Engine dies around 4500 rpm

[Blue]

I have a Microsquirt controller, code version 2.88, and I'm having trouble getting my engine to rev above 4,500 rpm. I'm attaching a datalog of the problem. It shows up at time = 515 seconds in the datalog.

Initially, the engine wouldn't rev beyond about 3,000 rpm, but I have been able to improve that some by putting in some very large numbers in the VE table. I had to put in VE values between 150 and 200 around 15 - 30 kPa and 2500 - 4500 rpm. Actually I got those values using Auto Tune.

The car idles well hot, and I'm testing with a water temp of around 195 deg.

The car is a Dodge Omni with the 2.2 N/A engine. Injection is TBI, one low impedance (2 ohms) injector. I have the Innovate LC-1 Wide Band O2 Sensor. I've tried 1, 2, and 4 pulses per engine cycle, without much difference. I'm using MicroSquirt to control both fuel and ignition.

Since it's a MicroSquirt unit, I have disabled PWM by setting the threshold time to 25.4 msec and current limit to 100%.

I don't know how to determine the time it takes for the injector to open, so I have used the default value of 1 msec.

I couldn't find a specification for the flow rate for my injector, so I removed the throttle body, turned on the fuel pump and injector and measured the flow rate that way. It worked out to 63 Lb/hr.

I have a feeling that I'm making up for some other problem by using the high VE numbers to get enough fuel into the engine, but I'm stumped at this point.

I would be grateful for any insight or suggestions.

Thanks.

[Blue]

Since I haven't received any replies to this thread, I thought I would add some more information and a few questions in hopes of eliciting a response.

I'm attaching a sketch of the oscilloscope trace I get at idle. I'm no electronics expert. In fact, this is only the second or third time I have ever used a scope. The scope probe is on the negative lead of the injector, which is the side that MicroSquirt pulls to ground to open the injector. I inverted the scope trace because it seemed to make more sense to me that way. The thing that concerns me is the spike at the end of the pulse. It goes to 60 volts above battery voltage (14.5V + 60 = 75V). Is that a problem, or can I ignore it? The spike is only 20 volts above battery voltage when the stock Chrylser controller is used in place of the MicroSquirt.

The shape of the pulse doesn't change as I increase the rpm. I assume that is good. The pulse width does get longer, and seems to agree with the value displayed in MegeTune.

I have the car set up so I can easily go back and forth between using the MicroSquirt controller and the stock Chrysler computer. The cars has always run fine with the Chrylser computer, reving all the way to the rev limiter at 6,000 RPM under full throttle.

This car uses a Bosch throttle body with a single injector. Is there any way to get a specification for the injector opening time and flow rate? I've searched on the web to avail.

Lastly, there seems to be a lot more traffic on the MegaSquirt forum than here on the MicroSquirt forum. Is it OK to post my MicroSquirt question on that forum in addition to this one?

As always I would be grateful for any advice or insight to this problem.

[Blue]

I wasn't able to attach the injector pulse jpg the first time. Hopefully, it works this time.

[grippo]

When you get to 4500 the afr seems to go lean. Have you tried to increase your ve just to see if it is a fueling problem. Then you can worry about why you need a high ve with a map of of only 30. Maybe the Reqfuel is too low. The pulsewidth of about 2.5 ms is not excessive and you have plenty of duty cycle left.

[Blue]

Thanks for the tip. I will keep increasing the numbers in the VE table to see if I can raise the rpm threshold when the problem kicks in. I guess it's alright to use the VE table as a tool to determine where the real problem lies. Since this is all new to me, I was a bit concerned about having to tweak things so much. I'm used to carburettors where a couple of jet sizes is usually sufficient to make a noticeable change.

I now realize that the accuracy of the value for the injector opening time plays a rather large role in how much the VE table needs to be tweaked. I'll keep searching the web for specs for this Bosch TBI injector.

I can't get back to garage until Wednesday evening. I'll post my results on Thursday.

[grippo]

The injector open time is most critical at low rpm/ low pulsewidths. If you use 1 ms give or take a few tenths, you will be in the ballpark. Very few people take the time to measure their open time and have no problems. You may need to increase Req Fuel.

[Blue]

I'm still struggling with the problem of getting my engine to rev above 4500. I discovered that this problem exists even when I'm running the engine in the garage with no load.

It seems that there is no combination of required fuel and ve table settings that will allow the engine to rev above 4500.

I'm using an Innovate WB O2 sensor, and it starts to go lean and I approach the critical RPM. I should note that when I connect everything back up to the factory engine controller, the engine runs fine.

