Similar to a DAC but instead of providing a digital number, it would be a PWM waveform. I would normally just filter the PWM to get the DC voltage I want but in this case the filter would be seen as a load by the sensor wire I'm applying the voltage to. Is there a discrete IC that would be appropriate in this case, to take a PWM signal of varying duty and interpret that and output a voltage from 0-5 ? Basically I just want to eliminate loading effects through isolation and I want to start with PWM. I feel like using an amp in this case wouldn't work because the feedback loop connects the output side to the input side. I don't know, maybe I'm wrong about that but I just want to be 100% sure the source of my DC is "invisible" to the wire its connected to when I'm not driving it actively. Because it's a sensor wire, it already has its own calibrated voltage divider and I'm kind of sabotaging it by actively driving the voltage, so I want it to work that way but also work normally when the driver is off.
A non inverting amplifier would present a minimal load on your signal wire, especially if it was a fet input amplifier.
A quick google search for "single supply jfet op amp" resulted in a number of candidates. One I looked at was the AD824. The input impedance is 250M ohms.
Your description has all the indications of an x-y problem
Please describe what you’re trying to do, not how you think it should be done. What is this mysterious load?
Car has Intake Air Temperature sensor. Intake Air Temperature sensor is a resistor connected to the ECU forming an internal voltage divider. Want to trick ECU by telling it what the voltage (temperature) is but only sometimes. Want to splice into IAT wire and apply voltage to it using MCU via PWM. Need to filter PWM to get DC. Filter appears as a load (resistive network) to ECU. Need to know how to prevent IAT circuit from seeing MCU circuit as a load either when running normally or when being tricked by MCU.
Basically I just want to eliminate loading effects through isolation
That is where you are going wrong, electrically that makes no sense.
What you are trying to do is exactly the opposite of this, as I understand it you want to make a voltage source that will swamp another voltage source. For that you need your voltage to have a low impedance. Then you need to be able to remove that voltage or make it such a high impedance that it is swamped by the sensor voltage.
Off hand I can't think of a way to do this, but I will think about it.
Gahhhrrrlic:
Want to trick ECU by telling it what the voltage (temperature) is but only sometimes.
You can RC filter the PWM, use an opamp as a voltage follower to get a low impedance 'DC' voltage.
Connecting this DC level to the temp sensor input will depend on the impedance of the existing circuit and by how much you want to 'bend' the sensor value as seen by the ECU.
e.g. if the existing sensor is 1kohm, you might want to 'inject your DC' via a 10k resistor or higher
To isolate your circuit when running normally, use an analog multiplexer e.g. one channel of a 4051
Yours,
TonyWilk
Gahhhrrrlic:
Similar to a DAC but instead of providing a digital number, it would be a PWM waveform.I would normally just filter the PWM to get the DC voltage I want but in this case the filter would be seen as a load by the sensor wire I'm applying the voltage to.
Is there a discrete IC that would be appropriate in this case, to take a PWM signal of varying duty and interpret that and output a voltage from 0-5 ?Basically I just want to eliminate loading effects through isolation and I want to start with PWM.
I feel like using an amp in this case wouldn't work because the feedback loop connects the output side to the input side.
I don't know, maybe I'm wrong about that but I just want to be 100% sure the source of my DC is "invisible" to the wire its connected to when I'm not driving it actively.
Because it's a sensor wire, it already has its own calibrated voltage divider and I'm kind of sabotaging it by actively driving the voltage, so I want it to work that way but also work normally when the driver is off.
Gahhhrrrlic:
Car has Intake Air Temperature sensor.
Intake Air Temperature sensor is a resistor connected to the ECU forming an internal voltage divider.Want to trick ECU by telling it what the voltage (temperature) is but only sometimes.
Want to splice into IAT wire and apply voltage to it using MCU via PWM.
Need to filter PWM to get DC.
Filter appears as a load (resistive network) to ECU.
Need to know how to prevent IAT circuit from seeing MCU circuit as a load either when running normally or when being tricked by MCU.
Sorry, I had to make it easier to read.
Tom... 
Hi,
You are going to have to break the wire from the sensor to the ECU to put your signal modifying system in circuit.
The output LP filter to get your DC output will not appear as a load to the ECU as it will be providing signal to the ECU.
What you will have problems with is on startup and the ECU does its sensor checks.
There are some analog voltage rescaling circuits that do the job.
You need to check the car/automotive modification forums, and/or google.
Tom...
I'm assuming you are trying to increase power under certain conditions.
There are a couple of important questions:
- How long do expect the "override" to be engaged? i.e. seconds, minutes, hours.....
- Do you currently know the amount of temperature shift you wish to create (in volts at the current sensor)?
