*Note: not quite finished yet, pending more trip experiences (skip to 2022/2023 update?)
Most of the time I would charge the Kona at home, like most electric vehicle
drivers do, as it's decidedly the cheapest and most convenient option.
My electric service has time-of-use billing, so charging off-peak brings
running cost per mile to less than two cents!
Compare that to 5 or 6 cents per mile for gasoline, about the lowest you
can presently achieve in an efficiently-driven Prius, and the petro-price
only goes up from there.
Long roadtrips done electrically are a different story, and that's the most common situation where DC fast-charging is needed. This brings up the headache of finding out where rapid-chargers are, which billing network they're on and *how* they bill, as well as which charging protocols they support. I described the present state, as of late 2019, to colleagues thusly: take a fairly efficient 50-ish MPG car like a Prius, and outfit it with a 5-gallon gas tank instead of the stock 11 or 12. Now, take away 90% of the existing gas stations, and then try to plan your roadtrip. Fortunately, there are plenty of apps for that, but without the internet you're basically hosed. You can't just spot a charging place on the fly along a major highway and pull in. [It gets even more confusing during any conversation about this, where "charging" can interchangeably refer to 1> pushing energy into the car, or 2> pulling value out of your wallet.] But the most confusion we encounter is about the electrical standards used for high-power vehicle juice-up. Put as simply as possible, we basically find three different connector types in the US at this time: CCS, CHAdeMO, and of course Tesla's own proprietary thing. And that's just the physical plug, where the only feature common to all of them is a pair of big honkin' pins to send current straight into a car's battery. The data interactions are the other layer of incompatibility, so there's no such thing as a simple passive connection adapter from plug A to car B. Non-Tesla standards seem to be settling on CCS or the Combined Charging System, an extension of the J1772 connector that's been around for a while. It seems to be slowly gaining ground over CHAdeMO, but the process has not at all been a pain-free one. This is the result of a bitter battle back around 2010-2011, and it's still not clear how things will eventually settle out. As many other articles about this have mentioned, it's VHS vs. Beta all over again, and the standard that eventually prevails is likely to do so because of money far more than technical merit. Frankly, if everyone had adopted the European "type 2" variant of J1772 and its extension to CCS, then we could have traditional Level 2 home charging, faster three-phase AC charging such as in commercial fleet environments, and high-power DC charging all in a single connector. It is very telling that Tesla is moving toward type 2 CCS for superchargers across the pond, but are still hiding in their walled garden in the US. [Well, up until 2023, when Tesla started pushing very hard for their connector to become the "standard" instead, passing either Tesla-proprietary CAN protocol or CCS/ISO15118 via PLC over it. Developing story.] In the meantime, US drivers still face an unholy mess. Charging networks are trying to build out regardless, by simply putting two or even three types of plugs and protocol support on their high-power deployments. They don't know what else to do, but the infrastructure needs to be installed. As stupid as it sounds, it's as though you pulled up to a gas pump in your ICE car and the pump had two or three different kinds of fuel nozzles hanging off it, and you had to pick the right one for your filler and hope it's not the one that got vandalized last night. High-speed EV charging is also far more complex than opening a valve and letting liquid fuel out. Vice has a fairly poignant article about this clusterfuck which is worth reading. Connectors have to be able to handle the high currents involved -- 200 or 300 or more amps for a sustained time, and the protocol for any system includes all kinds of lock-step safety and capacity checks. Cables for power levels at 100 kilowatts and beyond are starting to include liquid-cooling in their design, albeit with a few growing pains. That's an *assload* of power that we so casually talk about in this field, and it's kind of amazing that a vehicle can just quietly sit there and suck it all up. [To put that in perspective, pumping 10 gallons of gas into a car in two minutes is equivalent to over three *megawatts* of power, which would need about 10,000 amps into a typical EV battery, and that's only counting the *propulsive* energy stored. We are *SO* spoiled in terms of speed and perceived convenience.] Generally the car is entirely at the mercy of the rapid-charger's power supply, which could easily wreak major battery damage if not controlled carefully. There also has to be a lot of protection against disconnection under current load, to