Mini-QC Rapid-Charger (RC) Project for LEAF QC Port

[garygid]

At this point we know that the car allows net power out of the battery pack
while AC "charging", for example to run the heater to pre-heat the car.

However, when using the QC port, it "appears" that the car monitors the
current, and it is not clear that the car will allow net current out.
Since we do not have access to the power-out process to examine it,
we can not confirm that the car supports operation as a backup
power source. The manual appears to forbid such use, but that
condition might not be enforceable?

 

[jclemens]

I believe the 2011 and 2012 leafs are programmed for chademo spec v0.9
2013 (could be same) is programmed for chademo spec v1.0. (required to be reverse compatible with v0.9 chargers)
I have not seen v1.0, but I think it has provisions for vehicle to grid.
If we are using v0.9 chargers and cars to gather canbus data, then we will never see the commands for pulling power out of the pack. This really is not a priority anyways, but you can dream! (once commercial products are available, the same techniques can be used to gather and decode the canbus data, but that will likely be a different group of people)

lets cross that bridge when we get to it.

 

[Ingineer]

The LEAF does allow power out via CHAdeMO, but I have not fully fleshed out how it all works. Nissan sells the LEAF-to-home kit in japan that is a bi-directional CHAdeMO charger that can also go off-grid and allow "islanding" so you can have backup power for your home.

-Phil

At this point we know that the car allows net power out of the battery pack
while AC "charging", for example to run the heater to pre-heat the car.

However, when using the QC port, it "appears" that the car monitors the
current, and it is not clear that the car will allow net current out.
Since we do not have access to the power-out process that examine it,
we can not confirm that the car supports operation as a backup
power source. The manual appears to forbid such use, but that
condition might not be enforceable?

 

[garygid]

Thanks for the info.
It would be nice to get a good description of just what the
Japanese "home power" accessory does.

We are concentrating solely on charging until we get that
well done. The details of the charging CAN communication
are becoming more clear as we are able to run little experiments.
Gradually our guesses are becoming quite-likely descriptions,
and even some good conjectures of functionality.

We are experimenting with a hacked power supply that we
would not recommend to others, at least for safety reasons.
However, in carefully controlled circumstances (indoors, no kids
running around, etc.) we are able to use this supply to get
about 28 amps DC into the car's battery, via the QC port, with
no modifications to the car. Yes, around 10 kW in, until the car
tapers off the charging rate.

This power supply might be able to provide a bit more power,
but we are using 240v via a standard 50 amp breaker, and that
is enough power to both get a rapid-charge from this "mini-QC"
system that I affectionately call "Robomo", and test the interface
meaningfully.

Yes, we managed to test the breaker too, and it does work!

We have several people that are already saying "I want one!",
but it is not wise to encourage anybody to duplicate this test
equipment. However, for those capable of contributing to the
project, with sufficient skills, and with sufficient safety
considerations taken care of, we will consider sharing more
details, and providing some guidance toward duplicating what
we are doing. PM me if you qualify and want to spend
two or three thousand, and substantial time, to participate.

However, it is premature, and might even be considered
a bit irresponsible to make our guesses and conjectures
fully public at this time.

Indeed, we could use some appropriate legal help to better
evaluate just how and when to share what we are lerning.
Any volunteers, please PM me.

We are still concentrating on the control of the charging session,
along with the necessary control of a non-optimal power supply,
as needed to do the charging-session experiments.

As we get more comfortable with our results, we will try charging
other LEAFs, then other year models, and possibly even other
QC capable makes and models. Sure, there really not many, yet.

So, a parallel effort should be the design of a suitable power
supply. One with up to 50 amps 240v (12 kW input) is the
first goal, with a 24 kW input version kept in mind.
Folks with suitable skills, like switching transformer design,
driving circuits for regulating the output voltage and current
of an efficient, air-cooled, compact power supply are needed.
Sure, we can bumble through it, but real experience would
be a substantial help.

We are carefully avoiding or regulating contact with
those that have access to, or appear that they might
or could have access to, any proprietary or "regulated"
Chademo information. The associated constraints
are just way too burdensome, even if they might
not be legally enforceable. Remember Tucker?

More later, as appropriate.
Cheers, Gary

 

[mwalsh]

I love that you guys are doing this!

