Hello all,

I am having trouble thinking of how to deal with an unknown voltage. I am creating a circuit for an experiement that will read an input voltage and that input is unknown. The problem is that this voltage affects how the output looks. If the amplitude is too high or low, I need to adjust the reference voltage at a different part of the circuit. However, no one will be around to adjust the reference voltage to account for the unknown voltage when the experiement is going on. So I'm not sure how to get around this problem. If anyone has any suggestions, I would appreciate it. Also if there is anything unclear in how I explained the problem, let me know.

Matt

 

Well it's pretty unclear to me! I get that you need an "auto ranging". Is the setting of your reference voltage continuous, or in discrete steps?

I guess what I'm asking is, what is the mapping from input voltage to reference voltage that you need to achieve?

Another approach to consider is to limit the input voltage to within an acceptable range.

 

Ideal job for a microprocessor (like an Arduino). Use a gain-settable amplifier between the signal source and the internal ADC.

 

Your description is not very clear. Begin by telling us more about your experiment, not about how you need to measure a voltage.

 

uhh yeah sorry, I was worried it would be unclear. I am going to attach a picture of my schematic and some outputs given on the oscilloscope. Perhaps that will help.

Anyways, as far as the experiment goes, up yonder in space, we have to measure a flame. This flame will heat up and then cool. The heating and cooling will be measured as a voltage. Now, seeing that we won't know what that voltage will be, I have to figure out how to get a good picture without saturation or anything weird.

One of the attachments is the schematic. Another is of normal output. The third is saturated/weird (it's got that hump in its belly) The output is taken between R10 and C3.

 

Also, reference voltage (on LT1167) should be fixed. That would be ideal.

 

Look at something like this:

If you are sampling this into a Data Acquisition System, with just a little extra code you can dynamically set the gain so that you utilize the full dynamic range of your ADC.

btw- I am a member of ANSR, and I help students get their instrumentation package ready to fly...

 

But why is this voltage unknown?
Sounds like you are saying (in a confusing way) that you are
"Attempting to capture/record a changing voltage (sensor output)"
This what you call unknown voltage must be coming from a temperature sensor of some sort or something like that.
That temperature sensor has a specified output range.. so the output is known.

Why is this not a simple datalogger/sensor hookup and done.

 

It is pretty simple, really:
Output=Circuit*Input
You know the Circuit, you know the Output, two out of three are known. Work your way backwards and you will find the Input.

If you only know one of the three pieces, then you will NEVER figure it out.

 

I guess the problem is the interface. There is only some analog stuff right now so that's my limitation. Othewise I would have used digital stuff and be done with it. Also, yes sorry, the temperature sensor has a range, but I am more worried about the lower area. Apparently, the LT1228 doesn't like when the signal doesn't cross zero for some reason (I need to look at the datasheet again).

You see, if the voltage is super high, that's okay. However, let's say the voltage is 0.09 volts and I set my reference voltage to -0.1, then the signal is shifted below the 0 voltage and the lt1228 starts to have problems. Again, I'm not sure why.

 

Apparently, the LT1228 doesn't like when the signal doesn't cross zero for some reason (I need to look at the datasheet again).

Have you dealt with the LT1228 issues in your previous thread?

http://forum.allaboutcircuits.com/showthread.php?p=747246#post747246

 

Yes Richard and I posted a reply on that post. I guess I will just have to work with the LT1228 limitation I have now. I am going to keep a negative 0.1 offset and if there really are problems, I'll come back to this problem.