[Reg in Bristol]

OK

as requested, heres the BASIC circuit

I didnt actually draw it out before just knocked it up on the breadboard as

its so simple and have drawn it here from memory as a guide.

Its based on the 'typical applications' example in the national semiconductor LM3914 data sheet.



I'm still intending to tweak mine and will also add smoothing caps and possibly reverse current protection



This circuit shown, crude as it is, works on my bike so should work on yours

I coloured the leds I use for clarity and as an example



What it is:

Its a simple 0-5V voltmeter, working in reverse for our purposes and in 'dot mode'

connecting pin 9 to supply for 'bar mode' would look odd due to the way we use the IC

and would light all the leds at Ign On and reduce the count as the voltage at A drops

You could do it but as I said, I think it woudl look odd and I wanted simple indicator

and nothing too distracting.

the blue arrow shows the direction the leds light as the temp rises.



How it works:

at Ignition on, the voltage at A is so near 5V that the green led on pin 10 lights

and this serves as a power indicator telling us its on.

As the temperature rises in the engine, the sender resistance drops and so does

the voltage at A

When the engine is fully warmed up, the sender is around 50 ohms which lights the led at

pin 13 and sometimes pin 14.

(based on a 200 mile ride to London and back one day. Non stop motorway on the way up

2 lane A roads on the way home)



My rough and ready guesswork/testing showed the led at pin 18 will light if the sender drops

to around 25ohms

Thats very very hot ! and shouldnt really happen unless you have serious problems.

I'd have stopped the bike and investigated if the leds at pin 15-16 came on.



It will and does, work as a simple indicator of engine temp

It take very lttle current so a heat sink on the 7805 isnt really needed

the 1K series resistor only has to supply one led at a time so a .25W one will do

Ditto the 100R after the 7805 and the two 2K2's.



It could be tweaked and finessed further of course but I have other jobs on and will

leave mine as it is for the time being but would be interested to see if others develop

it further.

testing against a reliable thermometer would be useful. Never got round to that myself.

As I mentioned before, the levels at say pin 15 onwards could be use to trigger fan controllers

or alarms.

 

[marshallf3]

Thanks, knowing that IC I can reverse enginer some data from your circuit to see what I get.



According to the Honda service manual and simple mental extrapolation of the points I have from it, 50 ohms on the sensor would equate to a temperature of around 180*F, just when the thermostat is supposed to start to open. 25 ohms would equate to around 220*F which means the thermostat would be right around fully open. If this is near accurate these bikes don't need much cooling from the radiator at all.



Too bad you don't have an actual stock gauge in there, it would be interesting to see what it was reading at your 50 ohm point.

 

[greghoro]

I'm assuming that this is instead of the stock gauge as opposed to in addition to.



I've been trying to figure out how to use the existing temperature sending unit to trigger a fan as well as to run the factory temp gauge so I don't have to attach a second sensor in or on the radiator. This circuit looks like it would work except that it would affect the temp gauge reading unless it could be isolated from the gauge circuit somehow.



Thoughts?



Greg

 

[marshallf3]

I'm assuming that this is instead of the stock gauge as opposed to in addition to.



I've been trying to figure out how to use the existing temperature sending unit to trigger a fan as well as to run the factory temp gauge so I don't have to attach a second sensor in or on the radiator. This circuit looks like it would work except that it would affect the temp gauge reading unless it could be isolated from the gauge circuit somehow.



Thoughts?



Greg

Easy enough. Eliminate the 7805 and 100 ohm from the schematic, that's simulating the 7V supply and the stock gauge. You may need to adjust the reference voltage for the LM3914 by changing the 2.2K resistors, either through experimentation or waiting for me to get my bike back so I can get the readings.



A new problem has arisen though. Due to what he's reported, consideration of 30 year old technology and knowing how little they paid to accuracy of these gauges to begin with I'm not so sure we're ever going to come up with something as accurate as we'd like it to be.



Remember some of those many discussions in the past about where everyone's temp gauge ends up during normal riding? We had reports all over the place as to where the hneedle would be sitting. The entire setup was only meant to show if your engine was warm or overheating.



