Kewet Buddy City-Jet 6 Restoration

[matzetronics]

Gee, i really wonder if that was made by Elbil Norge or if it was another 'modification'
In fact i cannot imagine that the manufacturer made those bad connections, and i'm certainly glad you have found those now and not later being the cause of a traffic jam.

I inherited not only both of my El-Jet 4s from an old friend of mine after his death and the subsequent dissolving of his enterprise but also a very good crimping tool, the Weitkowitz PW 6/70:

PW 6/70 - WEITKOWITZ Kabelschuhe und Werkzeuge GmbH

www.weitkowitz.de

Since then i have used it on electric propulsion systems for boats (sometimes i work on a shipping yard) and of course for my Kewets. In the Kewets i use 35mm² cable and the appropriate cable lugs, mostly tin coated copper.
For your purpose it is more than sufficient to borrow a good crimping tool. Remember to first put a shrinking tube over the cable which you can slide over the connection and shrink once its crimped.

 

[Stefanhg]

Gee, i really wonder if that was made by Elbil Norge or if it was another 'modification'

I do be worried it is made by manufacturer. The crimps are not the best quality crimps I've seen. One thing to note here is those crimps are in tight spots and the angle of the cables may result in stress on the crimps, so if they are not crimped completely it can loosen up over time.

Ah that crimp tool looks amazing! Gotta enjoy whe nyou get your hands on a proper good tool.

Remember to first put a shrinking tube over the cable

I sadly forgot it when i made the cable, but i used electrical tape on the cable and it ended up looking good, but surely prefer shrinking tube, specially since electrical tape just ends up being loose at some point.

 

[Stefanhg]

First mod - warning when light is on when ignition is off
I was testing my car a few days ago and realised that I forgot to turn off my lights on the car, with the big batteries it is not a big deal but burns the bulbs and is just a waste of energy.
I thought it would be a good idea to add a buzzer to warn me if i forget to turn off the lights.
I hoped I could just find some voltages on the fuses, connect a buzzer and be done.


Lucky me there was two fuses to use. The black wire is the +12V when car key is turned
The red wire is the +12V going to the lighting system when you turn on the light switch in the dashboard.
How it works is when you have car key turned there is +12V on the fuse and when the light is on there is +12V
but when the key is off and the light is on there is 0V on the key ignition and +12V on the light switch, and then the buzzer makes a sound.
Notice that the buzzer is directional, so If i have +12V on ignition and light is off, it will not make a sound.

So just some quick info on how it works.
Because the buzzer is so low current, you can use whatever current path there is in the system as a source. My buzzer has a 10K resistor in series to reduce the current to 1.2mA which is nothing. The buzzer is extremely loud at full power so needed to reduce it.


Then i soldered two wires on it, crimped two spade connectors on, plugged it in, jammed it behind the "glove box" and BAM it works.
quick and easy mod.

 

[Stefanhg]

I have been a bit inactive but I am finally at it again!

This time I looked at the interior lighting which was not working when opening the doors.
This was rather obvious where to work since the contacts were hard to push in so they were rusted. I took out the contact and saw it looked very much like the one you could buy in Biltema, the "door contact" Art. 43-011

I bought two of these and they did not fit completely and required a little bit of work. I both grinded a bit of the hole in the fiber body of the car and also a bit of the contact itself as you can see here

Left is the unmodified, right is the modified.

After that it was just about installing them.

I used a different screw here because the old had the mounting bracket completely rusted so i used a bolt instead.
Then lucky enough the new contact came with a spade crimp so I could connect it directly to the original wire going to the indoor lamp and then I could also use the original ring terminal used to connect to chassis/ground.

I also added some moisture protection to the wire connected to chassis because it previously was rusted a lot so to protect it this hopefully helps.
Art. 36-3014 in Biltema if anyone wondered what I put on the wire.

On top of this, while checking wiring I actually found that unknown wire i couldn't figure out here

Some day I'll take the panel off and figure out where that 3 wire cable goes to.

This was connected to the interior light. One wire was the signal from the reed relay that was connected to the trunk with no magnet going to the negative side of the light, the other two was the constant +12V and ground. I can't remember how it was wired, but I assume the relay was for providing a proper path for turning on the light since the bulb would draw more current than the reed relay could deliver.