avoid destructive arc-faults in connectors and relays. Tesla and CHAdeMO have adopted variants of CAN automotive networking for their car-to-charger control protocols, but the SAE and related consortia decided to extend the thinking for CCS to broadly cover the whole "vehicle-to-grid" concept. This made it inordinately complex [may redirect to "switch-ev.com"], which may be one reason it took so much longer to nail down a useful standard. It uses a type of data-over-power-line transport called "HomePlug Green PHY", except that in this usage the data doesn't actually travel over any power connections. The spec for the lowest-level protocol alone is *seven hundred pages*. I found a copy here, but it is a bit more manageable to dip into when rendered as a PDF [about 3 Mb]. The spec for "PEV - EVSE" interaction starts around page 671; have fun with that. The signaling is sent superimposed over the existing J1772 control-pilot line, so as implemented a vehicle supporting CCS wouldn't be able to participate directly in any V2G setups without the charging station doing some translation up to the larger conceptual "smart grid". Not to mention the need for either the vehicle or the charging station to have an inverter capable of supplying AC power back from the vehicle's battery in a grid-safe fashion! For all the blather about it through the 20-teens and into 2020 when we were all supposed to have our atomic-powered flying cars by now, we're still a long way from even basic V2G ever reaching reality. It's all a work in progress, as gasoline fueling infrastructure doubtless was in its infancy. |
That aside, we still need a translation layer between CAN from the Kona's BMS and any CCS DCFC that speaks IEC15118 over PLC on the CP line. How's that for some alphabet soup? It is the job of a special Charge Control Module in the Kona to help sort all this out, a unit dedicated only to the purpose of fast-charging. This is a description of it that I found on Techinfo [click for better detail]. Compare these pics to the unit I found in my own car -- it looks quite different. The service info is correct about the function of either type of box, however: it serves as the protocol translator between the car's battery management system and an external charger. And with the way all of the standards pile on top of each other, that is not an easy job. |
Theory into practice
There's a Chargepoint 50 kW deployment at a relatively nearby Hannaford's grocery store, so one evening I loaded up some test gear and tooled over there just to play. I pulled in next to a Leaf that was already using it, so I'd have to wait until its owner came back. Two plug heads, but only one power supply, usual configuration! It was a bonus that the charger was free, sponsored by the local power company or something. I would still have to activate it with my freshly-minted RFID tap-card, as it doesn't simply turn on for any car that plugs in. | |
Basically, as of 2019, an EV owner has little choice but to sign up with
two or three of the major
fast-charging networks
that serve a target area, and use their RFID card
or phone app to enable charging.
It's the moral equivalent of Mobil Speedpass, although then you'd also need
different cards for every other retailer brand you're likely to visit as well,
juggling a fistful of creds for Shell, Sunoco, BP, Gulf, Chevron, etc.
Another obvious downside is that the network always knows who charged
where.
We don't get the somewhat anonymous equivalent of pulling into a random
gas station and paying cash for the energy intake.
Some chargers also have credit-card readers, which *should* all be EMV
chip-enabled by now, but they don't always work and aren't anonymous
either.
There is some movement toward cooperative partnerships to allow credentials for one network to start a charge on a different one, but progress on that is slow. The golden dream of the Charin EV consortium is for all vehicles to support transparent "plug and charge", where network authorization happens automatically using unique data from the vehicle. This raises obvious security concerns, and the the most vocal industry players seem to have agreed that simply using something unique like the MAC address of the car's CCS interface isn't strong enough -- they've decided that it needs a full-blown PKI behind it, with all of the same problems and failure potential that has plagued the internet for years. Some of it seems overkill, like someone's CS graduate thesis grown into a monster, but it remains to be seen how necessary all that is given the relatively limited threat model. Obviously this will leave any existing vehicles out in the cold unable to participate, because the manufacturers aren't likely to offer compatible charge-control module upgrades for what's on the road today. "Autocharge" is a simplified protocol that can support many older vehicles, and again, the security risk in that is really vanishingly low. |
The Leaf was using the CHAdeMO plug, because that's what Nissan decided
to stick with for now.
