Tesla Model 3 Failed To Fast Charge At CCS Charger: Video

FEB 23 2019 BY MARK KANE 55

Probably software issues prevented fast charging

One of the first Tesla Model 3 drivers in Norway, Bjørn Nyland, encountered a problem with fast charging the car at one of the new 175 kW Hypercharger fast chargers installed by Fortum.

As demonstrated, the charging didn’t start after connection and authorization. According to Bjørn and Hypercharger, there are two issues with the charging process – one with the charger, which maybe requires a software update and a second with the car, which apparently do not fully comply with ISO15118 (CCS communication specification). Tesla’s over-the-air-update should solve the second one if that is indeed the case.

Interestingly, we didn’t hear about problems at other fast chargers, including Fastned ultra-fast chargers (175 kW). The car was able to recharge also at the 50 kW ABB charger next to the Hypercharger.

From the video description:

Response from Hypercharger:

“We are already working closely with Telsa Munich on this issue. There is part of a fix in the new SW which we are going to roll out. But still it could happen that you would have to try several times, until the charging session starts. That other half Tesla has to fix it on their side since they do not comply with ISO15118 (CCS communication specification). We are still waiting on their answer, what they are going to do about it. I will let you know as soon we get an answer.”

Categories: Charging, Tesla

Tags: , , ,

Leave a Reply

55 Comments on "Tesla Model 3 Failed To Fast Charge At CCS Charger: Video"

newest oldest most voted

From what I have heard, this sort of problem is common to many cars that use CCS with its so-called “radio on wires” communication protocol. Tesla’s “baseband” digital communication (SAE J2411 single wire CAN) is more reliable and simpler than PLC. And seriously, what percentage of the market actually supports ISO 15118? Single digits? Most CCS in North America are still using the older DIN 70121, which is probably what was used by the 50 kW ABB charger that did work. But it is all hearsay to me, I don’t have anything that fast charges, I am in the 3.3 kW AC ghetto 🙁

If you look it up, you will see that Tesla has problems with their own “super” chargers, wall and mobile chargers.
So if they can’t make everything work perfectly between just a handful of products, how do you expect hundreds to work better together?

Nonsense. Tesla superchargers are the only fast chargers that can actually be relied upon to work. And I say this as a Leaf driver in a area where there are quite a few public L3 charging stations, all of them with atrociously poor uptime.

There’s clearly a big difference between the experience in Europe and the US. In Norway, both CHAdeMO and CCS was awfully unreliable in the beginning. These days it is rare to have any problems on CHAdeMO, while CCS still has some way to go but is much improved compared to give years ago.

So maybe don’t be so confident that your own anecdotal evidence is universally representative.

Sure let’s blame Nissan.

Look up where? I have never had a problem charging either of our Teslas (S and X) using mobile chargers and superchargers, with more than 1000 charges between the two vehicles. Don’t make uneducated comments.


Don’t make uneducated comments, combined with “*I* have never…”

It is not mean to point out that drawing broad conclusions based purely on personal experience is, in fact, precisely that: uneducated.

Do I really need to guide you? Look it up on TeslaMotorsClub and see what fanboys are saying. Remember, more indulgent people you will not find. Look up “Tesla Supercharger problems”.
There you go, I can’t make it more simple. If you still can’t find it, I sugest you go take some web browsing lessons.

You can just go look at charger ratings on plugshare to verify this dude’s opinion. Superchargers are almost all rated 10 while most other level 3 charger networks, including the new Electrify America chargers, are hit or miss with their ratings. As someone that’s used both in abundance, I can attest to Superchargers being the most reliable US level 3 chargers I’ve used.

That doesn’t mean you don’t run into the occasional problem at Superchargers. And that doesn’t mean the other networks won’t get better. And of course EU, and especially Norway, have different networks that are very reliable (from what I hear).