I also took a look at the contact at the handbrake with by the way is horribly made because if you lift the handbrake too high, the position of the arm pressing onto the limit switch could move causing it to only enable at that position. I initially wanted to 3D print a proper solution but after a few tries gave up and fixed the arm by putting on a limit switch with a longer arm on itself, so it can handle being pushed in without damaging or moving the switch "trigger" point.

So now I should have covered most of the electrical part.

 

[Stefanhg]

Reverse engineering the Kewet Displayboard

Been wondering how to interface the display board as a fun project I might make my own car controller as it is relatively simple and I would love to make one that supports CanBus to interface to for example a JK BMS.

The display board is relatively simple or so I thought until I started reading the datasheet for the SAA1064.
Contains 8 ICS. 5x SAA1064 LED drivers, 1x24U02 EEPROM, a 74HC04 hex inverter and last an TLC271C opamp

Pinout:

Light sensor
The light sensor output is connected directly from the opamp to the connector, the opamp receives what I assume to be the VREF from the CarController board.
It is directly put into pin 3 (non-inverting input) of opamp. Then most of it I assume to be standard light sensor stuff.


EEPROM
I'll cover that later, needs to create a quick program for dumping it.

NOW to the fun part!
Display(s)

Trip button
Trip button connects to the pin 4 of the hex inverter, outputs on pin 5 and directly to the connector. Pullup and debounce capacitor + series resistor between button and capacitor to limit current when button shorts the charged capacitor is the parts surrounding the button.

So while reading the datasheet for the SAA1064 I realised some magic was going on. The IC itself is I2C based meaning it got a bus i can put multiple devices on parallel, and each device got a preassigned address. This turned out to make some fun stuff happen. I realised that the IC only got 4 selectable addresses by applying a voltage on a pin of the IC, but here we got 5 devices so something was going on.
Of course first thing I do is confirm they all were parallel and of course they were.
I knew that hex inverter was gonna be playing a role here, else no reason to have a hex inverter with that many pins.

So first I got all the LEDs to light up, this was a struggle at first before i realised the first search result for library for Arduino had fixed address line even when half way implemented for the IC.
After then trying out coolSAA1064 library, which worked for some cases but the IC itself is designed for driving segment displays, so ChatGpt helped me generating a simple example for directly turning on all LEDs. BAM done!
So of course i started to control them individually, and realised that when driving specific addresses, it would control two leds every time. So now it was time for digging!

Reverse engineering the circuit

Time to pull out pin and paper.

After poking around, I finally figured out the weird resistor network and the application of it.
So there are some simple logic here.

They use the resistor network to alter the address of the two ICs but pulling them low. Interesting that the ICs allows this on-the-go!
They have conneted two hex inverted in series to first invert the signal and then back to input state, so basically just a fancy buffer. Only time they don't is when the trip button is pressed.
So the way they use this is first they write to IC5 by setting the input pin connected to IC6 low, that causes the diode to pull the voltage low on the pin, then the address of IC6 is changed, then they write to the IC6 and afterward do the same but for IC5.
If I made this, I would just have connected IC6 diode between the two hex inverts, thus saving me one pin on the header and all the additional complexity.

So after figuring out all this I can finally start writing a test application for the display!

 

[Stefanhg]

Ha! It is now working!
I (together with ChatGPT) made a sketch for Arduino that loops each LED individally.

Setup is rather simple. I2C is connected to the two top left pins, and the two top right is the "resistor network" which is the enable/disable of the two ICs address line.

And then in the bottom of the connector I connected +5V and GND, of course also GND to the arduino. I uploaded the code to github for those who may be interested in playing around:

GitHub - Stefanhg/kewet_displayboard: Contains information and code for controlling the kewet display board

Contains information and code for controlling the kewet display board - Stefanhg/kewet_displayboard

github.com

I'll have to clean it up later since I don't like the structure. But next goal is map everything so I can start setting each display to an number or set the KM/h meter and so forth.

 

[L.S.]

I think this wasn't mentioned before and I'm not quite sure if you are aware of this:

Prismatic LiFePo4 cells really need to be compressed correctly in their enclosure. This is important especially for the first few cycles of new cells. Otherwise (especially with higher current application) the capacity might not last long and they might age prematurely. I don't know about Envision, but many manufacturers require it in the datasheets. It is usually 3000N per cell or 300kgf.
If you're interested in why this is necessary or how to achieve it, there should be a lot of info available on the internet even though not everyone on there has understood it completely, so be selective...

