VEMS Community Forum
Members => Projects & Installs => Topic started by: DButters on June 30, 2011, 08:43:25 pm
-
Hi All - this is my second VEMS install, first was a turbo M30 3.5 BMW motor in an e30 M3 replica.
This time I have taken this failed project from a previous owner from this:
(http://i796.photobucket.com/albums/yy247/dbut019/RX7%20build%20pics/IMG_0117.jpg)
to this:
(http://i796.photobucket.com/albums/yy247/dbut019/RX7%20Track%20Pics/IMG110.jpg)
so the wide body , suspension, wheels, cooling, injectors, exhaust, intake have all been done. Still standard twin turbos but a new setup awaiting install for if (when) they pop.
After doing a custom wire job on the BMW I decided to tackle a plug in adaptor this time, and here is where I am up to:
(http://i796.photobucket.com/albums/yy247/dbut019/VEMS%20plug%20in%20build%20%20RX7/IMG_0238.jpg)
(http://i796.photobucket.com/albums/yy247/dbut019/VEMS%20plug%20in%20build%20%20RX7/IMG_0239.jpg)
So as you can see the old ECU is gutted, and the vems is fitted into the old box, awaiting heatsinking and wiring to the factory connector.
I have built up a spreadsheet for the pin linking and will post it soon.
-
This is awesome. I always get confused about how to setup the trigger parts of new installs, I hope you post lots of info relating to that.
Devin
-
Update:
Currently working on the non-plugin peripheral connections - LCD, RS232, WB02. Made templates for fitting them into the existing box ready for routing.
Also ready to start wiring to the ECU connector - link to my proposed wiring scheme is here:
http://www.mediafire.com/?zt16khowwev3vo3
Thoughts/comments/queries welcome
-
Update:
wiring to the factory connector is complete. I have noted the pin matching that has been done on the standard connector diagram, the pin numbers match what is on the old ECU internally and shown in the mazda wiring diagram.
(http://i796.photobucket.com/albums/yy247/dbut019/VEMS%20plug%20in%20build%20%20RX7/ECConnectorsOut3_6.png)
This is also in spreadsheet form here:
http://www.mediafire.com/?27x3mxg0943ffnh
Now to finish mounting the extra connectors and closing up the box ready for testing!
-
Update:
- the VEMS unit is all wired up to the factory plugs from the old ECU, and tested all the outputs on the bench with LED's and the signal generator on C024. Not quite right but close to the 12+1 the rotary should be giving me.
- then went to the car, and checked the TPS, IAT, CLT, inputs all working well. Took a triggerlog and it looks bad, almost like the primary/secondary signals are reversed, and not very consistent:
http://www.mediafire.com/?n3f5y9b3ed34e2d
I tried reversing them but then it got even worse again:
http://www.mediafire.com/?61wuber582wt1ua
this time it looks like a 1+1 signal.
The car runs 100% OK when I plug in the standard ECU, and the VEMS runs great on the soundcard trigger. the two VR sensors use the factory wiring and both ground to the same ground point inside the ECU which is wired back to the main ground on the standard loom. I will try adding another ground but anyone else have any other ideas? My next test is to chuck the oscilloscope on it and watch.
Has anyone had a VEMS running on RX7 crank sensors?
-
Have you verified the wiring to the VR sensors? Sometimes the OEMs use the in a reversed manner. I have a fc3s running VEMS, but I set it up about 7 years ago, so I really don't recall any wiring changes I did to the VR sensors. It's at a shop getting a new exhaust now, so I can't even check easily.
In the first triggerlog, the primary and secondary are of course reversed. I can't think of an explanation for the second one though unless the wiring is not perfect yet.
Do you have the factory manual? Maybe the wiring diagram can give some hints? I've never worked with a fd3s so I have limited confidence in how to proceed.
-
Thanks for the prompt reply.
I have the factory manuals and confirmed the wiring.
The factory loom joins the sensor grounds together before the ECU, so there is actually only 3 wires for the 2 sensors at the ECU end.
The thing is I tried swapping the 2 signals between primary and secondary trigger, and they did not simply swap inputs, the signals completely changed as you can see between the 2 trigger logs.
So I will confirm with the oscilliscope but what I am saying is that the signals appear to have large amounts of noise in them.
-
OK an update:
I took a look at the VR sensor signals during cranking.
- noise: ~100mV pk-pk Secondary, ~50mV pk-pk Primary
- signals: ~1.5V pk-pk
Confirmed 12 primary, 1 secondary pulse. The issue is the noise, generating false triggers on the secondary.
