DC bias, single centre tapped transfomer

Tags: #<Tag:0x00007f6e19cb6ae0>

Hello. I am trying to build device like emontx but with ESP8266 so directly communicating via wifi. And I am interested in idea of just one transformer to provide power as well as voltage measurement source. Specifically in design with two diodes rectification and ground from centre tapped transformer as described on . But I am afraid that mentioned design is too much tied to specifics of atmega because

  1. suggested 120 uF capacitor is probably too small to safely hold as regulator source to handle any larger load (which ESP8266 can easily require)
  2. my bigger question is common to all emonTX design where AC input is biased to DC usually via resistor divider 470k + 470k backed by 10 uF capacitor. Unless that capacitor is fully charged there is not expected DC bias and attached AC voltage then goes nearly full halfwave into negative. Charging 10uF cap through 470k can take quite long. Then all that time are all ADC inputs exposed to negative voltages which is usually bad. Did you think about any solution preventing this? Or just relying on that atmega should be tolerant to -0.5V on inputs and probably can handle even deeper excursion (which under full load on sensor in the moment of powering on can probably happen) shortly as current is limited by relatively not so low impedance on input (which iself is for ADC also not good thing). Unfortunately I am considering using MCP3208 ADC which is not sub zero voltages tolerant and even with adding so callled negative tolerant single ended opamp (LM358) to provide some low pass filter and low impedancy ADC will be still exposed to short times of small negative voltages. Maybe it will survive, but maybe not :frowning:. Any sugestions? Thank you.

Welcome to the OEM forum, Petja.

If you have only one transformer, you are using it to measure the voltage and supply the current to operate the unit. The first question is, how much current can your transformer deliver without there being a dip in the output voltage. This is the reason we say that to add the ESP8266 module to our emonTx using the a.c. adapter from the shop, you must get the power from a separate 5 V USB power pack. Even with the current limited to about 30 mA peak while transmitting with the RFM69CW - which is a lot less than the 170 mA peak that the ESP needs, there’s about a 1% distortion in the peak of the supply voltage.

So - you will use the same configuration, but you must recalculate the values. You must get the transformer impedance from the data sheet, or measure it, and calculate the component values that will give you just enough current with the minimum distortion of the voltage wave. Note that in the emonTx V3, there is a series resistor to limit and spread out the current peak, and having a transformer that gives a higher voltage than absolutely necessary helps - because you can use more of each half-cycle to charge the capacitor. Then, if the distortion turns out to be unacceptably large, you know you can’t do it that way.

The easiest way to verify your design is to simulate the circuit in something like LTSpice.

You cannot have that configuration when you are using the transformer as the power source. Like my diagram shows, you must a.c. couple the voltage input in much the same way that you do with almost every audio amplifier.

It is not voltage that causes problems, it is the current that flows in the protection diodes. Ideally, you would add external protection that operates at a lower voltage to hold the input within the allowable limits. And of course you must limit the current in all the protection devices.

There have been many suggestions in the forums over the years.