Bench testing Lambda ECU

[Bitsyncmaster]

I've got two working spare Lambda ECUs and had two dead Lambda ECUs so I decided to set up a bench test.

Current draw by the ECUs is low around 40 ma. That's not counting the current draw of the frequency valve.

The two dead ECUs would not drive the frequency valve with any power (FV has 12 volts one side the ECU clocks the other side to ground). So I traced the fault back to the one IC in both units (LM2901). Replacing the IC got both "dead" ECUs driving FV power.

The two good units drive the FV at 68 Hz with a duty cycle of 50% with the O2 input floating and the thermal throttle enrichment signal floating. One of the "bad" ECUs is running 49 Hz and 68% duty cycle. The other bad ECU is 0 Hz (not oscillating). So it looks like my new IC may need some of the calibration resistors adjusted. I'm guessing there are 5 trim resistor values looking at the units. Probably the duty cycle when the thermal throttle enrichment signal is grounded which is also the wide open throttle switch. Then I'm guessing there is a trim to set the O2 sensor threshold voltage (should be around 0.50 volts).

Which I had a schematic. But this brings up a point I've always been thinking of. That plastic ICs are only guaranteed to work for 10 years. In reality they work a lot longer since the idle ECU has a lot of plastic ICs.

[DMCVegas]

Do you think that if we wrote to BOSCH, they would be willing to share the schematics for the various ECUs, as well as calibration procedures?

[Bitsyncmaster]

I would not think Bosch would help us out.

I found the one bad unit, I swapped the readings when I wrote them down so it seems to be working and calibrated the same as the good two units. So its looking like the new IC change did not upset those two calibrations. The other bad one that is not oscillating must have a bad capacitor.

[jamesvnz]

Do you think that if we wrote to BOSCH, they would be willing to share the schematics for the various ECUs, as well as calibration procedures?


with this sort of thing I tend to follow it up as there is always no harm in asking :-) and it also depends who you get on the other line.

maybe they will help a delorean owner and if they did they may get business from other delorean owners. who knows

[Bitsyncmaster]

Got the second "bad" unit fixed. One of the small signal transistors was shorted. I replaced it with a general purpose PNP 2N2907 and it also is working the same as the others, 68 Hz and 50% duty.

I grounded the WOT signal and duty went to 40% (my meter reads backwards) it reads 60% in the car with my Snap-On meter.

So now I just have to setup a signal input I can vary above and below 0.50 volts DC. Below it should swing the duty to the minimum duty value and above to the maximum duty value

[FABombjoy]

Not sure if it's useful for what you're doing, but grounding terminal 11 will get you nearly 100% D/C

[Jonathan]

Dave,

Are there any in-car tests one could do to determine if they have a failed Lambda ECU as you've described? I mean with as little removed or altered in the car as possible. Like say just by disconnecting the frequency valve and taking readings on that connector only?

[Bitsyncmaster]

Dave,

Are there any in-car tests one could do to determine if they have a failed Lambda ECU as you've described? I mean with as little removed or altered in the car as possible. Like say just by disconnecting the frequency valve and taking readings on that connector only?

As long as the frequency valve is buzzing, your driver transistor is working and your oscillator is working. Then you can just take the readings off the orange monitor wire circuit. I used my multi-meter to read the frequency of 68 Hz. Also used it to read the duty cycle but you get the same reading dwell but dwell is 90% of the value of duty cycle. So yes you can do all the testing that I have done in the car. You can also do the battery test on the O2 signal. I plan to actually read the threshold voltage of the O2 on my bench test but really the battery test will tell you if it's working.

Now there are lots of filtering for noise in the circuits and those won't get tested. But those design values are done with caps and resistors which have very low failure rates.

[Bitsyncmaster]

Not sure if it's useful for what you're doing, but grounding terminal 11 will get you nearly 100% D/C

I will test that. Wonder why they wired that to and external pin.

[AirmanPika]

Semi related question.

I've been rebuilding the electronics in vintage audio gear for several years now. Would the ECUs in our cars benefit from rebuilds also? I'm thinking of the aging capacitors and any other component that may have been strained over 30 years. Certainly reflowing of the solder joints but anything else?

I'm aware we could simply create new equipment to replace the old stuff (hence why I bought a new DMCH fuel pump) but if the existing setups are still viable but aged I don't mind doing the work to refurbish what I have...same goes for any other electronic component in the cars on a circuit board.