Improvement #1:
- I changed the 0R at R182 with a 100K (is 150K on primary VR circuit)
- added the solder blob to pull up pin 5 of the LM1815 to Vcc
- this got the car running, at 500-690rpm
New Problem:
I now cannot bring the engine up past this speed, it starts generating false secondary triggers, the noise gets worse with RPM.
I am concerned that my board has issues, as it is was supplied in October 2006, could this be a problem:
http://www.vems.hu/wiki/index.php?page=GenBoard%2FManual%2FInputTriggerHardWare%2FReplaceC
I will try this anyway, and with attempt to put further filtering on the input - are there any recommendations on this before I calculate a suitable RC filter?
trigger log here shows the false sec triggers:
http://www.mediafire.com/?4x2w8d0c963dyet
-
100mv is close to what the LM1815 triggers at, as the RPM increases I'm not surprised about the noise.
You need to look at the use of pull-up and pull-down resistors to reduce the noise.
Check the type of cable from the sensor, is it only grounded at one end? Does it have the right sheilding?
Rob
-
Hi Rob, thanks for the valuable advice.
You are right about the use of pulldowns - I have added a 1K which should assist, I am about to go and test it now. Also added 1uF to the time constant generation section of the LM1815 circuit, which should help.
Remember that I can plug the factory ECU back in and the car runs 100%, so the sensor is not at issue.
Will let you know how I go!
-
Hi Rob, thanks for the valuable advice.
You are right about the use of pulldowns - I have added a 1K which should assist, I am about to go and test it now. Also added 1uF to the time constant generation section of the LM1815 circuit, which should help.
Remember that I can plug the factory ECU back in and the car runs 100%, so the sensor is not at issue.
Will let you know how I go!
The 1K / 1uF did the trick! Up to 3K RPM anyway, I have a MAP sensor issue so not willing to push things just yet.
I cannot believe how rock solid and consistent the engine sounds right down to ~400rpm, its like a swiss instrument compared to my M30B35 motor. Completely different engine tone compared with stock ECU.
So I now have working:
- 2 VR triggers
- CLT
- IAT
- Fuel pump control
- Injectors (yet to test secondaries)
-ignition, yet to confirm trailing is actually going but appears OK
Still to test:
- MAP, needs fixing
- Fans
- Clutch Switch
- Wheel Speed
- Nuetral switch
- boost control
- idle air control
- Knock sensors (when they return in firmware)
Oh, and I wold KILL for OMP control - I think it would make such a difference to the RX7 uptake if we added just a simple stepper motor position vs. load table.
-
You are right when you say that the sensor is certainly not the issue, the key difference is that the signal conditioning electronics in the Japanese ECUs is markedly different to that in the VEMS and the European ECUs that I've seen.
It is almost as if the Japanese took a low cost route on the trigger hardware, converting their distributors to provide usable signals, and spent the money in the ECU's trigger filtering, while the Europeans spent on crank mounted precision triggers saving some of the complexity required in ECUs.
The 12+1 trigger has caused a variety of issues over the years mainly because theres a lot of time when the trigger's transmission lines are not being stimulated by a tooth - this allows the voltage on the line to wander and pick-up EM noise.
It may pay to use a pull-up to bias the line a couple of volts above 0.
If the rotary runs like a time piece with a distributor mounted trigger - what would it be like with a decent crank mounted trigger and coil packs? :D
-
You are right when you say that the sensor is certainly not the issue, the key difference is that the signal conditioning electronics in the Japanese ECUs is markedly different to that in the VEMS and the European ECUs that I've seen.
It is almost as if the Japanese took a low cost route on the trigger hardware, converting their distributors to provide usable signals, and spent the money in the ECU's trigger filtering, while the Europeans spent on crank mounted precision triggers saving some of the complexity required in ECUs.
The 12+1 trigger has caused a variety of issues over the years mainly because theres a lot of time when the trigger's transmission lines are not being stimulated by a tooth - this allows the voltage on the line to wander and pick-up EM noise.
It may pay to use a pull-up to bias the line a couple of volts above 0.
If the rotary runs like a time piece with a distributor mounted trigger - what would it be like with a decent crank mounted trigger and coil packs? :D
I believe you are correct in your observations about sensors there - certainly the hall effect I used on the BMW was much better than these japanese VR sensors, trigger errors were not seen.
I do have to correct you on the sensor type though - all Series 6 onwards rotary engines use dual crank sensors and toothed crank wheels. They stopped using the distributor setup with the seried 5 (FC) RX7's.
-
I know nothing about these heathen rotary engines! :D
-
Update:
- wired up WB02 and calibrated free air, value of 233 seems very high?
- tried getting some lambda values on car, could only get 1.29 when running (though my timing is waaay out, like 1 tooth out I think) or 0.7 when car is so lean it misfires.