Also, if you want to drive the car anytime of the year where it could possibly get close to 0°C in the night (I would even say everything below 10°C, in Germany this includes most of spring and fall as well), you definitely should fit a well dimensioned battery heating System and correctly configured charging cutoff logic. If LiFePo cells get charged close to 0°C (and of course below), they will already start to age prematurely (meanwhile discharging is not as harmful).

 

[Stefanhg]

I think this wasn't mentioned before and I'm not quite sure if you are aware of this:

Prismatic LiFePo4 cells really need to be compressed correctly in their enclosure. This is important especially for the first few cycles of new cells. Otherwise (especially with higher current application) the capacity might not last long and they might age prematurely. I don't know about Envision, but many manufacturers require it in the datasheets. It is usually 3000N per cell or 300kgf.
If you're interested in why this is necessary or how to achieve it, there should be a lot of info available on the internet even though not everyone on there has understood it completely, so be selective...

Also, if you want to drive the car anytime of the year where it could possibly get close to 0°C in the night (I would even say everything below 10°C, in Germany this includes most of spring and fall as well), you definitely should fit a well dimensioned battery heating System and correctly configured charging cutoff logic. If LiFePo cells get charged close to 0°C (and of course below), they will already start to age prematurely (meanwhile discharging is not as harmful).

Thank you for mentioning this! I can see there are many different approaches to this. Everything between just taping them together and actually crewing things together. This surely is a mess in my case to do but Ill have to look into if i can in any way make this possible.

About the charging part. Initially one of the advantages of my plastic enclosure solutions is that I can add a silicon heated mat under the batteries for that. Sadly the BMS i bought didn't have the feature for this, I wish I knew there was many other models so I may consider upgrading it to the BMS with automatic heating when charging! But thank you for mentioning it for others! I have done a little fiddling with the voltage but until now I have only monitored things manually and turned of charging when i felt it got too close to the cell voltages.

 

[L.S.]

Your BMS doesn't necessarily need to have a specific "heating" feature (for safe measure, I wouldn't want to directly use that internal relay for the heating supply anyway. Don't these have a rating of only 2 or 3A?). The BMS just needs to disable charging if the Temperature is too low.


The control of the heating pad can easily be realized with a very simple circuit of cheap standard electronic parts (thermal switches and a relay).

 

[tewek]

Ha! It is now working!

Following this with great interest
Had the same thought when I bought the car and realised the car controller is a piece of junk nobody seems to be able to fix. When a Arduino could do the same.
Retaining the orig. Display would be great.

 

[Stefanhg]

Your BMS doesn't necessarily need to have a specific "heating" feature

I know. But what I want to achieve is avoid human errors. I would hate having to check up on my charging batteries constantly.

Don't these have a rating of only 2 or 3A?)

Well in the BMS i looked at they are mosfet based. So being able to throw out 10A is no problem. Also 1A should still be PLENTY of heating at 72V!

The control of the heating pad can easily be realized with a very simple circuit of cheap standard electronic parts (thermal switches and a relay).

True, but there still is some logic needed to handle it like to detect charging.

Prismatic LiFePo4 cells really need to be compressed correctly in their enclosure.

I took a little time to sit down and understand it. As you said it is the first few cycles as the material inside starts producing gasses meaning after the initial cycles they operate like normal. They will still swell internally. But we are talking going from 2500 cycles to 4000 cycles WHICH is a lot. But charging the car every 3rd day means 20 years lifetime AND we are still only at 80% capacity at that point so compression does not really matter THAT much but of course it has other advantages like.. Batteries being fixed properly together.

 

[matzetronics]

Following this with great interest
Had the same thought when I bought the car and realised the car controller is a piece of junk nobody seems to be able to fix. When a Arduino could do the same.
Retaining the orig. Display would be great.

You know very well that this is not the case and your car controller was mishandled with water in your car. You never tried to contact me, and i'm of course able to fix it.

 

[tewek]

(Without wanting to spoil the thread here)

My cc was faulty from day 1. Even before there was corrosion on the PCB (and there is no proof it has anything to do with it) it was nothing but trouble. Sometimes it would work flawless, but most of the times it showed wrong values. If it did work at all.

