VEMS Community Forum
Technical => Fuel Injection => Topic started by: Sprocket on November 04, 2008, 06:13:07 pm
-
In the space of a week I have had two Injector FETS fail. They have failed to ground, so the injector is on all the time when power is applied to the injectors. This last one, I can see in the logs exactly when it happened, which was right before the tank ran dry, which im sure has no bearing on it ;)
I have pulled the board from the case. It is obvious that something had failed as when i opened up the case, you get that distinctive electronics burning smell.
Now the question is, what is the likely cause?
These FETs were on the old board and I unsoldered and re soldered onto this board, this could be a problem? too much heat on the FETS?
The other thought was that the Transient Supression diode was either the wrong one, or inverted. The diode is installed onboard with the cathode tail soldered to the board in the middle of the flyback diode array, and the anode tail soldered to EC36 pin23. The diode used is a 1.5KE30A, which is a unidirectional 30v diode. Everything appears to be in order. However, to eliminate this, I have cut this out, bridged the gap with wire and installed the 1.5KE30CA Bidirectional 30v diode in the loom just before the EC36 connector, as it was on the last loom. The flyback wire is direct from the fuse box, fused side for the injectors.
The injectors were cleaned and tested, they did stick, but are free now, all now work and all have a resistance of around 15 ohm.
The two FETS that failed are channels 1 and 3 (binary 1 and 4)
Is there anything else I should be checking.
At least i can now diagnose a smoked FET, fuel pours out the TB :D
-
Hi mate, those FQPF20N06L FETs are rated at 15A.
So something must be very awry, can you be sure that theres no shorting between power and ground?
Rob
-
There is a ten amp fuse feeding all the injectors and that is not blowing.
I dont think there are any ground faults, all the wiring is brand new. The live feeding the injectors is daisy chained between the injectors.
I was thinking flyback after reading this statement
'Flyback circuits are used to protect the switching equipment (the injector driver FETs) from extremely high voltages that could occur when the injector is switched off. While the load is being switched off, voltage builds up and must be dissipated somehow so that we don’t burn up the FETs'
http://megasquirtavr.sourceforge.net/manual/Detailed.Output.Flyback.html#Operate
I will tonight replace the two smoked FETs but use other channels for now, at least I can then smoke the suspect FETS :D. I just dont know if its because these FETs have been exposed to too much heat desoldering.
I might get the clamp on out and see if they all draw the same current.
On another note, I have been blowing fuses on the coils, but I think that was down to the dwell being too high, again, I need to get the clamp on onto those.
It does run and runs well then it suddenly takes a dive.
-
Flyback issues are possible - but why only on two channels? I'd expect them all to pop in quick succession - but its certainly worth checking.
Rob
-
I have experience with no flyback diode connected (my fault). FETs survives long time. And fails only one at time. At redline.
-
Right, something is disturbingly amiss :-\
I replaced the two smoked FETs, replaced the Transient diode, and re installed everything. Set up for Alpha N and then tried again.
Again #3 cylinder flooded within seconds of starting. Injector channel 3 again down to ground and that was one of the new FETs >:(
Something deffo amiss.
I am ordering some new injectors tomorrow, and I will tear the loom apart to check everything is ok there. The only other thing I noticed different is the injector settings. In the pic bellow the left settings are from a known good working set up with same family injector, but larger cc, and the right is the settings I have. Will any of these differencies cause the FETs to fail?
The loom is all new modern thinwall 1mm wire, the injectors were tested and cleaned last year, they did stick at first, but freed up, they all work fine, the coils measure 15 ohms.
This is going to be a painfull excersise :-[
If I need to, can I use IGTBs instead of FETs, not the cheapest, but, I have four spare IGTBs ond no spare FETs at the moment. Not that I want to get to the stage where I have to start pissing about with that.
(http://freespace.virgin.net/colin.milne/Injector settings.jpg)
-
Not about FETs, but config:
Digging through wiki and on bench tests, I found:
(http://www.vems.hu/files/GintsK/InjSettings01.jpg)
Injector opening time @13.2V usually sits between 500 and 1000us
Battfactor is added time @7V. Here 600us/6.2V=96us/V - still low setting. Usually 100...200us/V -> 620...1240us. Most of configs I ever see have less. It can cause bad idle!
Rampup settings here is disabled. Most ECUs haven't such setting.
=====
Are you fire all injectors simultaneously?
-
Thanks for that, I altered the settings similar to the above.
Checked the injector drivers, cylinder3 500 ohm to ground, cylinder 4 800 ohm to ground >:(
Well, the saga continues. I swapped out the injectors for some larger ones of the same family, recently cleaned and tested, re configured. Tested the wiring for faults, no resistance between 12v feed and injector channel, with ECU disconnected. With ECU connected, testing to ground, both the live feed and the injector channel read 2.5 megohm, I assume the reading on the live feed is due to the flyback? Flyback wire has continuity from the fuse to the ECU. Swapped round the injector channels.
While I had the fuel rail out, I checked the injectors for operation, all ok and firing as they should. Started the engine, and it ran, all be it all over the place, and rich, EGO helped when it warmed up. Logs show the pulse width every now and then drop to zero and imediately return to normal, for no apparent reason? things were seemingly looking good, the engine was warming up slowley which was better, but then it started running rough again, switched off and looked at the TBs, fuel pouring out od #2 TB >:( injector channel 2 now FUBAR'd at 13k ohm to ground
Please help. I am after hard advice here, as I just cannot see where the problem can be :'(
-
Not about FETs, but config:
Digging through wiki and on bench tests, I found:
(http://www.vems.hu/files/GintsK/InjSettings01.jpg)
Injector opening time @13.2V usually sits between 500 and 1000us
Battfactor is added time @7V. Here 600us/6.2V=96us/V - still low setting. Usually 100...200us/V -> 620...1240us. Most of configs I ever see have less. It can cause bad idle!
