Hi,

First post and it feels a bit of a daft question, struggling to find resources or answers though so here goes.

I have been looking at the Atmel M90E32AS Energy metering IC

The Application note hardware design shows the voltage divider network and the Neutral connected to ground

My first thought was this is just in reference to the Divider network - not actually connected to the PCB Ground plane.

However, on the main IC, V1N -V3N are also connected to ground via a 1k resistor with DGND also connected to the same point.

Clearly - I think a Neutral, or at least a 0v reference is needed for the ADC channels to function.

In an isolated class 2 environment with no actual ground connection and a fully insulated enclosure for the meter with the 3v3 being derived internally from one of the phases & neutral this may not present an issue.

If we have an external power source say POE or 24vDC for the controls and then external comms say Modbus RTU this ground connection is going to be distributed around via comms network or in terms of a 24vDC power supply the Ground may be part of a PELV supply - essentially at some point the internal Ground may get earthed - which is connected to Neutral at the Voltage divider network.

Application note attached

Spiralling brain now - I fear I may be over thinking this!

Thanks for reading!

 

Connecting a chassis to ground ( earth GND) through a resistor is done fairly frequently helps reduce interference noise and acts as a safety ground.
Normally a central-single point is chosen to connect any and all earth ground conductors to one point.

 

Hi,

First post and it feels a bit of a daft question, struggling to find resources or answers though so here goes.

I have been looking at the Atmel M90E32AS Energy metering IC

The Application note hardware design shows the voltage divider network and the Neutral connected to ground

My first thought was this is just in reference to the Divider network - not actually connected to the PCB Ground plane.

However, on the main IC, V1N -V3N are also connected to ground via a 1k resistor with DGND also connected to the same point.

Clearly - I think a Neutral, or at least a 0v reference is needed for the ADC channels to function.

In an isolated class 2 environment with no actual ground connection and a fully insulated enclosure for the meter with the 3v3 being derived internally from one of the phases & neutral this may not present an issue.

If we have an external power source say POE or 24vDC for the controls and then external comms say Modbus RTU this ground connection is going to be distributed around via comms network or in terms of a 24vDC power supply the Ground may be part of a PELV supply - essentially at some point the internal Ground may get earthed - which is connected to Neutral at the Voltage divider network.

Application note attached

Spiralling brain now - I fear I may be over thinking this!

Thanks for reading!d

On what page is the neutral shown connected to ground? If it is, that's the PCB ground, not earth ground. You're right, the circuit 'ground' should not be connected to the earth ground other than the neutral/earth ground tie at the main panel.

The solution is easy (in theory but not practical application), don't do it (connect neutral to ground) and check to make sure any external power supply or device doesn't do it.

 

You're right - it is non-isolated. If your 3.3V is reference to earth, you have either to isolate all the voltage and current inputs, or isolate the SPI inputs and outputs.
Another thing that they hide in the datasheet is that it is non-calibrated. You need a calibrated source to compare it with, then have to load lots of coefficients you have calculated into its registers.
I had a conversation about this type of device with Analog Devices. When I told the application engineer I was going to use the Atmel device because it was cheaper, his response was "You'll be back after you've read the datasheet".

 

Thanks for the replies really appreciate your time!

The UN > Ground link is on the Hardware design pages 4 & 6 for both CT & Rogowski

You're right - it is non-isolated. If your 3.3V is reference to earth, you have either to isolate all the voltage and current inputs, or isolate the SPI inputs and outputs.

Could get very messy trying to verify all external supplies for the Modbus network or TCP network are earth free / all insulated - I think not a viable solution.

Another thing that they hide in the datasheet is that it is non-calibrated. You need a calibrated source to compare it with, then have to load lots of coefficients you have calculated into its registers.

I looked at that and thought ‘how hard can it be?’ Figured it was a problem for another day, maybe with the grounding issue it might need to be resolved with a different IC!

I had a conversation about this type of device with Analog Devices. When I told the application engineer I was going to use the Atmel device because it was cheaper, his response was "You'll be back after you've read the datasheet".

Ah yes! The Analogue Devices ADE78xx series, same grounding issue if I recall correctly but yes, price is a real shocker in comparison.

Also looked at the Renergy 7302 but the documentation or lack of made that too

ST also have an offering but seems to be multi IC for 3 phase!

 

I think some of the AD parts have isolated SPI interfaces.
My finished design used a quad op-amp and a microprocessor all tied to mains neutral, with the output through and isolated CAN interface.

 

I’m pretty new or more accurately newly back to the electronics after a 25 year break, though I’m fairly sure I have forgotten virtually everything I knew, EDA’s have certainly moved on!

Is your project open-source or Commercial?

Did you find any other poly phase IC’s?

 

Sometimes were looking at small trace length on a pcb and forget about the issues of a long trace that can
be a coax or a transmission line. He shows the waveform he is in affect simulating. When you do that in a simulation
you can derive interesting numbers which he shows using a paper and pencil. If you understand the video
and you can simulate then some programming to tidy up the output.

Was this on the right thread?