I have a 15VDC analog signal, but I only need the top part of that signal for an ADC. That is, I only want the region from 10-15v, and I want to offset that to 0-5v.

I don't have an AREF pin available on the ADC. Just scaling down the signal with a voltage divider loses too much resolution.

How to do this? Is there some kind of analog comparator IC?

 

How about using a single-supply opamp to "subtract" 10V? The output can't swing under 0V so what's left on the output should be only the part from 10-15V.

 

Using Op Amp as omparator or comparator for 10V~15V.
Using Op Amp negtive Clamping circuit from 10V to 0V.

 

Below is a suppressed zero level shifter amp using a rail-rail op amp that converts 10V-15V into 0-5V, as odinhg suggested.

 

I am new to comparators. I tried a couple and the output just floated around zero. LM339 and LM2903. Checked ground , Vcc = 5v ,In+ = 13v, In- = 8v, Out = 97-150mV. What am I missing here?

 

Crutschow,

So V+ in the schematic wants to be >= V4?

 

Since I'm a beginner myself, I tried to come up with a working solution for my own learning.
Maybe it is a bit over complicated for the task, but it should work.

But like I said, I'm still a novice, so you should probably listen to the more experienced users like crutschow and take my solution with a grain of salt.

 

I am new to comparators. I tried a couple and the output just floated around zero. LM339 and LM2903. Checked ground , Vcc = 5v ,In+ = 13v, In- = 8v, Out = 97-150mV. What am I missing here?

You are missing a pullup resistor of about 1kΩ from the output to +V.
LM339 and LM2903 have open-collector outputs.

Your input voltages are too high. Keep the inputs between 0V and Vcc - 1.5V.

 

I am new to comparators. I tried a couple and the output just floated around zero. LM339 and LM2903. Checked ground , Vcc = 5v ,In+ = 13v, In- = 8v, Out = 97-150mV. What am I missing here?

LM330 and LM2903 are voltage comparators, There is only 2 output states for them, +ve greater than -ve and the opposite, What you want is an op-amp. The 741 will not work since it requires a positive and negative voltage supply. Also, you will need one that goes from rail to rail like the LT1366.

 

Crutschow,

So V+ in the schematic wants to be >= V4?

V+ just needs to be somewhat larger than the maximum output voltage 5V (say 10V),

 

Since I'm a beginner myself, I tried to come up with a working solution for my own learning.
Maybe it is a bit over complicated for the task, but it should work.

But like I said, I'm still a novice, so you should probably listen to the more experienced users like crutschow and take my solution with a grain of salt.

If you substitute a single supply op amp like the LM324 your circuit should work.

Note that your input shown is ±7.5V not 0-15V.

 

Great forum, great info. Thanks everyone for their posts. I am going to get some op amps and play with the circuits suggested.

 

Note that your input shown is ±7.5V not 0-15V.

Oh sorry it looks like that. It doesn't show on the schematic, but the signal source is actually +7.5V offset.

 

Currently, I am using a resistor divider to bring the voltage into the 0-5V range of a 10-bit ADC. I could nominally kick up the resolution by a factor of 3 if I could just look at the top third of the signal range. The voltage stays between 10 and 15V, but varies so slowly over time that frequency response is not an issue.

I presume I need some kind of a voltage reference, but I do not have a 15V rail. What is available is, at this point, are rails at zero and 5V. The 5V I can calibrate off of the internal reference in the ADC.

Cost and chip count are factors. I have 3 voltages to monitor on each PCB and there are multiple PCB's in the project. Digikey wants over $7 for an LT1366.

How about this: If I use the resistor divider to reduce the signal to 5V max, then grab the top 2 volts of that signal for the ADC, I think that should give me a resolution of better than 5mV, which I could live with. Can I do that with my 5V rail as reference?

Thanks for the help. I am a programmer, not an EE. The digital stuff is easy as Legos, but the analog makes for some long nights.

 

If you can give a low and high reference to the ADC it could get full resolution. I.e. reduce the 0-15V to 0-5V then use 3.3V and 5V as the ADC lower and upper reference.

Bob

 

BobTH,

Thanks for responding. I only have one input pin available on the ADC, per channel. I don't have access to a low ref on this ADC. The ADC by default uses 5V as the high ref.

 

With a single +5V supply, you absolutely have to have a rail-to-rail op amp. I simulated with LT1366, but there are many available.
Your resistor tolerances depend on the precision you need.

The circuit probably won't work well at 1kHz, due the time required for the op amp to recover from saturation.

 

The LT1366 was just a RR device I arbitrarily picked that was in the LTspice model library. There are many (and cheaper) RR amps you could likely use, such as the LT1637 or MCP601. The offset voltage is one of the primary parameters to note for your application.

 

Below is a suppressed zero level shifter amp using a rail-rail op amp that converts 10V-15V into 0-5V, as odinhg suggested.

Ah! The answer to my own need. I have a very similar requirement, and only recently began thinking about it.

 

I am ordering some MCP601's. Will let you know how that goes.