Emontx v4.0.4 Bias Selection

I am designing a homebrew energy monitoring setup with 32 A/D inputs. Upon looking at emontx4, I really like the idea of using a ZMPT101B for voltage monitoring (I also plan to use an LS05-13B05R3 for powering the device).

Looking at the emontx4.0.4 schematic though, I am confused by the bias resistor divider.

Why use a 180k/33k divider instead of the standard midpoint of the 3.3v rail? I planned to use a REF5025 buffered by an LM358, but I was very confused why the emontx voltage bias seems to be ~0.5v instead of 1.65v.

Welcome, Jeff, to the OEM forum.

You are right - it is. It is because the input range of the ADC does not extend as far as the 3.3 V rail. The bias must be the mid-point of the ADC input range, not the midpoint of the power supply.

The choice of the ADC input range - and the reference voltage - also depends on the output voltage of the sensors. We’ve chosen to use 0.333 V output c.t’s because there’s a wide range of ratings available (the two “burden” resistors shown, R3 & R8, are a provision only, no resistors are fitted) so it makes sense to only just cover the range of voltages that are produced in order to maximise the resolution.

That’s the obvious part. A lot of discussion went into the whole analogue front end in order to maximise the accuracy of measurement.

Ahh, I did not realize you were using voltage output CTs. Thank you that clarifies everything.

Even if we don’t, the burden will be sized to give 0.333 V at rated current - the holes are there for a wire-ended resistor.

1 Like

I incorrectly assumed AREF was set to 3.3V / 2, thank you!

You don’t even mean that! - it was obvious that you had assumed AREF was set to VCC, which is the case with the emonTx & emonPi. The effect of that is to restrict everybody to a very limited range of c.t’s that can generate an adequate voltage to use the full resolution of the ADC in the Atmel '328P.

We reduced the input voltage needed so that the much more readily obtainable 0.333 V output c.t’s can be used.

1 Like