Rampup settings here is disabled. Most ECUs haven't such setting.
=====
Are you fire all injectors simultaneously?
While I was looking at Robs guide, it says that Injector open time should be set to 0
I had previously thought that the ramp up time is the bit you tune for the characteristic of the injector, IE if it takes 9ms to ramp up, you would set that in effective ramp up
There is conflicting information everywhere, an i do not know what is right or wrong any more.
Rob, if you read this, can you please see if one of the powers that be, can suggest anything. Im close to throwing this in the bin. This is the new board and is its first use.
Im considering DTA, But i really do not want to go down that route unless I really really really have to. This is a great product, but information is pants.
Any of the development team thought about building circuit protection into these units ??? Simple current sensing circuit thats all.
Sorry, rant over,
-
As a rule of thumb in the guide I suggest:
Inj open time should be 0
Inj rampup_battfac must be 4080 (which diables the feature)
The wiki is full of contradicting information, and should largely be ignored, the User Guide has been peer reviewed by a wide range of people and I hope to keep it spot-on.
The values you had in the screen shot
Inj open time 0
Inj effectiv 912
injrampup_battfac 4080
The battfac should be higher than 0, try 512 - this scales the opening time due to low voltage when cranking.
The injector model cannot cause the FETs to draw more current from the injectors than the 15A they can handle.
-
Don't worry: these settings can't kill FETs. But if you want good vision about various injectors, you can download A E M tuning software and under setup->injectors->primaries->batt.offset wizard find good graphs.
As I understand rampup time is very small time when needle travels from closed to fully opened. Opening time is time between start of pulse and rampup start - much longer time. According to Andreys documentation in russian, sum of both settings (opening and rampup) is linear graph anyway.
Are you fire all injectors simultaneous? May be that is a cause? I have more than 20 VEMS installed and no fried FETs without a clear reason.
Gints
-
Thanks for the re assurance Rob.
I have All sunday now to get to the bottom of this, and if that means pulling every wire out of the loom, then so be it.
I know what you are saying. These FETs can draw 15amp continuous, just seems odd that the 10 amp fuse isnt blowing. I might try a 7 amp fuse or something in there tomorrow.
It isnt the config, its not the injectors, that only leaves the hardware and the wiring. anything else?
-
Don't worry: these settings can't kill FETs. But if you want good vision about various injectors, you can download A E M tuning software and under setup->injectors->primaries->batt.offset wizard find good graphs.
As I understand rampup time is very small time when needle travels from closed to fully opened. Opening time is time between start of pulse and rampup start - much longer time. According to Andreys documentation in russian, sum of both settings (opening and rampup) is linear graph anyway.
Are you fire all injectors simultaneous? May be that is a cause? I have more than 20 VEMS installed and no fried FETs without a clear reason.
Gints
Running sequencial injection.
This is a replacement board, and has not been tested in action, this is its first outing. I never had any FETs fail on the last board. Thats why I thought it was perhaps the excess heat from desoldering, but its smoked one of the new replacement ones already.
going to have to order a rack of replacements :-\
-
Hm. ??? Why alternate banks from h[0] is 0 only. With such setting only simultaneous firing is possible.
-
Hi,
The zero is from my settings on the left. I'm running Tbi ;)
Cheers,
Gavin :)
-
Hm. ??? Why alternate banks from h[0] is 0 only. With such setting only simultaneous firing is possible.
In the pic bellow the left settings are from a known good working set up with same family injector, but larger cc, and the right is the settings I have.
(http://freespace.virgin.net/colin.milne/Injector settings.jpg)
;) :)
-
If the settings on the right are yours, and left is a running setup, then the big differnece is the PWM setting. Yours is at 1ms, and the running is at 25.5 (ie, disabled).....this would give the fet something to work hard over!!!
-Gavin
-
If injpwm DC is 100% then there's no PWM occuring
-
Which really, when you think about it......makes sense lol!
Must rememeber to start thinking more :p
-Gavin
-
Sorry about 0 only: I have too small screen and I was lazy slide right side of picture visible.
If injpwm DC is 100% then there's no PWM occuring
Are you 100% sure :D
IMHO 25.5ms and 100% is safer. I will scope this on the bench, if find time.
Correct PWMing can be Ubatt dependent: More % at low voltage, less at high.
But anyway. First component failed must be TVS diode if PWMing is active.
Gints
-
Had a thought
Would an inadequate, rather than poor ground cause FET failure?
The reason I say this is the ground point I am now using, is not the same one as that I have used on the 1400 which was the factory ground. I did however check the reisitance of this ground point to the battery, and there was no resistance. But, im now thinking that even though the ground resistance is fine, when its loaded up with current, it becomes a problem. I say this, As I know that it is a riv nut that I fitted, and I used some wax oil on the mating faces when I crimped it in.
First thing I will do is relocate this ground
Could this be the answer?
Could this also be the answer to blowing fuses on the ignition coils with slightly higher dwell times, and why the stepper driver doesnt seam to work correctly ???
-
I do some labor work regarding injector settings.
