LTC1867 ADC works fine on 3.3V but not on 5V supply

Hello, DeBug .

We'll look into this and get back to you. 

Regards,
Jo

Many thanks, further, another experiment to rule out the main board power supply that feeds the hand-wired ADC board. The main board is 4 layer PCB with 2 dedicated ground layers and SPI/5V/3V connectors . The only change is to feed the LTC1867 and LM4040-4.096  with an external dedicated power supply.  This supply is a simple board with an 7805 and an LD1117-3.3 only.

The result is the same as supply from the main board 5V/3.3V. At 5V 1 bit resolution where values range 15K apart (values vary 15 bits) thus giving1 bit resolution. Only changing the power wire from the 7805 to the LM1117-3.3 results directly into 12-13 bits resolution. (ADC values vary 3-4 bits).

As another side question. The datasheet calls out for tantalum bypass capacitors, but the DC806A BOM capacitors listed are all ceramic. Does this mean tantalum is not needed for good performance ?

Reference Buffer Output Pin. Bypass to GND with a 10µF tantalum capacitor in parallel with a
0.1µF ceramic capacitor/ REF (Pin 10): 2.5V Reference Output. This pin can also be used as an external reference buffer input for improved accuracy and drift. Bypass to GND with a 2.2µF tantalum
capacitor in parallel with a 0.1µF ceramic capacitor

The DC0806 BOM uses: 06035A102JAT2A, 0603ZC104MAT2A, 0603ZD105MAT2A, JMK212BJ106MG, LMK212BJ225MG, JMK212BJ475MG. 

We procured TAJA225K010RNJ and TAJA225K025RNJ (Mouser states 7 Ohm ESR but it's not mentioned in the datasheet). Would these improve ADC performance ? What ESR is required as there is  also TACU series with 5Ohm,  F921 with 12 Ohm, TAJR 20 Ohm and TPC1C series with 16Ohm ESR.

Hi, DeBug .

Please see the following insights/questions below:

1) Based on how I understand, you are using internal reference all throughout your tests. Right? What are the values provided like Vref and Refcomp from each tests? Are these your measured values when using 3.3V or 5V to your VDD? And speaking of 3.3V/5V, are these connected to your VDD? 

2) Can you provide me a block diagram, or a schematic of your setup/board? This can help me assess your query better regarding the connections of your supplies and your configurations. 

3) Nevertheless, may it be a tantalum or a ceramic capacitor, these two components were tested both to contribute to the device to be in its optimal performance. You can use the BOM from its evaluation board as your reference.

Warm regards,
Jo

[1] Yes, internal reference. The values of refcomp and vref were provided in the original post above for both 3.3V and 5V Vdd power. These reference/comp pins are not connected to VDD. they are decouple with caps as in DC806A. The VDD (pin 16)  is connected to +5 or 3.3V. using 5V values are totally wrong, using 3.3V pretty good.

[2] since it is hand wired, no schematic. it follows the schematic of DC806A without using the external reference. I did test with an LM4040-4.096V but  makes no difference.

MCU => SPI => internal vref/refcomp => 8 channels unipolar =>  cap and serial resistor on inputs => 10K potentiometer on one channel, all others inputs  to GND. (via the cap/resistor). 5V from external USB or LM7805  or 3V3 from LM1117-3.3V or via 5V and an RT9059.

[3] Thanks

B.T.W this is not an LTC1876L (which is the 3.3V version)

Schematic as shown. Default:  JP8, JP10 short, others open. This is the 5V  supply situation and  has bad result, Change circuit to 3.3V (JP5 or JP7) result is good. Remove the PI filter has same 5V poor result.

Hello NathanT Any other ideas on this with the provided additional details above ?

Hi, DeBug .

Apologies for coming back late. How are the things going? I will look into this now.

Regards,
Jo

No further progress made. Any suggestions for the schematic as provided above? I'll put a scope on the power to see if there is anything special.

Hi DeBug, 

LTC1867 operating reliably at 3.3 V VDD but encountering instability at 5 V VDD suggests a logic level mismatch with the STM32F401 µ-controller.

To ensure proper SPI communication between LTC1867 operating at 5 V and STM32F401, it is necessary to implement level shifting (using a Level Translator IC), and adhere to the logic level limits outlined in the µ-controller datasheet.

Thanks,
Faizan