• State TI Thinks Resolved
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 60 views
  • Users 0 members are here
Support feedback
Options
Options
Related

LM331: F/V conversion non-linearity problem

József Hermann
Expert 1050 points

Part Number: LM331
Other Parts Discussed in Thread: LMC6482, , LMC6484

Dear Forum,

In a production tester I use LM331s to measure RPMs with a new encoder 360 inc/rev.

(Max RPM is 9000 FS to be measured). Previously I used 100 inc/rev encoder without any problem.

For new encoder Rt-Ct was scaled down: Ct=4n7 -> 1n5 while Rt=12K (Rout=18K || 330nF, Rs~15K trimmerable, VCC=5V, loading is by LMC6482 non-inverting amp x4 to produce standard 0-10V automatic signal).

Now, -with new encoder- when tester calibarted say at 4000 PRM, the 7000 RPM measured inaccurately (too low value by ~300).

Note, there are two '331 in my system, one is quite OK (~still acceptable, error is only 100 PRM) but the second one exhibits unacceptable error mentioned above) despite the circuits exactly the same (ICs are soldered onto the PCB so I could not swap them).

It was tried using signal generator instead of encoders but the result is the same.

Can the '331 handle such high frequencies as F/V (~50KHz)? It seems something extra delay/prop. time starts to be effective at higher frequencies. Or the problem caused by an improperly choosen Rt+Ct pair?

Can you pls give some advise?

Thanks,

Joseph 

 

  • 0
    TI__Mastermind 18680 points

    Hi Joseph, 

    Are you using this implementation from the datasheet or a different circuit? 

    Yes, the LM331 should be able to handle up to 100kHz frequencies; however, the device is only specified to 10kHz, so non-linearities and errors may be introduced at the 50kHz range. 

    Best regards,
    Ashley

  • 0 in reply to Ashley Kang
    Expert 1050 points

    Dear Ashley, 

    Thank you for your quick response. Partially: output filtering is different: 3rd order Sallen-Key Bessel fc~20Hz formed by two opamps  Everything else is the same. However, Rx =10K at pin 7 in my circuit. Maybe is to too low value: the differentiated trigger pulse may not have enough amplitude to overdrive the input trigger comparator (tpd higer->can degrade perfomance)? I change RX to 56K at least.

    Joseph

  • 0 in reply to József Hermann
    TI__Mastermind 18680 points

    Hi Joseph, 

    What power supplies are you providing to the LMC6484? Also, are you using low temp coefficient passives and NPO caps? 

    I believe Rt and Ct determine the timing and how often the timer will switch. I would try to decrease these values to be able to detect the faster encoder. 

    I am not as familiar with using the LM331 as a frequency to voltage converter so let me also look into this a bit more to see if I can find a suitable solution. 

    Regards,
    Ashley

  • 0 in reply to Ashley Kang
    Expert 1050 points

    Hi Ashley,

    thank you for the advise. I could fix the problem, now the RPM measurement in the full range within 1% accuracy. Thre were two problem:

    - Rx pull down at Pin#7 was equal to pull up (10K) - > bias voltage was very low, much lower than bias at Pin#6. - > Should be the same. I now use 68K ensuring near same bias at both comparator terminal. 

    This solved the problem partially:

    - Altough for new encoder CtRt product was decreased by the Ct lowering but not enough. For new ecoder (higher frequency) new ideal value is 1,3nF but I used standard 1,5nF causing problems again at top of the range -> Rt also lowered 12K->8,2K, CtRt < Tmin*0,75 required.

    I think you can close this thread.

    Regads,

    Joseph

  • 0 in reply to József Hermann
    TI__Mastermind 18680 points

    Hi Joseph, 

    Thank you for the update and glad to hear the problem has been resolved! I will close this thread now. If you have further questions, please let us know. 

    Best regards,
    Ashley