PWMing can't kill your FETs. It is voltage dependant. But PWM% can only rise from configurated, when voltage falls. When voltage rises PWM% stays as configurated. 100% in this case. 14V seems is breakpoint.
But surprise!!! injrampup_battfac 4080 did not disable correction. It is maximal possible correction in fact!!
works in very strange way. It describe some kind of coefficient for inj_eff_rampup time. No influence @7V
If inj_eff_rampup time is configurated as 1500, it adds 1000us @7V. But injrampup_battfac scale this value @13V by aprox. :
4080 - 0.78
2000 - 0.89
500 - 0.97
0 - 1.0
So if injrampup_battfac is 4080 and inj_eff_rampup time is configurated as 1500 we get 1000*0.78=780us adder to PW @13V.
Very confusing and user unfriendly! Why, why ??? ??? ??? ???
Both other settings are more straightforward:
Inj_open_time @13.2V is simple adder at any voltage. 1:1. 500us=500us
Battfact is PW added @7V It scales down to 0us @13.2V.
Sorry about offtopic! Rob, can you move it to appropriate place?
Gints
-
It sounds like I need to clean this up and add it to the UserGuide.
-
Its bolloxed >:(
-
I have calmed down a bit now :-[
I have just ordered, 13 new FETs, 10 for me, and 3 to replace the ones I borrowed from Bat. Also ordered two TCP259 chips, to replace the one that smoked today :-\ Nobady's fault but mine for that one :-[
Anyway. I am pulling the loom out tomorrow, and re hashing it with as short wire runs as possible, the ground will get extra care, and Im installing the ECU under the bonnet where it will not get wet.
I doubt the length of the loom is the problem as considering the size of the engine bay, its pretty small :D
The saga continues :-\
Edit, Im going to feed each injector seperately and fuse them, any idea on the size of fuse that would survive, 5 amp?
-
http://ourworld.compuserve.com/homepages/Bill_Bowden/ohmslaw.htm
Voltage (E) = Current (I) * Resistance (R)
Power (watts) = Current Squared (I^2) * Resistance (R)
Power = I*E = E^2 / R
So each fuse has to be 1amp.
12v / 16ohm injector = 0.75amp
The 12v to the injectors would be
0.75 * 4 = 3amp, I´d go for 5amp fuse
-
Hi Gavin.
I would be worried about two things when desoldering a FET, you will heat it up a lot when desoldering it. That in itself can damage the FET but you can also likely to damage it with ESD when you handle it. FET's are VERY sensitive to ESD. I would NEVER try to reuse a semiconductor.
When you have blown a FET, regardless of why, you should always replace the FET drivers. They are often damaged when a FET shorts out and even if they seem to work we often see the FET's fail repeatedly if the FET driver isn't replaced.
So if a FET blows, replace the FET, the FET driver AND the FET that use the same FET driver.
If it happens again you should definately replace the capacitors servicing the FET drivers.
Also get rid of the 1.5KE30CA bidirectional transient suppressor diode, I am fairly certain that they have caused problems now and then. Use the 1.5KE30A unidirectional TS diode.
It is also safe to use IGBT's instead of FET's but you need to use them on all injectors on the same engine as they behave slightly different. I have used IGBT's on one car that recived a set of broken SARD highZ injectors. After running for two years with different injector types the SARD injectors went in and one of the FET's started blowing. Nothing short of fitting IGBT's fixed it.
What kind of injectors do you use?
Note that a 10A fuse will not protect a 15A semiconductor from a low resistance short. Never use a bigger fuse then you need.
It's not related to the injectors, but for the ignition the biggest acceptable fuse is 7.5A, unless you really know what you are doing.
Check that the flyback wire is properly connected and that the ground wires are max 150-200mm long and that they are joined like this: http://www.vems.se/forum/viewtopic.php?f=2&t=460
Stacked ring shaped cable ends is NOT acceptable, the above connection is pretty much the only connection that works in the long run.
I hope that this will help you get the car running again.
Jörgen Karlsson
Gothenburg, Sweden.
-
Cool
a little more work :D
However, where do I get the FET driver chip from? They are not in the webshop?
I may well have some in the rescue kit from the first board but I hed nothing with the second board, not even the inductor
Also, which capacitors do you speak of, are they the ones between the driver chips?
Excellent help, keep up the good work ;D
-
Hang on, the driver chips are in the PWM Kit, lol, confusing :D
-
Oh, and Sprockets, is Gavin, Sprocket is Colin :D
Very confusing :D, I thought some one had hacked my account at first :D
-
http://ourworld.compuserve.com/homepages/Bill_Bowden/ohmslaw.htm
Voltage (E) = Current (I) * Resistance (R)
Power (watts) = Current Squared (I^2) * Resistance (R)
Power = I*E = E^2 / R
So each fuse has to be 1amp.
12v / 16ohm injector = 0.75amp
The 12v to the injectors would be
0.75 * 4 = 3amp, I´d go for 5amp fuse
True - but bare in mind that voltage is closer to 14v
14 / 16ohm = 0.875A * 4 = 3.5A
But you will find that your duty cycle is a factor too @ 80% Duty
3.5 * 0.8 = 2.8A
If you duty cycle is less than 80% then you can use a smaller fuse yet.
-
Jörgen, how physically FET driver can damage FET?
Yes it can open FET forever, when damaged. But how it can be a cause?
When I had fried FETs, gate resistors was fried too. It works like fuse protecting driver chip. Check resistance!
Gints
-
I do some labor work regarding injector settings.