I began to suspect that the MicroSquirt controller was unable to properly open the injector, so I connected an oscilloscope to the negative lead on the injector, which is controlled by th eMicroSquirt controller. At idle, the pulse width shown in MegaTune is about 3.6 seconds, and the trace on the scope is in agreement. I can increase or decrease the value in the ve table to make the mixture richer or leaner. I can see the change in AFR displayed by the WB O2 Sensor. There is a "sweet spot" at about 13:1 AFR, and the rpm drops off on either side of that setting.

Here is the symptom that seems to hold a clue to the problem. As I approach 4500 RPM, MegaTune is displaying a PW of about 8 ms, but the scope indicates that the injector is only being turned on for about 2.5 ms. Furthermore, the scope begins to display an additional short injector pulse abour 1ms in duration. This might just be a result of bad triggering by the scope, but I find it interesting that the problem kicks in about the same time that the engine begins to go lean.

I'm going to try to attach a datalog of my most recent run. This was obtained while the car was running in the garage, with no load. I'm also going to attach a copy of my ve table. I've only been adjusting it in the MAP range below about 40 kPa, so don't pay attention to the values above 40 kPa.

I would be grateful for any suggestions or advice.

Thanks.

[grippo]

The only thing I see on the datalog is that you are getting jumps in PW that seem to be triggered by noise on the tps line. This appears to be causing the pw to jump as high as 20 ms, even though the datalogged tps is not jumping that high, the jumps in pw are definitely correlated with jumps in tps and it is likely there are larger but very short jumps in tps and they are not being caught by the datalog. But the PW put out by the processor is otherwise consistent and stays near 8 ms and the afr stays near 13. So the injector can not possibly be in reality 2.5 or even 3.5 ms. Your PW also sems to be following your VE which is going from 40 to 55 while the PW goes from about 6 to 8 ms. If this is a fuel setting problem, then you should be able to go higher than 4500 rpm by changing your ve to go from say 50 to 65-70 in this region from 4000 to 4500 rpm and by raising it comparably above that. If it still stays at 4500, then it would appear to be some sort of driver problem, since you say the injector works fine with the stock computer. It would be nice if you could get a scope shot of the stock computer at 4500 and see what the pw looks like - both in length and whether it displays the breakup you saw with microsquirt.

[Blue]

Thanks. I'll follow your suggestions and respond here with the results.

[Blue]

I tried raising the ve in the 4000 - 4500 RPM range, but it didn't seem to have any effect on the problem. I asked my neighbor to come down with his digital camera and stand last night. While I raised the RPM in increments of 500, he took a shot of the scope trace and recorded the PW as I called it out. I did this twice, with seven images each time at 1400, 2000, 2500, 3000, 3500, 4000, and 4300 RPM. The first set of images are at scope settings of time = 1ms/div, and amplitude = 20V/div. The second set is at 5ms, 20V. I will attempt to attach those images to this message.

I've been concerned by three things that I've see on the scope trace. The first is the rather large voltage spike when the injector closes. It is in the range of 70 - 80 volts. Second is the second injector pulse that appears to the right of the main pulse in the 4000 RPM range. The third is the apparent discrepancy between the PW value as displayed in MegaTune and the length of the pulse in the scope trace as the RPMs go above 3000.

I also have scope traces with everything hooked back up to the stock Chrysler computer at idle and at 4000 RPM, and a datalog of the MicroSquirt test. I will attach those to another message following this one.

It looks like I may have a noise problem at this point. I have been careful to use a common ground point on the engine for all MicroSquirt and WB O2 sensor grounds, and the ignition control and fuel injector wires are routed separately from the signal wires.

I had to have this controller repaired twice by Peter Florence. Peter is a great guy, and he has gone out of his way to help me with this project. My initial plan was to use the MicroSquirt to control fuel only. I was using the signal from the Hall effect sensor in the Chrysler distributor to drive the ignition module from an 80's VW Rabbit, which in turn controlled the ignition coil. I attached the Opto in + to the negative ignition coil terminal and Opto in - to ground. I was able to get the engine running, but when it died, a spark would jump from the coil tower to the - terminal of the coil, which fried the opto in circuit on the MicroSquirt. Peter fixed it. Despite additional precautions to avoid a repeat of the problem, I fried the opto in circuit a second time. Peter fixed it again, and I gave up on using the VW ignition module in this fashion, allowing the MicroSquirt to control both fuel and ignition.

I have no way of knowing if these episodes caused additional damage to the controller, but Peter said that he thoroughly checked out the controller before returning it to me on both occasions.

I will add more images and datalog files later today. In the meantime, I would be grateful for any insight or advice.

Thanks.