[Bitsyncmaster]

Semi related question.

I've been rebuilding the electronics in vintage audio gear for several years now. Would the ECUs in our cars benefit from rebuilds also? I'm thinking of the aging capacitors and any other component that may have been strained over 30 years. Certainly reflowing of the solder joints but anything else?

I'm aware we could simply create new equipment to replace the old stuff (hence why I bought a new DMCH fuel pump) but if the existing setups are still viable but aged I don't mind doing the work to refurbish what I have...same goes for any other electronic component in the cars on a circuit board.

I was thinking the one chip in the lambda ECU maybe should be replaced with a new one (later manufactured chip). The LM2901N is a quad voltage comparator which may make aging change some things but I found the new chip did not change calibration. But even with a lambda ECU failure on the road, you could just adjust the mixture to get you home or carry and extra ECU. The lambda ECU does not heat up much so soldering seems to hold up well. The units are conformal coated on the bottom of the PCB.

When I first got my D, I had a high RPM issue and I replaced all the ICs in the idle ECU in hopes of fixing it but it made no change. That problem turned out to be bad deceleration springs.

[flux3d]

I remember changing all the IC chips in my idle ecu when I got the car. I put in IC holders to hopefully help easy swap out in the future. My idle ECU wasn't controlling the idle valve at all and replacing the IC's worked wonders. Three years on still working perfectly. I've done similar upgrades to the door lock module as well to reduce standby current. At the end of the day these electronics are from the late 70s, in many ways the infancy of modern electronics. It's no surprise some of these units are now giving up. In fact is a miracle so many are still functional.

[Bitsyncmaster]

Now this is interesting data. I marked serial numbers on my units.

SN 001 is a good silver unit I had for a spare.
SN 002 is a good black unit I had for a spare.
SN 003 was a bad silver unit I fixed changing the IC.
SN 004 was a bad silver unit I fixed changing the IC and a signal transistor.

SN 001
Dropping below 0.50 volts on the O2 input drives the duty cycle to 96%.
Rising to 0.51 volts drives the duty cycle to 50%.
Rising to 0.56 volts drives the duty cycle to 12%.
Dropping to 0.55 volts drives the duty cycle to 50%.

SN 002
Dropping below 0.44 volts on the O2 input drives the duty cycle to 96%.
Rising to 0.45 volts drives the duty cycle to 50%.
Rising to 0.52 volts drives the duty cycle to 13%.
Dropping to 0.51 volts drives the duty cycle to 50%.

SN 003
Dropping below 0.50 volts on the O2 input drives the duty cycle to 97%.
Rising to 0.51 volts drives the duty cycle to 49%.
Rising to 0.56 volts drives the duty cycle to 13%.
Dropping to 0.55 volts drives the duty cycle to 49%.

SN 004
Dropping below 0.50 volts on the O2 input drives the duty cycle to 96%.
Rising to 0.51 volts drives the duty cycle to 49%.
Rising to 0.56 volts drives the duty cycle to 13%.
Dropping to 0.55 volts drives the duty cycle to 49%.

So it looks like the black unit is calibrated differently or it's changed it's calibration. Since the three silver units are the same, I'm guessing that is the calibration you want. I wonder why they made a dead zone running 50% rather than just one set point without hysteresis. Maybe that stops hunting idle or causes it.

[FABombjoy]

I will test that. Wonder why they wired that to and external pin.
A few different enrichment strategies, including:
-Boost enrichment on certain turbo cars / options
-Volvo removed the vacuum based enrichment plumbing to the WUR on 1982+ B28-based cars and went with an electrical switch to provide cold load enrichment. Details are documented in the B28 greenbook.

[Bitsyncmaster]

It's pretty much accepted that the narrow band sensor swings very sharply at 14.7 AFR at 0.50 volts. So even the black ECU I have would still hold 14.7 AFR. I'm running wideband into the stock lambda unit and my wideband is set to have that sharp curve at my set AFR. I don't see anything to gain with the threshold working like it does on the stock ECU but I don't think it would cause any problem unless your not adjusting your dwell to the stock 40 deg.

What I'm guessing is they had to make the ECU "deadband" to get the preset 50% when the O2 sensor has not warmed up and is high impedance.

[SS Spoiler]

I love it when Dave M talks dirty....

