Increasing watts available from a reference voltage

Discussion in 'General Electronics Chat' started by bwilliamson, Jun 23, 2016.

  1. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    I have a reference voltage that doesn't quite have the wattage required to energize a set of sensors that I would like to power with the reference voltage. Is there a way to increase the wattage available at the exact voltage (10.00111V) of the reference voltage? I need to get ~200mA out of it.

    I have a "noisy" 9V/1.5A power supply integrated into this project that I can draw from.

    Thanks for any help!
     
  2. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    You will need to buffer your reference with a precision op amp. You should supply details of the reference you are currently using, and others will be along quickly to help you.

    But, I have a question:

    How will you know if your reference is providing 10.00111V and not 10.00110V or 10.00112V?
     
  3. EM Fields

    Member

    Jun 8, 2016
    452
    107
    You're asking, basically, for a power supply with an output voltage sporting an accuracy of about 1 part per million, which is like asking for a million volt power supply with an output that varies less than one volt, no matter what.

    That's a tough one. What are you doing, anyway, where you need a reference that tight?
     
  4. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    What he is asking is not impossible. An LTZ1000 reference provides a tempco of 0.05 ppm/°C. If he maintains ambient within a couple of degrees and follows with a high quality chopper stabilized amp, the stability part is there.

    More importantly, though, is how he intends to measure and ensure the output is set accurately to within 5uV at 10.00111V. Without the ability to measure the output, the project is already a bust -- if the stated exact output is really a requirement.
     
  5. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    You guys are fast.

    I'm looking to use this supply to excite some load cells. I need to get at least 4 sig figs out of the excitation voltage. 5 would be better. The device I'm measuring the response has a spec of 0.01%. So I guess I need to maintain the voltage to the millivolt from the reference voltage. The reference supplies ~ 200mW (if i measured that right...!).
     
  6. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    Why? What is the load cell configuration? Normally, they are run ratiometric, and then reference accuracy or stability doesn't matter.
     
  7. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    Ahh, I am low on channels for the data acquisition system and would like to have a reliable V so I can assume the value vs having to always measure it!

    I can continuously measure the incoming "dirty" excitation V and then calculate from that but that bogs down the system and eats up 1 channel of capacity.

    I had this nice Vref and thought it could be used to supply a nice, constant signal that I could just count on.
     
  8. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    Here's an easy solution: eliminate the reference entirely and use something like this.
     
  9. EM Fields

    Member

    Jun 8, 2016
    452
    107
    Well, there are cal labs equipped with exquisite instrumentation traceable to the US's NIST's primary standards, and if that's not good enough, to the NIST standards themselves or, if that's not good enough, to any of the other international organizations dedicated to mensuration.

    Tricky, though, since they 're not likely to move their facilities to his basement / attic / wherever just to prove that his widget works in his basement / attic / wherever.
     
  10. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    https://datasheets.maximintegrated.com/en/ds/MAX44250-MAX44252.pdf

    19nV/°C Vos drift max. Combined with a pass transistor, it would do the job easily.

    Of course, feedback resistors would have to be sourced for low or matching tempco, and the environment would need to be controlled, as I already said.

    But that is neither here nor there. His original requirements weren't.
     
  11. crutschow

    Expert

    Mar 14, 2008
    14,779
    3,794
    How do you expect to get 10+ volts from a 9V supply? :confused:

    Note that for 1mV of accuracy at 200mA, the total impedance between the source and load, including the ground resistance, can be no more than 5mΩ.
     
  12. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    Joeyd- both great suggestions that are apparently targeted at what I'm trying to to (energize sensors with a high precision source).

    These will be used in a "room temperature" environment with real time temperature monitoring inside the case so I can back out temperature related drift if needed. I didn't think to include that in my original question.
     
  13. dannyf

    Well-Known Member

    Sep 13, 2015
    2,196
    421
    Put a current amplifier on it.
     
  14. crutschow

    Expert

    Mar 14, 2008
    14,779
    3,794
    I repeat: Where is the supply for the 10V coming from?
     
  15. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    This won't be possible. It is one thing to know the drift spec (min,max,typ). It's another to know how much and which direction. This can change for each device. It's better to just control the (local) environment, if possible.

    Further, if you use an integrated part, like the LDO I presented as an example, there will be "self heating" and it will be line, load, and ambient temperature dependent. In this case, it is best to ensure that the basic specs meet your requirement across the expected die temperature range.
     
  16. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,927
    3,528
    Why is this so important to you? He's got 13.5W available and only requires 0.2W. A basic integrated switcher can provide the ~10.5V minimum required for the LDO I referenced above.

    Edit: 2W ... forgot to shift the decimal. Other comments stand.
     
    Last edited: Jun 24, 2016
  17. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    Hi All.

    I wound up going with a dedicated MAX6176A for each device. On this piece I'm doing right now I only needed 4 so it wasn't too much of a headache to solder up.

    They are 1.5 ppm/C and +/-5mV at 10V delivery, so I get my 4-5 significant figures that I was wanting.

    Thanks for all your help.

    https://datasheets.maximintegrated.com/en/ds/MAX6173-MAX6177.pdf
     
  18. hp1729

    Well-Known Member

    Nov 23, 2015
    2,313
    279
    Super!
    A voltage reference is not intended to supply power, just be a reference. Thermal drift would seem to make the OP's goals difficult. And ... how do you measure something with that precision?
     
  19. bwilliamson

    Thread Starter New Member

    Jun 23, 2016
    7
    0
    This is being measured with a 22 bit ADC (24 bit nominal). https://labjack.com/support/datasheets/t7/hardware-overview

    The MAX specs say the chips drift 1.5ppm/C, so to get 4 sig figs on a nominally 10V signal the delivered voltage would have to drift by >10mV or 1000ppm... or >5mV since they claim the chips are +/-0.05% off from the stated rating.

    These things can supply up to 30mA each, and each sensor that I'm powering needs 20mA so that seems to be good.

    I'm probably overlooking lots of other sources of error that would creep into this, but as far as actual application I'm getting a good signal.
     
  20. crutschow

    Expert

    Mar 14, 2008
    14,779
    3,794
    It's important to me because it's a question I have.
    You have a problem with that?
     
Loading...