One downside with CHAdeMO is that the car needs two different connectors,
and thus a large charge-port door to cover it all.
Here's a better view of the two different inlets.
There seems to be some fairly nice gasketing around the essential
parts to seal them away from water and snow, a particular concern for a
car with its charge port right on the *front*.
The Kona has the same problem, and some owners have come up with some
creative foam-gasketing hacks
to seal the port door a little better.
I could see that the Leaf was almost full, because cars and rapid-chargers exchange a lot of information about battery and power state and generally display it to the user. Perhaps the Leaf's on-board telematics would also soon notify the owner's app that it was finished charging? Good guess, or fortuitous timing -- I had just set up the scope on my hood and clipped its probe to the pilot tap, when the Leaf owner walked up and unplugged. She was more interested in getting her groceries home than my technical setup, and just got in and drove away. But now I was free to play. |
While this charger was gratis, I still had to "authorize" to it, and that
of course depends on it having a working cellular-data connection to
talk back to the home office.
Sometimes those mechanisms fail.
Sometimes other mechanisms fail, such as protocol errors preventing
charger startup or flakey/misaligned plug connections.
No particularly useful information is shown to the user about what went
wrong, but it behooves anyone that runs into problems to call the support
number immediately and report any outages.
Sometimes the network operator can peek into the charger's log and
see what the actual problem was, and open a repair ticket if needed.
Sometimes they're just clueless, and spend the whole time "apologizing
for the inconvenience" but not able to fix anything.
And sometimes we can't even reach their call centers because their whole
phone system is down.
If this sort of crap happened to the trucking industry, we'd all starve. This is why charging network operators have to start thinking more like providers of "critical infrastructure", because if someone is stranded at a broken charger without enough energy to drive to another one, now you're talking potential life-safety issues or major inconvenience. Planning ahead and accounting for external conditions like weather obviously helps for a trouble-free driving experience, but things don't always work out as intended. As long as density of chargers remains so vanishingly below that of fuel stations while sales and adoption of electric vehicles grows, it really is kind of a critical-resource situation. I mention all this from a viewpoint of relatively little public fast-charging done myself, and this session at the Chargepoint was going to be my very first time trying it. But there are plenty of videos and horror-stories about roadtrips delayed or rerouted due to non-functional equipment, all of which lends perspective and helps set expectations, howver non-glorious those may have to be. Did I mention "work in progress" yet? |
Expanding the preamble or any other part of the packet doesn't tell us much; it's kind of a random radio-frequency mishmash running at several megahertz. PLC is basically a modem, with similar scrambling and modulation schemes as modems used on cable systems or phone lines [anybody remember those by now?]. Several OFDM carrier frequencies are dynamically assigned based on end-to-end conditions, so to get anything meaningful out of this you'd need a PLC and ISO-15118 analyzer to demodulate and sort out the transmissions. It is kind of interesting that the patterns here are regular at all, but it may be part of how PLC syncs up. |
That's where we get even more alphabet soup: V2G is actually a full OSI
network stack, with dynamic IPv6 addressing and then TCP carrying some
kind of "enhanced" XML between the car and the EVSE.
Not just that, but also wrapped in TLS with certificates.
All over the pilot line, as a one-wire transport network.
Dubious smart-grid needs aside, this kind of encapsulation overhead is what
happens when you let millennial "google kids" design network protocols.
They don't know any other way, and without their bloated web-kit type
development libraries they'd be completely floundering as to how to do
this.
