JCB 803 Saga

[awemawson]

So time for a little more investigation. The wiring diagram shows the gauge is in series with the transducer and the pair are across the 12v supply from the ignition switch - so no surprises there. Pulling it out of the dash and testing it on the bench across the connections purporting to be the only ones going to the indicator it measured 150 ohms - so reasonable. Applying 12v from a current limited power supply moved the needle to full scale - whoohay  But absolutely nothing short of poking the needle with a finger would move it back to zero from when the 12v supply would move it back 

Doing a bit of    and  Google gave me interesting facts about fuel gauges - seems that there are several variants - I'd presumed that they were basically a crude ammeter - but this style has a 'low' coil, and a 'full' coil as well as a third the purpose of which is a bit confusing 

About this time I thought it best to measure the tank transducer - or 'sender' as they seem to be called - basically a variable resistor operated by a float. In this case there is a vertical rod fitted from the base of the tank, surrounded by a float that I assume to be magnetic, that rises and falls with fuel level. All very well, but surely it shouldn't be INFINITE ohms  Yes the transducer is open circuit. I strongly suspect that the indicator has failed putting too much current through the 'sender'

So a parts hunt ensued. The 'sender' is available so I ordered one, but the gauge is made from unobtainium, I traced there to be ONE spare in the world, and the dealer somewhere in the US wanted $7K so obviously wants to keep it for ever !

The actual gauge module unplugs as per these pictures

[awemawson]

Try as I might I could not track down any vehicle gauge or even normal ammeter, small enough to fit in the available space - so what to do 

Well it's not rocket science - we need a bit of electronics that basically shows a reading varying with the variable resistance of the sender - in this case I measured 9 ohms with the tank dry, and 180 ohms with the float all the way up.

After a few false starts playing with miniature digital volt meters I thought it best to design something myself using the LM3914 Bargraph Driver. This chip accepts a DC voltage, and drives it's ten LED driver outputs in an increasing bargraph representing the voltage coming in. It also has provision for scaling, zero offset, and brightness, all in a single 18 pin chip for about a pound sterling !

So I set off pulling out bits of Veroboard to wire up my simple circuit and then thought - I bet someones done this before - go west young man and GOOGLE 

Sure enough, a nice man in Canada had used the self same chip, in his case to make a VU audio meter, and was selling the very nicely made PCBs on a well know auction site for absolute peanuts including postage, so I placed an order

[awemawson]

So by leaving out his op amp pre-amplifier, and putting links where he had capacitors for AC coupling I ended up with precisely the circuit that I had intended to glue together on strip board but in a much neater format 

Firstly I built it to run directly off the 12v vehicle voltage, but testing showed that as battery voltage varied from 11.5 to 14.5 there was an unacceptable variation in indication. As the PCB allows for insertion of a regulator I tried regulating down to 8 volts and running off that. Excellent results that are stable over the range of voltage that I'd expect in the machine.

Only down side of this is that the regulator dissipates a watt or two and runs hot, and will need a heat sink when mounted. I tried to reduce the dissipation by putting the LM3914 chip into 'single bar mode' so that a single bar 'walks' the indication, but there is an oddity with this integrated circuit that when in this mode, the tenth bar never illuminates 

At the moment I've set it up so that only 24 mA passes through the 'sender' - a nice low level but it might prove noise sensitive. Easy to alter as it means changing a single resistor and re-tweaking the twiddle pots on the PCB.

Here is my prototype running on the bench and using my wire strippers as a heat sink 

(The LEDs are green asnd although bright enough in real life don't show up very well in the pictures)

[awemawson]

I have some red bargraph LEDs and might try those.

All that remains is to physically mount it in the instrument cluster - there is enough room but will need a bit of cogitation to work out a best way - then make a mask to put in front to conceal all but the LEDs.

Then of course I need to drain the diesel tank and install the new transducer, but it's wetter than a wet thing with wet bits on it outside so that might have to wait until the Ark is built and all the pairs of animals are safely aboard 

[awemawson]

So how to mount it? I decided to copy the way that the hours meter is mounted, using 'posts' of thick wire soldered to the board, with tubular spacers separating the two pcb's.

