The Axino-tech appliance power meterJuly 2015
This presents the design of an 230V appliance power meter. The meter not only reads out the real time rms voltage and current values, but has isolated sample outputs of both voltage and current for viewing on an oscilloscope or measurement via any sound card based system.
Overall block diagram
The unit has two independent, separately fused 230Vac inputs; one is for the instrument itself and the other is the monitored circuit. Separating the sources allows one to check an appliance behaviour through a variac, for example. The left-hand section is the derivation of voltage and current samples. The voltage sample is via a small 1.5VA encapsulated power transformer. The current sample is derived from a Honeywell hall current sensor. The isolated, low voltage samples are brought out to buffers and scalers to feed the panel sample points. The idea for the current monitor and for the buffering/scaling circuitry is from a Rod Elliott project. To avoid rewriting, read about this part of the meter at Project 139. My version is constructed slightly differently and has some minor component variations, but is, in essence the same. The unit has additional circuitry for the voltage monitor as well as for the voltage and current metering built into the instrument.
The voltage sample feeds an rms detector and then to Input 1 of a Lascar Panel Pilot TFT meter for display. The current sample from the Hall sensor has an independent scaler and processing section to that which feeds the panel monitor point. The current sample feeds an peak detector and a 50Hz low pass filter . A panel switch then selects either the peak current value, the total rms current value or the 50Hz component of current. The dc output corresponding to the chosen current metric then feeds input 2 of the Lascar meter.
Below is the schematic of the voltage and current sample board.
The upper part is a simple linear PSU providing +/-12V and the unregulated + output. This is about 16.5V and is used to power the Lascar meter and drive the relay.
The monitoring section shows the input AC first feeding via relay contacts then through the Hall sensor. The N/C relay contact feeds the appliance current via 1 turn through the Hall sensor, while the N/O contact feeds the Hall via 10 turns. This technique is as per Rod Elliott's design and the x10 range is effective for monitoring low power appliances. The relay is controlled from a front panel switch. The Hall sensor dc output corresponds to 38.7mV per amp when using +/-5V rails, and becomes 387mV per amp with the x10 range selected.
For a voltage sample, the 230Vac is resistively dropped to a standard mains transformer of 1.5VA rating. The AC input to the transformer primary is only some 5V and the secondary, when loaded with 77.3 ohms provides 230mV rms output. This secondary load was set empirically. In fact any value near 100 ohms will be fine since there is an op-amp buffer with gain control after this point anyway.
V and I monitor section
Below are the buffer sections that provide the panel samples out to RCA sockets.
The upper section is a simple op-amp buffer for the voltage sample with independent outputs for the panel socket and the rms detector. The lower part is the current monitor per Rod Elliott. The differences from the original are: Use of 5V 3-terminal regulators in place of zener diodes, removal of the dc coupled feature and hard wiring one low-pass filter, and just making the other switchable. In this form the current probe has a 3dB down frequency of 96kHz with 'filter off' and 8.8kHz with 'filter on'. The latter removes a reasonable amount of any hf hash present on the current waveform. I have retained the x10 'boost' feature, which is additional gain independent from the x1/x10 afforded by the relay switching of turns through the Hall sensor.
The probe response and phase was measured using the Praxis measurement system, using a high power audio amp as the source. The audio amp was fed via the appliance meter and then to a 16 ohm resistive load. Below is the response of the current monitor with various settings:
The source voltage was of course nothing like 230V, I used only about 20Vrms to obtain these plots. So, the dB scale has no absolute meaning, but you can see that the three gain traces are 20dB (x10) apart for the three conditions of x1,x10 and x10,x10 where the latter has both the 10 turns and the 'electronic' x10 gain selected. The filter was then switched in for the latter setting, showing the 3dB point. The response is shown here only to 24kHz, however it seems the Hall sensor is flat to much higher.
Similarly, here is the response for the voltage probe:
Here the response is set largely by the transformer. It is however, sensibly flat for the application. Usually mains voltage harmonics diminish quickly from the 5th (250Hz). Note the gain at 50Hz is -60dB corresponding to 1000:1. Phase shift shows about 3 degrees advance, again due to the transformer. This minor error could probably be corrected with a network around the op-amp but at this stage I do not make any use of phase information so have not attempted to zero it out.