- wrong amp readings
- trip reading more than the odo
- cutting the car out randomly
- trip and odo randomly showed either zero or some Phantasy readings
- not able to shift in reverse
Etc. Etc.

And of course the capacity meter is absolutely useless.

Anyway, I am happy and reliable driving the car without this junk.

There is a reason old computers are worthless.
Even if I had a 120 years left I wouldn't waste a minute on this thing
A few bucks on China hardware make more than up for it.


*We were in contact if you remember...

 

[matzetronics]

*We were in contact if you remember...

Schick mir den Controller, ich repariere ihn dir.
Or in accordance with the language used here:
Send it, i'll fix it

 

[Stefanhg]

I haven't written much in a while but the car has passed inspection!
I had the known issue of getting the correct tie rod's and I actually had the car professionally adjusted and they JUST barely managed to adjust the wheels, the steering wheel is currently a few degrees off but It is fiiiine.
So plates on it now only to drive it a few days before weather got too cold.

I noticed that I struggled with turning on the diesel heater while driving so I need to figure out why, Of course i assume it is related to wind but it seems like if it fails after a while i have to completely turn off the car batteries to fix it.

Then also the cabine getting very cold when driving fast, seems like there is a lot of air getting inside the car so I have to figure that out too.
Lastly a very big problem. While driving I went shopping where i turned off the HV but not +12V battery, when i came back i forgot to turn on HV and then the car acted weird. Relay for high beam did not work and some other things too(didnt write details down). I tried turning off everything including +12V without it working. After the careful drive the car sat for a few minutes and somehow it works again.

So car is now sitting and waiting for better weather in storage but I still plan on working on it.
Currently I have the following plans
- Things mentioned in this post
- CarContoller V3(V4?) with minimal functionality including using BMS for battery management (already in progress)
- Control BMS output with ignition
- Look into issue where sometimes the MotorContorller does not change driving direction when pressing the button
- Isolation from road noise
- Look into low power electric heating solution, mainly for defogging
- Figure out the differential making knocking sounds when changing direction or driving from full stop.

I am also seeking feedback for the new CarController design, specially things like stupid faults that should be resolved. Already added a few fuses here and there including the +72V ground and +72V itself - I wonder why!
I will make a separate thread for this since it is not just related to my car.

But the car is working! Road legal and so fun to drive!

 

[Stefanhg]

- wrong amp readings
- trip reading more than the odo
- cutting the car out randomly
- trip and odo randomly showed either zero or some Phantasy readings
- not able to shift in reverse
Etc. Etc.

Now all this really indicates a bad ground imo.
Specially the wrong amp reading. Since the shunt is on the ground of battery, wrong readings indicate you have current going all over the place instead of only current for the motor is shown.

The CarController I use for reverse engineering right now Seems to have either a bad ground or somebody did a stupid and reversed voltages. Many of the pins where ground is has been exploded(literally), relays very dark, and the resistors connected to the shunt + and - are blown up(connects to ground on PCB on other side)
So one bad ground causes bunch of issues. I think one big flaw of the car's design is how there seems to be no isolation between HV and LV ground which already fails from the input jacks since there are EMI filters connecting to the chassis which many circuits are.

 

[Stefanhg]

Been a while since I have made a post, but I have been using the car a lot!
Sadly recently the car had a day with troubles..
I tried repairing the issue as I knew it had a bad connection on POT HIGH out on the curtis but it only fixed it shortly.

Sadly this didn't work and I had to get the car towed.

Now with the car home, i ended up figuring out the curtis had blown its logic and relay fuse which seemed odd to me thus I didn't catch a bad measurement when i initially measured it.
SO what to do next? I don't really condone the concept of replacing fuses unless i can reason it so I pulled out the whole curtis and threw it on my table.
Actually way more simple than i initially thought and was healthy to get an idea of the internal construction of it.

The controller PCB is on the top and then on the bottom there are more power electronics.
The model is a Curtis 1274-7401

From the looks of it there is a precharge circuit on the bottom, a 10R Dale power resistor, transistor and then I am unsure what drives it, what it looks like is this does not start powering up the secoundary side unless some condictions aremet.