[Bitsyncmaster]

I bench tested a lambda ECU sent to me for repair. The unit tested good so not sure why the owner was not seeing good dwell readings. I opened the unit to inspect for soldering or burned components and found nothing wrong.

Dropping below 0.49 volts on the O2 input drives the duty cycle to 95%.
Rising to 0.51 volts drives the duty cycle to 49%.
Rising to 0.56 volts drives the duty cycle to 12%.
Dropping to 0.55 volts drives the duty cycle to 49%.

I think I will put the ECU in my freezer and then do another test while it's cold.

[Bitsyncmaster]

I tested that lambda unit the owner thought was bad after 2.5 hour soak in my freezer and after a 2.5 hour soak in my oven at 170 deg. F. Nothing changed and testing was good. When we had intermittent electronic problems back in work the cold and hot testing almost always found the problem.

It must have been a problem elsewhere but he had swapped ECUs and his problem was solved. Maybe a bad connector.

[Elvis]

oh cool, you finally found defect lambda ECUs ? :-)

I got an ignition ECU that was said to be bad, but I could not find out what is bad...

car runs without a problem with it... hmmm

[David T]

Often just the act of plugging and unplugging connectors "fixes" things. Corrosion can occur and as you connect and disconnect things that action wipes the terminals and then things start to work again. Especially on old cars and especially if they were in damp conditions. And even more so where the voltages are very low like the O2 sensor to the Lambda ECU. One reason I say it is rare to see a bad ECU but they often get the blame.

[Horsebox]

I think I just got to the bottom of a fuelling problem today. I did the battery test on the lambda ECU and got the following results:

O2 sensor unplugged - 50% duty steady (off the orange wire)
WOT switch pressed - 60% duty steady
O2 sensor grounded to block - duty increased to about 97% and stayed there
1.5V on O2 sensor - duty started decreasing and got as far as about 30% before it increased back to around 50% again and stayed there.

I think my ECU is shot based on the last test above. I reset the fuelling and took it for a decent spin and it drives like a champ again so I'm tempted to just leave it as is and run open-loop.

Is identifying and replacing the failed components inside the ECU something that a mere mortal without oscilloscopes can do?

[Bitsyncmaster]

I think I just got to the bottom of a fuelling problem today. I did the battery test on the lambda ECU and got the following results:

O2 sensor unplugged - 50% duty steady (off the orange wire)
WOT switch pressed - 60% duty steady
O2 sensor grounded to block - duty increased to about 97% and stayed there
1.5V on O2 sensor - duty started decreasing and got as far as about 30% before it increased back to around 50% again and stayed there.

I think my ECU is shot based on the last test above. I reset the fuelling and took it for a decent spin and it drives like a champ again so I'm tempted to just leave it as is and run open-loop.

Is identifying and replacing the failed components inside the ECU something that a mere mortal without oscilloscopes can do?

Yes that last test suggests the ECU has a problem. It should drop way lower than 30% and stay there. Since the ECU is basically working, I would guess you may have a bad capacitor. You can often find bad diodes, resistors and transistors using an ohm meter and a good ECU to compair the readings. But a bad capacitor and bad ICs won't show up bad unless there is a hard short. There is only one IC in the lambda ECU but many capacitors.

[David T]

Yes that last test suggests the ECU has a problem. It should drop way lower than 30% and stay there. Since the ECU is basically working, I would guess you may have a bad capacitor. You can often find bad diodes, resistors and transistors using an ohm meter and a good ECU to compair the readings. But a bad capacitor and bad ICs won't show up bad unless there is a hard short. There is only one IC in the lambda ECU but many capacitors.

While the test you did points out that the ECU is not functioning as it should you have not ruled out the wiring harness. Because of the small voltages for the O2 sensor circuit a bad connection can make a big difference. Before condemning the ECU as the culprit rule out any bad connections first. Also verify you had a constant 12 volts feeding the ECU during the test. A fluctuating supply voltage can cause the ECU to malfunction. Running "Open Loop" is not a great idea because to do it you have to set the mixture richer than if it was to run in "Closed Loop". Also hard to pass emissions testing and you can burn out your catalytic convertor. One other possibility is the battery you used for your 1.5 volt source is bad and did not maintain a full 1.5 volts during the test. I point out all of these external possibilities because it is rare the ECU goes bad. Not impossible but rare.

