chris1howell
Well-known member
Darkstar, I did the EXACT same thing. It took me more than 2 weeks to find that part.
Weekend(s) Project - DIY EVSE
- Thread starter chris1howell
- Start date
Help Support My Nissan Leaf Forum:
Ingineer
Well-known member
GFCI sensing is simply a current transformer that passes both Neutral and Hot (or L1 & L2) through the same transformer. What this does is negate any current drawn from the load, but will detect any stray current that flows either in or out of either line and enters or returns through some other path (such as your body) rather than on the normal one.
It needs to be sensitive enough to detect a few milliamps of fault current. (6 to be exact) This can be done with a simple bridge rectifier off a current transformer and then by watching the voltage rise at the bridge output. For a Class-A designation, The whole system must function together to stop current flow within 6ms of detection of the 6ma fault. This is "sub-cycle" trip detection performance!
-Phil
It needs to be sensitive enough to detect a few milliamps of fault current. (6 to be exact) This can be done with a simple bridge rectifier off a current transformer and then by watching the voltage rise at the bridge output. For a Class-A designation, The whole system must function together to stop current flow within 6ms of detection of the 6ma fault. This is "sub-cycle" trip detection performance!
-Phil
garygid
Well-known member
The spec sheet you linked to says it should not be used for Safety-type applications.
Also, even though normal GFIs detect at something like 6 ma, that limit has been found to be too sensitive for the detection in EVSEs (and still have reliable operation). Apparenly too many false "trips". So it seems that NEC allow EVSEs to use a different limit, something around 20 ma.
I think that the J1772 standards also mention this higher GFI limit.
Also, even though normal GFIs detect at something like 6 ma, that limit has been found to be too sensitive for the detection in EVSEs (and still have reliable operation). Apparenly too many false "trips". So it seems that NEC allow EVSEs to use a different limit, something around 20 ma.
I think that the J1772 standards also mention this higher GFI limit.
chris1howell
Well-known member
Gary, The spec does say it should not be used as a Safety-type device... However it I also says "Typical Applications"... "Ground fault detectors". I'll keep looking and see if I can find a better sensor...
garygid
Well-known member
Detect a fault to stop a process that does not involve human peril, is fine, but perhaps the hall-sensor & associated circuitry can fail in the "no-fault" state, so using it for life-critical applications is not recommended.
Just making an EVSE circuit that "works" is not too difficult. However, making one that is robust to noise, brownouts, power spikes, etc. AND has fail-to-safe-mode operation, and good error detection and handling ... is a lot more difficult.
For example, if the micro-processor should "freeze" (possibly during power-up) for any reason, will the AC relay(s) go to the "Open" state?
Just making an EVSE circuit that "works" is not too difficult. However, making one that is robust to noise, brownouts, power spikes, etc. AND has fail-to-safe-mode operation, and good error detection and handling ... is a lot more difficult.
For example, if the micro-processor should "freeze" (possibly during power-up) for any reason, will the AC relay(s) go to the "Open" state?
chris1howell
Well-known member
Gary, You are right, the EVSE is very simple if you wanted to ignore all, or some of the safety features... (diode check, venting) .
The microprosser defaults all outputs LOW (off) until specificly written HIGH (on).
The microprosser defaults all outputs LOW (off) until specificly written HIGH (on).
garygid
Well-known member
After an output is set (perhaps High) to turn the Relay ON, and then the u-code freezes (or gets caught in a loop), the Relay might be "frozen" ON.
Driving the Relay with a "watchdog" chip/circuit activated by pulses from the u-processor MIGHT be better?
There are a LOT of ways to do things.
Driving the Relay with a "watchdog" chip/circuit activated by pulses from the u-processor MIGHT be better?
There are a LOT of ways to do things.
chris1howell
Well-known member
Gary, I will have to test that senario. I have not had any problems with lockup, but it is a possibility.