- Is the "override" always in the same direction? i.e. are you always going to reduce the temperature or increase?
- Does the "override" need to be continuous or could a couple of steps be OK?
I worked with a friend a while back to do the similar on a mass air flow sensor (he had installed larger injectors). This sounds similar.
Grumpy Mike: You are exactly right in your interpretation, although I wouldn't say I'm trying to do the opposite. It's only during a brief transient that I need low impedence and then I'm applying a greater voltage so the flow is away from the MCU.
Tony: Do you mean that the MUX would switch the circuit to open when I want normal operation so that the circuit sees high impedance?
Tom: I definitely don't want to throw off the normal calibration that exists in the circuit but I didn't know about these startup checks. Do you have any specific info on what the ECU does at startup?
John:
- On the order of 10 seconds, give or take a few
- I need to achieve -40C, which I can do with approximately 4k of resistance on the IAT sensor (has to be verified)
- Always reducing the temperature since my "special" override column is the -40 to -30 degree column in my tune
- It needs to maintain the DC voltage that would be the equivalent of 4k of resistance in the voltage divider for as long as I'm making power. PWM for example would freak out the ECU and cause "bad things" (TM) to happen.
Car has Intake Air Temperature sensor. Intake Air Temperature sensor is a resistor connected to the ECU forming an internal voltage divider. Want to trick ECU by telling it what the voltage (temperature) is but only sometimes
Injecting your intended solution into the problem statement is not helpful, it prevents you from seeing other solutions. You start out with helpful information and then you insert the word voltage into the problem statement and then add as an afterthought the proper word, temperature.
Then you go off the rails with your proposed solution:
Want to splice into IAT wire and apply voltage to it using MCU via PWM. Need to filter PWM to get DC. Filter appears as a load (resistive network) to ECU. Need to know how to prevent IAT circuit from seeing MCU circuit as a load either when running normally or when being tricked by MCU.
Why are you forcing the discussion to voltage when what you you want to modify is the temperature value? The temperature is in some way related to resistance, per your definition. Why aren’t we discussing resistance?
Take a step back from your voltage based solution and think about how you would change the resistance of the sensor. Would you need to raise the resistance or lower it? By how much? Let’s say you need to lower it. Couldn’t you simply parallel the exsisting element? If you had to raise the resistance, couldn’t you add something in series?
Please, let’s have a discussion about what you want to achieve, not the nuts and bolts of how you’re going to achieve it.
Gahhhrrrlic:
Do you mean that the MUX would switch the circuit to open when I want normal operation so that the circuit sees high impedance?
Yes.
That 4051 is one example, there's also the 4066 which is 4 separate switches.
Yours,
TonyWilk
avr_fred:
Injecting your intended solution into the problem statement is not helpful, it prevents you from seeing other solutions. You start out with helpful information and then you insert the word voltage into the problem statement and then add as an afterthought the proper word, temperature.Then you go off the rails with your proposed solution:
Why are you forcing the discussion to voltage when what you you want to modify is the temperature value? The temperature is in some way related to resistance, per your definition. Why aren’t we discussing resistance?
Take a step back from your voltage based solution and think about how you would change the resistance of the sensor. Would you need to raise the resistance or lower it? By how much? Let’s say you need to lower it. Couldn’t you simply parallel the exsisting element? If you had to raise the resistance, couldn’t you add something in series?
Please, let’s have a discussion about what you want to achieve, not the nuts and bolts of how you’re going to achieve it.
I understand your frustration about me leading the discussion rather than attacking the problem but what you don't understand is I have a fairly good grasp of the automotive application that this question pertains to and I have experience in the matter that you don't know about. You're assuming I don't know what I want.
I have tried the very solution you're hinting at. I had a resistor on a relay that would bypass the sensor altogether and give me the desired effect. I did this on 2 other cars and had the thing rigged up to my fog lamps. It works. However it's too crude for what I want to do. I'm designing a fully electronic control unit and want this function to be a part of that. I don't want mechanical switches that could fail or solenoids that require a lot of current. This is from experience. I had a mechanic working on my car and he yanked too hard on my relay and broke it. I'd rather just have a wire spliced, taped and covered in plastic tubing, it would be much more robust. I believe voltage control rather than resistance control is the way to go. Besides, the ECU doesn't care about resistance. In the end the resistors are serving to create a voltage, not the other way around. Any means to the end of creating that voltage, is a viable strategy.
I looked into the multiplexer and it seems like the perfect solution... a component I had not considered. Only thing I need to check is that the hookups needed can all exist inside my control box except for the one Vout pin. Any more wires than that and I'm stuck because I have a 15 pin D-Sub with all wires accounted for including this one.