RazMan
|
| posted on 4/1/07 at 11:42 PM |
|
|
Splitting temp sender signals?
Just curious really but while I was setting up my electric water pump controller and the rad fan settings in my ecu I noticed quite a discrepancy
between the temp sender readings for the Evodash and the ecu - they tend to bounce around a bit and can give a difference of 20 degrees at times 
despite having all the senders in the same manifold. Other times they are about 5 degrees different but it would be nice to get them both singing from
the same hymn sheet.
I was wondering if there was a little electronic gizmo which could split one sender output into two or three - it would simplify things a bit when
setting up temperatures.
Has anyone done this?
[Edited on 4-1-07 by RazMan]
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
|
|
nitram38
|
| posted on 5/1/07 at 09:45 AM |
|
|
Don't think that it will be that simple unless you go for some sort of logic circuitry.
Surely the ecu makes allowances for this?
(unless you have got a faulty sender).
I know it is not much help, but I always fit an oil pressure/water temp gauge that uses capillaries instead of electrics. They generally give more
accurate readings, especially if you are going to race or track your car.
|
|
|
RazMan
|
| posted on 5/1/07 at 10:31 AM |
|
|
I am not sure what you mean about the ecu making allowances but I think the problem was highlighted due to my slightly more complex setup.
Sender 1 - Feed to Evodash (purely a monitor)
Sender 2 - Feed to Pump controller
Sender 3 - Feed to ECU (viewed with mapping software)
With hindsight it would have been better to get the ecu to control the pump, but 'I've started so I'll finish'  so bear with
me.
The idea is that the pump is static until the engine jacket warms up and then 'pulses' water around the system until a working temp of 85
degrees (adjustable) is reached. It then switches on constantly and varies its speed accordingly, aiming for 85 degrees all the time. If the water
temp rises to 95 degrees, the ecu switches on the rad fan and cools everything down to 85 degrees, handing control back to the pump (hope this makes
sense)
I measured all the temps with a good IR thermal probe and found that each sender, although mounted within 100mm of each other, has a different log
curve which can confuse things a bit. The result is a kind of yo yo effect with the rad fan and pump fighting each other for control - a kind of
tolerance battle.
Ideally if just one temp sensor could supply all three systems, it would result in much better control. Maybe a bit of logic circuitry would do the
trick but I haven't a clue where to start 
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
craig1410
|
| posted on 5/1/07 at 10:49 AM |
|
|
Can't you just use three identical sensors? You might need thread adapters of course.
Otherwise to split the sensor signal you would probably need to convert the sensor resistance into a voltage and use this to either drive your
controllers directly or you could convert the voltage back into a resistance using a device called a digipot (or Digital Potentiometer). In this case
you would split the voltage signal and drive three separate voltage to resistance converters to give three identical resistances from one. In reality
you wouldn't always get exactly the same resistance due to tolerances between digipot devices but it would be very close.
Using identical sensors would be easiest if you can.
Cheers,
Craig.
|
|
|
nitram38
|
| posted on 5/1/07 at 10:54 AM |
|
|
Try swapping the order of the connections about. Trial and error may reduce the temp fluctuations.
|
|
|
RazMan
|
| posted on 5/1/07 at 12:31 PM |
|
|
Unfortunately the sender units are all totally different in their outputs as they are calibrated to their own system. So if I swapped them round I am
sure the tolerances would just go even further out. The same would of course apply if I used identical senders. The log curves are so different I have
a feeling I will just have to live with it or get the ecu to control the pump. The only down side of ecu control is that it will just be a simple
'on off' control and I will lose the sophistication of the controller.
Interesting stuff about the Digipot though 
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
craig1410
|
| posted on 5/1/07 at 02:43 PM |
|
|
When I was working for Lucas Automotive Electronics in Cirencester we used to use Digipots quite a bit for control logic and to simulate sensors (eg
coolant temp sensors). They are quite simple to use as a voltage to resistance converter but would require the addition of a comparator and clock
source as shown in this application note from Maxim:
http://www.maxim-ic.com/appnotes.cfm/an_pk/3284
Note that to provide three identical resistance outputs you could just add another two more MAX5160's in parallel with U2 in the diagram. Note
that U1 is used to track the input voltage so you can't use it for resistance output.
This type of circuit can also be used to convert between different ranges of sensor and for converting sensor polarity. For example, if you had a
sender which did 0-1000 ohm for empty to full respectively, you could change this to 1000-0 ohm empty to full or to 10000-0 ohm empty to full.
Cheers,
Craig.
|
|
|
BenB
|
| posted on 5/1/07 at 03:53 PM |
|
|
Yes it's possible but it's so not worth it!!!! IT would be heavier, more expensive and more likely to fail.....
sorry.
|
|
|
02GF74
|
| posted on 5/1/07 at 03:59 PM |
|
|
I started writing some stuff but then saw this:
quote:
Originally posted by RazMan
Unfortunately the sender units are all totally different in their outputs as they are calibrated to their own system.
which is what I was gonna write, albeit not in the same words 
To do what you want, you would need to calibrate the senders yourself - stick them in a pan of water that is heated up, ideally with an ammeter in
series for both and record the current and temp,. shown on the instruments - for added accuracy measure the water temp. (NB without the water temp you
will be able to get the displays to shaow same number although it may not be true water temp)
Once you have this data, you would connect the senders to 12 V with a series resistor and have a circuit that measures the voltage to get the current
flowing; a A/D converter can be used to read that and then have a table to conveert to an output current whcih is what y our guages want.