PWMing can't kill your FETs. It is voltage dependant. But PWM% can only rise from configurated, when voltage falls. When voltage rises PWM% stays as configurated. 100% in this case. 14V seems is breakpoint.
But surprise!!! injrampup_battfac 4080 did not disable correction. It is maximal possible correction in fact!!
works in very strange way. It describe some kind of coefficient for inj_eff_rampup time. No influence @7V
If inj_eff_rampup time is configurated as 1500, it adds 1000us @7V. But injrampup_battfac scale this value @13V by aprox. :
4080 - 0.78
2000 - 0.89
500 - 0.97
0 - 1.0
So if injrampup_battfac is 4080 and inj_eff_rampup time is configurated as 1500 we get 1000*0.78=780us adder to PW @13V.
Very confusing and user unfriendly! Why, why ??? ??? ??? ???
Both other settings are more straightforward:
Inj_open_time @13.2V is simple adder at any voltage. 1:1. 500us=500us
Battfact is PW added @7V It scales down to 0us @13.2V.
Sorry about offtopic! Rob, can you move it to appropriate place?
Gints
While Im waiting on the band aid turning up in the post, I will explain what I did with the injector settings.
As there is a general lack of explanation, data, and a starting point, I went about working out a theoretical good starting point for the injectors I am using. Now considering the VE table and EGOC will cater for much of the error.
I have some data on the injectors I am using. This was as a result of a fellow Mini enthusiast. He tested the injectors flow rates at various pulse widths using a signal generator. Starting with very low values and working up to a value that gave repeated flow results. 0.9ms was found to give the best repeatable results and as such could be considered the effective ramp up time. Basicaly the injector pintle has fully opened and at the point it fully opened the current was dropped, less than 0.9ms the pintle never managed to reach full open, and so the flow characteristics were inconsistent due to the pintle not opening the same amount each pulse. Effective ramp up time being the total time it takes from when current is applied to where the injector is full open. Open time is the time it takes from when current is applied to when the injector starts to lift the pintle off its seat, and ramp up time is the time it takes for the injector to go from full closed to full open excluding the initial open time. so, to recap, effective ramp up time is the sum of open time and ramp up time.
The flow rate of the injector at this minimum pulse width is a sum of the ramp up and ramp down flow rate, and the flow will obviously vary depending on the size injectors. This is why big injectors can have bad idle characteristics, when you start to get close to the minimum pulse width, as they flow more fuel than the engine needs at idle. I still think that correct sizing of injectors gets overlooked, and why injector staging is a desireble option. The injectors I have chosen, seem to give a pulse width of around 5ms at ilde, and have been calculated to be more than big enough for the expected output of the engine.
Not knowing the injector open time, I made an asumption of 500us and set the effective ramp up to 1500. then using the injrampup battfac scaling factor of (4080) 0.78, calculated the effective ramp up at 13v, this gave 1170us. Too high for the data I have. I then reduced both the open time and the effective ramp up time by 50us in each untill the calculation gave a close enough value to the 0.9ms (900us) the results I have are bellow.
(http://freespace.virgin.net/colin.milne/inj settup.jpg)
I am a little unsure of the battfac, as the name suggests its a factor, much like the injrampup battfac, but why units of us if its a factor. Is battfac a time value or a factor? For now I have set battfac as if it were a time value, and set it similar to the scaled effective ramp up time (1200x0.78=936us) minus the injector open time (352us) giving circa 600ms for the battfac value.
Now that may well be total rhubarb, but it looks and sounds feasable.
at 7v effective open time is 1800ms (effective ramp up + battfac) (note twice the scaled effective ramp up, coincidence?)
At 13v effective open time is 936 ( effective ramp up x Injrampup battfac)
So with voltage between 7 and 13v the values are scaled acordingly
This by no means an acurate way to set these, but due to the lack of a idiots guide on how to set this up, I applied some reasonable thought to get somewhere close, allowing some small error.
What do you think ;D
-
There is need for developers explanation...
My tests show something like this:
For injrampup battfact 4080:
PWefframpup=k*2/3*inj_eff_ramup_time
k=1.0@7V
k=0.90@10V
k=0.78@13V
k=0.737@15V
k=0.70@16V It is straight line in fact!
If injrampup battfact is reduced, we get k closer to 1.0. As example:
if 2040 used k=0.85@16V
if 0 - k=1.0 always
====
Battfact is in us bacause it is maximal adder to PW@7V. It linearly scales down to [email protected].
It is straight line too!
Now what? Straight line+straight line=straight line anyway. Why there is two straight line equations used for descscripting one straight line?
Here can be some answer: If active PW comes close to 2/3*inj_eff_ramup_time (at very low injection times), Inj_rampup becomes scaled down. Smaller PW cause smaller Inj_rampup adder.
It is my free translation from Russian from Andreys description in http://rus.vems.hu/punbb/viewtopic.php?id=42
Here is original text:
ЕÑли pulse_width < 2/3*Inj_rampup, то еÑÑ‚ÑŒ при очень малом времени впрыÑка идет дополнительное маÑштабирование Inj_rampup в меньшую Ñторону, чем меньше pulse_width тем Ñильнее маÑштабирование.
How it exatly works is not clearly understandabble. Need additional bench tests. And very preferable some description of original idea.
-
If active PW comes close to 2/3*inj_eff_ramup_time (at very low injection times), Inj_rampup becomes scaled down. Smaller PW cause smaller Inj_rampup adder.
