low power supply for energy meter

Hi,

 The reason that we limit the minimum input DC voltage in the design is as following: If the input ac voltage is too low, depending on the bus capacitor value used in the design and after taking into account the tolerance of the capacitor (normally at least 20%), the minimum dc voltage appears in the DC bus will be very low. To make the IC working, there is a internal regulator that powered from voltage between drain and sourse. If this voltage is too low, the internal regulator may not work correctly. We recommanded the voltage between drain and source to be minimum 50Vdc.

Coming back to your question, if you design the power supply at 30Vac minimum votlage, but  I would highly recommend you check if the BP voltage is correct especially during startup.

Hope this is helpful.

Dear Terry,

Thanks for sharing your views on the topic. I Apreciate it.

For the PSU design, currently I am also considering using any other device to get the 30 to 130V AC 50Hz operating range.

I was also looking for devices with constant PWM frequency operation from PI.  Any suggestions?

Vinod Tatti

Hi,

Would you mind if I ask what the reason of using a device with constant PWM frequency? The way that our IC works is variable switching frequency, this helps to improve the overrall efficiency of the power supply. Unfortunately we do not have any IC available with fixed PWM frequency.

Dear Terry,

Energy meter is very cost sensitive product. Hence one design catering to Whole current, LTCT and HTCT meters ishighly desirable. Our design was also intended to be like wise. However input supply voltage at 30V AC 50 Hz would be too low for the LNK304 to operate. To our surprise we found it is functioning even at that voltage. However the effective switching frequency goes up as can be expected. We found that the frequency reaches close to 4.5kHz, which is otherwise around 1.9kHz with 70V AC input. We also found that the out put is well regulated close to 1%.

The problem faced was that we had the switching frequency affect the metrology function of the meter.  Inspite of various filtering and snubber components selection, the error was going out of the acceptable range for the entire input voltage range of 30 to 130V AC 50 Hz.

Further investigation pointed that the Voltage measurement was erroneous.On the other hand, current measurement was almost error free. We found that current measurement was using differential signal for ADC whereas, voltage was measured using single ended ADC measurement. This lead us to investigate the voltage sensing circuit components and tracks layout. Once the tracks for the voltage sensing was moved away from the existing location we could see that the error came  down drastically. Almost unbelieveable!!.

Dear Terry,

the reason for considering fixed frequency PWM controller was to ensure that the frequency noise interference  is taken care of by a well designed filter for a known (also tunable) frequency of switching. This would make the meter immune to the switching  of the PSU for the entire range of operating volatge range.

This would mean that we have to redesign the PCB.

We have done some tests with the existing LNK 304 based PSU and by re-positioning the voltage sensing  circuit components of the meter with encouraging results. I hope we may not have to design a new board for the HTCT meter.

Some times it is nice to have discussion with some like minded  experts in the domain to ascertain our understanding of the technology and the underlying principles of physics.

Thanks for the support. I will get back once the mission is accomplished in entirity.

Thanks and regards,

Vinod Tatti

G.M. Technology and Projects,

Matrix Energy Pvt Ltd.

Bangalore, India.

Hi all,

 The problem of power supply noise pick up and variation of the errors over the operating voltage range was sorted out finally. Components used for the voltage measurement (resistive divider chain, ferrite bead and bypass capacitor etc) were mounted on a separate pcb and wired on to the main meter board. this gave me surprisingly good results. hence noise was the main cause of excessive error.  Variation of the positioning of the additional pcb and the connected wiring did not give me the required pin pointing of the noise pick up point.

A separate inductive pick up probe was prepared and used for measuring the noise level on two pcbs for comparison. One board was giving errors within limit, as outlined above and another board without any changes. the levels were the same.

This pointed that the board with excessive error had something different than the one with least errors. On comparison, it was discovered that the measured voltage was tapped from different ends of the surge protection resistor at the input stage. This was corrected on the PCB without any corrections or additional voltage divider board. The tests showed that the errors had dramatically come down and the problem was solved. Since the surge protection resistor would carry the pulsed currents of the switching PSU, the voltage drop across it would get subtracted from the measured voltage leading to errors in the measured voltage.

Once the repositioning of the point of measuring the applied voltage was done, the problem got sorted out, without requiring any pcb redesign or power supply redesign.

Thanks and regards to all who have seen this thread and found it useful.

Vinod Tatti