DDC264: Higher Johnson noise than expected

Antoine Letourneau1

Part Number: DDC264

Hi,

While using the DDC264EVM I am able to reduce the noise to around 6ppm with proper shielding around the circuit, matching the datasheet of the DDC264. This is done by wiring all of the 10Mohms resistors from the DDC264's input to ground. By lowering the resistor value, the noise is expected to increase because of the Johnson noise in the resistor. There should not be any 1/f noise as no current is flowing through the resistor. My problem is that when using a lower resistor value (228k, 56k or 18k) the noise is 3-4 times higher than expected. As I understand, the RMS noise should follow sqrt( (4*Kb*T) / (R * 2* Tint) ).

For a integration time of 160us and the gain set at 3 (37.5pF) the gain of the DDC264 is supposed to be close to 1040ppm/nA. For a 56k resistor, I should see a current noise of 30pA, corresponding to close to 30ppm. What I see is closer to 105ppm. The same pattern is true for other resistors values.

Can you point out how to solve this problem ?

Best regards,

over 4 years ago

0 Praveen Aroul over 4 years ago

TI__Mastermind 49666 points

Hello Antoine,

Welcome to the TI e2e forum!

Thank you for evaluating the DDC264 device.

Regarding your question on the noise, the device systems expert will get back to you tomorrow.

0 Eduardo Bartolome over 4 years ago

TI__Expert 4620 points

Hi Antoine,

All your explanation on the first paragraph is correct (and not trivial, actually). Only minor thing (doesn't affect anything) is that there is actually a very small current flowing through the resistor as the input bias of the integrator is not exactly zero.

The calculations on the 2nd paragraph are also correct and in fact I had to talk to the designer to figure where the difference was coming from. It looks like there are other sources of input voltage noise that as you make the resistor smaller result on larger input current noise (can't comment on the details). That integrated input voltage noise is about 6.2uV, which would yield about 110pA (6.2uV/56k) close to your measurement). If our theory is correct, you should get larger numbers using that same constant for the smaller resistors (inversely proportional relation). Changing the integration time should not make much of a difference either as this term doesn't depend on it (the 30ppm does but it is small in comparison).

We are curious here to know why you are trying the smaller resistors, though... I'll go ahead and mark this post as solved, but do not hesitate to comment on this or if you can actually verify (or not) the theory.

Best regards,
Edu

0 Antoine Letourneau1 over 4 years ago in reply to Eduardo Bartolome

Prodigy 45 points

Hi Eduardo,

The 6.2uV noise explains what I am measuring. Many thanks for your fast and helpful reply.

Best regards,

0 Eduardo Bartolome over 4 years ago in reply to Antoine Letourneau1

TI__Expert 4620 points

Great! Thanks for asking! I learnt something myself :)

As mentioned, please let us know if you ever find any other incoherences...

Best regards,

Eduardo

0 Antoine Letourneau1 over 4 years ago in reply to Eduardo Bartolome

Prodigy 45 points

Hi Eduardo,

I does the DDC2256 presents the same noise (6.2uV) at its inputs ? If lower, this part could also meet my requirements if noise is lower.

Best regards,

Antoine

0 Eduardo Bartolome over 4 years ago in reply to Antoine Letourneau1

TI__Expert 4620 points

Hi Antoine,

Will have to find someone here to answer that. I don't know... Let me check...

Also, as that part is under NDA, I'll PM you and take this offline.

Regards,
Edu

Data converters

Data converters forum

DDC264: Higher Johnson noise than expected