The only developers who appreciate compact and efficient data transmission
anymore might be the folks working on the CANbus side of things, still an
area where small bit-fields carry a lot of meaning and speed is of
prime importance.
Here is a printable graphic of this nightmare that I tape to the car at EV shows. |
That little CCM box under the car's dash handles all of this on the
car side.
Its physical design and inconsistent representation in the service data makes
me think the whole fast-charging capability was done for the Kona as sort
of an add-on late in the design, almost an afterthought, and with some
indecision whether to mount the controller underhood or inside the cabin.
What also supports this idea is that my New Hampshire buddy's Bolt from
around the same manufacture time does not have fast-charging at all -- just
a generic J1772 level 2 plug in
the side, and a space where the big DC contacts could have been.
In that year of Bolt, it was a model *option*, which needed quite a few
more parts added to the vehicle.
The guys at Weber State describe it rather thoroughly, starting around
11:00 in
this video.
If a Kona without fast-charging existed, the big relay box hung off the side
of the charger would almost be redundant.
The battery connection could go straight into the inverter, and the two
extra orange DC cables out to the charging port wouldn't be needed
either.
After that, the only real difference is the code running in various
modules that would either support fast-charging or not.
One downside of the overall V2G concept is that it can fundamentally put a vehicle on the open internet while it's plugged in. The security aspects of this have been talked about but not really worked out yet, and the more complex the protocols become the less likely real WAN-level security will be inherently baked in and NONE of it is likely to be under the end user's control. I already worked hard enough to take my own car OFF the internet; now every time I fast-charge it could be back on?? Hopefully the data exchanged by the CCM is *only* directly relevant to charging, but as these protocols develop I could easily see it leading to a more generalized victimization of owners. Cell data down? Oh hey, here's this other kind of Internet connection, allowing the Kona to phone home to Hyundai/Kia anyway and get a wad of unsolicited firmware updates stuffed down its throat. With the thought processes or lack thereof that often get put into these things, that's a frighteningly possible scenario -- just from an innocent attempt to add some range. |
My description and scope-shots here were actually assembled from two test charging sessions: the first at the Chargpoint, and another one a week or two later at an EVgo station in a local mall after my card for *their* network had arrived. The second shot is at the EVgo setup; I had taken a nice long hike in a forest I hadn't been to before that afternoon, and stopped here on the way home, so my entire day was a voyage of discovery. Just like back in the Prius days, here I was taking the o-scope out cruisin' on a Saturday night again and geeking out hard on car electronics. A few people going in and out of the mall were curious about what I was up to, because usually when someone is charging a car it's all closed up and the driver is nowhere to be seen. Instead, they saw the car with everything flung open, data packets flashing by on the scope screen, and a weird barefoot guy bouncing back and forth checking all kinds of numbers. Perhaps this was a glimpse for them into the incredible layers of complexity that would become part of their own automotive future, even if they'd ultimately take it all for granted. |
Because of the variability of vehicle charge rates, billing based on
time is flat-out gouging, a denial of physical reality, and
has to change, as soon as possible.
[Update: it is finally starting to, as larger networks such as
Electrify America change over to per-kWh billing where they can.]
The excuse for it originates from the fundamentally mistaken notion in
several states that providers of electric car charging are trying to be
public utilities, and are thus prohibited from reselling energy without
state regulation and all kinds of permits and special status.
This is wrong, since the energy is for a specific purpose and not being
re-sold as a generalized service.
It's a well-known problem, but as of late 2019 only about half the states had
clued into the truth
and clarified their laws and regulatory guidelines.
In the meantime, many of the charging networks had adopted the paranoid
and profiteering approach of billing by time, but they've screwed it
up in the process.
The Kona itself, in fact, is one of the vehicles at the center of a
raging controversy
over taking a "tiered" approach to billing, where more is charged per
minute for higher power level brackets.