 

[jclemens]

I have uploaded a newer spec for the power pin. This should make manufacturing it easier and cheaper.
Based on my design of the data pins (easier to build, but I have not created cad drawings yet), here is how to make a power pin. (still easier said than done)

1. start with a length of 2.75 inches of 3/8" diameter copper rod (alloy 101), available at mcmaster.com - part number 8965K141
(if 9.0mm precision ground raw copper rod is available anywhere, that would be much better, but even at mcmaster, their rod selection jumps from 8mm to 10mm)
2. drill a bore hole about 3/8 or 1/2 inch deep, this might be difficult depending on tools available, this hole is to hold a 2awg wire, so you might be able to go smaller than indicated in the photo. The diameter is ideally 6.8mm, any smaller and 2awg wire will be difficult to fit, any larger and the walls might be too thin (see next step)
3. Add threads to the rod at same end as bore with a 3/8-16 die, to slightly more than a 1 inch at one end. This step can be done at any time, but to ensure the contact area is not damaged by anything, I put it here.
4. Turn the piece and precision grind a length of at least 1.125inches from 3/8" (9.525mm) diameter to 9.0mm in diameter, using whatever method works for tools available. This is only shaving off a small amount. The tolerance for this 9.0mm +0 -0.1mm
It is also acceptable to have the entire piece at 9.0mm in diameter, the 3/8-16 thread should still work.
5. have the piece silver plated while not exceeding the tolerance specified above. The kind of home-use silver polish will not be sufficient as the layer of silver will be too thin. The only way to do this properly is electroplating somehow. There are many plating shops around that should be able to do this.

This pin will be held in place with a brass nut also available at mcmaster.com pn 92174A031 but should be available elsewhere such as fastenal.
Yes, the nut should be brass (or plastic) to avoid galvanic corrosion.

 

[KillaWhat]

how are you attaching the #2 wire into your hollow threaded base?

Oh, the orange plug housing looks great!

 

[jclemens]

how are you attaching the #2 wire into your hollow threaded base?

Oh, the orange plug housing looks great!

The wire/cable must be soldered to the pin before being mounted into the plastic, an industrial soldering iron or a torch would do the job (like like soldering copper pipe)

 

[Luft]

I believe the 2011 and 2012 leafs are programmed for chademo spec v0.9
2013 (could be same) is programmed for chademo spec v1.0. (required to be reverse compatible with v0.9 chargers)
I have not seen v1.0, but I think it has provisions for vehicle to grid.
If we are using v0.9 chargers and cars to gather canbus data, then we will never see the commands for pulling power out of the pack. This really is not a priority anyways, but you can dream! (once commercial products are available, the same techniques can be used to gather and decode the canbus data, but that will likely be a different group of people)

lets cross that bridge when we get to it.

I was talking to an AV tech and he told me that my latest software update for my 2011 LEAF made the QC Protocol the same as a 2013 LEAF's.

 

[garygid]

In making a power pin as described above, it is probably
better to swap steps 2 and 3, to do the threading before
the hole is drilled (because the thin walls might not be
strong enough to take a coarse threading tool)?

 

[garygid]

I got this question:

...
Are you willing/able to publish the precise sequence (msgId, etc) that resulted
in the successful charge (congratulations, BTW )?
Is the Chademo controller you developed generic enough that someone could
use it to control a variety of custom power supplies?
or will everyone have to write their own controller?

The car seems to use 3 MsgIDs, and the QC seems to use 2,in normal operation.
In addition there are 4 handshake signals, and the 2 high voltage connections.

The QC and LEAF need to be responsive to each other's messages, in a timely fashion.
We are still discovering limits to the messages and the related responses.
So, no not ready to publish.

However, the harder part is controlling a high-power supply, and that would
be unique to each different power supply design. For example, if the power
supply controller is told to ramp the voltage up to N volts, or down to zero
volts, how quickly can it do that? How about ramping the current up and down?
What noise and ripple is in the output? Can it maintain a good power factor while
in operation? Can it operate at high efficiency so that only modest cooling is
necessary? Are the grounds handled properly, and the HV truly isolated?

If you have, or are willing to buy, a suitable power supply, then we might be
willing to provide some reasonable suggestions to a collaborator who wants to help.
Safety is a must, and we welcome sincere offers to collaborate, from
persons who have priorities similar to our own. Safety First.
This is not ready for prime time.

We are currently using an AVR-CAN development board with some external circuitry
to interface and measure. Creating logs (usually CAN-Do compatible) of each
experiment helps us learn. Careful design of experiments helps broaden our
understanding of the process. Of course, we have to suffer with rapid-charging
the test LEAF, and driving it to deplete the battery for the next test.

My previous long post pretty much says the same thing.
Cheers, Gary

 

[garygid]

I was talking to an AV tech and he told me that my latest software
update for my 2011 LEAF made the QC Protocol the same as a 2013 LEAF's.

