When I got home tonight the KONNWEI dongle I ordered on August the 29th had arrived directly from China.

I ordered it here. Less than 9 euros.

It had the proper 120 ohm resistance on the CANbus. Next, I opened it up and it is the new design, i.o.w., it seems like the error was found and new production runs are OK. Of course we don’t know how many old supplies there are and if by chance I got the top of a just replenished stack. YMMV.

Note in the previous post that the problem started occurring already in April. I didn’t know this but at least there is a fair chance the problem dongles are sort of slowly going out of stock.

Short glossary: We have no ties with KONNWEI. Actually KONNWEI902s are knock offs of real ELM327 based hardware, but they have been the only consistently stable ones we have found. And we haven’t spotted “knockoffs of KONNWEI knockoffs”. Yet.

It seems the manufacturer has cost-optimized the hardware, which is not bad at all. However, as said in the previous post, there is a problem with one component. Recent deliveries have not worked and actually I heard of a few drivers getting weird error messages in the dash. Not fun.

Long story short: we would gladly recommend something that works, but we have no decent alternative. The solution at this moment unfortunately is tinkering. It’s not hard to do and you don’t need fancy knowledge or tools. Just a fitting hex tool and a sharp exacto style knife. I did this on a non working dongle and can confirm it works.

  • remove the 4 screws
  • gently pry open the case. Be careful that the connector and the front stay in the half that has the main PCB mounted on
  • check this post to determine if you have the old or the new design. Unfortunately there is no way to check this from the outside *)
  • if the design is “old”, stop here. If your dongle does not work there is another problem
  • if the design is new, locate the chip with 8 pins roughly in the middle of the PCB
  • just above and between the 2nd and 3rd pin of that chip, there is a tiny black component (resistor), see picture below
  • carefully cut away the middle part of the resistor. Ensure you do not damage the track going up on the right side. Damaging the track on the left, going to the second pin is not a big deal. Of course if you have soldering skill, you can also de-solder the resistor.
  • close the case and put the bolts back in. Note that one side has hex formed holes, these are for the nuts.

Done!

I drove with a modified dongle today without any problems. The only issue I had were more than usual ATMA timeouts and I had the impression it was all a bit slower than what I am used to. The former I need to investigate a bit further, it can probably be fixed with a modification in CanZE, the latter might be subjective, or maybe the processor is a bit slower in the optimized design.

We fully expect KONNWEI to fix the problem in the long run, and it seems there are already new designs shipped with the new resistor. part of current stock seems to be troublesome though.

*) If you have a multimeter there is actually an external way to check the dongle. Measure the resistance between pin 6 and 14. If it’s 20 ohms, you have a bad dongle. If it’s 120 ohms, it’s fine. BTW pin counting is from right (1) to left (8), top row, meaning in the wider part of the connector first, then the bottom row, again right (9) to left (16) in the smaller part.

EDIT: The leafspy guys reported on this last April. Oops! Oh well! More pictures there too BTW.

Today I received an email stating that some KONNWEI’s are shipped with a 20 ohm resistor across the CANbus. This is a massive production fault. First of all, normal bus termination is 120 ohms, second, the dongle should not impose a new termination resistance at all.

Now, I am not saying this is the root problem of the recent KONNWEI issues. I need to do tests, and I will try to do them tonight. It is still possible there are additional firmware issues, and in fact, earlier measurements indicate there may be indeed. This resistance problem might cause problems though with some ECUs and not with others, making everything transient and hard to tackle. I am extremely grateful to Amazon user for finding this error and SpeakEV member cDy for relentlessly pursuing the issue and following tons of reviews.

Stay tuned. If this is indeed the main problem, at least there is a path to a solution.

Latest KONNWEI guts with faulty resistor. Flat cable and crystal on the PCB
Old KONNWEI design with the wiring loom and the floating crystal

These picture Harm took for us “on request” and are all taken from under the car looking up with the cover screen and battery removed.

Battery cavity. Bottom of picture is front side of the car. Note the orange HV power cable, black control cable, shiny pipes running from the A/C to the battery evaporator, black brake lines and the cutout on the right side where the HV cutoff will go into the cabin.

 

Other end of “the cavity. Notice the extra space for “the chimney”, the three holes for cooling air (middle for air going into the battery, others for outflow), same A/C and brake pipes and the bowden cables for the hand brakes.

 

Same area but taken a bit front-to-back. Here you can see the very flat air ducts going from the top and over “the chimney” to the fan/evaporator unit behind it. The coolant pipes hop over the chimney too.

 

Front side of the battery, showing the HV receptacle and the control receptacle. The former is connected to the actual battery through a beefy relay, current sensor, 275 amp fuse and the HV disconnect. The latter basically contains just a 12 volt bus to run the LBC computer inside the battery and the Electro CANbus.