What I may need to do is figure out an easy place to add a second sensor that is:

1) Easy to add on

2) Reflects an accurate reading of the coolant temperature

3) Is fairly invisible as being added on



The thermistor bead I believe to be the best choice for durability & accuracy is only about 1 mm in diameter and if only used to control the fan it could be after the thermostat - after all if the stat isn't open you certainly don't need the fan running. It might also be to our advantage to have it monitor the water coming back from the radiator since the radiator already provides a fair amount of cooling when you're moving.



Alternately if you're wanting a true temperature gauge the speed control circuit could be expanded in design to convert the sensed temperature back into something the gauge would understand as well but, if mounted after the thermostat, it wouldn't provide a useful output until the thermostat is actually open and the radiator in use.



The only problem with not having the sensor directly in the water stream prior to the thermostat is you'd lose the fault detection capability should your thermostat get stuck closed. I suppose a little something could be added to the circuit that also looks at the voltage from the stock sensor for that.



Too many options and I should probably stick to my original objective of a variable fan speed control that keeps the water returning to the engine at a user-adjustable preset point.



Not having the bike directly in front of me is starting to waste my time, I'm just going to have to hhold where I'm at until I can sit down and do some actual hands on research into the finals on this one.

 

[nolavox]

adaptor bushings





Look for the metric to 1/8 npt adaptor.

 

[marshallf3]

adaptor bushings

Look for the metric to 1/8 npt adaptor.

They look nice, but what metric size does our temp sensor use?

 

[blue]

I'm trying to do this but the original schematic is a broken link - does anyone have it saved or a working link?

 

[Reg in Bristol]

That thing !
Blimey that was ages ago
I made one when my gauge failed and it worked well enough
and later another ( same basic circuit in a diffent case) which did me for a year or two until I bought a digital one
Not a thermometer by any stretch but as good as the orignal gauge anyway and more compact
I'll have a butchers and see if I made any notes or any pics survive.

 

[blue]

That thing !
Blimey that was ages ago
I made one when my gauge failed and it worked well enough
and later another ( same basic circuit in a diffent case) which did me for a year or two until I bought a digital one
Not a thermometer by any stretch but as good as the orignal gauge anyway and more compact
I'll have a butchers and see if I made any notes or any pics survive.

Thanks very much. My CX lacks any kind of temp gauge at the moment and I was looking for a way to make one with a few LEDs (e.g. on/cold -> warm -> warmer -> hot -> stop-and-check-before-kaput) and came across this thread... it sounds like pretty much what I'm after! I noticed in another thread you mentioned an LM3914 and a 7805 regulator, both of which are available locally... looks promising, anyhow

 

[Reg in Bristol]

I found this pic and know its basically correct and may have been the one referred to above,
but in reality I may have changed few values ( 2k for 2k2 maybe?) and not bothered to take notes,
so suggest knocking it up on a breadboard first to test/fine tune it.
Caps say 10 uF and 100nf mebbe

Function as when wittering on above in 2010
CX sender in themostat housing is bottom half of voltage divider
100R on top fed by 7805 and the tap to pin5 of the 3914 acting as an expanded scale voltmeter.
pin 9 left open so only one led on at a time ( dot mode)

I seem to recall the 1K in the led supply was just to reduce led current even more as I didnt want
the leds to be very bright and distracting at night.
Cant recall if I actually used it in the end though, the 3914 does control led current so its not vital

I do remember the 7805 likes to have a bit of a heatsink, a 30mm square bit of ally plate will do
I think the blue arrow is correct as the voltage drops as the engine heats but dont trust the led colours
looks wrong to me and I only used 7 leds
This should get you up and running but as above, test and experiment first to get it as you want it.

 

[Reg in Bristol]

Just remembered summat else
I omitted the heat sink on the second one as I was trying to keep it very small
and the gauge would cut out after a while.
After I switched off and later restarted it would work again, then I realised the 7805
was shutting itself down due to inbuilt thermal protection.
Added a modest heatsink and it was fine after that

 

[blue]

Thankyou very, very much! I will give this a go when I get time over the next week and see how well I can get it wrong