DIY charge point

[Glan gluaisne]

In this thread: External Garage - Charging

I mentioned that I was building a new DIY charge point, using off-the-shelf parts. Rather than take that thread off topic, I thought I'd start a new one, although I fully accept that not many are likely to be that interested in tinkering like this.

In that other thread @LukeUK wrote this:

I've found a couple of open source projects. Have you seen these, or any others?

OpenEVSE - Electric Vehicle Charging Solutions
Smart EVSE | Open Source EV charging

I know there's a difference between tinkering with your own software, rather than someone else's...

Luke

and @dakaix wrote this:

Be interested to know what you're using for DC protection. I built my own OpenEVSE charger (in a QUBEV box) during the summer but haven't installed it yet as I haven't sorted DC protection out, all the spark's I've spoken to so far were only interested in installing MATT:E units which I don't have space for.

I've opted to try out the new Viridian EVSE module: EVSE Protocol Controller 2.0 (EPC 2.0)

This comes with open PEN detection for £124.80 (inc VAT), and for another £50 plus VAT also includes the same DC leakage protection device I'm already using with my home brew charge points, the Western Automation RCM14, so has the required DC leakage current protection, so only needs a suitably rated Type A RCD or RCBO upstream, and doesn't need an earth electrode.

Other parts needed are a couple of 32 A contactors (I'm using a couple of Iskra ones I happen to already have as spares, but they are only about £12 each, inc VAT (see here: https://www.tme.eu/en/details/ika232-20_230v/contactors-main-modules/iskra/30-046-833/), plus some DIN rail contact blocks, some short lengths of 6mm² wire for internal connections and ideally some cable ferrules and crimp tool to make the wiring neater. The whole thing would easily fit in something like an empty garage CU box.

Because I need E7 off peak switching, I'm adding a DIN rail time switch to control the Viridian EVSE, plus I'm also adding variable current settings, plus the option to turn the charge point off, or run it only during the off-peak E7 period. Adding that extra functionality adds maybe £20 to the cost.

BTW, my reason for doing this is partly fun, and partly to remove my homebrew PIC based charge points with code I wrote years ago. They work fine, but use GPS modules to control the time switch function and rely on code I wrote years ago, and which I've modified a few times over the years. As I get older I'm trying to replace my homebrew controls with off-the-shelf stuff, that's properly documented, so that others have at least some chance of maintaining or repairing it in future.

 

[Glan gluaisne]

Just finished doing most of the wiring internally, just need to add the time switch and current select control, plus, possibly, an external LED indicator:

For anyone not needing time controlled charging, the DIN rail time switch could be left out, As this stands it would work fine as a 32 A fixed current charge point, just needs a supply to the DIN terminals on the left and a tethered lead connected to the contactor terminals on the right, with the Control Pilot connected to the bottom of the Viridian EVSE.

This unit is fully compliant with the regs, and only requires connecting to a supply with a normal Type A RCD/RCBO, with no requirement for an earth electrode etc, as it has built in open PEN protection.

Just need to wire up a weatherproof rotary panel switch to select operating mode (off - on - E7 timed) and variable charge current (handy to be able to turn the charge down sometimes). With a couple of cable glands on the base of the box it should be ready to install.

 

[Glan gluaisne]

Bench testing complete, using my charge point test simulator box, plus my multitester. All works exactly as it should. I managed to test the DC earth leakage protection, by deliberately injecting a DC earth leak, and that worked fine. Only thing I haven't yet tested is the open PEN protection, but the unit seems to go through a self-test on power on, so I assume it's able to do a health check on the open PEN detection system.