 

Underside of the motor peeking from the front side of the car. It can be divided in three parts. Left side is electrical inlet (you can just see the slip ring connector) and the modest water cooling in/outlet. Middle is the actual motor. Right side is the 9.8 reduction gear and differential. Top shows some A/C pipes. Bottom shows the axles to the front wheels.

 

Dutch ZOE driver and enthusiast Harm Otten was #3 in The Netherlands to have the battery of his Q210 upgraded to the ZE40 type. He was allowed to take pictures of the procedure done only yesterday and I want to thank him and the Arend Auto Eindhoven dealership for sharing them in public.

Battery as it arrived

 

Packaging opened. Note the HV disconnect plug that ends up under the foot well of the right front seat

 

Lifting out the battery using an engine lift. Note the “chimney”. More on that later.

 

Rolling it to the car on the lifting table

 

Lifting it in “the cavity”

 

Bolting it against the chassis with 8 bolts

 

New decoration 🙂

 

Done!

A few days ago I tested that dongle I received and things were not looking good. Flow control commands are not accepted, and while I assured it is able to both send and receive data on the CANbus *) it does not like the replies to diagnostic commands the car puts on the bus.

I opened it up and the hardware is definitely different. A better build in my view, but the firmware (reporting the same version BTW, lies lies lies) does not work.

The very cheap but very well build “HH” dongle I bought was total crap. Not only did it not work, the firmware was easily send to lala-land.

IN the mean time, as my own KONNWEI died on me (I abused it a lot), I ordered a new one. If it works, I will post a link to the item here. If not, we need to explore a few new paths:
– see if we can work around the issues in software (might prove impossible or very hard);
– hand pick dongles that do work (complicated to manage, and who can you trust sending exactly that device);
– build our own hardware (out of reach for 99.9% of the users, and we are certainly not going to market a new design);
– focus on the Freematics ( http://canze.fisch.lu/freematics/ ) which can run firmware we can make (relatively expensive).

All in all this really stinks and we are NOT happy.

This is an intermediate post. I got a report from a user who ordered what seemed to be a bona fide KONNWEI, but had no success. That was slightly alarming, but one error could be anything. He send me the dongle that arrived today. *)

I did a quick test, only to discover several fields are not picked up. Went back inside to check the version and ATZ reported a neat 1.5.

Now I am worried that we have no decent advice anymore and the dongles are truly hit-or-miss. Either there are bootleg KONNWEI’s on the market, or they did something to their dongles breaking CanZE. I will investigate further, and maybe I can make this thing behave. Stay tuned.

 

*) Sorry about confusion I caused earlier with this post. I had mixed up this dongle with one that is still on order. I will report on that one separately, especially if it works.

There are two ways to reset the trip part, which makes the car “forget” driving and battery behavior.

Reset “light” is through R-Link. This will reset the range indicator (GOM) to a value proportionally  related to the capacity of the battery. In effect it forgets how (un)economical you have driven. It was used in the old days to check how healthy the battery was.

There is also a harder reset, called the “two pedal reset”. Keep the driver door open, start the car, keep the gear in N, press and hold both pedals and press up key on the windshield wiper control until the averages message appears in the display. Now hold that button. First it displays the current values, then it starts flashing, then the average values display —. Now let go. This reset seems to also make the battery computer (LBC) loose quite a bit of acquired parameters and the GOM shoots up to a crazy value.

The car will rather quickly re-learn the state of affairs. Just don’t rely on the range indicator on your first trip after a reset.

Thanks to the kind people running the OVMS project, we have access to quite a bit of documentation on the Twizy. Unfortunately, the current developers are unable to implement this knowledge into CanZE, mainly because lack of Twizy to play with :-). If developing CanZE taught us one thing, it is that it is virtually impossible to add functions without the car available; remote testing simply doesn’t cut it.

So, if any volunteer is available, we would be happy to take him or her into the team. There is no hard prerequisite (hint: when I started I had not written a single Android program), but it would be fair to say the following would help, in diminishing order of importance:

  • Owns a Twizy, ELM dongle and an Android device, and is willing to play and experiment with it;
  • Can bear to stumble along and spend too much free time;
  • Has done some development in Java or C++;
  • Has done some Android development.

We would love to hear from you!

This post is more for reference. In the R models there is no BCB and no PEB. These functions (power distribution, inverters, motor controller) are all integrated in one huge box called the PEC. Inside are several modules doing the hard work. Here are a few pictures of the PEC,the R240 motor, and the complete assembly.

PEC

 

R240 / R90 motor
R motor / PEC assembly