[Horsebox]

Yes that last test suggests the ECU has a problem. It should drop way lower than 30% and stay there.

I did the same test 18 months or so ago and it was fine. Dropped down to 10% or so and stayed there. Something has changed...

Hmmmm.

[Horsebox]

While the test you did points out that the ECU is not functioning as it should you have not ruled out the wiring harness. Because of the small voltages for the O2 sensor circuit a bad connection can make a big difference. Before condemning the ECU as the culprit rule out any bad connections first. Also verify you had a constant 12 volts feeding the ECU during the test. A fluctuating supply voltage can cause the ECU to malfunction. Running "Open Loop" is not a great idea because to do it you have to set the mixture richer than if it was to run in "Closed Loop". Also hard to pass emissions testing and you can burn out your catalytic convertor. One other possibility is the battery you used for your 1.5 volt source is bad and did not maintain a full 1.5 volts during the test. I point out all of these external possibilities because it is rare the ECU goes bad. Not impossible but rare.

I'll look into the connections etc tomorrow, when it's not dark and 0C, but light and 0C!

I did richen it up a tad, to about 1.5% CO. No problems for the CAT as there isn't one and I can run to 4% CO before failing emissions fortunately.

Thanks guys, useful stuff to check.

[Horsebox]

I found the problem. Bad lambda ECU ground connection at the bulkhead connector. I cleaned it up and now it passes the battery tests. I tried to get it to run closed loop but the O2 sensor looks like it is toast.

Thanks for the help, yet another electrical problem chalked up to bad connections.

[David T]

I found the problem. Bad lambda ECU ground connection at the bulkhead connector. I cleaned it up and now it passes the battery tests. I tried to get it to run closed loop but the O2 sensor looks like it is toast.

Thanks for the help, yet another electrical problem chalked up to bad connections.

I have managed in the past to "rejuvenate" the O2 sensor by removing it and heating it up with a propane torch. It burns off some of the contamination. Doesn't always work and even when it does the sensor isn't that great but it may get it going a little bit. At least to test things out but don't make final mixture adjustments. Always install with some Never Seize on the threads but be careful not to get any on the probe. Many replacement sensors will require cutting off the wire from the old one and splicing it onto the new one because the wire is too short and has the wrong end connector.

[cis6409]

I found the problem. Bad lambda ECU ground connection at the bulkhead connector. I cleaned it up and now it passes the battery tests. I tried to get it to run closed loop but the O2 sensor looks like it is toast.

Thanks for the help, yet another electrical problem chalked up to bad connections.

Hi
Which bulkhead connector was that you cleaned ? I thought the lambda ecu earth was that single wire that connects to the side of the inlet manifold just above spark plug 1 ..
Is there another one on the bulkhead connectors ??

Cheers
Shane

[Horsebox]

You're right but it heads into the car via the bulkhead connector I guess. It was the 7 pin white connector, the one hardest to get at closest to the bulkhead. Can't remember which pin, it's on your circuit diagram or I can look it up for you later. Cleaning it did the trick anyway!

[Bitsyncmaster]

I thought I posted about the lambda ECU I now sell but could not find it with the search. Anyway my unit replaces the circuit board in your stock ECU box and uses the original connector. So it is plug and play.

However Mark suggested a while ago that I should make it in a smaller box so people could get more room in the cubby for a large speaker.

So the one problem putting it in a smaller box is the heat generated by a 10 ohm resistor that drives the FV to ground. So I was doing some testing of the function of that resistor in the OE ECU. The result is the FV only gets 3.0 volts of power when energized. That is the peak voltage so if you measure voltage with a DVM you will read about half that. Just something to note if your troubleshooting a lambda problem.

So anyway that resistor results in FV current of 1.08 amps peak. The resistor drops 10.8 volts so it's power is 5.8 watts at dwell of 45 deg. Now I have to dissipate that heat in my smaller ECU box so that limits how small I can make it.

If it was just something for my car, I would put a switching power supply in the FV cable harness to drop the 12 volt power to 3.0 volts and then I could run without the resistor in the lambda ECU.

[Nioreh]

I thought I posted about the lambda ECU I now sell but could not find it with the search. Anyway my unit replaces the circuit board in your stock ECU box and uses the original connector. So it is plug and play.

However Mark suggested a while ago that I should make it in a smaller box so people could get more room in the cubby for a large speaker.

