Post Go back to editing

LTC2323-16 temp co. fitting curve, INL curve?

Category: Datasheet/Specs
Product Number: LTC2323-16

Hi Experts,

I am using LTC2323-16, as I/V simultaneous sampling. For LTC2323, in spec page7, Offset error and Gain error temp curve are there, combine with Page11 fig2 transfer function curve to adjust system setting. Is it right solution for temp co. compensation? Besides, where I can find the INL curve? I mean, I need to find a solution to minimize the INL error using curve fitting. Basically, I want less than 2LSB INL. LTC2323-16 is good part for SPS and Bits demand. But INL is not good enough, or other new part can be introduced?

I found that, in small signal, or say the mV level input, the INL is not stable, if we amplify the same signal, say X2, or X4, the INL will be improved. For this part, in near GND swing, the C cap tree induced more error caused SNR is dropped? What is the Ideal input range to get best INL performance?

  • The typical curves shown on page 7 are just that, typical. Offset and gain error will vary from part to part. If you want to calibrate out offset and gain error, you will need to measure each part. INL error will also vary from part to part. If you want to calibrate INL error, you will need to measure INL for each part. A far less costly solution would be to buy a better ADC.

    You do not say what sample rate you need. Take a look at AD7380 and AD4630-16.

  • Thanks Ghoover!

    I need 4Msps, I learned that AD7380 and AD4630-16 before, seemed AD4630-16 is good, but this part SPI port is not good to FPGA ports, need more IO.

    If I change the input range to mid-REF level, it will help INL? Or the INL is not depended with voltage range, just vary with each part? If so, I will keep recent near GND range.

    More question, if error is vary form part to part, can we get the factory data of testing? say combined with physical goods?

    AD7626 is a good part, but single channel, I need Dual CH with Simultaneous Sampling.

    AD7380 is good for ADC core, but no S/H block, I need 4CH Amps to buffer, ADA4805 is good part I had learned, if for pre-amp, I am not sure ADA4807 or ADA4805 is best part to AD7380 buffer

  • If you look at the THD, Harmonics vs Input Common Mode graph on page 6 of the data sheet, you will see that if you can keep the input common mode voltage constant at about 2.1V performance improves noticeably. If you drive the inputs with a fully differential amplifier, it is possible to keep the input common mode voltage constant. You don't say what your input signal BW is but something like the ADA4945-1 might work for you.

    It is not possible to provide INL plots with each individual part.

  • Hi Ghoover,

    Thanks for strong support!

    I used InAmp in the front of LTC2323, I think that set InAmp REF to 2.1V will be great, but it will loss my small signal range, as my input is uni-direction, I will try to add MUX for REF to do range switching in further design.

    I have a question, if I do not change the recent design, as I have 2 set prototype is running, my system error will be +/- 12LSB in worst case? I saw page5 Note6 comments, I guess that maybe better @ My input signal is 10-20KHz, but I need to sampling the rising edge or falling edge, how can I use any tool to calculate the error?

    Thanks again!

  • If zero-scale and full-scale error are calibrated out, your max error should be +/-12LSB. If zero-scale and full-scale are not calibrated out, then max error could be as high as 114LSB. This does not include any error that could be introduced by the InAmp.