The problem is that on initial negotiation, CCS reports the *theoretical
maximum* of what the car is capable of accepting rather than what it's
actually *going* to accept at the time.
That highflown and usually bogus figure is used verbatim to set the
billing rate, and then the running reality of power level during charging
is usually quite a bit lower but the "tier" pricing never adapts to that.
If you went to the deli for a half-pound of tuna salad and they happened to be out of half-pound containers that day, how would you feel if they handed you your half-pound of food but charged you for a full pound simply because the container *could* hold up to a pound? How would you then feel if the deli owner got defensive about it when called on such bullshit dishonesty? |
On the road for real
On the way back home I decided to try Electrify America for the first
time.
This is the network
spun off from Volkswagen group
after the Dieselgate flap, with about $2B allocated to fund their atonement
in the US -- a big public EV charging buildout.
They were also pursuing big plans in vehicles on all fronts -- the Audi
e-Tron was already out and the Porsche Taycan was just
hitting the market, with the VW "ID" microbus-alike in the wings.
There were four or five charging heads here, connected back to a common power box. They were supposedly 50 kW, but I couldn't tell from the specification label on the box as the supply is probably shared across all the stations. In setups like that, more vehicles connected at once often receive less power each as they have to share the maximum load. Even Tesla does that, but only splits across two heads at a time. |
I was unable to order a tap-card for EA, because they don't have them, and never installed proximity readers on their head units. However, all of their stations have a "guest" credit-card reader on the front, and this group had proper chip slots so a card-only transaction would have a *little* more security wrapped around it. Chargers in out-of-the-way locations would be obvious targets for card skimmers, maybe even easier than gas pumps, but that's one of the things the EMV smartcards are supposed to help secure against. The first head I pulled up to didn't work; it read my card but couldn't negotiate charging startup for some reason, and at least assured me that I hadn't been billed. I moved over a stall, and that one seemed okay and started up. To fill time I called the bum charger in to EA support, went to find the loo at a nearby McDonald's, and then came back and talked with a guy who happened by and seemed generally interested in electric vehicles. He then wandered off, and I grabbed a snack and continued to monitor charge progress. |
EA doesn't arbitrarily limit session times, so since I still had over 200
miles to home I let the charge go a little over the recommended 80%.
It had become slow by then; with diminishing returns and the stupid
per-minute meter still ticking.
I wasn't at the really "controversial" tier because this was only a 50 kW
unit, but again, I never got more than about 42 kW out of it.
Even if I had cut off sooner, this was still rather expensive energy cost --
37 cents per kWh, 9 cents a mile, or about the equivalent of driving a
30 MPG gas car.
And that didn't even get me all the way home.  I had to find another waypoint about 40 miles out to make sure I'd make it; conveniently, another Hannaford's/Chargepoint deployment. Which I stopped in the proper way this time after imprisoning sufficient electrons, because sitting around in a desolate mall on a Sunday night in the cold is pretty boring. After finally getting home, I ranted about the experience into an appropriate forum. | |
At this point, electric vehicles used under the same fueling model as going
to petrol stations provide about zero financial advantage and take a lot
more of one's time to deal with.
Here's
another example
from a couple who did a longer roadtrip in their Kona and posted their
experiences to Reddit.
[Best tl;dr line: "Fuck EA - seriously"]
Roadtrippers and people who can't charge at home, who depend on public
infrastructure, are totally getting taken for a ride.
That's not exactly the way to help promote EV adoption in those markets.
It's hard to believe that charging network operators can't or won't
do the same simple math.
Well, maybe it's not so hard...
Electrify America, for example, exists *because* of corporate greed that got
caught out, and it wouldn't surprise me if many who work there feel
some resentment about that and less inclined to personally engage and truly
support what they're doing.
It might seem like the equivalent of forced community service in a way, but
they certainly earned it.
Big corporations like VW that have proven themselves *capable* of that kind
of arrogant and secretive disregard of people and the environment for the
sake of money,
still can't be trusted
to do what's best for the public long-term.