Not straightforwardd but doable (PICs are available with A/D converters and a resistor network could be sue to sum up currents.
Added=======================
forgot to say that have you heard the saying "ignorance is bliss?", well you are suffereing from the opposite - having too much info. and
worrying about t he discrepancy.
All you need as far as water temp goes is a red lamp that comes on when the temp exceeds a max, unless you are driving a deisel land rover in sub-zero
condtions, you needn't worry about under temp.
[Edited on 5/1/07 by 02GF74]
|
|
|
RazMan
|
| posted on 5/1/07 at 04:57 PM |
|
|
oooooooh my brain hurts now 
I agree that sometimes too much information can be a bad thing ..... I suppose I could always stick some tape over the Evodash temp display and just
trust the pump controller to sort it out. At least when I see steam coming from the engine bay I will know that it's at boiling point without
looking at a gauge
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
craig1410
|
| posted on 5/1/07 at 07:27 PM |
|
|
Razman,
After that last post by 02GF74 my brain hurts too...
The discrepancy you are seeing could be caused by sampling delays in the ECU. It is unlikely that the ECU will continuously sample coolant temperature
and it may only look at it every few seconds or so (perhaps even longer). If your engine was recently started then the ECU temperature might lag
behind the actual temperature by a few degrees which could easily explain the 5 degree difference you are seeing (not sure about the 20 degree
difference). Once the thermostat opens the engine temp will stabilise and sensor lag will not be such a problem.
I agree with the suggestion that this isn't something to worry too much about but it would certainly annoy me if I was in your position.
However, it is difficult to know how best to act unless you can tell which sensor is actually reading the correct temperature. Even then, if you found
that the ECU was reading 10 degrees low then would you really want to "fix" this? You would probably find that the ECU had been programmed
to take account of the sensor lag and wouldn't take kindly to being altered.
Cheers,
Craig.
|
|
|
RazMan
|
| posted on 5/1/07 at 07:55 PM |
|
|
If anything the ecu is ahead of the temp readings and looks as if it samples every two or three seconds from what I can deduce. There is no thermostat
either as the pump controller replaces it and takes over.
I'll try and ignore the fluctuating temps for now and maybe look at getting the ecu more involved on the next (inevitable) rebuild
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
craig1410
|
| posted on 5/1/07 at 08:46 PM |
|
|
Okay, looking at it the other way around, does the Evodash continuously monitor temperature or does it have a sample rate which might mean it lags
behind the ECU?
Craig.
|
|
|
RazMan
|
| posted on 5/1/07 at 09:12 PM |
|
|
Good question - there is nothing in the blurb about sampling rates but I might email the Evodash techie to see if he can throw any light on the
subject.
Actually on furrther thought the display does change fairly quickly - every few seconds anyway.
I checked the temp of the casting where all of the senders are mounted - it just seems that the ecu and Evodash readings are either side of a
tolerance range so maybe I can tweak something to bring them closer.
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
MikeR
|
| posted on 6/1/07 at 12:25 AM |
|
|
Daft question - which is wrong?
Could you take all three sensors out and put them into the same pan of water & then heat the water while taking temperature measurements.
If you know which one is wrong - its possible the sensor is a bit duff. If they all read the same temperature then its a design issue (ie temp change
between each mounting)
By doing this you'll also know hte truth - ie Dash is 5 degrees high, ecu is 5 degrees low and can alter settings to accomadate.
|
|
|
RazMan
|
| posted on 6/1/07 at 01:15 AM |
|
|
quote:
Originally posted by MikeR
Daft question - which is wrong?
They all are - up to a point. Bench testing them is not really possible as they output a resistance which is matched to each of the three systems,
otherwise I would have just chosen identical senders for all three.
For example on sensor will read 0-1000ohms, another is 0-200k and the other is 0-10M (hyperthetical figures but you get the idea)
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
chriscook
|
| posted on 6/1/07 at 10:23 AM |
|
|
Are you measuring the correct temperature with the IR device? General consensus seems to be that they can be a bit unreliable, particularly if the
surface is shiny/reflective.
Can you get hold of a thermocouple and reader? I araldited a thermocouple to my thermostat housing when I was sorting out my temp calibration.
Or can you get a jug of hot water to the car and do as MikeR suggests? You won't get to 95deg but 85deg should be easy enough.
|
|
|
MikeR
|
| posted on 6/1/07 at 10:49 AM |
|
|
Yeah, what i mean is input a known test value and see what the devices record it as. The known value being some water that is the same temp for all
three devices - ie pan of water.
|
|
|
RazMan
|
| posted on 6/1/07 at 11:28 AM |
|
|
I find the IR meter (just a posh multimeter really) is quite accurate as long as the surface is matt. I left the meter monitoring the common metal
manifold and it measured around 85 degrees. The Evodash measured 85 but the ecu software was moving between 80 - 95. This would suggest that the ecu
sender is a bit whacky so I'll change it when I get a chance.
Thanks for all your help guys
Cheers,
Raz
When thinking outside the box doesn't work any more, it's time to build a new box
|
|
|
chriscook
|
| posted on 6/1/07 at 11:53 AM |
|
|
Could it be picking up noise from somewhere? Have you got a 'scope?
Does the ECU value move about when the engine is not running?
[Edited on 6/1/07 by chriscook]
|
|
|
