24 Bit ADC selection for High Speed Step conversions

Dear TI Gurus,

Many thanks for the help in my single - cycle application. I have found that the ads1252 and ads1259 are the most suitable for me. I can see the time required for a step function to appear on the output for the ads1252 is about 120 usec, and the ads1259 about 424  usec. This is time it takes for the cycle to be done and I can readout the data. However, I do not see anything that clearly mentions the actual sample time. How long do I need to hold the analog signal on the ADS input with the ads1252 and ads1259. After the data stream from the modulator is fed into the filter, there should be no need for the analog signal input  to be held during the digital filter time for single cycle measurements.

Andrew Duran

Hi Andrew,

The input signal will need to be held steady during the full 120 usec or 424 usec conversion cycles.The filtering process is a running average type of computation and it is an insignificant amount of time between sampling and filtering after the input has been oversampled.

Regarding the accuracy of the application, you will be challenged to achieve the 18-20 bits ENOB for the very short step widths around 100usec. The settling time of the inputs should be added to the conversion cycle time (of 120 usec or 424 usec) for the best results. Otherwise your output data will reflect the averaged input signals as they are settling. Also any input filtering will increase the input settling time. Therefore while designing the input filtering, keep in mind that there will be a trade-off in the settling time vs. input noise.

Regards,

Chris

Thanks for the advice, I do realize it is a difficult task to get near 18+ bits, that is why I am struggling to find a fast 24 bit ADC. My main result from this will be increased dynamic range from the 16 bit SAR converter I am using now.

The 1259EVMPDK kit arrived today and I am looking for the 1252 EVMPDK, but I can't find it anywhere. Is the 1252 EVM still available?

I do need to synchronize 256 simultaneous 24 bit converters which are driven by 256 individual analog boxcar integrators. The integrators are integrating for about 3 usec, and then they hold until the ADC's are done. This repeats at synchronous 1ms intervals, but must be initiated by a sync pulse for each pulse at 1 ms. rep rate. A sync pulse arrives and starts the integration, then 3 usec later the ADC's need to start, while the integrator holds. It seems that the ADS1252 will require too much time to sync up for each pulse using the SCLK . The 1259 has a short response to the START pin. So it seems that the 1252 would be better (faster & less hold-droop time) if it could re-sync in a few usec. This then leaves me with the 1259 as the only converter that can handle this application. Do you agree? Please let me know if there is any other 24 bit ADC that can sync and cycle as fast as I need.

Hi Andrew,

We have not released an EVM for the ADS1252. Regarding ADS1252 synchronization...If you provide a single modulator clock to all of the ADS1252 devices, you will be able to synchronize them once on start-up and use them in a synchronized continuous conversion mode.

For other options, the ADS1258 would be fast enough but would be less advantageous in terms of size and cost because you would not be able to take advantage of the multiplexed inputs. 18-bit SARs would also be an option; they are more expensive. I'll keep looking to see if I can come up with other suggestions.

Regards,

Chris

Hi Chris,

Thanks for the help. The ads1252 datasheet list the eval board from 2001 http://www.ti.com/lit/ug/sbau042/sbau042.pdf.... However, I can't find it anywhere.

Regarding the ADS1258, I am reading through the datasheet now, but perhaps you know the answer...Since I need to hold my analog signal for at least 434 usec for the ADS1259, can the ads1258 cycle through all 16 channels in this time?  I have 16 channels per PCB, and 16 PCB's, so it may be possible to use one 1258 per PCB to do the conversions in the same time as 16 ads1259's. Will the ads1258 cycle through all 16 inputs in 500 usec or less and have an ENOB as good as the ads1259?

I would like to use the ads1258 controlled by the start pin, and have it convert all 16 channels, and then stop until the next start signal. Would this be the same or better than 16 ads1259's?

Andrew

Hi Andrew,

Thanks for pointing out that link! I don't know the full story behind the ADS1252EVM, but we did do a beta version that was not finished. If you decide that it would be invaluable to you, I can look into what I could provide you with.

The ADS1258 can sample each channel in auto-scan mode at 23.739 kSPS (42 usec / channel) and with an ENOB of 19.5 bits. Therefore, you will be able to scan through 10 channels in the 434 usec timeframe (or 16 channels in 672 usec).

Notice, that this assumes that the data at the particular input is settled before the MUX switches to that input. In your application, using boxcar integration, it seems like you will know when the integrator data is ready and settled (if input capacitance is either very low or accounted for). However, if you are unsure if the data has settled, you would want to run a second scan through all of the channels you're unsure of valid data to check if the data converter output changes (see figure 45 in the ADS1258 datasheet. The input signal must be settled for 1 whole conversion cycle for valid data to be ready at the next /DRDY pulse). You might need to consider adding a delay after the MUX switches channels (see "Switch Time Delay" on page 21). This will increase the time to measure all 16 inputs, but it might make you more certain that the result is correct and remove the need to measure all 16 inputs a second time.

If this seems like a good alternate to the ADS1259 (even if you had to use 2x ADS1258's to read all 16 channels; scanning through 8 channels on each) it seems like this would greatly reduce the component count, size and cost of the solution.

Regards,
Chris

I have my ads1259evm-pdk working and I purchased another evm without the mmb0. When I plug this new one into the mmb0, I get a check ID error

Error Code 85:  Scan From String (arg 2) in libmmb0.lvlib:EEPROM Check ID.vi->ads1259evm.lvlib:ADS1259EVM.lvclass:ADS1259EVMPanel.vi:1    1/17/2013:1:35:38 PM Eastern Standard Time

Is something wrong with the new evm card or is there a setup that needs to be done ? I have uninstalled the adcpro and reinstalled the adcpro and ads1259 plug in. I can run the evm board that came with the mmbo, but not the new one.