Some Help With Volt Scale Reversal Needed For An Idiot.

[Incony]

i want to use an existing circuit in my car.. to do more than one thing.

right now i have a coolant temperature gauge connected to a negative coefficient resistance - two wire temperature sender.

in effect , one of the two wires is ground ( car earth battery negative ) and the other goes to the gauge input..

the resistance of the sensor is low when the gauge reads high, and high when the gauge reads low temperature.

The gauge has three connection B+ ( @12v) and ground ( car earth battery negative ) and the input connection..

i cant change those... they must stay as they are.. i cant reverse the polarity of the gauge.. its all fixed.

the sensor to ground resistance varies from 1954 ohms ( 30 deg C ) to 132 ohms or less (100 deg C + ) its a closed pressurised system.

from that the voltage measured on the gauge input wire to ground measures 5V at 1954 ohms and 2,75 V at 132 ohms.

so the voltage input is high when the gauge reads low... and open circuit the gauge reads high.


i want to use a pressure sensor and switch between using the gauge to see temp, or pressure.

i can source a pressure sensor to do that task .. 0 to 30 psi. but its output is 0.5 to 4.5 V DC low to high.

the max pressure or the warning point for me is 15 psi.. so if the sensor is linear... 2V output is about the mid scale..

but there is my problem.. if i put 0.5 to 2V into the gauge input the gauge will never move from high.. and really what i want is the gauge to read max at 16 psi.. or above and half scale at 7 or 8 psi and @ 0 psi at low...

so need some help in making an interface circuit to do that,, it needs to reverse the voltage output to the gauge input so it reads in the right direction,,, and change the V output i want as max .. from @ 2V to 5V...


?

the existing wiring for the temp sensor is very close to the where the pressure senor will go... and thats why i wan to explore the options... this is a mid engined car.. the engine is behind the driver... hidden between the seats and the boot.. its access isnt easy.. secured in by bolted panels,,, etc. its not as if i can just change or alter anything easily.

fitting a different gauge just for pressure, isnt easy either... there is no room in the dash or nearby.. and why i explore using one gauge to see two things using a switch.

 

[Alec_t]

Sounds like an opamp could be used to invert and DC offset one of the signals. However, an NTC temp sensor is non-linear, so if the pressure sensor is linear you would have to put up with one or other readout being inaccurate. A micro with an ADC feature might be a better way, with all the conversion and linearising being based on look-up tables.

 

[Incony]

TY... very much Alec_t. ive used op amps like 741`s before but a long time ago now...
linearity, isnt a concern,,

the temp gauge sits at a " normal " point just under half way... once the engine is warm..and if it moves above that something is wrong..

there are three levels i need to see using it as a pressure gauge.. below 5 PSI.. 5 PSI to 7 or 8 PSI and above 8 PSI.. .... . Time... ( how long it takes due to nonlinearity ) is only critical below 8 PSI because when the engine is stopped it will hold pressure if the natural cooling and contraction of the engine components including the oil and coolant are slow.. If there was a leak in cooling system.. the pressure would drop quickly.. it normally takes over an hour to loose @ 6 PSI.. ive measured it..

ive considered using three leds instead... red green and yellow.. when i first considered the idea.. and then i was going to use two pressure switches rather than a gauge..

in reality i think i should just knuckle under, and do away with the inboard clock and replace it with a dedicated pressure gauge.. i can do that...

BUt.. if anyone reading this thread can provide a circuit using an op map. that addresses the problem i would be very grateful, my electronic knowledge is very limited.. though i can build and modify circuits if i fully know what is what.

theoretical circuit plans are no good to me... i need to see all the names of bits.. and the volt in and out points clearly..

 

[Incony]

i guess i could modify the above circuit and use comparators to switch in or out three resistance values to the existing gauge? ie.. volts out from a pressure transducer (0.5 to 4.5) set the switching points and take the gauge input wire to ground through the resistor values chosen... so as the values change the gauge changes position?


Could i use an LM3914 to do the switching? i then have ten switching positions from one chip reading volts..? ten positions each covering about 2 PSI takes me from 0 to 16 PSI and above - easily..

