Build a data logging system

Discussion in 'The Projects Forum' started by Dennisdb, Aug 12, 2015.

  1. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    I am building power supplies for different lighting applications: 230VAC in to 3, 4, 5 channels 24VDC out with PWM dimming.

    I would like to test these by logging the input and output current, voltage and phase as well as the temperatures inside and outside of the housings.

    I need to determine the systems efficiency and temperature-dependent behaviour (voltage drop, less power available, temp protection active etc...)
    I could put a massive amount of voltage meters on the system and log those, but I am looking for a more professional solution.

    I do not really know where to start.
    What do I need, what do I have to take into account when building this.
    The budget is preferably no more than 4-5000 euro
     
  2. TheButtonThief

    Active Member

    Feb 26, 2011
    219
    38
  3. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    Hey ButtonThief

    Thank you for your reply. This is a very nice device.
    Could you or someone else on here tell me more about data acquisition? Do you have any experience?
    I am reading articles throughout the internet to inform myself, so as to buy/build the right system that will suit my needs for at least some years. ;)

    I also have no idea how I will process the data. In what form is it acquired? .csv files?
    Can I connect every device to a Labview program?
     
  4. TheButtonThief

    Active Member

    Feb 26, 2011
    219
    38
    I'm an electrical engineer, yes I have experience collecting various types of data on lines around a pretty big plant. I've used everything from digital storage clamp meters and devices such as the one I just linked, to other more Robinson Cruso methods such as propping up a video camera to watch for mistakes that a machine operator may be making.

    What specifically would you like to know?
     
  5. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    I am a bit in the dark as to what parameters are important to compare and what makes these devices so pricey.

    The device you posted has 4 input channels, which is not enough for my needs. I need at least 10 channels to start with.
    Does the channel have different specs if I want to acquire reads from voltage, current or temperature? (I figured these are all scaled voltage measurements)

    Can I directly connect these channels to labview without the extremely accurate but very expensive hardware from NationalInstruments?
     
  6. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,770
    970
    There are tons of dataloggers out there...
    Labjack and Measurement Computing make some "low" cost versions
    Omega/NI/Agilent and others if you have money to spend and need more "accuracy",etc...

    Fully define your requirements and see what fits the bill.
    (resolution/sample speed/measurement type/number of channels, etc...)

    Do you really need Labview? OR maybe you just want to be able to export to Excel/csv which most if not all do... including the lower cost units as they all typically have some graphing/exporting software available..

    Heck..
    I bought an Agilent 34970A at an auction for $100USD which came with 2 x 40 channel multiplexing cards..
    And other cards are typically found on ebay,etc.. for cheap..
    Get it calibrated an away you go..
     
  7. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    I don't think that I really need Labview. I just remember Labview as a graphical programming environment specifically suited for data acquisition from my days as a student. I thought I would need it to scale and convert the measured signals before writing and analysing the datasets.

    I think a low cost version should suffice, though I am not sure about the needed budget. I threw in a number ( 4000 to 5000 euro) because the first devices that I found where readily available systems starting at 18-20.000 euro.
    Although I now think that even 4000 is still a tad bit too much considering the prices of different options that I came across.

    I want to test my power supplies. These as powered on a 230VAC 50Hz net and supply 3 or more channels of 24VDC with PWM. What I am interested in is:
    • The exact temperature where the AC/DC blocks go into thermal protection (should be at 70°C but I need proof)
    • The temperature difference on the components, in the cases and the ambient temperature
    • The behaviour of U and I related to the rising temperature of the device.
    • The influence of the temperature on the available power at each channel (Should be 150W, but practice has proven that at 80% of rated temperature the power drops to about 90-100W)
    • I also need to define the efficiency of the devices (Power out)/(Power in) and the power factor, but this is on the 230VAC side, so I think I better use my Fluke digital scope for that.
    Thus: 3x temperature, 3x times volts, 3x current makes at least 9 channels and I will develop some bigger supplies in the near future.

