiding
> the false positive water led that plagued me just shortly before.
>
> The above inconsistency in triggering water sensors errors seems to
> indicate that either my new LED is fluctuating in brightness (due to
> LED's low quality, or power fluctuation within Catgenie, or ambient
> temperature is affecting the LED brightness, or water sensor is
> progressively getting dirtier). My aim is to avoid cleaning the water
> sensor for at least two months without getting either false positive
> or negative error. So as my water sensor becomes slightly dirtier with
> usage, I will try decreasing the ohm by turning the potentiometer to
> make the false positive error disappear without triggering false
> negative error.
No, it'f probably due to differences in water sensor algos between CatGenius and GenieDiag, even though I wouldn't expect this.
> I used an extension cable to put the potentionmeter above the
> processing unit and put black tape to put it there so I can easily
> adjust the ohm. High moisture might degrade it but it will probably
> last a couple of years before having to be replaced. Right now, I
> noticed that the water pump seems to stutter (open then briefly close,
> then open fully again) so I am worried this might be the next part to
> fail.
>
> Anyways, that's it for now. If you got any ideas or tips, please do
> share.
Perhaps you were right the first time and it is the sensor that's at fault... Unfortunately it's no easy swap and it may even make things worse.
Cheers,
Robert.
Sonagi
my current attempts continue to fail for a few more weeks. So just to
recap, 1100 ohm results in false positive (too bright). 2100 ohm
results in false negative (too dim). So I was testing 1770 ohm, and it
worked fine for four days, and I got false negative (too dim). It was
a bit stranger this time as I got the error after two water/drain
cycles, and the program was stuck waiting for the water to be gone
while the bowl was rotating. So you could say I got a false drain
error.
So instead of cycling the machine, I manually rotated down the ohm on
my potentiometer, and the machine was able to start the 3rd/final
water/drain cycle. I turned off the power once the machine is fully
dry, and measured the ohm, and it was around 1550, so I turned down
further to 1390 ohm to make the LED brighter. So I think I am getting
closer to the magic spot where I won't get either false positive or
negative water error.... hopefully. :D
Do you think that the new LED pulsing might explain the inconsistency
of getting LED errors? Or maybe other than having somewhat defective
new LED, I have damaged some other electronics inside by plugging in
the pickit programmer wrongly.
I have not cleaned the water sensor for almost a month, and I want to
keep it this way until I fix the problem by just changing the ohm.
One strange thing I noticed, the ticking sound varies with the
resistance. Without any resistors, the tick was very strong. It became
lower with 2K ohm, but now at the current 1390 ohm, the ticking is
very faint.. fainter than either at higher or lower ohm. Strange. So
the ticking sound can be muted by resistors, but the behavior is not
linear.
On the other hand, you can still find the old pages (some of them at
least) and the firmware as they are still available as of this
moment.
https://groups.google.com/group/catgenius/web/release-history (look
to the lower right part of the page, and you will see some of the
files still linked at Recent pages and files). So at least you can
archive old firmwares, release notes, gardena hose instruction, and
water sensor info. If you know the page title, you can probably put
that into address bar and grab those pages as well. So all the files
are still there, but not linked except through the recent pages.
So https://groups.google.com/group/catgenius/web/(the page tile).
While dealing with my water sensor problems, I was thinking of
possible changes to the way the water sensor errors are handled.
First, the false positive water error: Right now, the water level is
checked right before the water pump opens, and the bowl keeps rotating
and catgenius is waiting for the water level to fall before
continuing. Though the catgenius can eventually recover if the false
water is due to water droplets, etc, but I think based on my
experience, it doesn't recover, and this can actually burn out the
motors. Also the catgenie is not useable for the cats at this state.
How about instead the catgenius checks the water level before the full
wash program even starts or even before lowering the scooper. If it
detects water at this initial check, the program is halted OR the
machine reverts to scoop-only. Of course, the error led will blink to
let the user know it has encountered the error. In other words, have
catgenius revert scoop-only function if it detects water before
actually running the full wash cycle. Even for users with good LEDS,
this change can allow users to avoid doing preventive water sensor
cleaning. They can just wait until the eventual error without any
adverse effects.
I am still thinking about the more rare false positive error....
Thanks.
Robert Deliën
I have found the bag with 3 additional LEDs. do you need one?
Cheers,
Robert.