Drilling holes matching those already in the display pcb placed 3 conveniently within the ground plane on the main pcb, but the fourth emerged right between a 12v trace and ground, so some careful soldering required so as not to short the supply out 

[awemawson]

I decided to mount the voltage regulator on flying leads off the board, and bolt it to the largish 0v copper plane on the main pcb as a heat sink - time only will tell if it is enough.

[awemawson]

I think I will just snip off the thick mounting wires and bend them over as retaining tabs rather than solder them, so that I can easily disassemble it in the future should the need arise.

I should perhaps point out that this is a different display board from the prototype one in my earlier posts that used ic sockets - I decided to solder in the display and ic as it's likely to get a bit of vibration, and anyway the lower profile was an advantage.

So a quick test to see it works - phew it does - then a tweak with the pots to calibrate 'empty' and 'full', and all that remains is to cut a bezel to surround the LEDs and hide the electronics

[awemawson]

So I cut a bezel - oh I like my Laser Engraver 

Popped it in the hole, secured by a drop of super-glue and put it all together for a soak test to see how hot it gets.

[awemawson]

That's actually the second bezel I cut.

Both were cut from some smoked brown perspex that I have squirrelled away that were water screens on my wire EDM machine that I replaced with clear polycarbonate. The perspex was a bit scruffy, so when it had been cut I cleaned it with IPA (Industrial Pure Alcohol - not the beer  ) and it instantly started forming stress cracks.

Now I've heard of this before but never seen it. The second bezel was cut from the same sheet, in fact with in a few mm of the first one, but this time I cleaned it with soap and water.

Have a picture of the cracks:

[Bluechip]

Oh yeah, done it myself 

One day at work on a IBM 1403 line printer, the print cartridge oiler reservoir was a Perspex block arrangement and it had some crap in it, so Dick-Ed here squirted it with AF spray .... 

Just like dropping an ice cube into water .. 

Didn't quite fall apart so I wrapped tape around it and ordered a new one. 

Dave

[AdeV]

IPA (not the beer) is Iso-Propyl Alcohol. Never heard it called Industrial Pure before - did you get the Chinese stuff? 

If you get bored of the linear bar graph, Sparkfun do some rather cool radial ones: https://www.sparkfun.com/products/11492

I bought a stack of LED products a while back, pretty sure I got some of the radial ones, you're welcome to one if you like.

[awemawson]

... heat ... heat ... heat 

This module only consumes 2.2 watts at the worst case of 15v battery voltage (on charge) and all bars illuminated. Yet on my soak test the regulator was getting unacceptably hot (48 degrees C after a couple of hours)

The dissipation is split about 50 / 50 between the regulator ( 15v -> 8v @ 150 mA = 1.05 Watts) and the LEDS & drive chip. I'm not too concerned about these as the dissipation is over a couple of square inches of board, but the regulator needs sorting.

I'd tried using a largish area of the 0v copper pcb, and when that was insufficient bent up a small sheet copper fin that the bolt fixing the regulator retained. This improved things, but all the heat to the heat sink had to pass through the 3 mm bolt so not ideal.

Not much point doing anything inside the case as the heat has to be got away. I hit on the idea of drilling a 10 mm hole in the pcb for the end of a cylindrical external copper heatsink to protrude into the case with the regulator directly bolted to it.

I turned a 1" bar as per the picture below, tapped it's internal end M3, painted it matt black and super glued it to the rear of the main pcb.

So at least now the regulator heat is exported fairly efficiently, and it's back on soak test to see the results

You'd not think that a couple of watts would be so much trouble 

[DMIOM]

Andrew,

If you want a cab heater, then ignore this .....

Otherwise, hot electronics => increased failure rates.  It sounds like you're using a linear regulator, which is just chucking the power out as heat. I only uses linears on very small projects now; there are pin-compatible switching equivalents for most of the 78xx/79xx series regulators, and they can approach 90+% efficiency.

If your board is now permanently populated with a linear regulator, I'd finding out what its minimum input voltage is and then running an upstream switching reg so that the linear reg does the least amount of work necessary.