Ignoring the time switch and variable current switching options, for someone using one of these to make a simple 16 A or 32 A tethered charge point, with integral DC leakage protection and open PEN protection, then the costs, assuming a 10m cable run from the charge point to the incoming supply, would be:

IP rated enclosure (I used an IP66 box because I had one, but any weatherproof small CU would do) ~£25
2 off 32 A DP contactors, 230 VAC coils ~£25
Viridian EPC 2.0 EVSE, with optional RCM14 DC sensing ~£185
DIN rail terminal blocks (I used 16mm² ones as I had them already) ~£5
Cable glands (assuming SWA for the inlet cable, nylon for the tethered lead) ~£4
Tethered Type 2 cable assembly ~£90

The grand total for a 32 A tethered charge point, with a 5m cable, DC earth leakage and open PEN detection/protection, comes to ~£334, inc. VAT.

To install this, assuming a 10m SWA cable run from the incoming supply, would need the following:

10m of 6mm² SWA ~£25
SWA cable gland ~£3
4 way metal enclosure, including isolator switch ~£18
40 A Type A RCBO to go in above enclosure at supply end ~ £25
Henley blocks ~£22
1m set of tails ~£10
cable cleats etc ~£5

So, a total materials cost for installation of ~£108

Accepting this needs a degree of competence in electrical work, and installing it would be notifiable work under Part P in England and Wales, this still seems to be pretty cheap for a reasonable quality non-smart charge point that complies with the current regs, has both DC earth leakage and open PEN detection/protection, and doesn't need an earth electrode or Type B RCD.

 

[davidmc]

Very neat

Could you do one with a Type 2 socket instead of a tethered cable? And how much would that cost?

 

[Glan gluaisne]

Very neat

Could you do one with a Type 2 socket instead of a tethered cable? And how much would that cost?

Yes, easy enough to do, just swap out the cost of the 5m tethered Type 2 cable and replace it with a pre-wired Type 2 outlet, which costs £96 inc VAT, so overall £6 more expensive. The enclosure would need a large enough flat area on the front to take the Type 2 outlet, but one of these, intended for use as a caravan hook up, for £20, should be OK: IP65 MCB, RCD Enclosure,Caravan Hook Up, consumer unit enclosure,adaptable box 660042416391 | eBay

 

[Glan gluaisne]

Time switch and IP65 rotary switch all connected and working. The rotary switch has six positions, OFF, 8 A, 10 A, 12 A, 32 A and E7 32 A. The reason for the three low current settings is really to roughly match solar generation.

On my original home brew charge point I used to use data from the house monitoring system to control the charge current, using an algorithm that selected a charge current that allowed the car to charge at the highest available rate when generating more electricity than the house used, whilst keeping the charge cost down below the off-peak rate. Although it worked, because of the inherent delay between the Control Pilot duty cycle changing (the signal that tells the charger how much current is available) and the actual change in charge current, I found that it was just as effective to just set a fixed low current when solar charging. At the moment, I have that fixed low current set to 10 A, which works OK, but ideally I'd like to be able to decrease that slightly for better utilisation in Spring and Autumn, and increase it slightly for use in mid-Summer, hence the additional 8 A and 12 A settings. We can export around 5.5 kW to 6 kW at times, but days when generation is that high aren't uncommon, as in really sunny weather we'll have the heat pump on cooling the house, so export tends to be around 3 kW or so, a reasonably good match to the 12 A setting.

I didn't see any point in making the off-peak E7 rate anything less than the maximum, as I've not yet had any need to reduce the charge rate when charging overnight. The time switch will be set to GMT all year around, as the E7 switching times are always GMT. Hopefully the clock in the time switch will be stable enough to not need setting too often. It has a battery back up, so retains the settings over power outages of up to a few days.

Just need to decide whether to add an external RGB LED indicator, or just fit a sealed window in the lid of the box so the RGB LED on top of the Viridian EPC can be seen from outside.

One useful option might be to use this Type 2 tethered lead: Unique Tesla charging cables - evChargeking in place of the cheaper one mentioned earlier, as it includes a Tesla button and transmitter, for both opening the charge port and unlatching the connector at the end of the charge. A bit more expensive, but perhaps worth it for the added convenience.

With luck I may get it installed soon, so can test it for real. If anyone wants links to sources for all the parts, yell, and I'll add them to this thread.

 

[LukeUK]

Your link to the EPC 2.0 has got me carried away thinking about putting something together.

