[arber333]

A bit like the Tesla small drive unit. Two small two pin connectors unobtainable so I desolder them from the Tesla board. Watching with interest. You guys are saving me a job

I am 3/4 of a Job into drawing of Lebowski design where i changed my mind. I will try to make 3x sockets for SMD version from Andy. It is really smallish board, only a bit larger than Olimex board. I would have three of them.
Also i would have another chip to controll perifery. Maybe Pic16F or Pic18F.
I need control of Inputs: Voltage sensing, Fault signal, Start signal, BMS signal... and Outputs: Precharge, DC contactor, Fault light, Voltage light.

Do any of you have some design on that? It would save time greatly. Specially if the firmware is allready written. I am no programmer.

A

 

[Tomdb]

Arber,

I am taking the same approach, I will design a 'generic' controller board that can drop onto an interface board. Or be connected using a ribbon cable, depends on what socket/header you solder onto the board.

Almost done with the generic board schematic, only the power supply is left.
I am thinking of using all the same headers for now so the total bom cost should be cheap to knock out the Lebowski boards.

What sort of functions are you looking for? Possibly a Teensy could form the extra control brain, however this might be overkill for a simple application.

However, I have more interesting projects in mind since the volt inverter contains two quite powerful stages that can drive two motors.

 

[Nuts&Volts]

It would be nice if that board had a resolver to encoder circuit/chip built in.

From the information I have seen the Volt IGBTs are 600A peak units.

I’m planning to split the Leaf motor phases and run each with a powerstage of the volt controller. This would require two Lebowski brain boards operating in a master/slave arrangement. Other option is one brain board that operates both powerstages with outputs parallled and use the motor as a standard 3phase machine. Either way I wouldnt be asking for more than 300-400kW peak to give some buffer. 500/1000A peaks (3-5secs). This is the lonnnnnng term goal.

And two Volt packs should be plenty of battery to do that I would think.



Sent from my iPhone using Tapatalk

 

[arber333]

It would be nice if that board had a resolver to encoder circuit/chip built in.

From the information I have seen the Volt IGBTs are 600A peak units.

I’m planning to split the Leaf motor phases and run each with a powerstage of the volt controller. This would require two Lebowski brain boards operating in a master/slave arrangement. Other option is one brain board that operates both powerstages with outputs parallled and use the motor as a standard 3phase machine. Either way I wouldnt be asking for more than 300-400kW peak to give some buffer. 500/1000A peaks (3-5secs). This is the lonnnnnng term goal.

And two Volt packs should be plenty of battery to do that I would think.



Sent from my iPhone using Tapatalk

No Ampera IGBTs are 500A 650V units. They are capable of 1000A for 1ms.
I stand corrected, IGBTs are Hitachi MBB650SV6A with 650A 650V rating. They should be capable of 1000A for 1ms. I think that is quite enough for normal BLDC motor.

I wouldnt recommend to use two brain boards with paralel inverters and a single engine, because of problems with sync. You do know that there is a signle capacitor and power line inside?
I would recommend to use two motors each on its own axis. However if you absolutely want paralel inverters then use a single brain to drive transistors trough an amplifier circuit. Connect phases in paralel before the cabels come to the motor. That way you dont get transient loops.

I am also in process of drawing three brained interface board. Third will drive AC compressor.

I am trying to use Arduino Nano to use I/O signals. Everything will be digital, except maybe Voltage sensing and Evaporator logic.

I will use next I/O connections:
- 12V power
- GND
- Precharge signal out
- DC switch signal out
- Enable/start for all units, Arduino will decide which should start
- Reset for all units on the outside + each own reset trough driver Fault line
- Reverse signal for main drive unit
- Throttle signals 2x
- Motor Hall signals UVW 3x

- A/C unit on
- Evaporator temp could be checked by Arduino?
EDIT: I may not need Evap temp probe. My car has automatic AC so i just take the wire that used to turn on the compressor and... what do you know, enable the lebowski to turn on the compressor! DOH!