- Anyone know any way to verify other lambda readings than just free air to confirm the sensor setup is OK? Sensor is brand new first time on a vehicle.
IAC valve is working now, using 200Hz to stop it being noisy and it works well despite my whacked out ignition timing...
-
233 is not unexpected, its when you get into the 250's you start to wonder...
1.29 is so lean as to cause misfires...
0.7 is so rich as to cause misfires...
I have, in the past, put the sensor tip in a butane flame, holding it out above the flame shows a lean mix - theres more air than fuel.
Getting it in the rich blue part of the flame gives a rich mixture reading, as theres lots of unburnt butane at that part of the flame.
Its not really calibration but shows something about combustion.
Get your ignition spot on before you worry about fuel - if some of the combustion is occuring in the ports then you're not going to get a proper reading.
-
OK so I checked the timing and it is 5 degrees off, not as bad as I suspected. I will change the TDC delay to 65 degrees to fix this (I wonder who verified the defaults in Vemstune for 2-rotor and what version 2-rotor they did it on???)
So my timing is not too bad, but my lambda is still toast - I either get full rich or full lean on the gauge. I free air calibrated while still in the exhaust pipe and it was showing ~20.4.
I will try the breathing on it trick (forgot about that, much safer than a blowtorch) and see if I get expected readings. Once I get the lambda working I can start tuning!
-
I used a butane lighter ;)
-
Given the amount of fuel I have around at the moment I would be hesitant to even use a lighter :)
I am still not convinced on the timing though - initially I only checked one rotor. Now it appears the primary coil may have blown during cranking, looks like it got left on somehow. Even going back to the factory ECU I cant get it started. Going to check spark on each plug tonight and diagnose to see if I now need some more coils!
-
So I was correct to not believe my initial timing setup. I was out by over 50 degrees! Now I am really unsure what type of rotary the defaults in VEMS were made for - definitely not FD.
Once I had that corrected the engine tone sounds a lot better, and I can get it to drive! The lambda readings also start showing relatively sane values while driving, though light throttle still jumps quickly into super lean regardless of VE settings.
I guess now it is time to tune and see what happens :)
Oh and I got a little external update too:
(http://i796.photobucket.com/albums/yy247/dbut019/2011-11-26201102.jpg)
-
Wow that car looks great! can you post one of your idling logs? I'd like to take to peek.
-
Had a track day today - sadly before I could get it on the dyno, but still good fun!
Here is the last log of the day, and the config:
http://www.mediafire.com/?o3qud3cq3gx6zeo (http://www.mediafire.com/?o3qud3cq3gx6zeo)
http://www.mediafire.com/?3wa86hg11vat7we (http://www.mediafire.com/?3wa86hg11vat7we)
If anyone cares to suggest some improvements I am all ears! I was very paranoid about detonation, and had convinced myself I could hear it, so turned the timing down a long way from other peoples configs I have seen. I really would like Knock back in the firmware, so I can confirm this.
One problem I can see is at the staged injector transition point - it suddenly gets a lot richer from then on, does anyone know how to fix this????
-
On a Rotary, the only use of a knock sensor is to tell you that you just blew the engine. Sadly. It wasn't knock you heard unless you have no compression now, which is unlikely. Those engines simply don't tolerate knock. Octane is a good friend.
I haven't gotten the staged injection tuned well yet, but as soon as my exhaust gets completed, I'll put it on my dyno and give it some much needed love.
-
On a Rotary, the only use of a knock sensor is to tell you that you just blew the engine. Sadly. It wasn't knock you heard unless you have no compression now, which is unlikely. Those engines simply don't tolerate knock. Octane is a good friend.
I haven't gotten the staged injection tuned well yet, but as soon as my exhaust gets completed, I'll put it on my dyno and give it some much needed love.
I agree that knock sensors are mostly the ambulance at the bottom of the hill for apex seals, but at low boost and mid-low rpm it would still be nice to know if there was anything, particularly given the severe lack of rotary base timing maps for VEMS :)
Although I did start with over 2 times the fuel in than what I ended up with in that log, and as you can see it is still running well rich! I was also running race gas as I am paranoid about the premix diluting the effectiveness of the fuel, but we also get 98 octane pump gas here.
With regards to the staged injectors, I am running 2 x 850 primary and 2 x850 secondary so the ratio is set at 100%. I guess I could lower the ratio to help even them out, but after looking at what other ECU's do i think the 80% throttle setting I took from your or the other rotary setup on the wiki is not matched with the actual opening of the secondary throttles, so I am going to test lowering this to match a more factory setting and see how that goes.
I also still have the issue with idle lambda values, they immediately run the gauge past full rich. The idle is however very stable, and if I drop the VE any lower it immediately starts to misfire.