AD7195 vs AD7172-2 in weighing applications

Hi,

following the thread "Precision Weigh scale Issues".....

I want to make a comparison between a weighing scale that we have operating with exceptional behavior (AD7195), and a new one that we are designing with the

AD7172-2.

1- I do not understand why the AD7195 behaves better than the AD7172-2 over the temperature range, in addition to all the improvements that I am supposed to

have made in the design with the AD7172-2.
2- I attached comparative tables of the 2 devices, and the AD7195 does not need temperature compensation in the entire range, even in no chop mode (VDC

excitation of load cell). With the AD7195 we do an internal zero and full scale calibration when turning on the equipment (only once), while with the AD7172-2 we

do the internal zero calibration when turning on the equipment (only once). Then we apply the equation of the line, with the points recorded with empty scale and

scale with known weight.
NOTES: The load cell is 30kg, the 2 in bipolar mode, load cell excitation in AD7195 is + 7.5V and in AD7172-2 is + 5V.

I was quite disappointed by the AD7172-2 regarding its temperature deviations, and I still have no idea how I can make that compensation in the microprocessor.

Even so, I will try the AD7175-2, which is pin-to-pin compatible, to see if it works better, although it is more expensive.



Tagged
[edited by: JValeriani at 4:40 PM (GMT 0) on 3 Oct 2019]
  • 0
    •  Analog Employees 
    on Oct 7, 2019 9:28 AM

    Hi,

    1. Yes, you are correct. So far the AD719x series offers the best performance in terms of rms noise. And this lowest noise value will lead to a higher system accuracy in a load cell application. But if you want alternative parts for high speed weigh scales, the AD717x series such as AD7172-2, AD7175-2, AD7177-2 are useful, however external PGA will be required. So, the part we recommend really depends on the customer's requirements. Take note that larger excitation voltage gives much higher performance resolution and lower noise will lead to much better performance. Noise is affected by ODR, Gain and Filter choice. So like I said, it would depend on your target application requirement. 

    2. In terms of temperature compensation, we really suggest to perform calibration when there is a change in temperature. Though AD7195 temperature drift is much better for Gain greater than 32, at Gain 1 to 16 it should have the same performance for AD7172. So I would really suggest to calibrate your system each time a Gain, Temperature, Polarity, channels is changed.

    Thanks,

    Jellenie

  • Hi,

    1- So, for getting a better performance, do you recommend applying a higher excitation voltage (although by doing this, I make a pseudo-ratiometric system), or as you recommend that I can improve the system, with the scheme I have made? The target is that the weighing scale meets the legal weighing requirements.
    2 - It really only occurs to me to calibrate each change of temperature the zero internal offset, but the gain offset is not how to do it. I guess I will have to do a lot of tests to see if it follows a compensation relationship and I can apply a compensation coefficient. And I hope it is the same for all weiging scales and not have to be done individually.
    I don't know why I changed something that worked very well .... :(
    Regards.
    Javier.
  • 0
    •  Analog Employees 
    on Oct 8, 2019 9:57 AM in reply to jartigas

    Hi,

    Like I mentioned for the Gain=1 they should have the same performance. It is probably the gain stage that can contributes to noise/errors for your AD7172. I'll check on again your schematic and details and see how can improve the system to meet your desired specification. I'll get back to you as soon as I can.

    Thanks,

    Jellenie 

  • 0
    •  Analog Employees 
    on Oct 16, 2019 6:24 AM in reply to jartigas

    Hi, Javier.

    Upon looking at your schematic, I am seeing AD7730 instead of AD7195 can you confirm this please? It might be worth comparing the two ADCs with the same setup and configuration. I'm seeing quite difference with the setup and circuit as AD7195 uses the internal Gain and AD7172 uses external amplifiers. Ferrite bead in supply is good but the ferrite bead in load cell can possible add errors though not sure how significant, could you just please remove this in the circuit?

    If you could provide us more details regarding the configurations such as Filter type, Output Data rate, Gain etc. so we can compare the two ADC core in terms of noise performance, then that would be great. It might be also worth it if you could isolate the gain stage (bypass) and then use the same configuration for the two setup just to compare the performance. If you could look at the noise performance of the two ADCs at a Gain=1 with the same configuration you will see that AD7172 is quite better. So we have to take a look at the whole system why the old system give you a better performance rather than looking at the specific part itself. 

    Are you also using the same load cell (sensor) for both system? In terms of target performance or requirements if you could tell us what counts are you targeting and then we calculate what excitation or configuration are needed to meet this specification. Probably we can start with the configuration on the ADC itself by ruling out some external circuit (bypassing the Amplifier stage) and when we prove that we are getting the target performance we can go deeper on the gain stage and see how can we still meet the same performance.

    Thanks,

    Jellenie

  • Hi Jellenie,

    Sorry, I send you the AD7195 schematic and test with AD7172-2....

    Of course, I remove ferrite beads of signals.

    The load cell is the same for testing AD7195 vs AD7172-2 (Mettler Toledo MT1041 Class C3, 30kg, 2mV/V). https://www.mt.com/es/es/home/library/datasheets/industrial-scales/MT1041_DL.html

    Scale with AD7195 Converter (Configuration):

    The scale in start-up does 2 calibrations (just in start-up) internal zero scale and internal full scale, and It´s working with the same configuration always. Excitation of load cell (7,5VAC).

    Enable Chop.

    Bipolar Mode.

    Buffer enabled.

    Gain 64

    Sync 4 filter

    REJ60 ON

    ODR 2,5

    The Test above does not need temperature compensation in the entire range, even with VDC

    excitation of load cell.

    Scale with AD7172-2 Converter (Configuration):

    Right now is in prototype stage. I attach the test again, without ferrite beads in signals, Rg=1K.

    Bipolar Mode.

    External Gain stage, Rg= 1 K --> Gain = 200

    Buffer disabled.

    Sync 3 filter.

    ODR 2,5.

    If you need more data, I will be happy to tell you.

    Regards,

    Javier.