ECT and voltage divider help?

[Mike R Brown]

Bernd Felsche wrote:
> 
> Daniel Houlton writes:
> 
> >I'm working on a schmatic for a small controller for an electric
> >radiator fan and I have some questions on reading the ECT sensor.
> 
> >Doing some testing on my truck I found that the sensor is fed 5V from
> >the ECM (measured with the wire dis-connected from the ECT), but
> >connected it only reads 3V when cold.  The voltage then drops to
> >around .65 V at normal operating temp.
> 
> Looks like nominal behaviour for any NTC device. You're probably
> seeing the result of a constant-current supply being used to
> determine the sensor's resistance - a constant current through a
> variable resistance will give a variable voltage.
> 

I don't think this he has a constant current source here.  Most likely
it is just what he describes below.  The 5v reference sourcing a fixed
internal resistance in series with the variable resistance of the
thermistor in the sensor.  The ADC is attached to the junction of the
two resistive elements and reads the voltage.  Probably a table just
translates the voltage into a temperature value.

> >It's not really important, but I was somewhat confused how hooking
> >the wire to the ECT sensor caused the voltage to drop from 5V to 3V
> >(or .7V or so when hot).  After some research on the net, I found that
> >it's set up as a voltage divider.
> 
> Test the sensor in isolation; switch off the engine, unplug the
> sensor and measure the resistance using a multi-meter. If the
> resistance falls with increasing coolant temperature, then it's
> probably an NTC resistor.
> 
> >So, my question is, how do I switch something based on this voltage?
> >Basically, I want to monitor the voltage to the ECT sensor and when it
> >drops to around .4 or .5 V I want to trigger an output (that eventually
> >drives a relay).  Then, while the output is triggered and the voltage
> 

	Don't know what kind of thermistor is in your sensor but these guys can
be (are) very nonlinear.  The common GM sensors used in the system I'm
running have very little delta_R/delta_T at the very temperature range a
person is most interested in.  You can see some graphs here where I put
sensors in a thermal chamber and plotted their resistance:
http://www.sonic.net/~mikebr/plots/mat_vs_res_adc.html
What this boils down to is that the voltage ranges you are most likely
dealing with can very small.  The difference between the fan on and fan
off voltages can be very small (in the order 10-100mv).  Before I got
into reverse engineering the EFI system in my car I built an analog
circuit that would turn the fan on and off at certain set points. 
Problems you will run into is that a car is a very electrically noisy
environment.  You need to pay attention to your input and supply
filtering.  Also watch out for ground loops.  My battery is in the rear
and the frame is the common return.  When the fans came on (about 30
amps inrush current) a few millivolts would develop across the frame
rails and that was more than the total hysteresis range I was dealing
with so the relay and fans would start oscillating.  Not trying to scare
you away here because an analog design is really not that difficult but
more to let you know that sometimes some weird and un-expected gotchas
are out there looking for you.  Knowing what I know now I would use the
filtering and ADC in the ECM to control these things.  A small PIC or
something like Bernd suggest might be the way to go. 


> First; you need to isolate your circuit from the other one using the
> same sensor. Using an op-amp is the easiest way of doing that - Op
> amps have input impedances in the megaohm range. Isolating the
> sensor gives you a buffered output signal - you might want to have a
> small gain set up on the op-amp for convenience - else keep it at
> unity.
> 
> You can then feed the signal into a "comparator"; an op-amp can be
> used for that as well. The comparator is basical a "switch" that has
> a low output when the sensed input is less than the reference
> voltage, and the output is high when the sense level is higher than
> reference. Depending on the op-amp, you can then drive the relay
> "directly" from the comparator output.
> 
> >goes back up to around .7 V to turn the output back off.  And I want
> >to be able to fine-tune the on and off voltages with a couple pots.
> 
> If you need hysteresis, then the circuit gets more complex. You'd need
> another comparator and then combine outputs to set and reset the
> corresponding driver output. There are dozens of ways of doing that; a
> flip-flop is just one way - using the high-level comparator output as
> the SET and the low-level comparator output to RESET the flip-flop.
> 
> The reference levels of the comparator would be how you set the
> on-off levels. Use the potentiometers in series to ensure that the
> high reference level will always be higher than the low!
> 
> >Any ideas how I can do this?  I don't know a lot of the solid state
> >devices.  Would I be using a transitor, op-amp, etc?
> 
> >Will tapping into the ECT wire to read voltage somehow screw up the
> >reading to the ECM by acting as a sink or source?
> 
> >Also, I'm including several inputs that can trigger the fan relay like
> >the A/C clutch, air compressor clutch, and a couple extras as well as
> >a manual on, manual off (which over-ride the other inputs) and an
> >automatic setting.
> 
> In that case, you may be better off with a small computer with an
> analogue to digital converter and several digital inputs. You then
> also have the ability to drive the radiator fan at a speed according
> to the load by using a pulse-width-modulated drive instead of a relay.
> 
> The computer (micro-controller) would eliminate all the small
> relays and a heap of nested wiring. Flash-programmable computers are
> very cheap - the micro-controller itself typically costs a few
> dollars; the big bucks arise due to sophisticated output (mainly
> driver transistors), housings and environmental requirements - heat
> sinks, etc.
> 
> In your application; if you're not worried about how it looks, you
> should be able to throw something together for about $50 - assuming
> you have access to a suitable desktop computer to write and compile
> the programs; and then download them to the micro-controller.
> 
> I know what the necessary circuits look like on my car when using
> "conventional" setups that do the same sort of thing == UGLY.
> 
> >My question is, that he low-amp signal relay (needs to handle about 200
> >mA max to trigger the fan relay coil) seems kinda big and bulky.  Is
> >there a solid state device that can replace this?  Typically, a relay
> >is used to drive a big load with a small signal, but I want to drive a
> >small load with a small signal.  Just wondering if there was some kind
> >of small IC that would do that instead of a big relay.
> 
> A high-power transistor. say a (MOS)FET - don't know how an IGBT would
> handle being on all the time.
> 
> >Oh yeah, one more thing.  I've found sockets for different types of
> >relays in Jameco and a couple other catalogs, but they aren't very
> >high amperage and I'd rather use a common automotive 30A relay.  I
> >can't find board mounted sockets for these automotive relays though,
> >just pigtail sockets with wire leads.  Anybody know where I could find
> >a board mount socket for these?
> 
> Turn the problem upside-down. See if you can fit your logic circuits
> into a relay case, find a relay plate and plug it in. A relay plate
> is just a moulded plastic section designed for you to plug in a
> relay - you clip a few together and wire the connections as needed
> using conventional automotive spade connectors.
> 
> Auto-electricians will never guess you have a computer in one of
> those tiny relay cases. :-)
> 
> --
> Real Name: Bernd Felsche
>     Email: nospam.bernie at perth.DIALix.com.au
>         http://www.perth.dialix.com.au/~bernie - Private HP
> ----------------------------------------------------------------------------
> To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
> in the body of a message (not the subject) to majordomo at lists.diy-efi.org
----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org