My input is an analog voltage varying between 0.5uV and -0.5uV, how can I convert this into a positive voltage so that I can measure it with a microcontroller? The voltage will have to be amplified by an opamp for sure, but I don't know how to work with the negative voltage. A negative power supply isn't an option either.

 

Hi jacky,
Do you already any OPA's you could use.?
+/-500nVolt signal would require a very low noise OPA.
E

BTW. what is the signal source.?

 

I think you will need an instrumentation amplifier for a signal at that level. After amplifying, you could use a precision rectifier.

Bob

 

What frequency (or time) for the signal? How long from -.5 to + and back to -5.?

Is this signal "DC" as in -0.5uV=cold and +0.5uV=hot or is it AC, where 0V=no signal and +/- 0.5uV=signal?

I looked up the specifications on my HAM radio receiver. It probably can not find uV signals in the noise. Certainly it can not measure accurately. So building a low noise amplifier is hard and there will be a problem with input offset. It will help if we know where this signal is coming from.

 

Hi jacky,
Do you already any OPA's you could use.?
+/-500nVolt signal would require a very low noise OPA.
E

BTW. what is the signal source.?

No, I haven't checked which one I should use yet and the source is a bit hard to give details. The source is a selfmade heatflux sensor and the voltage across the electrodes vary between the two given voltages depending on the amount of heat.

 

I think you will need an instrumentation amplifier for a signal at that level. After amplifying, you could use a precision rectifier.

Bob

I checked it out, but is this possible without a negative power supply?

 

What frequency (or time) for the signal? How long from -.5 to + and back to -5.?

Is this signal "DC" as in -0.5uV=cold and +0.5uV=hot or is it AC, where 0V=no signal and +/- 0.5uV=signal?

The fastest changing frequency is around 10Hz I believe. The signal is 0V at standby.

 

around 10Hz

Ok so the amplifier can include a 20hz low pass filter. That will certainly help the noise.

 

Is the signal something like a heartbeat?

 

Is the signal something like a heartbeat?

Let's assume that it's random.

 

I think drift and 1/f noise might be the biggest problem at those frequencies.

 

Apart from the noise problems, is there any way I can convert the negative into a positive voltage without a negative power supply?

 

https://www.google.com/search?q=thermopile+preamp
https://www.azosensors.com/article.aspx?ArticleID=597

 

Apart from the noise problems, is there any way I can convert the negative into a positive voltage without a negative power supply?

sure, need a positive voltage reference and a pair of resistors...
just bring signal to R1, connect other side of R1 to both R2 and your analog input. connect other side of R2 to reference voltage.

but your signal is very small and needs amplification

 

An inverting amplifier will work. Most rail-to-rail input op amps will work correctly provided that the inputs don’t go below -0.3V

 

Don’t let the lack of a negative supply voltage hinder you. You can generate a negative voltage from a positive supply.

 

All op-amps have "input offset error". Amplifiers are not perfect. Looking at the amplifiers at digikey.com I see the input offset error can be anywhere from 0.1uV to 900mV. This error is a function of price and complexity. It is easy to make an amplifier what does not know where ground is by too much.

I know you will not tell me what the signal looks like. The amplifier could be complex and expense or simple depending on what is really happening.
I am going to make up a signal. It spends much time at 0V. It has about the same energy&time at +0.5uF as it does at -0.5uV. What I am describing is a signal where the average is 0V. In this case we can "ac couple" the signal and thus use low cost amplifiers.

By AC coupling and building amplifiers that only look at 1hz to 20hz we can remove much noise and amplifier offset error. The first component is a capacitor and a resistor that moves the "0V" point to exactly 1/2 the supply voltage of the micro. Now almost any amp will work. We will have several stages of amplifications (along with ac coupling). (Assume 5V supply) 0V=2.5V which is the center voltage of the ADC. +0.5uV=4V and -0.5uV=1V.
ronsimpson

 

https://www.analog.com/en/analog-di...lue-from-absolute-value-circuit-diff-amp.html

Looks like they're working with much larger signals, but it is an absolute value circuit.

 

I don't see the value in adding hardware to rectify the signal or do any "thinking" in opamps and diodes. The signal is so low there is no problem, just reading the voltage and doing simple math. If you read 30 times a second you should get several readings for the "+" and sever for the "-".

 

To amplify the signal, you could use a low-noise op amp such as an OP-07 or NE5532 in a AC coupled circuit with a low-pass filter to reduce wideband noise.