Hi,

I'm building a solar panel installation in Africa this summer and I'd like to be able to monitor its usage in the succeeding months.
The solar panel installation itself has been more or less designed. I'll need to measure 6 to 7 different signals.
I want to measure DCV (up to 150V) twice, DCI (up to 120A) twice, ACV (around 230V RMS) once, ACI (around 30A RMS) once and optional the AC frequency, but that's rather redundant.
This year, a team already tried, using Arduino Mega and dedicated circuits they designed on their own. However, the price turned out rather high, and the circuits were not too accurate (read 1%-5% deviation). Other small issues popped up as well.
I was thinking to use a Raspberry Pi Zero, to cut the costs, with a MCP3008 A/D converter, because its gsm modules are cheaper (the used connection to send the data). However, the circuits are too big costs for the low accuracy. Is there some kind of ready-to-go solution on the market (dedicated voltage and current meters), which would turn out relatively cheap? Most provide unnecessarily high accuracy for an unacceptable high price.
I'm aiming at 250-300 euros max for all circuits or multimeters together.

Thanks!

 

https://openenergymonitor.org/emon/

Maybe some info here you haven't found yet?

 

Thanks! That's definitely the kind of thing we're looking for. However, this is quite limited in sensors (no DC) and I believe it will turn out too expensive as well. I'm looking for a similar, but more low-level and more customizable solution. Any other suggestions?

 

I think the Raspberry Pi is more than you need, unless you want it to function as a free-standing computer with keyboard and monitor, and that requires various accessory cables. My suggestion would be an Arduino Nano or something similar, where you could visit the installation with a laptop and plug in a USB cable. (Which you didn't ask for, but I think it would be useful.)

If you're willing to tackle the hardware design, a quick suggestion is to have as many isolated subsystems as you need. One to measure high-voltage dc and another for high-voltage ac. Have a single 5V dc supply running off the ac output, with a battery for reserve use if the system shuts down, and run the processor from it (or main processor, if the subsystems have their own processors). Use isolating dc-dc converters to power the high-voltage subsystems and optically couple the data back to the processor. My instinct is against the single a-d converter because you've got various different and dangerous voltages to deal with, and bringing them down to usable levels just seems too difficult. I'd say digitize them in their own environments and then couple them back to the processor via isolating barriers.

 

You will need the added resolution, possible added accuracy and measurement repeatability of a 12+ bit ADC for accurate long-term data trends on raw panel and power usage data. Your 1%-5% deviation likely won't be solved by a different processor core.

You need good sensors and converters to make consistent and accurate measurements. The processor in the solar monitor could easily be a 5vdc 8/16-bit micro with 12-bit adcs (with a precision vref) that would more likely to survive an Africa summer vs the Raspberry Pi Zero with a SD card slot.

 

I think the Raspberry Pi is more than you need, unless you want it to function as a free-standing computer with keyboard and monitor, and that requires various accessory cables. My suggestion would be an Arduino Nano or something similar, where you could visit the installation with a laptop and plug in a USB cable. (Which you didn't ask for, but I think it would be useful.)

If you're willing to tackle the hardware design, a quick suggestion is to have as many isolated subsystems as you need. One to measure high-voltage dc and another for high-voltage ac. Have a single 5V dc supply running off the ac output, with a battery for reserve use if the system shuts down, and run the processor from it (or main processor, if the subsystems have their own processors). Use isolating dc-dc converters to power the high-voltage subsystems and optically couple the data back to the processor. My instinct is against the single a-d converter because you've got various different and dangerous voltages to deal with, and bringing them down to usable levels just seems too difficult. I'd say digitize them in their own environments and then couple them back to the processor via isolating barriers.

We picked the Raspberry Pi over any Arduino, because an arduino compatible sms-shield costs 70 euros, whereas a Raspberry Pi can connect over USB to a 10 euros gsm modem. A raspberry might be overkill, but the overall price tag is more appealing.
I don't completely understand what you're describing. You'd use different A/D converters for the high voltages and currents? Such as these: http://www.analog.com/en/products/a...cision-adc-10msps/isolated-ad-converters.html ?
Currently, I've got circuits to pull down the high voltages to the [0V,5V] range, however, these circuits are the unreliable and expensive part, so this might indeed be the way too go here. Thank you very much!
Do you know if there are similar solutions for current measuring or will we have to map the current on a voltage range?

 

You will need the added resolution, possible added accuracy and measurement repeatability of a 12+ bit ADC for accurate long-term data trends on raw panel and power usage data. Your 1%-5% deviation likely won't be solved by a different processor core.

You need good sensors and converters to make consistent and accurate measurements. The processor in the solar monitor could easily be a 5vdc 8/16-bit micro with 12-bit adcs (with a precision vref) that would more likely to survive an Africa summer vs the Raspberry Pi Zero with a SD card slot.

Hi! You suggest I should use PIC's? I was thinking about the Raspberry because it's very well documented and I haven't got any experience with PIC's really. Additionally, it's quite a big step (for me) to link a gsm module to a low-level PIC. Are the PIC's that much more robust to high temperature and humidity than the Raspberry? Thanks for the information!

 

Hi! You suggest I should use PIC's? I was thinking about the Raspberry because it's very well documented and I haven't got any experience with PIC's really. Additionally, it's quite a big step (for me) to link a gsm module to a low-level PIC. Are the PIC's that much more robust to high temperature and humidity than the Raspberry? Thanks for the information!

If you don't have PIC experience then it could be a challenge.

The Rasperry PI is an educational system that people have used for projects like this. The general tendency is for RPi projects to work great indoors but to fail when exposed to harsh environments because the sockets (USB and SDcard) and electrical specifications for I/O are not industrial quality. The Pi was not designed to be an industrial device with guaranteed reliability and guaranteed availability of identical hardware. It is an excellent educational product but you can't just put one in a box in the hot sun with little or no electrical protection and expect it to be a reliable product. You can make anything survive any environment as long as you package it right.

 

I admit that I didn't think about the phone interface. The first thing I'd look for would be a unit that interfaces via a UART or SPI. Is that possible? Maybe an Arduino shield isn't such a bad idea.

Yes, I did mean independent a-d converters in each subsystem. I'm not surprised that translating the voltages down to the main system level is a major hassle! PICs and Arduinos come with 10-bit conversion on board; I doubt that you need more precision, but you haven't said what you actually do need. Or possibly you could do without processors in the subsystems, and use SPI-connected a-d's alone, with the timing coming from the processor via isolators.