DIY Home EVSE

[W.Petefish]

Traced some of my GFCI troubles possibly to my Roadster. (noticed strange activity on the HPC during charging. E-mailed Tesla and awaiting a reply.) I am going to test that theory though with a J1772 to 14-50 box that I have.

 

[Chris1howell]

I have decided to open source the EVSE I am working on. DIY EVSE goes open source...

http://code.google.com/p/open-evse/

I just started the site, there is much more to come...

Chris

 

[Kevin Sharpe]

I have decided to open source the EVSE I am working on.

many thanks Chris... this is an important step towards low cost Charging Stations everywhere :smile:

 

[mpt]

Add my thanks too; these projects are how we get to the $100 EVSE at Home Depot.

 

[W.Petefish]

Question to ya'll: Should I go ahead with publishing an instructable on how to make one w/o the GFCI (or as I call it, the GRRRRFCI)?

 

[Tommy]

Question to ya'll: Should I go ahead with publishing an instructable on how to make one w/o the GFCI (or as I call it, the GRRRRFCI)?

IMO, a DYI EVSE must mimic electrical requirements of commercial version of EVSE. In other words, if would be a mistake to "trick" the vehicle to charge with DYI EVSE without including all the safeguards a commercial EVSE provides. One of the requirements is to have circuitry that continuously monitors safety of ground, I don't know how one can do that without a GFCI as part of the circuit.

 

[hcsharp]

Question to ya'll: Should I go ahead with publishing an instructable on how to make one w/o the GFCI (or as I call it, the GRRRRFCI)?

I would suggest not. Get it working and test it first. You may discover the problem with the GFCI is related to something else even though it works fine without the GFCI. Just my .02

 

[TEG]

For what it is worth, I think even the big manufacturers (EVSE/cars/etc.) have struggled with getting the GFCI to work unobtrusively.

 

[efxjim]

One could always wire your circuit into a heavy duty GFCI breaker to provide protection. Please proceed without GFCI as a first step. I would like to purchase bare/populated boards and programmed chips as well.

 

[W.Petefish]

What I fear is that the GFCI problem is related to the roadster from my HPC's GFCI tripping. (or at least acting like it is tripping):cursing: By that logic, do ya'll think that I should just write the instructable while it is unhooked and hook it up for some photos? Or should I specify a GFCI breaker in the parts list instead of the 8895 that I have now?

I don't think it is the GFCI itself that has a problem. I think the charging circuit in the Roadster has, or at least had, an intermittent ground fault that has been tripping GFCIs. The other day I had a ground fault happen at a ChargePoint station.

 

[Chris1howell]

GFCI for an EVSE is a bit tricky. Most GFCI units trip at 5ma leakage current, EVs need a higher trip point. 20ma works very well and is used by most EVSEs. In my prototype EVSE I also added a filter to avoid trips from high frequancy line noise.

I plan to add the GFCI details info to the Open EVSE site soon. The site is growing quickly,currently there is a basic schematic, board file, several wikis and the source code was just posted.

 

[W.Petefish]

If you get a GFCI designed that'll handle the leakage current we could compare notes. I haven't yet been able to solve this problem and taking apart one of the 8895s didn't yield any workable solutions.

 

[Chris1howell]

I have posted the basic experimental GFCI circuit I have been using with the CRMagnetics 8420-1000-G CT coil. http://code.google.com/p/open-evse/downloads/list

To change the trip point you can adjust R17, 330 Ohms sets a 20ma trip with the 8420-1000-G CT. C8 is to filter high frequency noise and may need to be adjusted if the circuit is not tripping quickly enough.

Chris

 

[hcsharp]

Chris this is the best design and implementation of GFCI that I have seen yet on an EVSE! I would give up trying to use those other expensive units and go this route. Fully adjustable, easy to test, cheap and simple. Thanks for posting it!

I have posted the basic experimental GFCI circuit I have been using with the CRMagnetics 8420-1000-G CT coil. http://code.google.com/p/open-evse/downloads/list

To change the trip point you can adjust R17, 330 Ohms sets a 20ma trip with the 8420-1000-G CT. C8 is to filter high frequency noise and may need to be adjusted if the circuit is not tripping quickly enough.

Chris

 

[W.Petefish]

I'm starting to get the Github urge for this project. Maybe we could produce a better product faster.

 

[widodh]

I'm starting to get the Github urge for this project. Maybe we could produce a better product faster.

Yes, please do so! Github is much easier.

 

[Fuzzylogic]

This will not solve the GFCI issue, but i noticed something else.

In the schematics a LM358 or LF353 is used as driver for the Pilot signal.
I have not measured the outputs of these Opams yet, but according to the datasheet, they are not rail-to rail opamps.
So in theory the voltage swing might not be -12V to +12V but less then that.
This might give some connection problems, as the Car might not be able to sense it's connected to a EVSE.

My DIY EVSE is partly based on the Roadster Foundry Mobile Charger
M. Eberhard used a LT1498, dual rail-to-rail opamp. So i decided to use the same ;-)

The EVSE i made also measures (with the pulses coming from a kWh meter) the other loads on the same phase, in my case i'm limited to 25A max. Any higher then that and my main fuse will blow.
So if for example the Oven and Microwave are on, the kWh meter will tell the EVSE to limit the max current for the roadster temporarily (by adjusting the width of the Pilot signal)

 

[Chris1howell]

The LM358 is used for GFCI not the pilot, it is far too slow for the 1khz square wave. The pilot uses the LF353 which is a much faster opamp, it is not a rail to rail device so it does have a small voltage offset (spec sheet says 10mv). I only have access to a LEAF, so far the pilot with the LF353 works perfectly. If other vehicles have an issue with the slight voltage drop, an LT1498 would be simple to drop in...

This will not solve the GFCI issue, but i noticed something else.

I am not sure why it will not solve GFCI (typo maybe). GFCI is performing quite well with the EVSEs I built so far.

 

[W.Petefish]

Chris,

Do you mind hosting this project in line with your own as you already have a page setup?

 

[Chris1howell]

Not at all, I will be happy to add any contributers to site, send me an e-mail with your google username to chris1howell83 at gmail dotcom. I welcome anyone to post schematics, code and wiki pages...