If you have tried to tune the a/f on a dynojet with the dynojet brand a/f sensor, you probably have thought the VEC2 (or other)wasn't working, or working incorrectly. You can make large changes in injector duty cycle and you do not get corresponding changes in a/f, this is especially true at lower rpms.
This is not a criticism of the dyno, the hp data is not the issue, just that a/f tuning is is made difficult (especially at lower rpms) due to the slow sampling rate of the O2 sensor.
Here is a very simple answer from the chief engineer at Innovate.
"From Klatinn:
These discrepancies have been reported many times, but always with a Dynojet WB and Dynojet Dyno. Without a time based recording (not RPM based) it's difficult to see. But we did a test a while ago with another wideband. On an inertial dyno, where the RPMs rise during the test, the engine showed very lean in the low RPMs and then converged (richer) with the LM-1 readings at higher RPM. Taking the same car to a brake-type dyno, with the same wideband, same tune, that wideband showed very lean at high RPM, but converged with the LM-1 at low RPM. The car was tuned to a flat AFR. But on the brake-type dyno the engine was revved up without load and then load increased while RPM dropped.
With time-based recording it was easy to see what happend. The car was in closed loop (14.7) at the start of either run. The other wideband was so slow responding that it took some time for it to go to the real AFR from the closed loop AFR of 14.7. You can check that for yourself easily. Do the run with both LM-1 and Dynojet WB. In the middle of the run, switch the ignition off while keeping your foot down. The injectors will shut off and the engine will just pump air (infinite lean). Observe how fast each wideband reacts. It won't show on an RPM based recording, you have to read the widebands directly.
The exhaust pipes to the WBs should fill with air within about 5-10 engine revs. Because of the location of the WB's as you described, it's even less, 2-3 engine revs. At an assumed 3000 RPM shutoff, that would be within less than 100 milliseconds.
With a slow wideband and an inertial dyno you can't tune AFR because the slow response time "smears" the initial lean AFR from idling over the initial RPM range. Because of that, fueling changes at those low RPM numbers will have very little effect. By the time the wideband reacts the engine is already out of the RPM range where you made the changes.
A slow wideband will give you smooth AFR plots, but they are usable only if you can keep RPM and load steady. "Slowness" is basically a response that's the running average over the run.
A lot of WB manufacturers give the response time as a "t63" number, for example t63 is 100msec. This is an engineering term which means that for a sudden step-change in AFR the wideband takes 100msec to traverse 63% of the distance to the new number. Then it takes another 100msec to traverse 63% of the remaining distance, then another 100msec for the then remaining and so on. The net effect is that it takes about 5 times t63 to travel 99% of the way to the correct number. The 100msec number I gave is typical for a very good, fast wideband using the conventional analog pump current measurement method.
Example: For a real AFR change from 20 AFR to 10 AFR, the distance is 10 AFR (20 - 10). A t63=100msec wideband will show 13.7 AFR after 100msec, 11.4 AFR after 200msec, 10.5 AFR after 300msec, 10.2 AFR after 400msec and so on.
But in this example the AFR after the step is constant. If the AFR changes, the wideband will always play catch-up but never reach the real number.
We don't spec a t63 number because we think it's deceiving and confusing for a typical user. We tested the LM-1 to have a response of < 110msec from free air to correct Lambda 0.8. This is MUCH faster than almost anything out there. The LC-1 is even faster.
Regards,
Klaus"
This is not a criticism of the dyno, the hp data is not the issue, just that a/f tuning is is made difficult (especially at lower rpms) due to the slow sampling rate of the O2 sensor.
Here is a very simple answer from the chief engineer at Innovate.
"From Klatinn:
These discrepancies have been reported many times, but always with a Dynojet WB and Dynojet Dyno. Without a time based recording (not RPM based) it's difficult to see. But we did a test a while ago with another wideband. On an inertial dyno, where the RPMs rise during the test, the engine showed very lean in the low RPMs and then converged (richer) with the LM-1 readings at higher RPM. Taking the same car to a brake-type dyno, with the same wideband, same tune, that wideband showed very lean at high RPM, but converged with the LM-1 at low RPM. The car was tuned to a flat AFR. But on the brake-type dyno the engine was revved up without load and then load increased while RPM dropped.
With time-based recording it was easy to see what happend. The car was in closed loop (14.7) at the start of either run. The other wideband was so slow responding that it took some time for it to go to the real AFR from the closed loop AFR of 14.7. You can check that for yourself easily. Do the run with both LM-1 and Dynojet WB. In the middle of the run, switch the ignition off while keeping your foot down. The injectors will shut off and the engine will just pump air (infinite lean). Observe how fast each wideband reacts. It won't show on an RPM based recording, you have to read the widebands directly.
The exhaust pipes to the WBs should fill with air within about 5-10 engine revs. Because of the location of the WB's as you described, it's even less, 2-3 engine revs. At an assumed 3000 RPM shutoff, that would be within less than 100 milliseconds.
With a slow wideband and an inertial dyno you can't tune AFR because the slow response time "smears" the initial lean AFR from idling over the initial RPM range. Because of that, fueling changes at those low RPM numbers will have very little effect. By the time the wideband reacts the engine is already out of the RPM range where you made the changes.
A slow wideband will give you smooth AFR plots, but they are usable only if you can keep RPM and load steady. "Slowness" is basically a response that's the running average over the run.
A lot of WB manufacturers give the response time as a "t63" number, for example t63 is 100msec. This is an engineering term which means that for a sudden step-change in AFR the wideband takes 100msec to traverse 63% of the distance to the new number. Then it takes another 100msec to traverse 63% of the remaining distance, then another 100msec for the then remaining and so on. The net effect is that it takes about 5 times t63 to travel 99% of the way to the correct number. The 100msec number I gave is typical for a very good, fast wideband using the conventional analog pump current measurement method.
Example: For a real AFR change from 20 AFR to 10 AFR, the distance is 10 AFR (20 - 10). A t63=100msec wideband will show 13.7 AFR after 100msec, 11.4 AFR after 200msec, 10.5 AFR after 300msec, 10.2 AFR after 400msec and so on.
But in this example the AFR after the step is constant. If the AFR changes, the wideband will always play catch-up but never reach the real number.
We don't spec a t63 number because we think it's deceiving and confusing for a typical user. We tested the LM-1 to have a response of < 110msec from free air to correct Lambda 0.8. This is MUCH faster than almost anything out there. The LC-1 is even faster.
Regards,
Klaus"
