DC2094A-A Evalboard

I have just reviewed the company and found a DC590 interface card. 
According to demo Manuel CD2094A this device is supported, unfortunately the PScope does not find the DS590.
Can I do anything, the DC590 is already a few years old (it is a SW update required)?
best regards 
Thomas

Hi Thomas,

I am the applications engineer for the LTC2348/DC2094.

You can use QuikEval directly with the DC590 and DC2094. To use PScope you need the DC890 controller. With the DC890 controller you can also use LinearLab Tools and write your own data acquisition application.

Let me know if you need any additional help.

Guy

Hi Guy,

Do you happen to know how I can connect the SPI bus of the LT2348 directly to my FPGA without disturbing the "on board FPGA (Altera)"?

As already mentioned, I have already taken some measurements with the Eval board and the DC590. For this the following modifications / settings have been made:

1) R105 removed and R114 equipped -> differential input for AIN0.

2) JP5, JP6 and JP9 closed -> inputs on GND.

3) J7 Jumper for AIN0 on VIN1, AIN1 on VIN2, AIN2 on VIN1, AIN3 on VIN1, AIN4 on VIN2, AIN5 removed, AIN6 on VIN2, AIN7 on VIN2.

4) Temperature sensor (10mV / ° C) to AIN5 via JP10.

Subsequently, the data for temperature response of about -30 ° C to 65 ° C were recorded. The data recording took place with 1s sampling time over 2 hours and 40 minutes. In the evaluation, I noticed that more than half of the readings show a value that is less than -4LSB, how is this explained (my setup?) ??

AIN0       AIN1        AIN2       AIN3     AIN4       AIN5        AIN6       AIN7

sum ((Dat.data) <- 4)
ans =

        5871       8617       4847       3518       5580       611       0       204

even values ​​smaller than -8LSB appear very often.
sum ((Dat.data) <- 8)
ans =

         848       4203       41       10       75       608       0       0

LTC2348_18_Log_16_Ta_1sec.txt

Hi Thomas,

Based on your description, you have grounded all of the ADC inputs except for AIN5. If you look at your data log, you will see that except for AIN5 all of your readings are very close to zero. This is what I would expect.

Looking at the AIN5 data, it starts at -3882 and rises to 10003. This is a change of 13885 codes. The data log says that AIN5 was using the +/-1.25Vref range. With a 4.096V reference this means the input range is +/-5.12V. Your temperature range was -30C to 65C with a tempco of 10mV/degC. This will cause a 0.95V change in the sensor output. 0.95V/10.24V full scale * 2^18 codes results in an ideal change of 24320 codes. You are only seeing about half that many codes, so you have a significant gain error. Did you measure the output of the temperature sensor with a DVM when you took your data to see how much it actually changed? If the temperature sensor has an output impedance greater than 10kohms it will be necessary to increase the capacitor across the analog inputs to at least 680pF (Use a C0G of film type capacitor) to get an accurate reading.

To communicate directly with the ADC, it should only be necessary to tie WRIN to ground. WRIN2 should by default be held high. The CNV, SCK_IN2, SDI and SDO0 signals on J4 connect directly to the ADC. By default the LVDS/CMOSL line should be low (CMOS mode) but you might check just to be sure.

Hi Thomas,

Based on your description, you have grounded all of the ADC inputs except for AIN5. If you look at your data log, you will see that except for AIN5 all of your readings are very close to zero. This is what I would expect.

Looking at the AIN5 data, it starts at -3882 and rises to 10003. This is a change of 13885 codes. The data log says that AIN5 was using the +/-1.25Vref range. With a 4.096V reference this means the input range is +/-5.12V. Your temperature range was -30C to 65C with a tempco of 10mV/degC. This will cause a 0.95V change in the sensor output. 0.95V/10.24V full scale * 2^18 codes results in an ideal change of 24320 codes. You are only seeing about half that many codes, so you have a significant gain error. Did you measure the output of the temperature sensor with a DVM when you took your data to see how much it actually changed? If the temperature sensor has an output impedance greater than 10kohms it will be necessary to increase the capacitor across the analog inputs to at least 680pF (Use a C0G of film type capacitor) to get an accurate reading.

To communicate directly with the ADC, it should only be necessary to tie WRIN to ground. WRIN2 should by default be held high. The CNV, SCK_IN2, SDI and SDO0 signals on J4 connect directly to the ADC. By default the LVDS/CMOSL line should be low (CMOS mode) but you might check just to be sure.

Hello
I'm not so worried about the temperature measurement (ADI5), 
but rather I wonder about the results of the short-circuiting inputs. 
Here are a lot of values ​​with a high deviation to see (more than 4 LSB and 
even very many with more than 8 LSB) according to the data sheet, 
the should stay between 2 to - 4 LSB (data sheet page 11: 
"Zero-Scale Error vs. Temperature and Channel ").
 Why do not I receive such a measurement result?
best regards
Thomas

Hi Thomas,

If you look at the histogram on page 9 of the data sheet, you will see that there is noise associated with that offset. Even with the inputs shorted, you should expect +/-4LSB of noise around the offset. So with a 2-3LSB change in offset due to temperature drift in addition to the +/-4LSB of noise that is 6-7LSB of change. The inputs are not really shorted in this case. You are going through LT1355 op amps which will add additional noise. If you look at the 1/f noise curve of the LT1355 you will see that the longer you wait the higher the potential noise will be. If you actually ground the ADC inputs to the PCB ground plane, you should be able to get the noise down a little bit more but I don't think the results you are getting are that bad.