[gpounce]

Hi all,

I built a version of this for my old Bandit, and used it on my R6 as well, for about 10 years all in. It was all discrete components and worked fine. It went with my R6, so I designed a version 2 to go the gsxs- this one using a cpu (Microchip pic) which adds a bit of flexibility. Pics follow;

The reservoir is an HDPE bottle which goes under the pillion seat. A submersible micropump (TCS micropumps, out of the UK) goes into the bottle, along with associated plumbing. The external hose is the air vent into the bottle, with the pump leads running in it. The unplumbed external fitting leads down to the chain. When rigging the outbound tubing, it should lead up above the level of the bottle with a vent, then downwards to the chain so as to avoid siphoning.

The pump wants 3v or so- a bit less- so the board runs it via a resistor, to prevent overdriving. TCS makes an interesting variety of pumps. My previous design used a self-priming diaphragm pump, I wanted this one to use one of their submersible impeller designs, to reduce the plumbing and hopefully avoid the need for a check valve.

Having a display is one advantage; the leftmost digit shows the count up to the wheel rotation limit in 16ths, so progress of the wheel tick count can be shown. The middle two digits are the wheel limit in hex, meaning from 3 wheel counter ticks up to 255. The rightmost is the pump run-time in seconds from 3 up to 15. The two multiturn pots beneath the display are sampled via the pic's adc to facilitate adjustment. The software computes a running average of 8 samples taken on 100ms intervals, so the least-significant bits are more stable.

Operationally, the box goes under the pillion seat next to the bottle, with the top easily removed to afford adjustment of the pots. I like to get the settings adjusted so after a 30 min ride on the highway no more than 1 drop will fall onto the ground from the chain after a few hours of sitting. Once thats dialed in the only maintenance is checking the bottle's oil level and occasionally cleaning accumulated dirt off the swingarm.

Inputs are accessory power and a 3 wire connection from the ECU's sensor power, rear-wheel rotation sensor, sensor ground. As seen in the schematic, I run the ECU signal thru an optoisolator to a 4020 counter tapped so the pic gets a wheel counter tick at a rate / 256- the wheel sensor is quite high resolution. The only output is a wire pair going to the pump.

When I integrate it in the bike, having the logic in software makes it easy to adjust the tick counter divisions. I certainly could change the 4020 tap also. Once the software is stable I'll burn it onto an OTP device... these eprom chips are great for development but its better leave them in the project box when not in use...

I got the software working properly tonight driven by my signal generator, so its nearly ready to test on the bike- next is to get the board mounted in the project box and the harness finished. I'm using Sumitomo connectors- a number of HM and MT's to spare so it will be nice to use them up. The only big mod on the bike is to tap into the ECU wiring which will be a careful bit of work- the idea being to cut off a short bit of insulation from the given wires, then solder on the leads and re-insulate.

 

[gpounce]

Between one thing and another I made slow progress with the unit, but finally got it buttoned up. A diaphragm pump in the small black box sucks oil from the upper reservoir (which is stored under the pillion seat), and pumps it out thru a delivery tube with spring ball check valve which cracks at 1psi, so there is no difficulty with siphoning.

The control unit counts pulses from the rear wheel speed sensor, and based upon interval and duration settings, runs the pump. I was hoping to tap directly off the wheel sensor by introducing a man-in-the-middle connection but unfortunately the sensor isn't obviously ground or +12v referenced so is troublesome to use directly. Instead to keep things simple I spliced into the harness at the ECU, which helpfully supplies the wheel sensor signal as a convenient 5v square wave, and a +5v supply rail for sensors and ground. Accordingly, the signal is easily propagated thru an optoisolator without introducing ground loops.

I opted for no external controls since the unit is easily unplugged from accessory power if it needs to be turned off, and the box's lid is easily taken off (when I make the 4-48 thumbscrews) for adjustment and priming.