    As I use PWM, I would think that I need a sampling rate of less than 1ms (is that correct?)
    The resolution should be 0.5 V/I/°C
    Portability is not so much of an issue.
    More important is an easy set-up
    What do you mean with measurement type? (by guessing, I think these are Temperature, current and voltage)
     
  8. Reloadron

    Active Member

    Jan 15, 2015
    963
    232
    Here is what I would suggest. Go about this exactly as you have began in determining how many channels of DAQ (Data Acquisition) you will need and then add several channels for future expansion. Next determine the uncertainty you can live with. For example the tolerances of your measured data.

    I would likely use a PC in conjunction with my DAQ, something like a laptop might be suitable, so I need to decide on how my DAQ device should interface to my PC. I happen to like USB and here are some examples of USB DAQ devices.

    Based on the DAQ you choose you will likely need some signal conditioning. For example a typical DAQ will not measure 230 volt mains voltage or current so we use some signal conditioners to take the signals we have and convert them to the signals we want. For example AC Voltage Transducers or AC Current Transducers. Note the links provided are merely examples. Unless the chosen DAQ has Temperature you will need some temperature transducers and to decide, based on temperatures involved, a choice of sensors for temperature.

    Most data logging DAQ devices include software where the user can designate and range channels. Pretty much a matter of what you need and any budgetary limits. You build your system almost like using building blocks.

    What will the ranges be? For example the approximate values for volts, current and temperatures plus any others? You have resolution and you also have the allowable error or how accurate.

    Ron
     
  9. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    I will overthink the system further and then try to find suitable devices.

    You are correct, I did not really think thoroughly.
    The current fluctuates from 0-24mA's up to 6 or 7A's. I would think 0.1mA would suffice for me
    The voltage is measured with ranges from 0-10V, 0-24V I was thinking about 0.05V resolution
    The expected temperature range is 0 - 150°C with a resolution of 0.5°C
     
  10. Reloadron

    Active Member

    Jan 15, 2015
    963
    232
    OK for example let's say I have this DAQ device. When we look at DAQ devices and their A to D conversion a big concern is the bits. The device I linked to is a 16 bit A/D device. Actually 16 bit is pretty good. More on that shortly. This unit offers 8 Differential or 16 Single Ended analog inputs. Each analog channel then offers input ranges of:

    What this means is That any range for example 0 to 10 volts will have 2^16 quantization levels (16 bit). So 0 to 10 volts becomes 0 to 65536 quantization levels. So as an example if I use a 0 to 24 volt input range using one of these transducers then 0 to 24 volts = 0 to 10 volts = 0 to 65536 bits. The A/D will give you a resolution of 10/65536 = 15.2 uV. While not that accurate you would well exceed what you need. Most DAQs like I have linked to include very good charting software and storage. You are only limited by the hard drive in the PC. When we save or collect data we need to put it somewhere. :)

    So you sit down and look at all the data you want or need to collect and decide which transducers will do the job. The transducers I have linked to are just examples of some I used in my career before I retired. Dozens of companies make these things and outputs are typically 0 to 5 volts, 0 to 10 volts or 4 to 20 mA. Voltages AC or DC, Currents AC or DC, Temperatures and other engineering units by ranges.

    Ron
     
  11. Dennisdb

    Thread Starter New Member

    Jul 14, 2015
    10
    0
    So,

    I found this system from Keysight which appears to almost suit my needs:

    The interesting thing is that each module can work on independently on USB without mainframe, but if I have more modules, I can install them in the mainframe and they are all synced together and transmit over a single usb connection.

    However, I still didn't find a fitting solution for the 26VDC to 10VDC conversion, or the scaling of 0-10Ampere current on 10VDC.
    Those transducers are a very expensive solution. (I think I did not fully understand the concept of those things).

    What do you think about this?
     
  12. Reloadron

    Active Member

    Jan 15, 2015
    963
    232
    That system is pretty much like what I linked to. Yes, it should work fine. Before I forget, when buying units like this, make sure you get the connectors you need. Units like this that do not have screw connectors do not always include the needed connector(s). Also make sure some charting software in included, unless you plan to write your own. The modular 6 slot unit is also a good way to go if more channels may be added at a future time.

    Yes, voltage and current transducers can get expensive. Less expensive approach is to build your own but unless you know what you are doing and all the parts to buy a simple ready off the shelf is the best way to do it. The units I linked to were CR Magnetics. Possibly with some luck less expensive units can be found. What you want are called Voltage and Current Transducers.

    Ron
     
Loading...