Dave

[John Rudd]

Minimum delta-v for the 78 xxxx series is 2.0 volts.....
There are drop in switchers on a pcb with a TO220 pin out....but expensive ( Ebay - typically £4 ish for a 7805 replacement) oops DMIOM already said....

Can't you reduce the led current a bit?

[Pete W.]

SNIP

Can't you reduce the led current a bit?

It's surprising how dim an LED can be out of doors when the sun is shining!

Well, to be pedantic: the LED is just as bright but the high ambient sunlight causes the eyes to accommodate, so it LOOKS dimmer!!!!

[awemawson]

Thanks chaps, but I'll stick with the linear regulator, as the new heat sink is doing it's job 

The rain actually stopped after lunch, no longer giving me an excuse not to drain the fuel tank and fit the new sender. Draining the tank was easier than I'd expected as I managed to disconnect the feed to the main filter and extend it into a 25 litre drum. When I got up close and personal with the sender, it turns out that the original has a whacking great 2" AF nut, and although I have spanners that size there's no way I could get past the gubbins such a slew rams and pipes, so the tank had to come out.

Actually not too bad a job. flow and return pipes of course, and four bolts and out she came. Fortunately the thread on the new sender is also 2" bsp, but the sealing methods were not the same. Original tightened up on a flange with a sealing washer, replacement was intended to seat onto a female taper.

No way I'm cutting a 2" taper on the tank boss, so I've put my trust in Loctite 577 pipe seal, which is pretty fantastic stuff for threads over 3/4" bsp. If there's a puddle of diesel under the machine tomorrow morning I'll eat my words 

So putting it all back together, I firstly checked that the gauge showed empty, then I slowly syphoned the diesel back into the tank while carefully watching for drips. None so far, and we have 'three bars' on the new fuel gauge.

A quick trundle down to my 'red diesel' bowser, fill her up sir ? - yes please, and lo and behold we have all 'ten bars'.

So at long last we have a working fuel gauge - what a palaver 

Only one more issue to sort now on this digger. One track can track at slow and fast rate, whereas the other will only do slow rate, which unless you know, leads to some interesting driving. There is a spool valve bolted on each motor that controls oil flow (either fast low flow, or slow high volume) and I'm pretty sure that one is stuck - it's another crawl about on the ground job I keep putting off 

[AdeV]

After all that work, don't tell me you're not going to de-rust & repaint it? Or will you leave it with that "rat look"?

[appletree]

After all that work, don't tell me you're not going to de-rust & repaint it? Or will you leave it with that "rat look"?

Yea I was just about to say the same LOL

[awemawson]

Well it's a working tool. It needs to function and be reliable  but doesn't need bling 

If for instance I removed the dipper arm to weld up the worn bushes and re-machine them, I'd probably sand blast it and respray, but I've no intention of doing a cosmetic makeover - can't have the local travelling fraternity thinking it's worth nicking 

[John Rudd]

Well it's a working tool. It needs to function and be reliable  but doesn't need bling 

Says he just adding an led fuel gauge..,,

Pimp my JCB.....lol....

[PeterE]

Ah, so paintwise you are going down the "too much used camouflage" look then ... 

[awemawson]

To be honest, unless you totally dismantle a machine like this for re-painting, the 'repaint' looks a mess. Just a 'quick blow over with a spray gun' may make it 'ten yards pretty' but the charm is lost when you get close 

[PeterE]

Of course, I agree. Like most  farm and industrial machinery it will get that "used" look and there is no way of keeping a polished one without a recurring stripdown - and that does not improve the machine performance so it is just as good to skip that.

But keeping a tin or so of the main colour for patching is useful, as it will make it more agreable to have a machine that looks maintained.

I am impressed by your skills, fearlessness and ingenuity to take on that wide variety of projects    I always learn a thing or two for each thread I read.

[awemawson]

Thank you for your kind words Peter. It's reassuring to hear that someone is reading my drivel 

[Stilldrillin]

I've been here, from the start.   

I am impressed by your skills, fearlessness and ingenuity to take on that wide variety of projects    I always learn a thing or two for each thread I read.

My thoughts, too....... 

David D