It is my free translation from Russian from Andreys description in http://rus.vems.hu/punbb/viewtopic.php?id=42
Here is original text:
ЕÑли pulse_width < 2/3*Inj_rampup, то еÑÑ‚ÑŒ при очень малом времени впрыÑка идет дополнительное маÑштабирование Inj_rampup в меньшую Ñторону, чем меньше pulse_width тем Ñильнее маÑштабирование.
How it exatly works is not clearly understandabble. Need additional bench tests. And very preferable some description of original idea.
I think I understandthis part.
Consider the effective flow rate of the injector at minimum effective pulse width (effective ramp up time) (in my example 0.9ms). The PW at idle is a calculation of the required fuel PW divided by the VE, plus, the scaled effective ramp up time. You need this effective ramp up time for consistant injector opening irrespective. If the injectors are so big that the PW at idle is such that PW minus effective ramp up time is too small, the effective ramp up time itself is scaled, so to increase the required fuel PW (VE value)(the bit in the middle of ramp up and ramp down). Otherwise VE at idle would be near zero?
So much going on in my head, but with all this rambling im starting to understand this more.
Next lesson, idle control PID ;D
-
Right need a little help
I have replaced the FET driver and P259 chips. I came to replace the driver chp capacitors. The shcematic shows 0.22uf capacitors, and the rescue kit says there are 8*220nf capacitors, but the most of one capacitor in the reskue kit is 6 :-\ I have assumed for now that these are the 220nf capacitors PLEASE CONFIRM
Also, I measured the resistance of the FET gate resistors all except one read 18 ohm, which acording to the code 180, this is correct. The other one, read 39 ohms and the code was unreadable, just black. I removed it and measured again, still 39 ohm. It also had a black line through it so it has obviously been smoked. Now there are no 18 ohm resistors in the rescue kit, but there are 33 ohm, which, are the same ones available in the PWM kit using the same FET Driver chips. Do I try and source an 18 ohm resistor? (dificult) or do I replace all the resistors with the 33 ohm ones? PLEASE CONFIRM
All the old FETs whether they were good or not have been removed ready for all new one.
I dont think I did too bad a job. Obviously not as pretty as the flow solder, but everything looks to be a good contact and no tracking.
Also, can anybody see any issues with the grounding wires I remade?
Nearly there :D
(http://freespace.virgin.net/colin.milne/DSCF3454.JPG)
(http://freespace.virgin.net/colin.milne/DSCF3455.JPG)
(http://freespace.virgin.net/colin.milne/P259 chip.JPG)
(http://freespace.virgin.net/colin.milne/FET Drivers.JPG)
-
Nothing wrong with trying the 33ohm resistor, if the original was a 33 and you were suggesting using an 18 then I'd be wary, but you're not so that's all good.
-
Thats what I ws thinking Rob. But what about those capacitors?
-
Good question and unfortunately one I cannot answer :(
I'll check my rescue kit this evening and see if I can find an answer for you.
Rob
-
Bump ;D
-
Bugger :-[
-
Bump ;)
-
The ones I have are all in strips of 6.
I've put a question on the wiki - I'll be asking the developers...
-
Jörgen, how physically FET driver can damage FET?
Yes it can open FET forever, when damaged. But how it can be a cause?
When I had fried FETs, gate resistors was fried too. It works like fuse protecting driver chip. Check resistance!
Gints
I have yet to find a broken gate resistor in any of the boxes I service and I have serviced a fair number of boxes. Is this something that is exclusive to the recent batch of boxes using the 18ohm resistors?
I can only guess why a damaged FET driver kills the FET but Sweden is a large market for VEMS and I have seen more then a few broken boxes over the years. But as I'm one of the original developers of the v3 I tend to get involved sooner or later whenever a box has anything but trivial damage. For example when the FET's fail repeatedly. The initial reason for the failure has usually been because of a fucked up install but in many cases the boxes has continued to fail after being repaired. A few years ago we found that the FET driver is the reason for this and that replacing it fixes the problem. Most other cases can be explained with broken external hardware like injectors, flyback or that the install is still fucked up in some way or another.
A FET will often fail when it's turned on or off slowly, it wouldn't be far fetched to guess that a FET driver could fail in a way that makes it turn on and off slowly. Stranger things happen, like for example FET drivers exploding with enough force to make a crater in the board if the capacitors for the FET drivers is left out.
Sprocket, I have verified that the strip of 6 in rescue kit 2 are 220nF capacitors. I did this by measuring them on a recent kit.
The 33 ohm resistors is a good replacement for the 18 ohm resistors.
Jörgen
-
Jorgen.
Thank you, that is exactly what I was after, I didnt want to power up the board before I had this info. Time will tell how I get on now.
I still dont know what cause the failure, but i will assume it was the grounding, that has been re done, as you can see above. However, i have one last thought. The injector and ignition power supply comes from the main relay which is switched when the RPM rises above 100rpm. would this cause any problems?
Thansk for everyones help
Sprox
-
Where does the flyback connect to with regard to the relay? If it goes to the battery then its possible that when the engine speed falls below 100rpm after running there's a few injector events that have no flyback path...
-
Flyback is connected at the fuse that supplies the injectors, fused side
-
and what side of the relay is that?
-
Its the switched side of the relay, so when the relay drops out so does the 12v to both flyback and injectors, as per the recommended drawings ;)
Battery - Relay - fuse - flyback/injectors
-
Cool stuff - it always pays to check these things :D
-
No worries mate, i am open to suggestions ;D
-
All this effort and a FET still died >:( >:( >:( >:( >:( >:( >:( >:( >:( >:( >:( >:( >:(
HAMMER TIME!!!!