 

[Incony]

here is how the temp gauge operates...

i think that if the outputs of the LM3914 go low when on.. thats what i need? can i just tie resistors to the gauge input from the 3914 ?

 

[Incony]

i.e if my pressure transducer gives 0.5 volts when low.. the lm3914 turns on the first output.. ( 0.5V) that is connected to a resistor of @ 1954 ohms... to the gauge input.. and it will give a low reading... As the pressure increases.. its volt output rises.. so at @2V (15 PSI ) on say - lm3914 output 8 , i have a resistor of @ 160 Ohms... and the gauge would read "high" and any resistor value in between determines the scale based on what lm3914 output is low? i guess the lm3914 outputs are not true ground though... and its not quite as simple as that.. but if i can go that road.. its a good solution to a complex problem... since the 3914 can be scaled to suit the pressure sensor output i want to see... 0 to 16 plus PSi...(0.5 to to 2 plus VDC ) it depends on if i can use those switched lows to give resistance values the temp gauge will see,?

scaling isnt a problem i have a mechanical pressure gauge connected to the system in the engine bay... so i already know the pressure values for any situation.. since the pressure seen is also the effect of temperature, natural expansion and the pump that drives the coolant..

the 3914 will allow me to reverse scale the transducer output. if i can use resistors to feed the temp gauge input.

 

[AnalogKid]

Note that 450 is not half way between 132 and 1954 on a linear scale. The sensor is non-linear, and the gauge has a complimentary non-linearity to produce a linear display. While it is possible to turn your pressure sensor into the correctly non-linear analog voltage to exactly match the gauge (with many opamps), how badly do you want to go through that?

Since you already brought up a more simple LED tree, that can be done with one LM393 dual comparator and a few resistors.

oops, just saw the 3914 part. A good piece-wise approximation. The output transistor saturation voltage is 0.15 V at 2 mA, which is pretty low, but your worst-case current is 20 mA, which is greater than the 3914 output rating of 10 mA typ.

To get around this, consider operating the 3914 in bar graph mode. As the pressure increases, another resistor is added in parallel to the ones already on. In this way, that 20 mA is shared among multiple outputs. I have not run the numbers, but it might work.

ak

 

[Incony]

Ty AnalogKid if you can show me a basic circuit as a pic,,, that would help me. can i use resistors in place of the leds on the 3914? if i dont have a base circuit i will just have to experiment... and buy several chips in case i naff up.. is there a ten transistor or some other ten point switching ic i can use to switch the resistors in? i.e then i wont rely on the 3914 outputs, i drive another ic that does that..

 

[AnalogKid]

Yes, you can connect resistors instead of LEDs.

My mistake - the 3914 outputs are programmable up to 30 mA per pin. These are constant-current outputs, so if your external voltage and resistance cause a current less than the programmed value, the output saturates like an NPN transistor.

ak

 

[Incony]

TY. AnalogKid. my assumption this mo, is that i can directly replace the leds, with resistors of known values. determined by my mechanical gauge pressure.. and toms scale resistance values.. as ive shown in his pic..


so each output is connected to a resistor of determined value.. in ohms. every one of those ten resistors will connect to the temp gauge meter input.. since only one is fully on or low at any time.. even though the hysteresis means that the next is starting to turn on.. or off ( either side of the one chosen )

i.e ten into one.. via a switch.. i switch between the two inputs i want the gauge to see.. temp.. or pressure..

i dont need blocking diodes?

will there be likely to be anything on the gauge input that will affect the 3914 outputs

i have thought that i need to make the outputs of the 3914 one way... ie its only an " out signal" nothing goes into the 3914 that might give it bad news, a diode would block it.

and a diode in series with each resistor chosen,, would do that?

 

[Alec_t]

since only one is fully on or low at any time.

That is the case if the LM3914 is in dot mode, but I don't see how it would work, regardless of the resistor values, because all ten outputs sink the same current. The IC needs to be in bar mode instead, so that an increasing number of current sinks become active as the IC input voltage rises. The combined conductance will, however, vary linearly with input signal voltage, hence not in the same way as the NTC resistance changes with temperature.
An LM3915 might give a better result (but I can't get my head round whether you need a logarithmic or anti-logarithmic response ).