Another method is to add another (smaller) microprocessor and give the main processor a heartbeat, if not detected within a certian period activate the reset pin on the main processor.
Another method is to add another (smaller) microprocessor and give the main processor a heartbeat, if not detected within a certian period activate the reset pin on the main processor.
AndyH
Well-known member
While I understand the concept of 'backward compatibility' in general...garygid said:
The current EVSE is a J1772-2009 device. There are no vehicles on the road fitted with a J1772-2009 connectors that also carry flooded batteries - so by default the Leaf's L1 EVSE would work with all other "J1772" vehicles (J1772-2009).
Yes - there are still a small number of legacy EVs with lead in the battery box, but they can't use the latest EVSE. Same for the DIY conversions. Both of these groups will need to do DIY upgrades to J1772-2009 and the person doing the retrofit would be responsible for ensuring their EVSE choice is appropriate. This isn't Nissan's problem.
edit... J1772 revision started in 2009 but it wasn't ratified until Jan 2010. The legacy vehicles using conductive (Ford Ranger) and inductive (RAV4, S10, EV1, etc.) were designed around J1772-1996. Calling modern cars "J1772-2009" compliant is sloppy by heartfelt.
DarkStar
Well-known member
FYI, the SAE updated the specification to J1772-2010 in January of last year...AndyH said:
chris1howell
Well-known member
Gary, I did a bit of research on the possible issue you brought up, the possibility the microprocessor could lockup while the relay was on. The ATMEL 328 microprocessor has a built in hardware watchdog timer. I can enable the timer and add the code to reset the timer every second or so. If the Microprocessor stops responding the timer will expire and Watchdog will reset the chip.
Ingineer
Well-known member
A hardware watchdog is a must, and all relays should be open when the micro is in reset.
Another thing that should be done is to check the integrity of the GFCI circuit. The Panasonic even does this!
You have to make sure it's working. For example, what happens if the CT fails open?
You should also verify ground.
-Phil
Another thing that should be done is to check the integrity of the GFCI circuit. The Panasonic even does this!
You have to make sure it's working. For example, what happens if the CT fails open?
You should also verify ground.
-Phil
garygid
Well-known member
What is a good way for the circuit to "verify" the Ground?
Ingineer
Well-known member
garygid said:
Attempt to pass a small current through it. This could simultaneously test the GFCI and the ground at the same time.
Ingineer
Well-known member
Oh! Gary you'll love this: Today I checked the AV EVSE, and it too fails the diode check!
It promptly enabled output with just a resistor.
-Phil
It promptly enabled output with just a resistor.
-Phil
AndyH
Well-known member
Right - ratified in Jan 2010 - thanks.DarkStar said:
J1772 Issued: Oct 1996 Revised: Nov 2001 Revised: Jan 2010
DarkStar
Well-known member
Is there something we're missing in the specification? It seems kinda strange that both Panasonic and AV have created EVSEs that don't do diode checks... Isn't the whole purpose of that for the "mud puddle" safety check?Ingineer said:
chris1howell
Well-known member
J1772 is just a recommended practice. Adding the diode check (-12V) adds a bit of complexity and cost. Some manufacturers probably did not feel it was significant or that other safety features adequately mitigated the risk.
AndyH
Well-known member
In the "what it is" VS. "what some think it should be" department...
J1772, per the SAE, is a "Surface Vehicle Recommended Practice" - it's not a requirement or regulation. From the center of the front page of the J1772 Jan2010 document:
J1772, per the SAE, is a "Surface Vehicle Recommended Practice" - it's not a requirement or regulation. From the center of the front page of the J1772 Jan2010 document:
garygid
Well-known member
Phil,
Yes, thanks, ... (big grin).
That is exactly what I wanted to test on the dealer's AV EVSE this Saturday!
Yes, thanks, ... (big grin).
That is exactly what I wanted to test on the dealer's AV EVSE this Saturday!