I have since calmed down and replaced four good NEW FETs with four IGTBs and it seems to work.
I have no faith :(
-
Since the FET´s are these
http://www.fairchildsemi.com/ds/FQ/FQPF20N06L.pdf
I can´t even see a reason they should crap out.
Maybe there is problem with the voltage to turn it on?
http://en.wikipedia.org/wiki/Power_MOSFET
Maximum drain to source voltage
Power MOSFETS have a maximum specified drain to source voltage, beyond which breakdown may occur. Exceeding the breakdown voltage causes the device to turn on, potentially damaging it and other circuit elements due to excessive power dissipation.
I´m just not good enough to actually make sense of the datasheet to figure out a potential problem to why they fail.
Maybe an electrical engineer could really discover the problem and consult on a possible solution so that the FET´s are protected at all angles?
-
I didnt know that IGBT's could be used instead of fets. Also, I agree that something needs to be looked at.
I work with many brands of commercial ECU's and on none of them did I have to run a flyback wire to the ecu. Also I did not have to lie sleepless in my bed at night worried that a strand was going to break on one of the 5 ground wires.
This product is a very powerful tool with some great features. It would be great to match that with the same level of support and reliability. Thanks to Bob for this forum, hes spending a great deal of his own time helping people out.
-
The Nissan's I've done the plug and play harness adaptors for have, in all but one case, flyback wires, I've seen it on other OEM ECUs too. The ones that do not seem to have some built-in power controller that allows the ECU to latch its own power.
The reason for the ground wire setup (and the flyback) is to keep the sensor signals clean, when the injectors are firing, and the coils discharging, with a mass of extra current load (relays, wideband etc...) the small resistance in the wires, and the high current load result in a rise in potential - which will send all your analog readings up the chute.
-
I too know that the Rover ECU has a flyback path, as the ECU is fed with a live from the same source as the injector.
Anyway, I really dont know where I would be without this forum, as the Wiki is so bad. Big up for Rob ;D
-
Since the FET´s are these
http://www.fairchildsemi.com/ds/FQ/FQPF20N06L.pdf
I can´t even see a reason they should crap out.
Maybe there is problem with the voltage to turn it on?
http://en.wikipedia.org/wiki/Power_MOSFET
Maximum drain to source voltage
Power MOSFETS have a maximum specified drain to source voltage, beyond which breakdown may occur. Exceeding the breakdown voltage causes the device to turn on, potentially damaging it and other circuit elements due to excessive power dissipation.
I´m just not good enough to actually make sense of the datasheet to figure out a potential problem to why they fail.
Maybe an electrical engineer could really discover the problem and consult on a possible solution so that the FET´s are protected at all angles?
Some additional background. I knew ECU where http://www.irf.com/product-info/datasheets/data/irf7380.pdf is used for injectors. With no additional flyback protection. Driven with logical level AFAIK. And with no single damage. It is local (Latvia) construction ECU based on same Atmega128. One-men project from hardware up to GUI. Wise man! Scientist in nature.
Here is some description in Latvian:
http://www.elfa.lv/forum/viewtopic.php?f=32&t=1984&st=0&sk=t&sd=asf
Gints
-
well the forum has been down for over a week and a lot has happened ;D
It took me two days to tune the idle and then to get it to a point to drive, all be it with the aid of the wideband and EGOC ;D
Engine idles at 850 - 900 rpm
Been for an MOT today and passed. Emissions of 0.6% CO and 222ppm HC ( basic test, its only a Mini)
Only managed to get it to 49 % throttle but I really dont think I need any more for general driving :D
It growels under acceleration, that'll be the ITBs ;D
Dyno booked for next tuesday, just going to put some miles on it this week, nice gentle driving as the map is rather rough and the wideband is working alot, but its driveble without damage to the engine as long as you take it reasonably gently. Its difficult to hold back though :D
http://uk.youtube.com/watch?v=sqbJhBPS6-w
This topic is now closed, and I still have no idea what the problem really is :-\ but it works, I couldnt have done it with out help on here, Thanks.
Sprox
-
Sounds nice and smooth.
Glad that you got it sorted, shame it took as long as it did and that you didn't get to the bottom of why.
-
This is the resulting carnage ;D
(http://www.turbominis.co.uk/uploads/i/wh909-oe1/sprocket-1221.JPG)
-
Well, at least the fets make an interesting key ring.......well, interesting to the people who read these forums at least :p
-Gavin
-
I had customer who installed vems and had permanent fet failures (HighZ injectors). After second repair (I have replaced fets and drivers) i decide to modify vems to be "undestroyable".
I decide to use omnifets (VNP14N04). These FETS are very cool, have very good parameters, are thermal protected, short circut protected, ESD protected, and what is important they have integrated voltage protection so they can work without flyback wire :) (internal clamp at 42V). I have tested this feature and works perfect (i made some scope records) however in filnal configuration the vems works with flyback. The only disadvantage of vnp14N04 is that maximum input voltage (gate voltage in standard fet) is 18V. So the protection diode 1.5KE18 should be replaced by 1.5KE15.
I suggest this setup for HighZ injectors. It works great, can't be easy destroyed and VNP14N04 is quite chip (about 1Euro)
And what is most important the cars still works :)
Kuba
-
Why is it that every person here has a home-brew remedy that is more reliable than the FETs that are supplied with the vems ecu?
I had a Nissan KA24 install that ran fine for two weeks, then started losing a fet every week for 3 weeks in a row. I was told to "check the grounds" of course, did nothing to help.