I recall a presentation a few years ago at the Infrastructure Working Council, where someone from PlugShare compiled user comments that indicated a much higher success rate with first-try charging events at Tesla DC chargers (high ninety % range) compared with all other DC chargers (around 85% if I recall). I have not seen such a study since, and admittedly this was not scientific by any means. At the time, “all other DC chargers” might have been mostly Chademo, where my biggest problem was always that huge, grouchy connector. But I have seen many blog postings in the past complaining that only 3/4 of CCS stations works with any given CCS EV in the US. I hope that is improving, there certainly has been lots of work in this area. Argonne National Labs in particular has been helping.

Ditto on LS’s characterization. When I researched local charging spots, Tesla charging infrastructure rarely has down-time, while other networks tend to have a lot of it.

What is “down-time” as you define it here? Of course it’s easier if the number of models that can charge is very limited so like no workarounds for bugs in EV software are necessary or like only one authorization method of one customer card provider is possible – like for the SC network…
What are “other networks” for example? Names? Or do you mean chargers that were mounted but then aren’t connected to anything for weeks for some reason?


Normally CAN bus uses differential signal, i.e. use 2 signal wires, for electric noise protection. Using single wire CAN you loose this noise protection, not good when running several meters of wire right next to power wires that will be noisy.

CCS uses OFDM signaling, i.e. a frequency modulated signal. That is robust signaling and better protected against electric noise while only using a single wire.

ISO 15118 was published in 2013. DIN 70121 is from 2012 and has a smaller scope than ISO 15118. The EVSE from HyperCharger supports both.

The problem we see in Norway is likely due to communication protocol error as the charger does see a car connected. Perhaps the EVSE or the car does not follow the specs.

This is why companies test their equipment with other companies. There were a test symposium like that in Arnhem NL in 2018-11.


Char-In is organizing similar events. Most recently 2018-10 in South-Korea.

Good points. But they should test with reference testing equipment that exercises lots of possible paths in the protocol, not a subset of other products. Otherwise it’ll be like the web mess, where browsers “helpfully” tried to display invalid documents instead of showing an error message. Therefore the people making invalid documents never notice, never correct their mistakes, and it becomes almost impossible to recover! Companies, then, get stuck with the burden of testing their pages in a huge number of different browsers — Chrome, Firefox, internet explorer, opera and so on, on Windows, on Linux, on Android, Safari and many of the others on iOS and Mac OS, and still there’s lots of combinations of devices and browsers it may not work correctly in. That may not be a big problem for the publisher, since these combos are much rarer, but it’s no fun for those with those devices.

To illustrate the legacy, I ran the InsideEVs front page through the W3C validator. It reports a long list of warnings and errors:


You can try this on almost any web page with similar results.

“Good points. But they should test with reference testing equipment that exercises lots of possible paths in the protocol, not a subset of other products.”

Yes, see Arnhem video at 2mn19s

You are correct in the sense that high-speed CAN networks on cars (usually 500 kbps) and heavy trucks (often 250 kbps with longer data-links) always use differential (2-wire) CAN for better noise immunity. Tesla’s J2411 single wire CAN runs much slower (33.3 kpbs) which is another way to improve robustness. Charging is a rather low-data-rate business, so lower data rate and fewer connector contacts improves costs and experience shows that there is no problem with noise. The problem that has plagued CCS is that the radio frequency signaling does not stay on the wires, it is transmitted to adjacent cars, and figuring out which vehicle is connected to which station is not trivial (I hear this referred to as the “association problem”). The PLC data rate has been reduced which helps but it can still take some stations up to a couple minutes before they can even begin charging. Most people have walked away by then, only to come back and find out that maybe their car never started charging. There are efforts underway in the engineering standards groups to improve “association time”, but controls might move to WiFi first (which is also supported by 15118) if a baseband “association”… Read more »

“The problem that has plagued CCS is that the radio frequency signaling does not stay on the wires, it is transmitted to adjacent cars”

Do you have a link to this?

Remember that the OFDM signal is on the separate CCS Control pin, one wire used only for communication and only for the connected car. It would be very strange if the OFDM signal can be detected on a different cable or a different charger.