So the one problem putting it in a smaller box is the heat generated by a 10 ohm resistor that drives the FV to ground. So I was doing some testing of the function of that resistor in the OE ECU. The result is the FV only gets 3.0 volts of power when energized. That is the peak voltage so if you measure voltage with a DVM you will read about half that. Just something to note if your troubleshooting a lambda problem.

So anyway that resistor results in FV current of 1.08 amps peak. The resistor drops 10.8 volts so it's power is 5.8 watts at dwell of 45 deg. Now I have to dissipate that heat in my smaller ECU box so that limits how small I can make it.

If it was just something for my car, I would put a switching power supply in the FV cable harness to drop the 12 volt power to 3.0 volts and then I could run without the resistor in the lambda ECU.

Where can one read more about this lambda ECU? I can't find it on your web page. Does it do anything new/cool compared to the stock one?

[Bitsyncmaster]

I've not updated the web site yet and only did minor instructions. It now is plug and play and does the same thing as stock but it does stop the idle hunt if you get that hunt right when the lambda starts working (about 3 minuets after cold start). It also reads a more stable dwell reading than a stock unit. I did add an LED but you can only see it if your drivers side plywood cover is off. My plan is to make it smaller so you have more room for a speaker and don't need to return your old one. But that would require you to wire the new connector.

[WHO1DMC]

I've not updated the web site yet and only did minor instructions. It now is plug and play and does the same thing as stock but it does stop the idle hunt if you get that hunt right when the lambda starts working (about 3 minuets after cold start). It also reads a more stable dwell reading than a stock unit. I did add an LED but you can only see it if your drivers side plywood cover is off. My plan is to make it smaller so you have more room for a speaker and don't need to return your old one. But that would require you to wire the new connector.

What's the price on this unit?






Dave B.

[Bitsyncmaster]

$ 155 with your old unit returned.

[Bitsyncmaster]

I've been playing with the frequency valve power. When I found out that 10 ohm resistor in the lambda ECU produces a lot of heat (6 watts normally). I did some testing and found the stock lambda ECU running through the stock harness produces 2.56 volts on the FV when powered by the lambda (it switches on and off at 68 Hz.) My lambda ECU is a little higher at 2.76 volts because I use a MOSFET to turn it on and that has less voltage drop then a transistor.

You all probably know it bugs me to waste electrical power and produce heat. So I put together a 3.3 volt switching regulator and shorted that 10 ohm resistor. I now get 2.86 volts peak on the FV. The voltage looks cleaner than the resistor and 12 volt power.

However I noticed my dwell reading climbed about 7 degrees. So I learned that the FV could change the mixture if your car voltage has changed or is not stable. Just something to note when an owner says his dwell setting is not holding.

[Bitsyncmaster]

After thinking about it, I suspect the cleaner voltage is what caused the dwell to shift a little. When looking at an inductive switching voltage it's very messy. So the cleaner voltage is probably driving the FV a little harder.

I don't think anyone else will do this change but here is the info. Very cheap part converts the 12 volts to 3.3 volts with 91% efficiency. I just cut the red/purple wire at the lambda relay and that drives the 12 volt in and the 3.3 volts out goes to the the other end of that cut wire. You do need ground also. I mounted it on a hot start relay PCB and will pot it. I put a 10 uf 16 volt cap on the input and a 10 uf 10 volt cap on the output but it may not need those. Using the blank light relay socket.

[Bitsyncmaster]

After doing more testing, I decided the 10 ohm resistor going 12 volts is probably the way to go.

The current does not get a chance to peak. Although it seemed to work holding dwell but that explains why my dwell change. I checked the current through a 0.1 ohm resistor for the 3.3 volt system. The current for the stock system is checked on the 10 ohm resistor.

For you guys into electronics, I've attache scope photos.

[Bitsyncmaster]

I'm back thinking of any other ways to improve the lambda. I wonder of clamping those inductive spikes at the FV would make any improvements

I guess the best way to monitor it is to look at the O2 voltage. The narrow band voltage does swing drastically. That's how all narrow band systems work. I do have a test point on my lambda ECU that is driven by the op-amp I use to buffer (gain of 2) the O2 voltage into my micro.

[Bitsyncmaster]

Here is the O2 voltage times two and the FV voltage on my lambda ECU. As you can see the spikes due to the inductive load are -300 volts and +120 volts.