- MB generator on
- BMS input for the Arduino to control MB motor generator regen
- manual Throttle for regen trough Arduino
Probably we wont use UVW signals for generator since torque is more or less constant.

 

[arber333]

Arber,

I am taking the same approach, I will design a 'generic' controller board that can drop onto an interface board. Or be connected using a ribbon cable, depends on what socket/header you solder onto the board.

Almost done with the generic board schematic, only the power supply is left.
I am thinking of using all the same headers for now so the total bom cost should be cheap to knock out the Lebowski boards.

What sort of functions are you looking for? Possibly a Teensy could form the extra control brain, however this might be overkill for a simple application.

However, I have more interesting projects in mind since the volt inverter contains two quite powerful stages that can drive two motors.

Tom do you use voltage sensing with Lebowski? As i understand that would remove isolation. Bas does not recommend it. I will use arduino to sense preset voltage and trigger DC switch, power reduction or PWM stop etc...
How do you intend to observe temperature of IGBTs? Is there an algoritm with Lebowski brain or is it just digital safety? Are sensors inside IGBTs digital? That we could use then...

EDIT: Furthermore i propose that you test aux inverter with your remy motor. You could limit aux Lebowski to less than 3kW or 10A. I think you should test it with some load. I am sure that small inverter can survive that.
You could simply desolder the PCB to see the rating on the chip. But you would risk damaging the PCB.... In any case it was meant to be sacrified for knowledge.

A

 

[arber333]

Ok i sent interface board to manufacture.
I will be getting 5 pieces back.

Idea here is that i use Johannes master board to control precharge and DC contactor and of course the start pin. That way i dont need other controller chips in case i use only one motor.

However in case of multiple motors i prepared interface through Arduino that would control generator and compressor throttle action. That way one could have one inverter to drive primary motor and rest would be slaved to Arduino. They would be turned on by a switch from inside car.
AC would simply be controlled by internal AC controller that used to trigger compressor clutch. AC compressor than would work very much analog to old belt driven unit. I dont think actively varying RPM of compressor would bring any energy benefit, besides car already uses ON/OFF control. I just have to find how to fine tune it.

Idea is to ramp up RPM of generator/compressor to some 5000RPM in say 3s and keep constant torque with max at 75%. That would not stress mechanical parts too much and allow gasoline genset to follow electric machine. When at borderline conditions arduino would have a hysteresis that would eliminate fast on/off state change.

 

[Tomdb]

Very nice work, how much do you think it will cost get the pcb made?

Have you found a good source of the connectors?

 

[arber333]

Very nice work, how much do you think it will cost get the pcb made?

Have you found a good source of the connectors?

Not exactly .
It is a long wait to get those connectors from Mouser. I see a best bet to gut the existing master board and get them from there. It is not that inverter will need another brain board after my surgery .
But you have to be really carefull applying heat when removing connectors. They have a lot of gnd pins and you can burn yourself if you are not carefull.

EDIT: 5 boards were some 40€ with shipping, so lets say 10€ each board. I also ordered 5x Lebowski boards for about 5€ each. Those i made for DIP chips. It is simpler that way, 5V tolerance and supply...
Then there are parts to populate. I will try to populate all and see then how much it was. My design uses Johannes board as mandatory. Everything else can be omitted.
Later i can make another revision to use Arduno nano to observe precharge and actuate DC contactor and enable pins in case someone would want 3x Lebowski boards.

 

[arber333]

Surprise surprise!
I got my boards today and i tried to fit the main board to inverter. Like a glove!!! I would have to cutoff a little on the side so cables will fit better.

Also i have Johannes board almost assembled. I may remove 5V regulator from Johannes board and only use one on the main board. What do you think, could two 5V buck regulators coexist? They would be supplied from the same 12V source and will have 5V as well as GND in common. Also there is hefty capacitance near both.

I also received my processor boards for lebowski design. It is simple sensorless design with a possible use of UVW hall sensors reserved.

I now have to decide how to fit everything together. If i try flush fit i maybe can get away with original aluminum cover. This is not good for experimenting though, since i may have to remove boards to try another approach.
Probably would be better to use fitted headers. Board could be dissassembled, but it would be too high to install stock cover. I will try to get something 3D printed...