Note that way OFDM is used in CCS is very different from PLC (Power Line Communication) network where the OFDM is deliberately modulated on the power wires to spread as far as possible in the power cables grapevine of a house.

SAE J1772 links to all of the PLC documents, I think it is in the J2847 series (maybe -3?). They are behind a paywall but not expensive (<$100). Isn't one of the functions of the SLAC protocol (Signal Level Attenuation Characterization) to get the RF signal level to a normalized range which helps figure out which EV is connected to which station? And then they blink the baseband PWM (5% duty cycle if I recall) to confirm? Sorry, I am not a PLC expert, I just hang out with them too much for my own good apparently. But I understand that CCS does use PLC, "HomePlug Green PHY" to be specific, except they run it on the control pilot line now instead of the actual power wires. So the RF carrier (in the vicinity of 30 MHz) rides on a 1 kHz square wave at 5% duty cycle that swings from -12 V to +9 V or so. Now I don't know about SAE J3105, they might run HomePlug Green PHY on the power wires with that automated overhead pantograph for the buses, I have not kept up with what they are doing. There was talk about using WiFi, but… Read more »

CCS simply appears to be more complex, or maybe just more poorly documented, than CHAdeMO. I can’t see any other reason why there should be so much more trouble with the CCS implementations than CHAdeMO. It’s not a particular charging network, nor a particular car manufacturer, it really seems to be the protocol.

Whether nobody supports the full ISO standard or not, I don’t know. But clearly everyone needs to follow whatever standards it is they adopt.

There are now specialized suppliers that make testing equipment to thoroughly test all possible “conversations” for different protocols. Perhaps Tesla should buy a few of these devices.

Testing equipment will never be able to test every possible corner case that might occur. Anyone who has worked with computers for a while knows that seeing compatibility problems between devices from various vendors using supposedly standardised interfaces is simply unavoidable in practice. Though the frequency of such issues depends a lot on the complexity and quality of the particular standard in question…

Do Not Read Between The Lines

E.g. Bluetooth stacks.

Chademo uses a very simple and straightforward CAN protocol, I have heard there are only a few messages. ISO 15118 is notoriously complex by all accounts and a lot of it has to do with billing systems and energy management, not just local charge control (all good things, no doubt). 15118 does not actually require a specific physical layer if I understand, but CCS today uses PLC, and modulation schemes do add another layer of complexity. One of my German colleagues says there is an idiom in German for over-engineered solutions which he thought applied to CCS PLC implementations. The idiom translates roughly to a pig that gives milk and wool and lays eggs. By implication, it is no good for sausage apparently? It seemed funny at the time, although German humor usually escapes me. Maybe it makes more sense after drinking a beer.

If 175 kW is hypercharging or ultra fast charging, then how do we call the 350 kW charging rate?

I don’t know who tried to label charging at 175 kW “ultra fast charging”, but that’s nothing but hype.

Superfast or ultrafast charging for a BEV passenger car would be charging at close to 1 MW or above. We likely won’t see that for a few years; perhaps not for several.

These labels are completely arbitrary — and so is your postulate what they should apply to.

At that high c-rates i wonder how much IT gonne harm the battery over time.
Model 3 has the best batterycooling of any EV, but 350 Kw sounds imo like to much heat to be healthy

Safe C rates vary from one cell type to another. (And better cooling can only help to some degree.) Going beyond the safe C rate doesn’t just shorten life somewhat — it totally destroys the batteries very quickly. Sans major engineering mistakes, you can be pretty confident that no car maker would allow a higher C rate than what is safe for that particular battery.

>=150 KW is HPC – High Power Charging.

Imho it’s like this: regular, super, hyper, ultra, mega…
350 kW = the Ultracharger
1000 kW = the Megacharger (Tesla Semi)

1000 kilo Watt is 1 mega Watt btw.

It is a company name. It hasn’t got anything todo with the charging speed or the charging technology. They do chademo as well.