The pump is a TCS Micropumps diaphragm unit, driving at 3.8v or so, it is capable of up to about 10psi- kind of a surprise given its small size. Its a nice high quality unit, very happy with its performance. Being able to pump through a ball check valve avoids having to run the tubing high up in the tail with a t-connector to provide a vacuum breaking vent. It does have some difficulty pumping air bubbles through the check valve, so when I put in the tubing I will include a t connector with a manual valve that can be opened to relieve the backpressure when priming. I hope that works out- the vacuum breaker setup leads to lots of tubing all over and is something of a nuisance.

The TCS pump adds substantially to the cost of the unit OTOH is very conveniently purchased as compared to most other micropump vendors who make finding their products a pain due to old-fashioned distributor models. I found some small far-East diaphragm pumps on ebay... they might work out, or not. The TCS unit certainly does.

I did a bit of pipe strangling with the copper tubing, which is forked down at the delivery end to drip onto the o-rings on both edges of the chain. I'm not convinced about this arrangement but its a good start. The pic may give the impression the delivery tubes are very close to the chain, but they are at least 1/2" above several inches before the front sprocket- so the chain <should> clear them even when it whips. I took the time to make brass tubing adapters to make it easier to hook up (and take them apart).

I put the board in a box w/ transparent lid so operation can be ascertained without taking it off. Adjustment of interval and duration is via small screwdriver adjustment of the two pots. Green led blinks with the sensor ticks input, red when the pump is being driven.

Leftmost digit is the realtime count up to the oiling interval in 1/16ths (in hex). Middle two digits is the oiling interval in 1024 counts of the wheel sensor (in hex, 8 bit divisor via the CD4020 and 3 bits of software prescale). The rightmost digit is the number of seconds the pump will run when triggered, from 2 to 15. The software constantly reads the two pots at 10hz, averaging over 4 samples, and updates the display as they are adjusted. The button directly beneath the display is to manually run the pump, for testing/priming etc.

As a heartbeat, the software toggles the display brightness at about .5hz.

The only snafu with integrating it on the bike was finding the sensor +5 rail on the ECU connector. The service manual says that is a Red wire on the S1000 or Blue on S1000T. I have a S1000T but the Blue is Neutral sensor, and the +5 sensor rail is actually on Red. So that took a bit of messing around but the harness is all done up, my spliced in wires running down thru the vacuum line. The wheel sensor counts fine, so the final step tomorrow is putting in the unit and tank, and plumbing the oil lines- and then a test.

I find splicing into a new bike's harness is a bit uncomfortable but the harness is well laid up; not too right and easy to fish through to find the wires of interest. I did not have to disassemble the shroud. I found it satisfactory to strip the harness wire ends about 1/4" and the wire being spliced in is stripped about 3/4", then the two harness ends are spliced to; the spliced-in wire adds length so overall tension and layup is not affected. I soldered each splice, coated with dielectric grease and heatshrink tubing. Then a few turns of electrical tape and a wire-tie to close up the bundle.

 

[dekker]

Seems to me you should copyright it quick before someone steals it.
Even if you're not going into production.

 

[gpounce]

👍 nothing much original or innovative in the design.. and would be impossible as a product...I just really dislike chain maintenance 🤣

I think a design win would be to use a gps or imu chip to determine speed. Thought about doing that here but was not convinced on performance, or that the pic had enough i/o and program memory for the math. I could move to a fancier cpu but that means a comprehensive redesign, so I opted for simple instead.

 

[Delbert]

Fair play to you , a lot of work and thought gone into that!!

 

[Bruce00]

This is so far above my head, it might as well be a moon landing....LOL
Kudos to you sir!
I haven't dealt with chain maintenance since the late 70's, as all my motorcycles have been rear belts or shafts.
And here I was thinking a couple squirts as part of my saturday morning garage rituals, but I see the 20th century oiling's have taken a giant leap forward

 

[gpounce]

Just for the record, I finally found an oil reservoir that fits well under the pillion seat; a Dubro 12oz model airplane fuel tank- the 14oz <might> fit but the 16oz is a bit too long. The volume under the seat is just a bit too small for the usual sort of 500cc (16oz) HDPE bottle- so the 12oz tank is better than going down to 8oz, and it has the necessary dual tubes built into the cap with a flexible pickup tube inside the tank.