Arduino nano at the top will be a throttle controller for AC and can also run generator or second motor.

 

[Tomdb]

Arber, very nice work.

I am very glad I was able to help out get you to this stage. Now we got to find out if it actually works, but then again the pcb is practically a wiring loom

Lets hope you can find a source for the remaining connectors, because I think the dual socket connector will be a pain to remove.

Just a thought on the lid, create a simple aluminium spacer ring and if you are done testing remove headers to use the pins to solder to the main board.

 

[arber333]

Arber, very nice work.

I am very glad I was able to help out get you to this stage. Now we got to find out if it actually works, but then again the pcb is practically a wiring loom

Lets hope you can find a source for the remaining connectors, because I think the dual socket connector will be a pain to remove.

Just a thought on the lid, create a simple aluminium spacer ring and if you are done testing remove headers to use the pins to solder to the main board.

Yes i think there is no way around it, Delco main board will be destroyed when pulling connectors off. But in reality that is about everything we need from it. I will try to develop procedure to get connectors off the board relatively easy. Probably will include heatgun, but no open fire.

I intend to try also Lebowski brain in solo config. This is where Arduino will come in use. To enable precharge. I dont want to just put the precharge on timer. HV has to be verified otherwise no DC contactor and no inverter operation.

I am not sure if large discharge resistor will interfere with precharge. I will use 470R like before. Maybe 2 of them in parallel?

A

 

[arber333]

I got connectors safely off the main board.
That was more effort than expected. 4 layer board sinks heat and GND pins dont seem to budge. There is a danger you pull the rest of the connector off the board while one pin remains soldered still. I managed to get it off with 420°C soldering iron with fine point. I kept heating the pin from the reverse side with my point as well as heat station. At some point i pushed on the stub and kept pressure so the pin slided in and i rammed the point through the hole so connector fell out. Wheu!!!

I got the most problems from 18p gate connectors, rest was not problematic. Main board is scrap now though...

Another question to EEs here.
What will happen if i use PSU on Johannes board and use another PSU with the same regulator and more caps? Basicaly it would be the same as one on the old boards. I would run them in paralel. Would it cause a problem?
They are quite close together and i can use monster cap on 5V side. To smooth out ripple i can put in 100n film caps across all elcaps.

Then there is arduino whith its own 5V/3V regulator onboard....

 

[arber333]

I had some hairy moments yesterday.
Just when i had everything set up i did a triple continuity check with my multimeter. I measured signals as well as power supply. Well darn... +5V plane was shorting to GND. How can that be, i checked just that once before.

I just installed Johannes board and short was coming from there. I did one thorough check and after some bickering i found Johannes made a mistake of not notifying me of different JP4 connector layout and i made a mistake of just copying the rev 2 board configuration, DOH.

Now everything else fits. But nontheless i will make another interface revision ASAP. It seems a shame though to wait couple of weeks with inverter sitting there. So I decided to put some dual row headers on both sides and make connector that would connect correct pins together. It will take me couple of days to get everything, but in any case i need to get voltage sensor parts, so i think untill thursday or so...

I also did some beginner mistakes on Lebowski circuit, but nothing i cant repair.

I can see the problem of closing the lid after everything. It will require some 3D sculpturing, maybe 3D printing a flange...

EDIT:I was bored in the evening so i went and milled the alu shield so that there is at least 2mm clearance from lowest part of PCB to alu shield.
I intend to put thin isolated sheet under interface board to further separate the two. Caution! Alu shield is soft, but it is not perfectly straight. Mine was somewhat warped and i ended up milling too much. Not a problem though...

 

[arber333]

I got the sensor connector pinout now and i can see lights are on . Now lets see if someone is home...

 

[Tomdb]

Very nice progress, hopefully you end up with something that others can use.

 

[arber333]

Very nice progress, hopefully you end up with something that others can use.

Lets see... hah!
1. one guy suggested to use two ampera 6bridge modules on a single Leaf motor. Of course both should run off a single Lebowski brain.