Then the right way to reffer to it is: the 175 kW charger from Hypercharger and not the 175 kW hypercharger.
Anyway, ultra fast charging isn’t though.

I would think that for most people, 175 kW charge rates are “fast enough” and 350 kW charge rates are “faster than I really need but nice to have!”
Seriously, if you don’t run into significant tapering, you can get nearly 100 additional miles of highway range in 10 minutes at 175 and 5 minutes at 350. So worst case, you can get 2 hours of additional range in 15 minutes at 175 kW and in the latter case, 7.5 minutes at 350 kW.
175 is good enough for most cars that will be built in the next couple years.

It’s not quite that simple. 175 kW chargers typically seem to have a max current of 350 A (though Fastned claims 375 A for theirs), at a max voltage of 500 V. While that comes out to a 175 kW theoretical maximum, that’s not actually achievable in practice. A typical 350 V nominal EV (~400 V when full) will only get something in the range of 125 kW out of that (or a tick more on Fastned) — the exact value depending on where tapering sets in. (That’s the power delivered to the vehicle. The actual rate at which energy is stored is somewhat lower, due to heat losses.) The few existing 400 V (nominal) EVs could take a bit more; and a hypothetical ~435 V nominal EV (for 500 V full) yet a bit more — but still less than the theoretical 175 kW. 350 kW charges on the other hand seem to go up to 500 A. (Though some might only have 400 A?… Hard to find definitive information on that.) 500 A allows actually achieving something like 175 kW with a 350 V EV — though higher voltage is still required to get more than that. (An… Read more »

This question is exactly what all drivers will need to know in the future when it comes to explaining and using charging options. Words like “fast”, “super” or “hyper” are too similar and don’t translate well to all languages. If anything it’s confusing.

A solution like http://www.chargeway.net can help a lot. It captures the various ranges of kW into sequential levels that can be more easily explained and used for both station power output and vehicle charging capability.

Average people will not learn to understand kW quickly, but they fully understand that “6” is greater than “5”. The higher the number, the more power you can get. And for plug type differences Chargeway uses different colors.

We need a solution like this to make it easier for non early adopters to understand if EVs are truly going to compete for larger market share.

Why do you think that people won’t understand that 6kW is greater than 5kW?!
E.g. electrical devices also for domestic use (especially heating devices) must have a label with their max. power to not exceed the value of the circuit breaker.

There’s also the following problem:
1phase: 11kW becomes 3.6kW, 22kW becomes 4.6kW due to max. 20A unbalanced load (at least on the German market) if the charger doesn’t limit to 16A then .
2phase: 11kW becomes 7.2kW, 22kW becomes 9.2kW (see above but never saw an EV with 2phase charger and >16A).
3phase: Full 11 resp. 22kW.
But in all cases the EVs can also have a limitation to e.g. 16 or 32A (additionally to the 20A unbalanced load limit).

Luckily they obviously didn’t made the same mistake as AWG or the European efficency label. At least in last case, A was the best an F or G the worst. But then there were devices which are better than A, because A didn’t even meant 100% efficiency -> A+, A++, A+++, … Some years later, they changed the criteria so old and new labels are not comparable… 🙁


While most people might have trouble figuring out the real charge rate from confusing voltage and amperage limits, I’m sure anyone who finished grade school should be able to understand the general relation between kW and charge speed well enough… Surely better than from some opaque and arbitrary “classes”.

Frankly, all these labels are stupid, and attempt to dumb down the actual relevant info. Sorry, but it doesn’t work.
Even the L1/L2/L3 labels cause more harm than good.
You should refer to the charger by electricity type (AC or DC), actual numeric max rate in kW, and protocol/interface supported.

Spaceballs provides the ideal answer, already used elsewhere by Tesla.

350 kW and beyond is PLAID.

This guy got CCS working on his Model 3. Got better charging speeds than Tesla Supercharger even. However, it wasn’t 100% successful.

