Cheap Charge Meter Design

[user212_nr]

I did some research on a hypothetical design for a cheap charge meter and billing system. Basically, charge-point costs $6,000 which seems like far too much.

This is not something I plan or want to do, just a hypothetical. Just because it could be build doesn't mean it should be.

The way it would work - you pair your phone with the charging station, and use an app to authorize the charge. The device uses an electromagnetic relay switch rated for 240V 30A to open an outlet or wire to a charger. The current is measured using the ACS712 current sensor for Arduino and sent over ethernet to the billing system.

So in theory (not 100% familiar with bluetooth specs), you walk up to the charging station that you've already authorize, then open the app and press "charge. The unit itself would be covered somewhere - no physical interaction.

Parts:
Arduino w/ Ethernet ($40)
Electormagnetic relay 30A (MGN2C-AC240) $50
HiLetgo 2pcs ACS712 30A ($7.19)
Bluetooth (DSD TECH HC-05) $8.49

Total $105.68 plus assembly, plastic cover, misc parts. Maybe sells for $200-$250, plus 5-10% commission to support billing infrastructure. Target would be apartment building owners, workplace owners, and workers/tenants looking to convince. They'd be able to make a profit off the electricity to pay for the installation and just to make a profit.

I'd have the skills to make it work, but I'd rather someone stole the idea. The business details and liabilities are not small, and I have my full time job which is enough. I suppose if I keep thinking about it, I'll have to make a prototype and a kickstarter.

Also, I'd posted something (not the same exactly) in "supercharger infrastructure" but seems no one there interested to respond. This is after all for the M3 primarily in mind.

 

[ewoodrick]

There is so much more to make a reliable solution. I don't see any of the hardware included, which is quite expensive.

 

[user212_nr]

There is so much more to make a reliable solution. I don't see any of the hardware included, which is quite expensive.

Its just a switch with a current measuring device, connected the internet and phone via bluetooth. What gets plugged into it just has to be under 30A - the rest of the wiring including charger/outlet/breaker would have to be done by an electrician.

So lets say that costs $700 to run a wire put a new 6-30 outlet on the device. Add $250 and that makes $950, better than 6k. If they are doing multiple charging stations, then their incremental cost is only $250 per-station and they can run it from a sub-panel located near the stations.

Imagine you are apartment owner and you want to install 10 stations. You put a sub-panel on the wall, 10 30A breakers and a 120V circuit for the units. $2,500 for the units, $1,000 for the wiring between the sub-panel, unit, and the outlet, and then fixed cost for the wiring to the sub-panel. That's less than 6k for 10 charging spaces vs. 6k for one space.

Obviously, they will eventually want to install the HPWC or some other unit, but they can do that upgrade 1 at a time as they get tenants w/ electric cars.

Its a good point - why not embed the hardware, but the goal is to make the unit as cheaply as possible, and that means taking advantage of existing hardware. The HPWC and other units are manufactured at scale, as are the parts listed above.

It also might make sense to make a 10-switch version.

 

[ewoodrick]

So what if power restarts? What it multiple people are plugging in at the same time and Bluetooth is being heard everywhere. What happens when the PI doesn't restart. How do you do the back end billing?
Go for it, but there's a reason why things cost a lot.

 

[user212_nr]

So what if power restarts? What it multiple people are plugging in at the same time and Bluetooth is being heard everywhere. What happens when the PI doesn't restart. How do you do the back end billing?
Go for it, but there's a reason why things cost a lot.

As to the bluetooth, that's why I was thinking it might need a 10x switch version to keep the signals down, but I think it should be fine to have even 10 devices near each other, just like the way we have 40 wifi networks in some areas. It only uses the bluetooth for the user interaction when they are near the car, after that goes over the internet via wire.

I thought about adding WiFi, but there are plenty of $20 devices that let you extend a WiFi network to ethernet and they have their own software written already for setup.

As to power failures, the device would update a backend every 1 minute or so, so when the power cuts, would complete the billing if it doesn't come back online. Restarting a session should be easy enough - notification to phone or auto-restart option.

As to payments, Paypal offers a feature that can split payments, so that a commission is collected to the software company (ex. eBay) and then the rest of the payment forwarded on to the "real seller" and goes right into their account.