Some more pictures can be seen attached
So what is the fault condictions?
High inrush currents reaching above 1A, it does latch the HV relay.
What i found was the diode D4 was shorted, which actually connects B+ and KSI pin together meaning that the KSI sees the full set of capacitors on the board causing the inrush current.
I partly reverese engineered the circuit to understand what was going on:

As you can see, D4 shorted connects B+ and KSI, when it is working what D4 does is power the logic from the B+ instead of KSI pin.
Then there is the transistors which i think is a current limited transistor used for pre-charge of B+ through D33.

So after replacing D4 with a beefy whatever-i-had-laying-around diode It powered up just like before but this time it did not latch the HV relay and no voltage on B+!
FOR those following along; D4 is the MELF diode placed closest to the board-to-board connector.

So now It was missing important things as i did not have a motor connected.
After throwing it back in the car I was left with a new error code I never have seen before 5-2 which is KSI voltage problem. BUT it was not enough to stop the car from driving!
I actually can see that, as if i measure KSI i measure 71.3V and my B+ is at 72V. Yeah it is .7V but that means roughly 3.5 ohms of resistance in the wiring.

Sadly I still don't have the diagrams for the citi-jet 6 so I am a bit screwed in terms of troubleshooting, but I do have some guessing on where it comes from. I have measured the resistance of the wires in the speeder switch and the 72V relay and those are good.
Actually while writing, I think the +72V coming in splits out on the relay and I actually remember measuring a rather high resistance there (2-3 ohms) so that may be the area. Need to check that out tomorrow!

SO; I said I did not throw a fuse in unless i can give a good reason:
Due to D4 being shorted, the massive bank of ~40x330uF Nichicon PJ capacitors was directly connected
With an impedance of 0.062 ohms each and facing 13.200uF and roughly 0.00155 ohms impedance total
an theorical ESR limited inrush is 64A, YES it is an short pulse but it is a lot of energy it needs to sustain and i would suspect over time this would have damaged it.

 

[matzetronics]

I don't know the 1274 Controller, but every other Curtis requires a precharge Resistor across the main relay to prevent abrupt charging of the elektrolytes in the controller when switched on. The manual should mention this resistor in the example circuit diagram.
The same is true for the BMS. If you switch off the discharging MOSFets and switch them on again, excessive current flows into the controller and into the DC/DC. On the old El-Jet 4 this would often blow a MOSFet on the DC/DC converter, something which hopefully doesn't happen with your Brusa model. I bridged my ANT BMS with a 1k/15W resistor to reduce the precharge current.

Considering the state of crimping you showed a few posts ago, i'd inspect all power cables for bad contacts - and probably you already have found one with this high resistance.

actually remember measuring a rather high resistance there (2-3 ohms) so that may be the area.

Make sure the main relay doesn't develop heat after a few kilometers of driving and your battery connections don't do that ether. That would point to unwanted bad contacts. And the motor cables surely as they are exposed to weather.
Good luck

 

[Stefanhg]

but every other Curtis requires a precharge Resistor across the main relay to prevent abrupt charging of the elektrolytes in the controller when switched on

I do believe mine also needs it, but for some reason there is none on mine.
Planing on buying one just to be sure.

flows into the controller and into the DC/DC.

Never actually thought much on the DC-DC converter needing this but a very valid point you have! I always turn off my HV so it is always being switched on.

i'd inspect all power cables for bad contacts

Have been through a lot of them already since many connections are bad. Specially the wires under the hood, for some reason many are completely corroded 2-3 meters up the wire.

Make sure the main relay doesn't develop heat after a few kilometers of driving and your battery connections don't do that ether

I actually figured out it was a bad measurement. I made a USB wire so I could check the curtis and it actually was a 2-5 error code which is the brake. I went and reverse engineered that too and it turns out they disconnect the regen potentiometer when brake pedal is not pressed so the input for regen signal is floating! So error disappears when i press my foot pedal so all error codes are gone now!
I am planning on properly making the regen part later on but for now I understand how it works and can accept it.
And about the relay, Mine has some damage in the contacts and needs replacement. I found an replacement kit for the relay contacts. They sell a full contact kit which i am planning on buying
2180-42 Albright SW180 Series Contact Kit

 

[matzetronics]

If you plan regen, always think about not overloading the current flowing through the brushes of the motor. Keep it moderate as the brushes are much more expensive and hard to change than a few kilometers more range.