EvoScan v0.98 Now with Power & Torque Graphing - Map Tracing completed also!
Evolving Member
Joined: Dec 2006
Posts: 276
Likes: 0
Second, comparing serial sample rates is totally a red hearing. A LC-1 typically takes discrete measurements about 400-500 times per second. Every measurement is converted into lambda, and there is no PID control loop involved. We use a patented measurement principle. When you are monitoring the serial output at 12 Hz, every *lambda* measurement in that interval is averaged and the result is output.
So, if lambda changes, the response time of the instrument is a worst case of about 83 mS serially. And, because it is an average of lambda measurements, abnormal combustion events show up as expected. You can demonstrate both these for yourself. A single miss (air, no hydrocarbons, infinite lambda) will show up as a one sample spike when logging serially. Similiarly, if you cut your ignition switch WOT, the log will shoot to O2 in a couple of samples, which is exactly what the engine is really doing.
On the other hand, a conventional, current based wideband does not take lambda measurements 400-500 times a second. It typically takes 1/lambda measurements about 4-10 times a second and feeds them into a PID style control loop. Because of the way the sensor works, the feedback in the system is not used for measurements (it is actually always dithering around the same point, equivilency), the output of the *control loop* is used as a measurement. This means that when lambda changes, the new value is not reflected at the sample rate (normally referred to as tao), but typically at 5 tao (.5 seconds on a good controller, .9 to 1.2 seconds on a poor one).
In other words, the speed that readings are reported serially is irrelevant, the limiting factor in the response time of the instrument is the measurement principle and the implementation of the PID style control loop. Again, you can prove this for yourself. The single puff of unburnt mix (infinite lambda) that showed up as a spike on the LC-1 shows up quite differently on other controllers. It is treated as 1/lambda (effectively 0) and fed into the control loop. Instead of a spike, you get a long hump of artifically lean readings. If you are getting occassional misses (idling too rich, or too cool for the points), you don't even get a hump. You just get steady readings that are 1.0 to 2.5 AFR too lean.
Similiarly, try killing the ignition switch at WOT and watch the output of a current based wideband. The engine filled up with O2 in a couple cycles, but the wideband will, as designed, take .5 to 1.2 seconds to catch up to that state. In both these cases it does not matter if you report the .5 to 1.2 seconds (or the hump) 10 times a second, 50 times a second, or a 100 times a second over a serial port, the values are still wrong. So the response time of an LC-1 reporting at 12 Hz is still, worst case, 6 to 14 times faster.
If you need faster measurements still, say for a specialized application like race engine valve tuning, you can switch to the LC-1's analog outputs.
-jjf
Thread Starter
Former Sponsor
iTrader: (4)
Joined: Dec 2003
Posts: 723
Likes: 1
From: TGA, New Zealand
jfitz, you're cool, chill out. the innovate products are awesome and fast, I agree, thats why I bought one. Do you get paid by innovate per line for forum posts or something? lol.
Evolving Member
Joined: Sep 2006
Posts: 152
Likes: 0
From: Germany
Come one. Even if you stand with the back to the wall it is not a good style to bite around.
...
Thread Starter
Former Sponsor
iTrader: (4)
Joined: Dec 2003
Posts: 723
Likes: 1
From: TGA, New Zealand