I did three things at the same time to fix it (had no other choice, I do not have any spare FETs left!) -
1) Replace Honda resistor box (with another one that was in a running Honda, the box from the car was put into the running Honda and it is still running.
2) Replaced 20awg flyback wire with an 18awg flyback wire. The 20awg flyback wire was previously tested with a battery and a string of light bulbs.
3) The Honda resistor box had no connector, instead used ECxx terminal pins, pushed onto the resistor box terminals and strain relieved. This time we found the correct connector and installed it.
I think #3 was the problem.
-
I spoke too soon, this morning I get a call that the car is broken again.
It is happening on Cylinder #2 everytime.
1) If it is the injector shorting out, the resistor's 6.8k ohms should be enough to protect the ecu?
2) If it is the wiring to the injector, same as above, even with a dead short across the inj wires, the resistor is still 6.8k.
I'm going to make a new injector harness just to say that I did it.
I don't know what can be causing this. In the previous 8 years and 100+ installs/harnesses, I've never had any other brand of ECU (Autronic,Electromotive)with a failed injector driver. This ecu has a nice features-to-dollar ratio but I am really wasting a lot of time with the failures, one failure is too many. ???
-
Are you changed also FET driver chip?
I think damage comes from gate side. From power side everything seems strong enough.
Personaly I use other FETs than from webshop. STF20NF06.
I had very few problems with these. But was.
-
I have had only one FET failure and that was due to broken ground wire. Never since (around 30 VEMS installs).
When FET fails for some reason, it's usually replaced together with the driver as Gints suggested...
-
I haven't replaced any FET's, I simply use another available driver.
I don't (want to) work on the ecus, and I'm weary (and wary) of selling ecu's that have such a high failure rate.
I know the average customer wiring is often poorly done and can cause problems. But I still don't see the failures with self-installed other brand ecus like I do with these. And the other brand ecu's I install do not have these failures.
No person or product is perfect. You will have faIlures sometimes. That is not the problem.
How the company reacts (or doesnt react) to the failure, then it becomes a problem..
All you ever hear is to check the ground and flyback wires, and the injector ohms - if you even get a reply at all.
After that, nobody knows what to do.
I'm about ready to jump ship.
-
From a commercial point of view I can see your problem.
Fact is that a FET won't blow if its protected by a fuse and the flyback is correct.
In reality people hack OEMs (painfully badly fused), or make their own with usual results - then you're in a world of pain with a disgruntled customer.
-
From a commercial point of view I can see your problem.
Fact is that a FET won't blow if its protected by a fuse and the flyback is correct.
In reality people hack OEMs (painfully badly fused), or make their own with usual results - then you're in a world of pain with a disgruntled customer.
Thanks for the reply, you are a voice of reason as usual Rob.
What amperage value will each FETs handle? 1 amp? 5 amps?
It's definitely not feasable to fuse each individual injector.
In what scenario would the fuse protect the FET?
1) If the flyback was broken? or -
2) If there was another problem?
And if the flyback is OK, and the fuse protects the FET from (some other reason for) death, the customer is still left sitting with a dead cylinder.
I'm still trying to figure out the scenario that is causing injector #2 FET deaths on this Nissan.
I need a new ECU now because I'm out of drivers and have still not found the problem.
-
Those FETs are good for about 10Amps.
Sure the customer is left with a dead cylinder, they run on 3,5 or 7 but they're not getting petrol pouring into their sump, but they get to see a dead fuse which they can replace and slowly it can dawn on them that they have a problem without the VEMS needing to be returned for resoldering of components.
Its then the wiring thats the fault, and not the VEMS...
-
Interesting what is historical reason why VEMS don't use Shottky diodes in each FET flyback path, but usual "fast" diodes??? Shottky is way faster IMHO.
From other point of view FETs are self protected. There is something like TVS diode inside each FET. If FETs are avalanche rated.
It can't be that current killing FETS. Fets has much of safety margin for that. Failure should come from voltage. Question is from what side: gate or power side...
-
<<<Its then the wiring thats the fault, and not the VEMS...>>>
So can anyone explain exactly what would cause the fets to die like this?
1) Dead short across injector? (resistor box still provides 6.8 ohms to ecu)
2) Short injector power to ground?
3) short injector signal to ground?
4) short injector signal to non-resistor power?
I'll be removing the car's harness for complete disassembly to find the problem, if there is one.
-
Here's the datasheet (http://www.fairchildsemi.com/ds/FQ/FQPF20N06L.pdf) of the Fairchild FQPF20N06L.
It's good for 11A even at 100°C. The gate voltage maximum rating (VGSS) is +/- 20V so I doubt that causes the failures.
Over-current condition:
- If an injector driver blows and you use another available driver location (i.e. different EC36 output pin) but the same thing happens then it's clearly a wiring problem, probably a short.
Over-voltage condition:
- If you replace the driver and it regularly blows (i.e. same EC36 output pin) even if wired to a different injector, I would suspect a high voltage flyback problem. Check the 2 fast diodes from the driver output (ex. diodes D7 and D79 for INJ Q1) to the flyback pad. One of them might be destroyed or the soldering might be faulty.
(http://e23.montecastro.com/images/vems/vems-flyback-diodes.png)
The flyback diodes are SS24 (http://www.vishay.com/docs/88748/88748.pdf), 2.0A 40V Schottky diodes.
-
Again: V3 do not use Schottkys for some reason. What reason?