Sonagi
I removed the Radioshack 950 nm infrared LED. Instead, I put High-
Output 940nm infrared LED (part no. 276-143) without any resistor. It
has been one week since replacing it and I did not get any water
sensor errors, so I think the 950 nm LED was either defective or not
compatible with my catgenie. In fact, the 950 nm LED didn't work more
than a few days at a time without me fiddling with the ohm
potentiometer (where I kept decreasing ohm to clear dirty sensor
error). So if I don't get any error for a few more weeks with the new
LED, I have fixed this particular problem. Anyways, I repaired the
ripped out detergent tube with aquarium silicone tubing and two
silicone tubing connectors bought from a pet store (Petco).
So for those wanting to replace their infrared LED's, I suggest trying
the Radioshack 940nm (part no. 276-143) first instead of the more
expensive 950nm LED that comes with a matching infrared detector.
Robert, thanks again for your help.
Sonagi
replacement after first LED gave me false negative error), I started
getting dirty sensor error (constant and regular blinking) again
starting last night. I cleaned the water sensor and used vinegar to
remove any lime, and tried it again with same result. I also even
tried Beta 1 to see if the older constant LED would prevent the
problem. No go. I switched to completely 100% water solution (no more
sud), but same error.
The interesting point is that the water sensor comes up after first OR
second water filling (where water is dispensed in the proper amount,
the bowl keeps rotating, and water does not drain, and the program is
stuck). So unlike previous water sensor error, where water does NOT
get dispensed after scooping, and the bowl rotates while it's in dry
condition. I am really puzzled by this error as water has been already
dispensed and stopped at the proper level (water has been detected
successfully).
I guess it's possible that my processing board is somehow killing the
infra LED's (3 thus far) due to too much power to the LED. I guess
it's like water sensor suddenly stops sensing water just after
dispensing water to the proper level.
Sorry for the rant but would appreciate any insight.
Sonagi
problem. It takes 59 seconds for the water to start seeping through
the bowl, and almost 2 minutes to fill the bowl(the water sensor
trigger line). Sometimes it takes more than 2 minutes, and this trips
the timer, which stops the water flow and make the red led brink
constantly. After scrubbing the entire catgenie clean along with
cleaning the water sensor four times, I remembered that the usual fill
time for me is 25 seconds (Robert's asked about this time before).
I am not sure whether this is caused by faulty water pump, or low
water pressure in my apartment. I have never used hot water, but I
guess the pump could still fail. On the other hand, I noticed that it
takes 60 seconds to fill my water-saving toilet tank (which is 1.6
gallon/6 liter per flush) which seems too slow. I read on the other
forum that Catgenie uses 5 gallons of water each cycle, so if this is
the case, my water pressure is half the usual for some reason.
Hopefully the apartment's water pressure will go back to normal.
PS. While cleaning the catgenie inside-out, I noticed that one of the
rollers popped out and was in the drainage compartment.
Robert Deliën
Hi. After almost a month of trouble free usage (after my second LED
replacement after first LED gave me false negative error), I started
getting dirty sensor error (constant and regular blinking) again
starting last night. I cleaned the water sensor and used vinegar to
remove any lime, and tried it again with same result. I also even
tried Beta 1 to see if the older constant LED would prevent the
problem. No go. I switched to completely 100% water solution (no more
sud), but same error.
The interesting point is that the water sensor comes up after first OR
second water filling (where water is dispensed in the proper amount,
the bowl keeps rotating, and water does not drain, and the program is
stuck). So unlike previous water sensor error, where water does NOT
get dispensed after scooping, and the bowl rotates while it's in dry
condition.
I am really puzzled by this error as water has been already
dispensed and stopped at the proper level (water has been detected
successfully).
I guess it's possible that my processing board is somehow killing the
infra LED's (3 thus far) due to too much power to the LED. I guess
it's like water sensor suddenly stops sensing water just after
dispensing water to the proper level.
Sorry for the rant but would appreciate any insight.
Robert Deliën
> problem. It takes 59 seconds for the water to start seeping through
> the bowl, and almost 2 minutes to fill the bowl(the water sensor
> trigger line).
Did you figure that out using GenieDiag? That would actually be a good test. GenieDiag will stop filling if the sensor has been triggered, but not after a specific timeout. So if the time betwee valve-open and vavle closed (by water sensor) is greater than 2 minutes and 15 seconds, it's a water flow problem. CatGenius uses this 2:15 as a timeout because we have found out that even a low pressure water system can fill up a bowl in that time, and a high pressure water system will not flood you bowl in that time (even though water will be way too high after that time).
The filling time of a typical water system should be under a minute, from valve open to valve closed.
> Sometimes it takes more than 2 minutes, and this trips
> the timer, which stops the water flow and make the red led brink
> constantly.
Well found; 2:15 is the magic number here.