I thought that the 2011 used the 0.9 version of Chademo, and the 2013
used the 1.0, which is apparently "required" to be backwards compatible
with version 0.9 chargers.

We assume, at this time, that we are working with the 0.9 type protocol.
But, we will eventually design some experiments to try and detect the
car's protocol version, and put it into backward-compatible mode,
if necessary.

But, for now, with the ability to charge, we are trying to get better
control of the power supply, and prepare for doing experiments
with other cars, including logging the session for analysis.

I am modifying CAN-Do to automatically produce multi-variable graphs,
over various sections of the log session, and produce (at the moment)
". bmp" files, for printing or email, or eventually, for posting here.

Cherrs, Gary

 

[thorx]

Thanks for the info.
It would be nice to get a good description of just what the
Japanese "home power" accessory does.

Not much information, but at least I found something on the topic.

V2H system:
http://news.nost.jp/2012/07/japan-sells-first-smart-house-for-electric-vehicles/" onclick="window.open(this.href);return false;
http://www.nichicon.co.jp/english/lib/new122.html" onclick="window.open(this.href);return false;

 

[garygid]

The EV Power Station described at the end of the 2nd link
is apparently the V2H (Vehicle-to-Home) unit. If we could log
its operations, both charging and in V2H mode, we might be
able to design an L2L "rescue" attachment, where one QC
vehicle (V2H compatible) could charge (thus rescue) a depleted
QC-compatible vehicle.

The L2L device would just plug into both QC ports, and "rescue"
the depleted vehicle by sharing the charge remaining in the
"rescue" vehicle. This would be a third project, and I have
not seen the discussion of such a device. Have you?

A third project is a passive QC-Tap (socket and plug), arranged to
plug into the vehicle's QC port (socket) very much like the
front "barrel" end of our 3D printed plug does, but have the
"handle" end replaced with a "QC socket", to accept a QC Plug.
The internal pins might be one piece, and slightly longer, but
with a female receptacle on the back end of each pin to
accept the pins of a standard QC plug.

When deployed, this QC-Tap would include a "tap" for monitoring
the CAN bus, the control signals, and even the HV voltage, perhaps
all through optical isolated A-to-D sensors or converters.
This would allow easy monitoring of the performance of
any device attached to the vehicle's QC port.

Who wants to tackle that project, initially with just pins and
a 3D printed part (perhaps in two or three sections), but with
no included electronics?

As we have more power supplies configured as mini-QC devices,
and wish to try them on different vehicles, this QC-Tap could be
used for monitoring or logging, to better investigate the
process occurring between the "charger" and the vehicle,
without having to modify either device. Very Useful.

Any volunteers?

 

[garygid]

I got this question:

It's kinda chicken and egg.
I was asking about the details so I could start dreaming up
my power supply. Yes, I do expect to spend some money
to make it work. Unfortunately, I do not full the bill as an
experienced high power designer.

I have designed and built one 5kw battery charger but that
was only 36v. 400v scares me. As I mentioned earlier,
I am trying to see if I could make a direct charger from
my solar and bypass the whole dc-dc step ( it outputs ~400v
at 10kw so maybe a much simple linear regulation would suffice).

Perhaps add a spec sheet for the power supply as far as you know it right now?

I will attempt to describe a simplified (paraphrased) overview of what
the car appears to expect during normal QC operation.

The car and charger exchange some handshaking to "come alive"
and begin the CAN communication. The charger says what current
it is capable of providing, and the car indicates the voltage that it
might need, along with some other information.

The car asks for "proof" expecting the charger to ramp its output
voltage up and back down, when not yet loaded by the car's battery.

Assuming that ground fault type tests are passed here, the charger
and car can mutually close the relay to expose the car's battery on
the high voltage pins, no current yet.

Then, the car asks for current, starting at 1 amp, and ramping up.
The LEAF seems to give requests at 10 per second, to ramp up
at about 20 amps per second. The car expects the charger to follow,
but we do not yet know how closely, but it has been reported that it is
possible to offend the car sufficiently so that it will refuse to charge,
even with L2.

Assuming that the car is sufficiently happy with the charger's
delivery of current, the car eventually reduces its current request,
and eventually drops to request to zero amps. It expects the
charger to stop charging.

Either side, the charger or the car, can initiate an end to the
charging session.

So, connecting an uncontrolled current source, like a solar array
would be, would probably offend the car. DC to DC regulation
would appear to be necessary, and linear regulation would
seem to require a LOT of cooling. A switching regulator might
be "required".

How would you proceed with such a project?

 

[garygid]

The car appears to produce three different CAN Messages,
and the charger appears to respond with two CAN Messages.
Call this a "packet".