Instead ES2J is used. It is ultra fast p-n diode with recovery time 35-50ns (1000 times faster than 1N400x). But anyway Schottkys are faster. Or no?
P.S. here is a quote from Jörgen regrding my PWM peak and hold experiment:I'm afraid that you simplify this too much, it took a long time for us to find the actual problem with the PWM solution and your powerflyback variant doesn't address the main problem at all. It's pretty much identical to the prototype I built a few years ago (also without knowing what caused the random problems we sometimes saw). As you are obviously willing to experiment and have access to a number of cars to test on you should do some modifications that will address the problem. You need to replace a few diodes in the ECU with faster units. As it is VERY possible that the diode modification can cause other problems we decided to use the safe and in many ways better resistor solution instead.
The current diodes we have are about as fast as a common diode will get, this "ultra fast" diode is however sluggish compared to a Schottky diode. The Schottky type diodes will not survive as much voltage as the standard diodes but they are a lot quicker and they will prevent the very high inrush current that is the cause of most of the noise and that kills the FET when you PWM them. You should start replacing these diodes on all boxes you plan to PWM: D7,D8,D10,D16,D17,D25,D28 and D29. For added safety you should leave the original diodes in the D79,D169,D259,D289,D299,D179 and D89 positions, that way the box will pretty much go back to normal if the schottky diode would fail. You will need to examine the Schottky diodes on the test cars periodically to see how well they survive the beating they get. The behavior of the car will not change if the Schottky fails but the FET's will start to take a severe beating again.
I must say that I'm suprised that you used a FET for this, even if the part you use is fairly rugged I find that most FET's are much more fragile then an IGBT or NPN.
Note that any tests done with the 1600cc Bosch injectors are not likely to test how rugged the box is after it's modified. The 1600cc injector hardly put any load on the electronics and it's always used without a resistor. - Jörgen Karlsson, Gothenburg,Sweden.
http://www.vems.hu/wiki/index.php?page=MembersPage%2FGintsK%2FFlybackForLowZ
Gints
-
All 8 injector drivers on my V3.5 board (purchased May 2009, #2362) have the dual SS24 Schottky diodes installed per driver. They're DO-214AA package manufactured by Taiwan Semiconductor, with the S-shaped diode logo on them.
-
Interesting detail! I did not note what diodes come on latest boards.
Does anyone had FET problem with latest boards?
One more detail. Old Megasquirt V2.2 use ultra slow 1N4007 in this place. and I can't remember any damage of FET. Usually one chanel is loaded by 2 or more injectors.
-
Today i just had FET failure on newer (v3.5) board, two cylinders went for a diving at a time. Everything was just perfect, car run very nice, V3 harness used, XXX times checked grounding, 30V TS flyback, and they still died, just in front of a man who wanted to buy a car, while he was testing launch control... :-[
-
If you replace them like-for-like then you should replace the FET drivers too. It seems that the FETs damage the drivers somehow when they blow. Everything still works fine, but there is a vastly increased chance of the FETs blowing again.
The other solution is to replace them with IGBTs. They won't blow then, even if you chose to not replace the drivers. The downside of this is slightly increased power consumption and the additional cost of the IGBTs vs the FETs.
Check your injector config carefully - make sure the PWM is disabled (set to 100%) and the peak time is set to 25.5ms if you are not using PWM.
-
As the begining of this topic shows, I suffered numerous FET failures without any real reason, and no one can to this day tell me what the cause was other than the box being a self build. With the advice of one of the developers, I installed the IGTBs and been running since ;D
I did everything recommended to replace blown FET devices, I changed all the FET driver chips, 4 in total, the FET gate resistors and the FET driver capacitors. The loom was completely pulled apart to check for faults. The loom was ALL new wire and built as per the guide lines, injectors were changed, grounds were remade closer to the ECU, and it made not one blind bit of difference.
IGTBs would appear to be a quick solution where there appears to be no answer to the FET failures and when all other avenues have been exhausted. You MUST use ALL IGTBs for ALL injectors on the same engine. I would expect that if IGTBs start to fail in this situation that not all probable faults have been eradicated.
I have just replaced the last of the FETs in the box with IGTBs (excluding WBO2). I am now only 4 short on the ignition chanels for a full house :D
Looking back, I can see possible where the prime mover for these failures came from, but I am not clever enough to actualy understand it if it was. I used what I thought was a bipolar stepper motor, when infact it was a DC motor and a switch (four wires) I did not check the reistance. This smoked one of the stepper chip drivers. this was at first start at the time i suffered the first failures. While mucking arround, I split the grounds as strongly suggested NOT to do and then smoked the P259 chip, at which time I think it damaged the Mcp ADC chip, which showed up as EGT chip failures or at least I thought.
My black box has had a hard time of it, but its still going and I have become quite good at replacing surface mount chips with only the basic tools :D And I dont think its going to be the last I do :-\
-
My FET's stoped to blown after i installed better grounds...even after installing good grounds i think 2 of my IGBT's blown,one igbt destroyed my msd gm coil and i was out of wastedspark,i got myself 3x2 coil and i was back and running on wastedspark again,but the happiness didnt last long,again one igbt was dead...so i switch back to some VW/Audi single coil and i`m running it like that for a while now and didnt had a single problem....
-
I had more than 30 units installed and two cases with dead FETs. And both using Low-Z with built in PWM-ing.( It is strongly sugested not to do from developers).
But I have two differences in most of my installations: FETs itself - I mostly use STP20NF06.
And grounding. Usually I use separate wire for sensor grond. It comes from same place as power grounds, but from other bolt.