2. i got idea to put one Leaf motor in front and one Volvo hybrid 60kW motor on the rear axle

3. I will try running two Emrax motors like on a front axle.

4. There is an option of one motor in the back and second motor paired with gas engine for range extender

All the while there would be AC compressor connected to the third inverter. And that is your work Tom, thanks.
Do you think you could connect your small inverter to a Prius AC compressor and run it with port closed? That would tell us if small inverter can be used to run serious AC at all.

tnx

Arber

 

[jhuebner]

Nicely done. Will check that header when back home. Wasn't aware it changed.

 

[arber333]

Since more stuff went wrong on my inverter interface board i decided to step back and redesign new rev. PCB.

So today i let myself go and measured Ampera inverter temp sensors. I heated it up to 70°C and let the pump slowly cool it down.

Ampera 1/3 inverter temp sensor
20°C ---- 5,9K
25°C ---- 4,3K
30°C ---- 3,9K
35°C ---- 3,2K
40°C ---- 2,7K
45°C ---- 2,3K
50°C ---- 2,0K
55°C ---- 1,8K
60°C ---- 1,5K
65°C ---- 1,3K
70°C ---- 1K

While i had heater ready i also measured my Leaf motor and its sensor. This one is not linear untill well into 50°C. Because my heater has bimetal thermostat i couldnt go beyond 70°C. Maybe later when i hack the heater to 100°C.

Leaf motor temp sensor
20°C ---- 18K
25°C ---- 17K
30°C ---- 16,4K
35°C ---- 14,2K
40°C ---- 12.3K
45°C ---- 11K
50°C ---- 9,2K
55°C ---- 8,3K
60°C ---- 7,6K
65°C ---- 7,2K
70°C ---- 6,6K

EDIT: I got IGBT datasheet very kindly from mr. John Kelly and there is good table for thermistor in the last page. Thank you.

 

[arber333]

Like i said procedure to remove connectors from Volt/Ampera main board:

1. Prepare solder station, use fairly pointed solder iron. set it to 350°C.

2. Prepare one flat scredriver cca. 3mm across chisel.

3. Heat up the screws that keep connectors fixed from the underside. You can use heatgun, but i figured 350°C of soldering station quickly heats up screw.

4. When solder inside screw is soft insert scredriver and try turning it. Dont overheat screw since you can melt plastic inside connector.

5. When screwes are out prepare soldering station + vacuum pump.

6. Heat each pin from the underside and suck solder out of the hole. Make sure to force the pin to move in the hole after you suck solder out.
Afterwards try moving connector a little to see which pins need more persuading . That way there will be less resistance when pulling connector out.
GND pin solder cannot be vacuumed off since GND plain is hidden inside PCB.

7. When only GND connection remains put more solder on that one and heat it up. When solder is soft push that pin in its hole (that is the reason for pointed soldering iron).

When connector falls out the front you are done. Make sure you have lots of hands available. I needed at least 3 to succesfuly perform this.

Enjoy!

A

 

[arber333]

Hi Tom

Yesterday i had a new development. I had made small interface boards that can connect Ampera inverter to Johannes main boards. They include signal inverting and current sensor connectors. But there is also inverter temp output. It is compact and suits well to single inverter use.

However i got OClimit error on trying to connect with inverter. When trying to find error i pulled driver cable and OClimit stopped. When i investigated i found out that pin i thought as driver Fault pin is automaticaly connected to GND. There is like 47R difference from Fault pin to sensor GND.

Do you know how Fault pin 4 (white wire) of driver connector is connected? Can you confirm it is a GND connection with some resistance on your inverter?
Do you think i would need to connect a PNP transistor to sustain enable signal? Johannes has a Fault scheme so 12V signal is always present on working inverter. Similar to Lebowski Enable pin. If Fault pin is triggered 12V goes to GND and trigger OClimit error that shuts down PWM.

Do you have an idea how to use your scheme with that signal connection? Probably this pin goes high when Fault condition, but i am not sure how to test it.

tnx

EDIT: All posts from this thread are back!!!