AMC7834: Should the ADC be stopped from autonomous conversion mode before reading conversion result?

Hi John,

What are they using for the AVDD supply? A low noise LDO? or some kind of switching supply that can be introducing ripple noise?

I have seen very noise supplies result in noise local temperature measurements, especially if they are switching a frequency similar to (or a harmonic of) the LTS sample rate. 

Thanks,

Paul

Hi Paul,

Thank you for reply.

AVDD, AVCC, and DVDD are all powered from a 5VOUT LDO, LT1764. They also tried to power AMC7834 with a separate 5V desktop power supply (to isolate the digital noise). There is not any improvement. So it seems it is not the result of power supply noise. Is there any other potential causes?

Best Regards,

John

Is this measured on the EVM? Or their custom hardware? Is there some kind of oscillation or external current load? Is the device hot? Can they share the schematic?

Are they seeing this on just one board? All boards?  Could there be some grounding issue with their assembly? 

Hi Paul,

It is measured on customer board. No oscillation, there is a RF PA on board while the PA is not in operation mode. The device is not hot. Attached pls find their schematic.

They only have one prototype in hand. It seems not a grounding issue. It is a 8 layers board with a complete ground plane.

Thank you!AMC7834_SCH.pdf

John

Hi John,

I suppose I should clarify that I am suspecting an assembly issue in regards to the grounding.  Does the remote temperature sensor see similar noise?

If all other ADC input channels are disabled except the local temperature sensor, is it improved? What level of noise are you seeing (how many LSBs)? 

Hi Paul,

What do you mean about assembly issue?

It seems the ADC itself is correct. They can read correct value from ADC2 if a voltage is applied on ADC2 input. Readout is 0.798V to 0.805V if 0.8V is applied, readout is 1.897v to 1.909v if 1.9V is applied. While the LT readout fluctuated dramatically, such as 5 consecutive reading is: 0x85, 0x81, 0x66, 0x97, 0x4d. While if 0.8V is applied on ADC2, the max reading is 0x527, the min reading is 0x51B. 

Thank you!

John

So my concern is that if the ground pins (or even a single ground pin) has a bad solder joint, then there might be some kind of current-dependent offset voltage on a localized ground point in the device.  The temperature sensor uses of a series of different excitation currents and measure differential currents.  So some assembly issues can cause unexpected behavior.

I see you are listing just 8 bits. Is that the lower 8-bits? or upper?

The remote temperatures sensors have similar noise?

Can you look at ripple on the DVDD after the ferrite bead, B8? Does removing B8 with a short help?

Hi Paul,

It is lower 8-bits. The upper 8-bits are all '0'.

The fluctuation in remote temperature sensor reading is much smaller than LT. The fluctuation in LT reading is more than 100 LSBs, the RT is around 20 LSBs, while the real temperature is stable.

Regards,

John

Hi Paul,

They have tried to short B8 ferrite bead, no effect. The ripple on DVDD is 59mVpp.

Thank you.

John

Is there a tone in the ripple? Maybe a switching frequency?

No significant tone. Here is the scope waveform of DVDD ripple.

Looks like there might be tone there, but you would need to look at it from a spectrum or change time domain around until you see a pattern. 

Otherwise, it is difficult to say why you are having this noise.  The next step might be to try a different AMC7834 or try another board.  It is important to determine is this is a systemic issue with the design, a one-off part issue, or something else.

Hi Paul,

Yes, this customer is not familiar with testing of power supply ripple. There should be some pattern due to SMPS ripple. Anyway I think 60mVpk-pk of pattern or ripple has been low enough. It is typical for a system. I am afraid it is not the root cause. Is there any other possibility? such as incorrect configuration or operation mode of device?

Thank you!

John

Hi John,

I am not aware of any other issue that could cause this.  To clarify, this device might have worse local temperature sensor performance if the switching frequency is around 175kHz or some harmonic of that.  I think it is worth the customer to confirm the switching frequency.