As observed, 10 packets per second have been reported, and
the CAN bus appears to be similar to the other 3 CAN buses
in the LEAF, running at 500000 bits per second. This bus
appears to be devoted exclusively to the QC process.

The car's 3 MsgIDs seem to be 100, 101, and 102 hex, which include
3 x 8 = 24 bytes of data sent to the charger.

The charger appears to generate MsgIDs 108 and 109 hex in
response, soon after the car's messages arrive. These 2
messages provide the car with 2 x 8 = 16 bytes of "response".

We are making good progress on understanding the data.
However, many "discoveries" are just conjecture, since no official
description of these bytes exists in the open literature, as far
as we now at this moment.

For those interested in the details, I suggest to use my CAN-Do
program to examine the one QC log file that is on-line, at
http://www.wwwsite.com/puzzles/cando/" onclick="window.open(this.href);return false;

We discourage public conjecture about unverified "possible"
meanings of the data bytes, to avoid misleading others
into this dangerous High Voltage area. However, we hope to
be able to release some preliminary conjecture soon,
based upon the project's observations.

Cheers, Gary

 

[garygid]

Another challenge:
3D Print a surface-mount J1772 socket (look on thingiverse for J1772),
and supply it with female pins (two power pins, one ground pin, and the
Control Pilot signal pin), suitable for an experimental 30-amp connection.
Getting suitable female pins is the more difficult part, I suspect.

We could configure our Robomo mini-QC unit for the J1772 car-side
handshake (one resistor and one diode would be sufficient, I suspect),
to get 30 amps at 240 volts AC. That 7.2 kW might allow us to demonstrate
charging at near 6.6 kW via the QC port using a public L2 EVSE power source.

One person has a 75 amp EVSE that they really anxious to try out, so a similar
socket with pins suitable for 75 amps would allow a higher-power test.

Eventually we will try using a 100-amp breaker, but the present power
supply is not suitable for much over 12 kW, as I recall.

 

[mwalsh]

^ Ooh. J1172 use sounds like a compact (and inexpensive) solution to faster charging!

 

[klapauzius]

Another challenge:
3D Print a surface-mount J1772 socket (look on thingiverse for J1772),
and supply it with female pins (two power pins, one ground pin, and the
Control Pilot signal pin), suitable for an experimental 30-amp connection.
Getting suitable female pins is the more difficult part, I suspect.

We could configure our Robomo mini-QC unit for the J1772 car-side
handshake (one resistor and one diode would be sufficient, I suspect),
to get 30 amps at 240 volts AC. That 7.2 kW might allow us to demonstrate
charging at near 6.6 kW via the QC port using a public L2 EVSE power source.

One person has a 75 amp EVSE that they really anxious to try out, so a similar
socket with pins suitable for 75 amps would allow a higher-power test.

Eventually we will try using a 100-amp breaker, but the present power
supply is not suitable for much over 12 kW, as I recall.

YES!!! Cant wait to see this working. This would be a very elegant solution.

 

[garygid]

The normal QC process that has been observed
might be described as eight different Phases:

Phase 1: Come Alive Handshakes.
Here, the four Handshake lines are used to Wake Up the car.

Phase 2: Getting Aquainted Communication.
Here, initial CAN bus communications advertise the car's wants and
the Charger's capabilities, to see if charging should be attempted.

Phase 3: HV ramp and leakage tests.
Monitored by the car, the Charger ramps up the output voltage
to the voltage requested by the car, and Ground Fault and HV
leakage tests should be performed by both sides.

Phase 4: Starting to Charge.
The Battery is exposed to the Charger, and current requests from the car
begin, typically ending at whatever value (or the car's max) that the charger
advertised that it could deliver. From 28 to 120 amps have been observed.

Phase 5: Maintaining the Charging.
This is the first long-duration phase, where cooling of the Charger
becomes important. Usually, the car maintains an almost constant
amps-requested level.

Phase 6: Tapering Off.
The car starts requesting less amps, usually as the battery pack gets
fuller, or hotter. The requested amps can vary as the car desires.

Phase 7: Terminating Charging.
Either the car or the Charger decide to end the charging session.
The battery might be sufficiently full, a timer might have expired,
some unusual condition might have occured, or the user decided
to quit charging. The car's request for amps goes to zero, and
the Charger is expected to stop charging. The battery relay opens.

Phase 8: A Friendly Goodbye.
A final communication and expected reversal of the handshaking,
if any is appropriate, would be done here.

Hopefully, by this point, you have not "bricked" your car, or
created "magic smoke" from anything.

----------
In the future I will try to describe each Phase in more detail,
and provide a link here to each Phase-specific post.