That’s a long video, want to give a hint where he gets to a non-Tesla CCS station? I doubt that Tesla uses 15118 protocols at their own SuperChargers, even if they now use the 8 mm pins.

The first non-Tesla station had some signs up, maybe “out of service” at about 6:03 (can anyone read German?). So that is not fair. There were several Tesla stations, apparently they all work? One was quite slow until the Tesla next to him left and he restarted and got a better charge rate. (The auto-translated captions are not much help, I could be wrong.) At 10:30 he finally finds a non-Tesla DC charger that he gets to work after a couple tries, looks like 120 kW into the car with a 350 kW capable station. Apparently it was quite noisy. At 17:30 he finds on also labeled Hyper Charge and the screen looks like the one in Bjorn’s video, but he cannot get it to work even though it has tow CCS connectors? He always has trouble getting the connector out but this time he has to use the emergency release cord in the truck, not sure why.

Shhheeeit !!!

Lol. I love Bjørn videos

The German magazine Auto, Motor und Sport also had similar problems: https://www.auto-motor-und-sport.de/fahrbericht/tesla-model-3-performance-2019-test-1/
At first Ionity said like: You’re the first trying a Model 3 at our chargers. Yes, charging is possible but just 24kW max.
After charging aborted multiple times (probably communication problems between the Model 3 and Ionity), 74kW came out. But at 69% SoC it aborted again.


The VOX Auto Mobil team had a similar issue with their Model 3. It would charge for 10-15m on a fastcharger and then stop. They made a kind of funny musical chair session then moving the car from charger to charger every 15 minutes.

Tesla MUST update it’s vehicle charging stack in the firmware to the latest revision of ISO15118.

According to a little birdie.. they are still using DIN SPEC 70121:2014 , ISO/IEC 15118-2:2014 ED1 (<350 kW) and ISO/IEC 15118-3:2015 ED1 (<350 kW).

The latest revision is available openly at CharinEV.org

Why “must” Tesla update? Most CCS stations in the US are DIN 70121 and except for Electrify America, they might well all remain DIN 70121. The US may never use OCCP either, despite what Europe wants. It is a different model for use cases, the US preference is for billing to go with the driver, not the car. Europe is not (yet) Tesla’s primary focus. China is the largest market for EVs.

AIUI the DIN one only works for <80 kW? So it's not a question.

As for OCCP, I'm not hearing people complaining about Tesla's model… I hear people complaining about having to use different payment approaches with other chargers. I'm pretty sure OCCP will catch on in the US just fine.

future revisions are backwards compatible. Totally up to US Preference if it wants to stsy that way, but the rest of the world wants progress. It’s like saying we made an iPhone and we’re not going to provide firmware updates.

he s probably right about the handshake error, which is fine, these things better be done properly, 175kw is a lot of energy to manage safely.

74 kW sounds like it used only CSS 1.0 (200 A limit), rather than CSS 2.0 needed for full power.

Seems like everyone developing charging control protocols have forgotten some basic 60+ year old engineering principles which would eliminate all these problems.

Current loop signaling is far more reliable over variable distances than voltage based signaling like the CAN bus. And it is obvious that differential signaling over twisted pairs is more noise immune than an unbalanced single wire signal. And there is no need for high speed data to perform some simple identity and control commands, low speed is less error prone and more than sufficient for the purpose. And suggesting the use of disassociated wireless links for a dedicated critical power cable connection is just asking for identification trouble and validation delays.

Sadly the current trend among young engineers to use an overly complicated solution for a simple problem inevitably leads to reliability issues and higher costs. Proven knowledge should be leveraged, not ignored.

“Current loop signaling is far more reliable over variable distances than voltage based signaling like the CAN bus.”
But you will have more crosstalk and it consumes much more energy. Ok, last is probably more interesting for devices with like 6Wh battery than with 16-100kWh battery…


A 20mA current loop control does not consume any significant energy. Nor is it subject to crosstalk. Proven in industrial control applications for decades.

Firmware update and it will be okay