> After scrubbing the entire catgenie clean along with
> cleaning the water sensor four times, I remembered that the usual fill
> time for me is 25 seconds (Robert's asked about this time before).
This 25 second isn't the complete filling time. It's the filling time for the base, below the bowl. After the water valve opens, the water should run out through the bottom of your bowl for a little over 25 seconds. After that, the level will be high enough to start flooding the bowl. If your bowl starts flooding within this 25 seconds, it's a good indication that your bowl's drain slits may be obstructed.
I needed this 25 seconds for a new feature: Flushing before washing. It basically means that we first flush out the waste before we start cleaning the box, preventing the waste at the bottom to contaminate the bowl. I will introduce this feature 'soon'. I have just tried the pause function and it didn't resume properly, so I have to figure that one out first.
> I am not sure whether this is caused by faulty water pump, or low
> water pressure in my apartment. I have never used hot water, but I
> guess the pump could still fail. On the other hand, I noticed that it
> takes 60 seconds to fill my water-saving toilet tank (which is 1.6
> gallon/6 liter per flush) which seems too slow. I read on the other
> forum that Catgenie uses 5 gallons of water each cycle, so if this is
> the case, my water pressure is half the usual for some reason.
> Hopefully the apartment's water pressure will go back to normal.
Jaime has discovered that hot water can damage the water valve in a way that it will obstruct the water and hence will make filling time longer. You can test this by filling it with a garden hose. The water sensor is a fairly universal type, as often used is laundry machines and dishwashers.
Also check the slits at the bottom of your bowl: If they are clogged, the water flow to the base is delayed and this may just be enough to trigger the timeout. Especially if your pressure is already low.
Why do you think your pump has failed? Your CG has two pumps: A really small dosage pump, to pump in cleaning fluid. And a big draining pump, to grind and pump out the waste. For filling, the box is entirely depending on your water pressure, like nearly all appliances.
> PS. While cleaning the catgenie inside-out, I noticed that one of the
> rollers popped out and was in the drainage compartment.
Just pop it back and you'll be fine. It's inconvenient, yet harmless.
Sonagi
considering that it takes twice as long to fill my toilet bowl as well
(63 seconds to fill 1/6 gallon). I guess the apartment is really
trying to save money on water or it's just a transient problem.
Anyways, if the water pressure doesn't increase, can you please give
me the hex address to modify the water filling time on the next
version? I think changing it to 2:45 minutes will fix my problem.
Thanks and have a great weekend.
Robert Deliën
> considering that it takes twice as long to fill my toilet bowl as well
> (63 seconds to fill 1/6 gallon). I guess the apartment is really
> trying to save money on water or it's just a transient problem.
> Anyways, if the water pressure doesn't increase, can you please give
> me the hex address to modify the water filling time on the next
> version? I think changing it to 2:45 minutes will fix my problem.
> Thanks and have a great weekend.
It's not that easy. Currently the maximum fill time is defined in a C macro, so the actual value will end up somewhere in the assembly code. I will see if I can find it.
Perhaps a better solution is to store this parameter - and other box tuning parameters - in the EEPROM area. This way you can easily tweak them using your PICkit programmer, without reflashing the whole firmware.
I can make a version for you later today, but to keep my workspace organized, I would prefer not to go back to version B03 for that. That means that you will have a borked pause function.
Cheers,
Robert.
Sonagi
just the hex address would be really sweet. Are you in the process of
parameterizing the firmware, or is this on the future todo? I don't
mind the borked pause function but it would be interesting to test out
the new pre-flush routine. Do you find this improvement useful? I
remember there being a big bug fix contributed by the new contributor,
so a changelog would be appreciated ;D
Thanks.
Robert Deliën
> just the hex address would be really sweet.
I will build you a personal version tomorrow: Today the weather was too nice to sit inside. I will mail it to you directly, because I don't want it to wander around on this group: Such a long filling timeout will be harmful for most other users because it could cause a box to flood in case of a dirty water sensor.
> Are you in the process of
> parameterizing the firmware, or is this on the future todo?
Yes, but that's not a lot of work. Getting it to work properly after updating your box is. My highest priority is to get the pause function to work properly, because the improved error handling is depending on it.
> I don't
> mind the borked pause function but it would be interesting to test out
> the new pre-flush routine. Do you find this improvement useful?
Your personal version will have the borked pause function and the longer filling timeout. Pre-flush will come later. I think it will be useful because it will flush out the waste before it can contaminate the bowl and granules.
> I
> remember there being a big bug fix contributed by the new contributor,
> so a changelog would be appreciated ;D
That fix will be in, but it only hits the new compiler version I'm using. The one I used for the version distributed so far didn't suffer from it.