Time for some news with updates on several points:) (a year later as I see now)The car
This summer the car has been approved by the national car department (RDW in dutch) for import (formally German car), for MOT and for the engine swap!
So it is fully legal now to drive
This is a big plus for testing FreeEMS more than an occasional revving in the driveway.
It has some minor flaws to iron out like a squealing rear axle, some rattles from the plastic interior and some hesitation from the engine while driving closed loop on the OEM motor management.
I thought it was a good idea to get everything running smooth first on the OEM parts before commencing the FreeEMS adventure but it takes a lot of effort finding the cause of this hesitation.
Which brings me to my current side project:CAN based sensor network
I'm a CAN nerd during the office hours which I enjoy very much. To gain some more insight on what is going on in and around the engine I needed some more data:
Oil pressure and oil temperature sensors, EGT, wideband, fuel pressure, MAP and an output to control the boost control valve. Of course is this all build on vero prints for proof of concept so I can draw a PCB in a later stage when all (or most) of the bugs are out.
Boost, EGT interface and wb input:
The middle pcb is a 4 analog input node, the left one receives all can messages and puts it on an LCD for now. When the network is operational I will continue developing the lcd project
(the lcd project: https://www.youtube.com/watch?v=pjkqkh1QTpo
Bought some housings:
And experimented some with a stand alone CAN i/o module. This is just a dumb unit that interprets can messages to controls its outputs and can send some messages regardings its inputs:https://www.youtube.com/watch?v=T7mEAe95WTQ
All this testing on the driveway could annoy the neighbors with the deep humming from the exhaust so I built a new one. The goal was to get it as silent as OEM.
The loud setup:
Replaced with a BMW e92 3.0 6 cyl rear silencer. 2,5" inlet, dual 2" out of which one has a butterfly valve to regulate the noise output https://www.youtube.com/watch?v=GmnbwFkxKSw
The end result:
With the valve closed, it sounds like the orginal car did, but now with a full 2,5" exhaust and a 2.0T instead of a 105 hp 1.6L engine
Happy times. the spartan
After this little distraction I continued building and testing the CAN sensor network to monitor the hick-ups in closed loop. At this point I picked up my Spartan Lambda sensor which was laying on my desk for about a year and put it in the car to see if it was leaning out to much under partial load. Connected the narrowband simu to the OEM ecu, started the engine and connected the dealer monitoring software to the OEM ecu: lambda sensor at 1,25v (capped), oops! Picked up the multimeter and saw a whooping 5v at the simulated output, also at the wideband output, a bit weird, but I thought it wasn't heated up yet. After driving around the block it stayed the same so I took it out of the exhaust to prevent any damage to my ecu.
Oops! (brown = narrowband output)
Contacted Alan, I explained the weird behavior of the sensor and I could ship it back to him. Great! The mail took its time but after a while: "I got the package, I will send out a replacement next week." Great again! Finally some progress
Then, after a week: "I tested your unit with my lab gas, the unit is working perfectly". Alan is getting famous for his one of two sentence replies at this moment.
So I asked him to check the narrow band output for voltages in excess of 1v;
"I just retested the unit, both the nwarrowband and the linear output are working correctly. The narrowband voltage should never exceed 1v, I have confirmed this in my testing. I can send the unit back to you, or I can refund your purchase less shipping. Let me know what you want."
Well, then the fault must be at myself not capable of wiring up 4 wires but whatever. If it works, it works (even there was no evidence sent it actually did work) and I still needed a (working) sensor so I asked for shipping it back. After a good two weeks I've got the sensor back. At this stage I'm almost euphoric to test it again. Secretly I hoped that the unit was defective after all and that Alan just replaced it with a new one without admitting it that it was broken. Opened the package, same unit. Ok, fair. Wired it up on the bench: same $@%^ 5v at the narrowband output. No 1.66-,3.33 startup sequence, same faulty unit as 6 weeks before. Sad panda
So I e-mail Alan once again that it is still faulty, his reply: "Send it back to me again, and I will refund you purchase. I tested the unit twice, both times it behaved as I designed it. I have no idea what the issue is."
At first sight this is a fairly reasonable deal. From his standpoint I would blame me for not being able to get a unit working right and then even offering to take it back: great service.
Although, refund is without shipping (initial + taxes), returning and a second return, so $105 refund against $usd21 shipping towards me + 30 euro taxes & handling (customs)for the first time, 16 euro shipping the second time, and the final return shipment of 16 euro would have an outcome of:
$105 - $21 - $(30*1.25) - $(16 * 1.25) - $(16 * 1.25) = $6,50 and no sensor. So I have a $183 paperweight at the moment.
So I proposed several other outcomes to this situation:
- Let me help you to debug this issue remotely, I have a scope and a logic analyzer. When it is fixable I repair it myself
- Let me know how I can test if the sensor itself is still ok and send me a new controller or controller pcb
- refund $105 and if you want it back for research you pay the shipping
So I start measuring: no PWM traceable from the heater FET, solid supply voltage. No pwm at the input of the FET. So everything looked stuck at some point. Emailed these findings.
So I noticed that the power converter was getting hotter every time I was measuring the device to the point it was untouchable. The output on the narrowband and wideband output were climbing also: first 4.97 as pictured, afterwards 5.15 and even 6v just before it crapped out on me. End of controller. Mailed findings
So I ordered a new controller and lambda sensor from a local spartan supplier just to check my own sanity: they arrived, I hooked them up: 1.66v, 3.33v, 5v. Whoa
used some lighter gas at the lambda nozzle: quick drop in voltage, and rising again after releasing the gas.
Emailed findings, begged for a deal on a new controller (like this guy/guys which was even out or warranty http://volvospeed.com/vs_forum/topic/15 ... 4/page-163
), and a reply from Alan! It must be my lucky day
"Sorry I am not following what is happening. Did you order another unit from me?"
Ok, maybe not so lucky after all.
Maybe it is me not being a native english writer, maybe I'm cursed for driving a 309, maybe it is because we celebrate this uber racism Sinterklaas-fest, maybe I'm just too pretty. I don't know. Hoped for a nicer outcome than this. $183 down the drain, another $180 (controller + sensor + shipping) spent on a fully working unit. It stings. In the wallet. In the heart. (Being dutch, this is more or less the same).
In the meanwhile I was building my 0.6 board and two 0.7 boards to keep my mind away from the story above.
Hoping that the hickups are a) gone with FreeEMS or b) traceable with the logging feature of FreeEMS.
There are almost done now, except for the 3 * 3w zener and the 3 * 604 ohm resistor in the power protection circuit
Yesterday I scoped the idle air valve to get an indication for setting the correct parameters later on:
here I have offset the second input to get a clear "inverted" pwm signal
Here it is at the same level
So I will make an control circuit like this:
Except using VND14NV04 I think, because I have them already.
I hope to get my CAN network to a level so that can provide me some usable input within a few weeks and finish up wiring a 0.7 board within one or two weeks.