 

[AnalogKid]

That is the case if the LM3914 is in dot mode, but I don't see how it would work, regardless of the resistor values, because all ten outputs sink the same current.

Only if the load resistance is low enough to pass the current. With a high value resistors load, a 3914 output will try to sink the programmed current to the point of saturating. At that point it looks like a saturated switch.

For example, with a 10 V Vcc and a 1 K load, the max possible current is 10 mA. If the 3914 outputs are programmed for 5 mA, they will saturate.

ak

 

[Alec_t]

Point taken. Thanks.

 

[Incony]

ive ordered a very simple pressure sensor. it gives a voltage output dependent on pressure, 0,5 to 4.5 V DC. with a scale of 0 to 30 PSI. ( 0 , as far as i understand being air pressure at sea level @ 1 bar and disregarded )

so i can connect up a 3914 to read the pressure 0 to to 16 PSI or more, full scale ( 0 to 2V plus ) , it doesnt matter if the voltage isnt linear if i can use resistors to set the scale the analogue meter scale reads..

since this mo, i dont know if the 3914 out[put signal alone on each of its ten channels can actually do that, i am just going to buy a couple, and make a circuit, since at least the voltage scale input set up is well known.. if.. i cant feed the 3914 output into resistors directly connected to the meter via it common input.. i will find another way of interfacing the 3914 outputs, i dont think this idea wont work.. i do think it might need more than the 3914 alone to do it... i will post whatever circuit i find that does it... its an unusual thing to get a led driver to do.. become an analogue meter driver..

 

[Alec_t]

Here's a 3914-based circuit I designed a while back for a guy who was modifying a fuel gauge. Pull-up resistors in conjunction with resistors R11,12 inside the gauge provide a non-linear response to the input signal. Perhaps you could play with their values to get the response you want for your pressure gauge.

 

[Incony]

i think i might need a transistor interface... its cumbersome.. but i turn transistors on.. instead..? my problem is the meter needs reverse scaling, and so i use the resistor values to do that..
it saves me inverting the voltage output of the pressure sensor. i only need one IC. but if i cant
just use the output of the 3914... i can use transistors to do the switching..? i can set the current output of the 3914... to the base requirement of the 2N3906`s they turn on then one at a time and take the meter input on a reverse scale output.. dependent on the resistance value to ground.

My problem Alec_t is that the meter reads high when its input is low,,, and i cant change that.. its also reading the temp, and so i want to be able to simply switch its input between pressure and temp.. using a mechanical change over switch... therefore its input must be the same for both ... electrically.. and the temp sensor is resistance to ground,,, so thats what the pressure input must be... just different scaling to make it traverse full scale and read pressure instead.

linearity isnt a problem.. provided i can make the resistance values make the meter show me what i want to see.. in some sense. like low normal and high. i think ten different values for a max two volt input or thereabouts.. (i dont know exactly yet) is more than enough to do that.

 

[Incony]

this is where i am right now. i have the resistance values that deflect the gauge from low to high... i need help in deciding what RV1 and RV3 should be in value... equations are no good to me , i just need values etc..give me those as a sole means.. my maths is crap and i fall over using calculators... i just want values in ohms.

 

[Incony]

and before i have to tell folks.. the max i want to see.. is about 15 or 16 psi... and why a need the upper limit to be around 2 to 2.5v thats the reason.. i need a "! window" to set the scale too..

 

[Alec_t]

To get the voltage range you want (~0.5V - ~2.5V say) :-
Remove RV1,
Drive pin 5 directly with the pressure gauge output,
Replace RV2 with ~800Ω (1k trimmer) and ~330Ω (500Ω trimmer) in series to ground,
Connect pin 4 (Rlo) to the junction of those two resistors (trimmers).
However, the circuit in dot mode doesn't do what you want, since whichever output of the LM3914 is active draws about 12mA current regardless of the value of its connected resistor (133Ω ...... 10Ω) .

 

[Incony]

so.. it wont take the resistors to ground alec_t? can i use transistors to do it like 2N3906`s ? ie each output turning on a tranistor? its so close to being a good solution with minimal components...