Actions
Forum Thread Details
  • Replies 19 replies
  • Subscribers 300 subscribers
  • Views 3484 views
  • Users 0 members are here
Related
See a helpful answer?

Be sure to click 'more' and select 'suggest as answer'!

If you're the thread creator, be sure to click 'more' then 'Verify as Answer'!

amplify small dc signal sitting on large dc offset

davebullockmbe

Weekend brain teaser?

Hi experts,

I have a sensor that gives a varying dc output with sensed input.

My problem is that the sensor's output has a 'static' dc offset of around 1V d.c. but I need to use the sensor at the low end of it's range meaning that it's output will only swing a couple of millivolts d.c.

The couple of millivolts will then need to be amplified by something like x1000 to fully exercise an arduino ADC.

I am not sure how stable the 'static' dc offset is with time/temperature or how noisy the output is yet as I am just starting to tackle this problem.

Initially I am thinking differential amplifier and cancel the unwanted offset with a separate 1V input, but I imagine device dc drift will come into play?

Clearly it would be ideal if the dc offset could be tracked by the cancelling voltage but the 'wanted' dc signal will be cancelled too as they are both pretty static.

Before I spend time inventing the wheel or disappearing down rabbit holes.......

Has anyone solved any similar problems?

Discovered any suitable low drift devices etc.

Thanks in anticipation......

Dave

Parents

  • By far the best way to characterise the device is with an optical chopper as suggested by Shabaz.

    The data sheet is very light on useful detail (which usually means that the useful details of the performance are not controlled very well !) but it looks as if the response time is < 1ms. This is the response to an enable signal  but I would expect the optical response time to be at least as fast.

    An optical chopper can easily be made, small motor with a disc with holes in it. 600 rpm with 20 holes would give you 200Hz which is a nice frequency to work with.

    You can AC couple the signal and filter it with simple analogue band pass filters (followed up by any amount of fancy digital filtering in the processor.

    If you use a tiny stepper motor (might need to go for less than 600rpm)  and drive it with pulses from a micro-controller and use it's built in ADC then everything is nicely synchronised. The bits will cost < £20.

    If it's pure research and money is no object then you can just buy an optical chopper and a nice lock in amp from Stanford Research but it will cost a fortune.

    You may have a problem untangling drift and noise from the offset voltage, the chopper technique allows you to eliminate the drift of the offset and by making the bandwidth of the amplifier very small to get rid of sensor noise as well.

    Of course it's one thing using a mechanical chopper in the lab - quite different in the field or mass production.

    MK

  • in reply to michaelkellett

    This looks interesting, although it's just 10 slots. It might be about 50-60mm dia, the description doesn't specify, unfortunately.

    I was wondering if multiple sensors could be placed, for instance, two could provide a differential output. Maybe it's diminishing returns to add more sensors though.

    image

  • in reply to shabaz

    OK, So I just did some tests taking voltage measurements with my Philips PM2534 and pin 10 has a very stable 0.999V on it.
    However noise from the DMM's leads does affect the ML8511's output voltage, which I expected and wouldn't necessarily be a problem in a final design.
    The voltage is slightly affected with PSU voltage variations though my ML8511 module does come with a cheap 3v3 regulator on board
    Pin 9 only has a low voltage which I guess is affected by the amount of UV striking it's sensitive area so wouldn't be of use as a cancellation voltage.
    The more I have considered your suggestions the closer I am warming to dougw's suggestion of a wheatstone bridge, as the sensors are low priced but (hopefully) pretty repeatable as regards offset voltage stability. The offset should be balanceable out and as suggested should minimise external supply fluctuations.
    I now have a second module in-the-post.
    Enjoy the rest of the weekend :-)
    Dave

Reply Children
  • in reply to anniel747

    Hi anniel747 good question :-)


    The original ML8511 module came with the 'sort it out' challenge that I foolishly accepted. (It's a case of "If I were doing this again I wouldn't start from here!")
    However, the reason for using the ML8511 was that it has a fairly wide UV spectral response, unlike devices like the LTR390 or VEML6075 etc that have very sharp responses at specific nm wavelengths.
    The project was to compare the amount of UV light emitted from cheap UV led's whose specific wavelength could not be relied on. (hence needing a 'wider' bandwidth).
    Of course these sensors are designed for measuring the UV radiation in sunlight which at it's highest can be as high as 6.5mW/cm2, whereas the UV from LED's and small 'actinic' fluorescent tubes is in the 10's of microwatts.
    The sensor displays a useable varying output at these very low UV levels but can't just be 'dc' amplified, due to this annoying 1V dc offset.
    So that's where I appeared on this forum..lol!


    The project was started to demonstrate that 3W UV LEDs (as advertised) do NOT emit 3W of UV light on the (for instance) 365nm wavelength but may indeed consume 3W of dc power.
    The project was NOT to measure exact UV output, but a cheap and cheerful way to compare the output of different UV LED's in order to fine the most efficient. A spin-off from a 'moth-trapping' study.
    Great project, big headache :-)
    Dave