noise on LTC2053 output

Hello RMC123,

I am moving this query to the Instrumentation Amplifiers community. Someone there should be able to assist you better.

Regards,
Paul

Hi RMC123,

Thank you for reaching out and sending your inquiries.

To further assess your problem and be able to share the correct recommendations, we would like to know more about the intended applications and design parameters.

Please share the following items:

• The Intended applications: what is the purpose of this circuit? i.e. as a Low Noise Reference source? And to source her discrete parts? ADC, DAC, etc..
• The target reference output is +1.25V DC?
• What Supply instrument are you using for +5V_ISO? What is the settings for Current, Voltage Clamps, etc..
• Can you share a scope shot of the +5v coming out from your supply instrument using oscilloscope.  Noise on your supply can degrade the performance of your reference circuit.

We can isolate other parts of the circuit to figure out where is the noise coming from, I prepared the following items to check:

1. Disconnect Pin9(IC16) connection to GND.
2. Ensure C53 is placed as closed to VI(Pin4), C67 is placed as closed to V+(Pin8)
3. Adding 10uF Tantalum Cap to make the supply clean; both VI(Pin4) of IC6 and V+(Pin8) of IC16
4. Please disconnect the connection from Pin6 (IC6) and check the output of the LTC1790 (desired is +2.5V DC)

      5. Disconnect R79 (including +5V pullup) and C52 and check the VREF Test point using Oscilloscope (desired is +2.5V DC)

      6. Check VREF /2 using Oscilloscope (desired is +1.25V DC).

Please let me know if these suggestions help.

Regards,

Arsol

Hi Arsol, All the DC levels in the circuit are correct. And conveniently, I've already done most of what you suggest.

5V_ISO = 5V + the AC ripple I shared in the original post.

VREF = 2.5V.  This voltage is actually very clean and flat.  I see very minimal noise on the oscilloscope.  

VREF/2 = 1.25V + the "steppy" AC noise at 3kHz shown in the original post

Re exposed pad - I can't disconnect from the ground connection on the footprint.  Quickly looking through the datasheet, I don't see that it says it must be connected to anything.  Is the pad really best left floating?  I don't quite understand.  Can you please explain?  

5V_ISO is provided by a TRACO TIM 2-0911 module. 

Adding a 10uF cap does not help the ripple on 5V_ISO, nor does it affect the output of the LTC2053.

Removing R79 does not affect the output of the LTC2053

When you say : "Noise on your supply can degrade the performance."  yes.  That's what I'm asking about.  Does it make sense that the noise seen on the output of LTC2053 is caused by the 1kHz 50mVpp ripple seen on the 5V_ISO power rail?

So far, it seems like reducing power supply ripple DOES improve LTC2053 output, but the LTC2053 still has "steppy" noise on the output.  Is there anything else (besides ripple on the power supply) to suspect as the cause?

I have another theory that's a little difficult to test in the existing design : improving the placement of bypass caps and power vias.  The cap right now is not in an ideal location, but it doesn't seem awful.  I figure we can improve placement on Rev B of the board, but I'm not sure that'll have any noticeable impact on the current issue

Hi RMC123 

Thank you for your response and for confirming these suggestions.

Based on your responses, the Low Noise Reference circuit you are working with is functional.  There is a high possibility that the 5V_ISO(having these AC ripples) is causing the Noisy Vref/2 .You mentioned that the TRACO module is sourcing the 5V_ISO; I recommend using another Power Source for 5V, which is clean and free from Noise (AC ripples). Then, you can re-validate your setup if the Vref/2 output is still noisy.  You are right, and I agree; this might help the problem, as I have suggested earlier, to verify if all bypass caps are placed closely in the supply pins. For the short term, you can place a through hole (TH) Cap component and solder one leg directly to the supply pins and connect the other leg to a near GND(via or test points).  

In summary, the priority is to validate the circuit using other Power Supply Sources for 5V_ISO (you might try bench tap instruments).

Regards,

Argy