My PV is only 3.5 kWp, so charging at the minimum 6A on a cloudy day is not efficient at all. I fancy trying to use a weather data API to look at the day's forecast weather and then decide whether it is worth trying to go for some PV, or whether to just leave the PV to top up the hot water via the immersion heater, and leave the car to charge during the Octopus Go window overnight.

The same data could be used to decide what to do with the Go window on a Friday and Sunday morning, based on the weather during the day. I could get carried away and try to access my calendar to determine if I'm going to be at home to use the PV or not.

The other function that would be useful, based on quite a lot of driving so far, would be an evening trickle charge function. I often arrive home with ~10%, and an unhappy car that will be waiting until 00:30 to start charging in the Go window. It would be nice to use the Tesla API to look at the SOC and start charging at 6A to stop the battery going too low, whilst waiting for 00:30. I think this will become more relevant in the colder months as well?

 

[MrBadger]

Unfortunately neither the app nor the API give any way of setting the charge rate. Hopefully it will come one day although having seen recent information mentioning (cannot locate the article right now) a reliable battery source that charging at lower rates is a bad thing for the battery has changed my view on the overall usefulness of setting charge rate below 10A. A number of projects have acted upon this information and set a lower 10A limit on their charging. What impact differences between US/UK voltages has on this I do not know.

 

[tsh2]

I rarely generate more than 3kw (needs to be clear and cool, I think the panels are rather temperature sensitive), and find there are few days when it is sufficiently clear to even charge at 6A. There are days when I can push to 7 or 8, but even a bit of cloud makes a difference.

I found that the 'cloudy' indication on HomeAssistant was quite pessimistic, so at the moment I have automation to switch on if it is not cloudy and generation is high, and to switch off if generation drops too low - so I need to manually start charging - but I don't often need a full charge...

 

[LukeUK]

Unfortunately neither the app nor the API give any way of setting the charge rate.

But the EPC 2.0 does, and the charge rate PWM to the car can be controlled by an analogue voltage that I could adjust via a uC.

The EPC saves the hassle of writing the PWM myself, and getting it accurate enough. I was at the point of it being too much hassle to develop and test. (For me, and the time I have. I know it's not that much hassle/time, if you've got time available).

Also a time saver in that it handles the PEN side of things as well.

 

[Glan gluaisne]

The way I used to control charge rate with the previous version of my homebrew charge point, was to take the average export power over the previous 30 seconds (initially, later changed to 5 minutes), compare that to the set charge power, and if the charge power needed to be adjusted up or down make a change. When the charge current dropped to 6 A, after export being below 6 A for five minutes the charge point turned off (went to State A, a steady +12V on the Control Pilot).

This sort of worked, but most of the time when we were exporting power the charge current tended to sit at around 10 A in practice, and as I like to try and keep things simple if possible, I just ripped the telemetry stuff out and added a switch to select 10 A or 32 A. In practice this works OK, except for a few times when I could either have charged at a bit higher current, or would like to have charged at a bit lower current. 6 A is a bit too low, IMHO, both because the car charger is horribly inefficient at that low power, plus by the time export has dropped down this low it will be fluctuating a lot up and down, anyway.

The Viridian EPC 2.0 has an RS485 control capability, although that's not documented in the (rather sparse) data sheet. That may allow a fair bit of additional smart functionality to be added, so if someone wanted to produce a custom control system it may well be fairly easy to do using that. Could be quite cheap and easy, too, something like an ESP32 would have more than enough processing power, the ability to interface with a WiFi network and more than enough spare IO lines for all manner of sensors or controls to be added. Be nice if Viridian published the interface spec for this.

Control via the IC connection is pretty simple. If switched to 0V it turns the unit off gracefully, and switching different resistors between IC and 0V, or applying a control voltage to IC, allows the charge current to be changed. This seems to work well, as it's how I'm varying the charge current and using the time switch to control the charging time.

 

[LukeUK]

The 'smart' charger that I have only uses PV export. I think it would be far more useful to manage the charge rate to maintain zero import at the grid connection.

