Alternative water sensor mod

[Paul Gumerman]

Over the course of a week or so, my CG120 started throwing the 3-beep error.  It was puzzling, because the litter was clean and dry.  Further observation revealed that it was only dispensing water on the first of the three wash cycles.  Thinking that it was the usual dirty sensor probe problem, I cleaned the probe, and even polished the ends, but the problem only got worse, to the point where it refused to dispense water at all.

Doing research led me to this group, and I eventually found the schematics as well as a description of the failure of the IR emitter.  My problem looked suspiciously like an emitter failure.

When I disassembled the CG to try to get at the emitter, I ran into the fact that both the emitter and the sensor were glued into place.  I wanted to remove both so that I could let them "see" each other, and so that I could test the few IR LEDs that I had on hand to get one with the right wavelength.

Unfortunately, I broke the sensor while removing it.

So, I thought I'd try to make the existing comparator circuit work with actual metal probes dipping down into the bowl.  Electrically, it wasn't too hard.  First, we need to bias the input toward ground, putting an 18K (approximately) in place of the sensor, between the comparator input (J10 pin 1) and ground (J10 pin 2).  Extend the sensor wire from J10 pin 1, that will be one probe wire.  J10 pin 3 will be the other probe wire, so extend that one too.

I found some plastic tubing that fit the holes for the original water sensor lightguide, cut it to 4.5" length, fed the wires through, soldered on some tiny stainless steel bolts that fit the tubing, and then epoxied the tubing into the small holes on the bottom of the main unit, where the top of the lightguide used to sit.

It's been working flawlessly now for the last 3 days.

A couple of notes: first I tried 22K, and that worked for ONE cycle, but when the CG heated up, it failed.  Apparently it was just too close to the hairy edge, lol.  I put in a trim-pot with an access hole, and adjusted it on the bench.  I didn't measure the value, but I'm betting 18K should do just fine.

Also, I realize that the heavy main unit with two 4.5" long, slender plastic tubes glued to the bottom is just waiting to be broken.  I now need to be extremely careful when putting the main unit back onto the base, feeding the probes into the lightguide slot and keeping everything aligned while lowering the unit into place.  If I were to do this again, I would use stainless steel TIG welding rod, insulated with cocktail straws, glued into the base where the lightguide was.  I would then make some spring-loaded contacts in the space where the emitter and sensor were, which would make the mod much more robust.

[Flyboy BC]

I am impressed! Such an elegant solution.

I will replicate your modification and report back.

[Paul Gumerman]

Thanks for the compliment :)  Happy to report that we're still running just fine.

In this case necessity was very much the mother of invention.  If it wasn't clear from my original message, the presence of the water is sensed by the +5 from J10 pin 3 being conducted by the water to the other probe, and that additional voltage being enough to trip the comparator.

Also, to save on frustration, you can bench-test the resistor value.  Jam something into the power switch on the bottom of the main unit, and place it into tech mode.  Then select test 3 to run the water.  When you hit start, you should hear the water valve relay and the water value click on.  If you don't, then the resistor value is too high (or you have some other issue).  If the valve is open, touch the probes to water and it should close.

I might have another CG coming from a friend, in which case I'll definitely figure out a way to make contact with probes that are mounted to the base using some sort of spring contacts.  I already have a good mental picture of how to do the probes themselves; it's just the spring contacts from the main unit to the tops of the probes that I am still pondering.

Let me know what your thinking is on that subject, and don't hesitate to reach out if you have any questions.

[Paul Gumerman]

Just had  my first error on the CG since my mod. Single-beep (it thinks it still sees water when it should not) error. Pulled main unit w/probes out, to find some cat hair bridging the stainless probes. Wiped it off ... back in business.  

Based on this, if I were to this mod again, I would have the stainless probes be of slightly different lengths, perhaps 3/16" or so.  Also, I would not have them extend past the insulating straws, but have the probes extend just to the end of their respective straws.  This would lessen the chance of any debris creating a conductive path and fouling the sensor.

[Michael Conner]

I've used Honeywell LLE liquid level sensor in a tet CG-120 with very good success., These LLE sensors been around for decades and are quite reliable. 

I bored a hole and mounted a LLE (with Schmitt Trigger output) on top of the Pump / Macerator assemblyHousing with a jacketed cable and disconnect. You can also wire in a LED to indicate if the sensor needs cleaning.

The drawback is the CG's have a weak 5 Volt supply, so for longevity you'll need a solution.

Second,  To clean the sensor you must remove the pump, open it up for cleaning, not quite as handy.

The probes were common in the Marine and RV marketplace, but require heavy maintenance, cleaning. 

You can read the spec sheet o the LLE sensors and find their Tech notes that will provide you will a solid understanding of their design.  

One of the CG102's has a Flaky and Intermittent +5 Volt supply.  I'm planning on repairing this power supply and installing a Schmitt Trigger Output replacements, with an LED on the Control Box panel by the buttons,, indicating the State of it's output. 

Michael..

--
You received this message because you are subscribed to the Google Groups "CatGenius" group.
To unsubscribe from this group and stop receiving emails from it, send an email to catgenius+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/catgenius/61a49363-331f-4a6f-adce-2dce7948c5d1n%40googlegroups.com.

[Paul Gumerman]

My previous mod to the water sensor worked fine for about a week or so, then seemed to become unstable; the CG would throw either a three-beep error or a one-beep error, I'd    adjust the variable resistor to make it go away, it would run for a day or throw an error on the next cycle.

I got tired of that, and decided to deal with it.

See my new mod here: New ACTIVE water sensor mod