As long as they can get away with underhanded practice and focus on profit
uber alles, there will never be any true transparency or fairness.
Here's another great example of what we're going to find in fledgling public networks, before the organizations behind them manage to get their shit together. This was me trying a new network, with no expectations and enough charge still left to move on and get to a known-functional location. Which, obviously, I needed to do after that morning's frustrating conversation. The trend toward "app-only" is disturbing and flat-out discriminatory, and represents seriously short-sighted thinking on the part of people implementing these networks. Why it's discriminatory is explored further in my rant against Blink, which made the "we only have the app" mistake a couple of years later. They took away the web portal I had originally signed up through, leaving no other options for account management. My relationship with them was terminated with extreme prejudice shortly thereafter, especially considering that even by 2022, they basically had no CCS infrastructure at all. |
So how did I know exactly where the next working charger was that morning? A typical one-word answer could be Plugshare, but as I did more road-trips, I developed an enhanced system to have that information always at hand without external dependencies. Basically, I would research in advance, scan Plugshare and look up the locations close to my usual routes and more likely to be reliable, and then transfer those as waypoints into my GPS app on a tablet I brought along. That way, I'd always have the data in local storage, without any need to be on the internet, and only need to give "location" permission to the on-board GPS app which couldn't leak data up to "the cloud". Almost like modern airplane pilots do with their nav-charts, I could plan each leg of the journey and then have my navigation helper in my lap as I approached and "landed" at the next charging spot. | |
And a nice perk is that many of the new Electrify America deployments are
at Wal-mart or other big-box locations, which are also convenient places to
overnight in the sleeper-berth setup.
Basically, pull in and do the day's last charge while the pack is still nice
and warm, then move a few parking spots away and settle in for the night.
I often reflect on just how much computer code and related technology is involved in traveling this way. We depend on the GPS infrastructure and our ability to receive and use it, the cell-data network when we do want to go online for more information or "check in" at a charger, the network protocol between our vehicle and the charger, plus the network from the charger to the back office, the banking-transaction mesh behind how we pay for the energy... And how many little computers work together to simply run the electric vehicle itself? All told, likely millions of lines of code just to plan and execute a trip. And as consumers, we *demand* that it all work perfectly. |
UPDATE 2022:
While the number of deployments is going up, the quality and reliability of
the experience seems to be heading in the other direction by the end of
2022, with many public rapid-chargers simply not working and not getting
repaired in a timely manner.
Kyle Conner, one of the more prominent Youtube uber-geeks on EVs, recorded an
excellent
long and detailed diatribe
on this, which is well worth watching.
In 2022 I had to live through a lot of this same pain, with *four* trips to
Florida and back over the course of the year.
After the big heat waves in July/August that year, for example, Electrify
America chargers were dying right and left and then the needed repair parts
got snarled up in supply-chain delays.
As a result their support callcenters were slammed, and it took 20 or more
minutes to even get someone on the phone to complain about or report broken
hardware, get help starting charges, or whatever.
Across all the charging networks as well, those little surface-mount
card readers from Nayax
are, in Kyle's words, "total garbage", and keep going offline or totally
dying.
This is a particular problem at EA deployents, where the only other choice
for starting a charge is their stupid phone app.
In some other networks it's also the prox reader for their tap-cards instead
of having a separate RFID reader, so that failing can render my little satchel
of per-network cards useless as well.
Kyle and others who are able to make a lot of noise in the industry need to
keep after the networks to get their act together and start thinking of
themselves as the critical infrastructure that they *are*, and do
their part to discourage the public from getting disgusted and going back
to petrol.
Maybe Tesla getting into the game (2023, the whole NACS / CCS furor) will improve things and raise everyone's bar as far as reliability, but especially with Tesla, there's still the app problem to solve. |
_H* 191115, 200228, 220806, etc