Here we have a capture from the voltage and current probes with an incandescent lamp (resistive) load. The small phase error between the probe outputs is about 5 degrees.
In addition to the voltage and current probes on the front panel, there is a dual channel panel meter Lascar SGD-28M. This is programmed to scale and display the voltage and current values.
The voltage metering side is relatively straight-forward. It comprises an AD736 rms detector. For 230mV rms input, the chip provides 230mV dc out, which is fed directly to the Lascar meter input 1. One could argue that use of such a chip is overkill for measuring the voltage. Since the ac voltage is sinusoidal with typically less than 5% THD, a simple average voltage detector, scaled up by 1.11 to get rms would be sufficient. However, I had some AD736 chips to hand, so elected to use one anyway.
The current metering side is more problematical. The problems lie with the limited dynamic range of both the rms detector and the 50Hz low pass filter sections on the next diagram. To solve this issue I have a current sample scaling arrangement above, whereby if the x10 turns range is selected, the gain of the scaler is unity, and if the x1 turns is selected, the scaler gain is x10. Thus the output of this section is equivalent to 387mV per amp whichever turns ratio are in use at the Hall sensor. The tiny relay on this board is a DIP reed relay type.
The remaining metering sections are shown below:
Along the top of this schematic is the peak detector. The fullwave detector ensures that the highest of positive or negative peaks are measured. At bottom left is an implementation of a 50Hz low pass filter to enable the measurement of just the fundamental current value. The LTC1062 is a switched capacitor 5th order Butterworth filter with internal oscillator. The clock to cutoff frequency ratio is 100:1. With an eye to future upgrades I noted that the phase shift through this filter was close to 180 degrees when the cutoff frequency was set to be perilously close to 50Hz. The final value was 52.6Hz. This was of concern because I used the internal oscillator, however with judicious use of capacitor types, the clock oscillator has shown to be remarkably stable and hence the cutoff frequency of the filter does not waltz around noticeably. Loss at 150Hz is over 42dB, so the system will read the 50Hz component of current with better than 1% accuracy. With 180 degree phase shift, a following inverter brings the phase shift at 50Hz back to zero. At some future stage I might want to measure displacement power factor. The LTC1062 handles only about 1.5Vrms at it's input, hence the need for scaling. The gain of the whole section is restored to unity at the inverter stage.
Finally at bottom right is the rms detector for the current measurement. Another AD736, however this time used in it's low impedance mode because we need highest dynamic range. If I measure a 10 amp load, the voltage into the chip would be 3.87V, which is higher than recommended, but it is safe and measures correctly. Most of my appliance measurements are expected to be under 1 amp.
I have a 3 position switch for the current metering mode. Pos.1 is peak current and in this case, the output of the peak detector is fed direct to the panel meter. Pos.2 is total rms current. One pole of the selector feeds the total current into the rms detector, while another pole feeds the meter. Pos.3 is 50Hz rms current. One pole feeds the output of the 50Hz LPF to the rms detector, and the other pole feeds the Lascar meter.
The Lascar is a versatile dual input voltmeter and is supplied with a number of configuration options as well as provision for booting up with a logo on screen. Configurations are set on a windows pc then uploaded to the meter via USB. The scale factors can be configured, as can the display units. It was a relatively simple matter to scale the voltage reading and display whole volts. The current reading was also scaled so that 387mV read 1.00 amps. The final scale factors and overall calibration was by comparison of the readings with a Yokogawa WT210 AC power meter. This meter also confirmed the peak current readings and the 50Hz current readings are correct.
The Lascar SGD-24M panel pilot does lack the ability to display different units by external signalling. I would have liked an annunciator input, to allow a switch from peak, to rms total current and rms 50Hz current , but have used an external LED bar for that purpose. Further, it would have been useful if the Lascar could display a third parameter, being the power, by calculation. There is in fact a configuration mode that does this, however it is constrained to displaying power from the applied voltage and current through a resistor. In my case, the meter would need to make the calculation from the scaled values.
Being technically minded, I can mentally calculate power from the voltage and current readings easily enough, so not having a power readout is hardly limiting. In my case the V and I monitors will be of more use, but a future upgrade path includes the ability to acquire V-I phase and hence displacement power factor and I may also like to add logging to a SD card using an Arduino or similar.
There are no comments yet
Make a comment via the contact page