I do iOS and Android apps and have done payments as well. It wouldn't take much time at all, but its disheartening when nobody buys your product/software/app due to lack of interest. That's where something like Kickstarter comes in, can make a prototype and then say "I'll only do this if pre-orders reach 2,000 units (500k goal)".

Definitely would need a LLC and lots of other things so its not really a get rich quick scheme.

 

[RandyS]

Or you can buy some Hialeah refurb utility meters for about $25 and wire one of those in front of each EVSE. Read the meter and bill accordingly. Just bear in mind that in an apt/condo scenario, that means the EVSE will probably be on a commercial rate (and may include demand charges).

Single Phase Watt Hour Meters

Or for a simpler solution yet...just charge a flat monthly rate for EVSE use and add it to the rent/HOA fees for each participating driver...

 

[user212_nr]

Or you can buy some Hialeah refurb utility meters for about $25 and wire one of those in front of each EVSE. Read the meter and bill accordingly. Just bear in mind that in an apt/condo scenario, that means the EVSE will probably be on a commercial rate (and may include demand charges).

Single Phase Watt Hour Meters

Or for a simpler solution yet...just charge a flat monthly rate for EVSE use and add it to the rent/HOA fees for each participating driver...

I'm not so much solving a personal problem as presenting a design for a device + system. The solutions you mention make sense vs paying 6,000$ for a chargepoint system, but vs. paying $250 for a unit for a billing system is not bad.

 

[user212_nr]

Not saying that its a good idea or that I would do it.

I ordered some parts for playing around with,

 

[Rottenapplr]

I just bought a dryer buddy for my car. Measures kwh. I’ll just track usage weekly as opposed to checking my stats app every morning to monitor my m kwh usage.

 

[user212_nr]

I just bought a dryer buddy for my car. Measures kwh. I’ll just track usage weekly as opposed to checking my stats app every morning to monitor my m kwh usage.

Please read the post before commenting.

 

[user212_nr]

I ran a test of the current measuring module on a tea kettle (https://www.amazon.com/gp/product/B07SPRL8DL/ref=ppx_yo_dt_b_asin_title_o03_s01?ie=UTF8&psc=1) and it gives some interesting results - The current measures between -16.181 and 15.811.

I'm guessing that this is due to alternating current and that the amount of power used by the kettle is a constant 16A.

Measured at 2.5 seconds intervals (from start)

Measured at 500ms, at end

 

[user212_nr]

Photos

 

[Missile Toad]

Another current sensor, FYI: Non-Invasive Current Sensor - 30A - SEN-11005 - SparkFun Electronics

 

[user212_nr]

Another current sensor, FYI: Non-Invasive Current Sensor - 30A - SEN-11005 - SparkFun Electronics

I'd need something that interfaces easily with the Arduino board. Not just w/ pins but also with software examples / pre-calibrated.

There is a nicer looking
https://www.amazon.com/Electronics-...=ACS758&qid=1570897902&sr=8-9#customerReviews

for $20, which is 6x more expensive. I paid $7 for 2 of the other 30A sensor.

Next thing to integrate is some kind or relay - I ordered this one:

https://www.amazon.com/gp/product/B07QS35SR9/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

 

[user212_nr]

Solid State Relay works. Now just needs to put together some demo.

 

[user212_nr]

Figured out how to calculate the current properly. Now its withing 0.1 of the kill-o-watt wall meter value.

I also realized that the bluetooth is useless. Both devices would have internet, so they can scan a QR code.

Of course actually mass producing this or doing a kickstarter is a terrible waste of time.

 

[earthwormjim]

That current sense IC has pretty huge error, that is awful for doing energy measurements. That error just compounds over time.
If you're going to charge based on usage, you need 0.1% to 0.5% error or less.

How are you measuring input voltage too? A rectifier then a resistor divider?

How are you calculating real power? Are you accounting for power factors less than 1? You will have a phase shift, and higher harmonic content in the current waveform than the fundamental in the voltage waveform. That content does not contribute to power, but does show up in the RMS value. If you're just doing magnitude current times magnitude voltage, that's apparent power, not real power. You can't charge based on current magnitude either, again because of the difference between apparent power and real power.