It's all well and good me trying to charge for free either side of noon on a sunny day, until my good lady wife decides to put the oven on. The PV can't supply both and the car is happy charging in the middle of the night, whereas I don't think I'll convince my wife that 00:30 is a good time to start cooking.

 

[LukeUK]

The Viridian EPC 2.0 has an RS485 control capability, although that's not documented in the (rather sparse) data sheet.

Do you have any pointers at all?

 

[Glan gluaisne]

Do you have any pointers at all?

I'm afraid not. I've used the contact section on their website to ask for more info, but not had a reply. Neither the datasheet or the manual ( https://viridianev.co.uk/s/Viridian-EV-EPC-20-Datasheet-Iss-11-tak3.pdf https://viridianev.co.uk/s/EVSE-Protocol-Controller-20-EPC-20-Manual-Iss-12.pdf ) are helpful, they just mention that the RS485 connection is there on the plus model, but that's it. I'm assuming they included that connectivity to allow this module to form the basis of a smart charge point, by allowing fairly straightforward control of charge time and current. I can't see any other reason for including it, TBH, although RS485 seems an odd protocol to choose, as that's more normally used for long distance data links. Something like I²C would seem more suited to a short link between this module and some sort of smart control board.

Might be an idea to ring them and ask about the RS485 functionality. Perhaps they might respond more quickly to a phone call, or it may be that they are short staffed, perhaps with people working from home at the moment, so unable to deal with enquiries. I'd also asked them about the connector type used for the external LED, but I've popped the lid and its a 4 way JST XH connector, 2.5mm pitch. If using the external LED then a hole needs to be drilled in the lid of the box for the cable, which seems a bit messy. The internal LED is very bright, though, so I'm going to try making a light pipe to direct that to an external lens of some sort, perhaps a cut down indicator light bezel. If fitting the unit into a weatherproof CU type box, perhaps a caravan hookup type box, then the light should be visible through the clear panel. The box I have is a heavy duty glass reinforced one, with no clear panel, hence the thoughts of playing around with some sort of light pipe.

 

[LukeUK]

It's quite easy to buy clear acrylic rod, a bit of polishing both ends should do the trick?

 

[Glan gluaisne]

It's quite easy to buy clear acrylic rod, a bit of polishing both ends should do the trick?

Yes, that's the plan, with the bit of perspex poked into the back of this indicator lens, that has a small O ring to seal it to the panel:

 

[bitmanEV]

And this was done in 2013...
EVSE Build Day

 

[Glan gluaisne]

And this was done in 2013...

Yes, saw that probably around that time, back when Kevin Sharpe was a regular on the SpeakEV forum, as it was 2013 when I built my first charge point. I did look at using the original Viridian/Mainpine EPC, as shown in that video, back then, but instead opted to build a charge point from scratch, using a PIC to generate the CP PWM, monitor the CP voltage etc, with an op amp driving the CP signal. I used the same Western Automation DC sensitive leakage sensor that Viridian are now using with version 2.0 of their EPC, to provide the charge point with the required DC tolerant leakage protection, and connected it as TT, to provide open PEN protection.

The original Viridian EPC was OK, several manufacturers, including Rolec, used it in their charge points, but it didn't include any form of protection, and relied on the installation being connected as TT, with an earth electrode and expensive Type B RCD. The big difference with version 2.0 of the Viridian EPC is that they've incorporated both open PEN protection (so no need to connect the installation up as TT with an earth electrode) and they've also included provision to use the Western Automation RCM14 module for DC leakage sensing. This makes it straightforward to build a charge point that is fully compliant with the regs (and the one in that video is not compliant, it requires an earth electrode together with a Type B RCD, neither of which was fitted as they dodged the issue by making it portable).

 

[pdk42]

Very interesting. Thanks for documenting. I think if I were starting again I'd seriously consider building my own after the agro I had with off-the-shelf units and dodgy installers.

 

[bitmanEV]

This is my portable box build years ago