Here's an IC that does all the metering work for you:
STPM32 - STMicroelectronics

 

[user212_nr]

That current sense IC has pretty huge error, that is awful for doing energy measurements. That error just compounds over time.
If you're going to charge based on usage, you need 0.1% to 0.5% error or less.
STPM32 - STMicroelectronics

Does it compound over time or does it stay within +/- 1%? I've not done any power kWh calculations yet.

I was just looking at voltage, and found that the sensor I bought needs calibrating (w/ no instructions). I found online a meter from ekmmetering "120/240 Volt Pass-Through kWh Meter" and it just assumes 240v and does not do voltage sensing. I don't see a disclaimer on their manual that its not a "real meter", though they must be off. So I'm done with voltage sensing for this prototype.

The "120/240 Volt Pass-Through kWh Meter" is $90, and $60 if you buy 100+, and it has a pulse communication system that would work with Arduino to take their value. Its just that $10 vs. $60 is a lot when the goal is to make it cheaper.

The "bare minimum", "legally bulletproof" way is just to bill based on time, but that is the worst most inefficient way possible.

 

[earthwormjim]

Error should stay within the same relative percentage, but over time the absolute number can become quite large.

If you're looking for an economical solution, you need to buy an accurate all in one IC, like the ST one I linked. That can be easily integrated into an Arduino or Raspberry Pi, or STM32 hobby type kit.

The key point, is that a battery charge controller is not a resistive load like the hot water kettle you tested with. It's going to have power factors below 1, possibly closer to 0.9 when at 240Vac, and it's not necessarily going to cause a displacement in the current relative to the voltage, so you may not even be able to properly account for it with just using the displacement power factor. A charge controller is a complex (has inductance and capacitance) load, which will have a lot of higher harmonic content in the current. An FFT needs to be performed to break the current down into its harmonic content, so you can then properly compute real power. Meter ICs do this work for you, and send out nice values over the isolated digital communication bus.

If you're not accounting for power factor, and utility line regulation on the voltage, which can be +/-10%, you will have a pretty large error compounded on top of the current measuring error.

You can see about buying eval boards:
EVALSTPM32 - STMicroelectronics

Looks to be around $45.

There's other brands that have metering ICs too from the big players, Infineon, Microchip, On Semi, etc... I just work with ST microprocessors regularly, so I'm familiar with their part numbering.

 

[user212_nr]

Error should stay within the same relative percentage, but over time the absolute number can become quite large.

If you're looking for an economical solution, you need to buy an accurate all in one IC, like the ST one I linked. That can be easily integrated into an Arduino or Raspberry Pi, or STM32 hobby type kit.

The key point, is that a battery charge controller is not a resistive load like the hot water kettle you tested with. It's going to have power factors below 1, possibly closer to 0.9 when at 240Vac, and it's not necessarily going to cause a displacement in the current relative to the voltage, so you may not even be able to properly account for it with just using the displacement power factor. A charge controller is a complex (has inductance and capacitance) load, which will have a lot of higher harmonic content in the current. An FFT needs to be performed to break the current down into its harmonic content, so you can then properly compute real power. Meter ICs do this work for you, and send out nice values over the isolated digital communication bus.

If you're not accounting for power factor, and utility line regulation on the voltage, which can be +/-10%, you will have a pretty large error compounded on top of the current measuring error.

You can see about buying eval boards:
EVALSTPM32 - STMicroelectronics

Looks to be around $45.

There's other brands that have metering ICs too from the big players, Infineon, Microchip, On Semi, etc... I just work with ST microprocessors regularly, so I'm familiar with their part numbering.

I ordered this part earlier today, should come tomorrow.

https://www.amazon.com/gp/product/B00A4VF7U8/ref=ask_ql_qh_dp_hza

It outputs a digital signal 800 pules per kWh, which appears to be a standard, and is very easy to measure. Its in the $20-$30 range, so that works very well.

BTW, I can only get the SSRs to open if I connect ground to "-" and 5V to "+". It does the same thing with a 9v battery, requiring that I connect "-" to "+" to open and light up. Any idea why that is?