Widebands That You Would or Wouldn't Buy

[HelmutVonAutobahn]

Thanks.

On the tail-pipe sniffer; the Innovate part looks mostly good. But, It seems like the intake ( straight inlet ) will be against the tubing wall. Which is the lowest gas velocity area.

A little more on the new AEM units. From poking around a bit...

1. I had a bit of a user error. The CAN addresses are NOT 0x1d01, 0x1d02 ... etc. It just looks that way on the gauge UI. it's trying to say "ID_01" "ID_02" etc. The addresses are 0x0180, 0x0181... etc. Once I got that right, the CAN data came through, no problem. Maybe light the middle decimal point guys; so it doesn't look like a perfectly valid hex number ? Or, I can just RTFM

2. The heater really is a DC voltage. I measure 7.85v with the sensor in free air. It does change when you blow on it, etc., as it should. And, it seems to follow the warm-up ramp in the sensor datasheet. Probably done to get the high-current switching noise away from the sensor/cable.

3. The little switching regulator is for the LEDs. It runs at 3.5v The linear regulator runs at the normal 5.0v.

4. It is EXTREMELY fast to respond. Especially given how clean the signal is. The instruction sheet says 20ms. But, it covers the span of 20.9% o2 to Lambda 1.000 in one or two samples ( 100s/s on the CAN). when hit with a spray of CO2. So, the t63 time is more like 10ms. Which is how it looks on the DAC ( 500s/s ). But, the DAC only goes up to ~ Lambda 1.22. So, going by slew rate. This is faster than an LC-1 in "instant mode"

5. There does not seem to be a provision to increase the CAN data rate. It is obvious, looking at the DAC line, that there is more resolution to be had there, time-wise.

e.g.
Here is a shot of the DAC output when the sensor is hit with a blast of CO2. You can see the individual 2ms (500Hz) steps. It covers the entire span in 3 steps ( 6ms ). Allot of that detail is lost in the CAN output updating at 10ms per sample ( 100Hz ). It, obviously, knows more than it's telling Hopefully, there will be a firmware update. I'd rather have the CAN at 500Hz and the DAC at 100Hz. Although, that may flood the bus.

[HelmutVonAutobahn]

oops... Please excuse the ~40mv offset

[Fred]

On the tail-pipe sniffer; the Innovate part looks mostly good. But, It seems like the intake ( straight inlet ) will be against the tubing wall. Which is the lowest gas velocity area.

Yes, I was thinking that as well (but neglected to type it). Especially considering how small it is. If it were larger and against the wall, it'd be less of an issue.

[HelmutVonAutobahn]

Maybe some sort of S-shaped tube pressed into the end, to get the inlet closer to the centerline would help?

[DelSolid]

Sorry to revive a thread from a couple months ago but this thread is actually pretty good so it wouldn't have stayed down for long anyway.

A little more on the new AEM units. From poking around a bit...

4. It is EXTREMELY fast to respond. Especially given how clean the signal is. The instruction sheet says 20ms. But, it covers the span of 20.9% o2 to Lambda 1.000 in one or two samples ( 100s/s on the CAN). when hit with a spray of CO2. So, the t63 time is more like 10ms. Which is how it looks on the DAC ( 500s/s ). But, the DAC only goes up to ~ Lambda 1.22. So, going by slew rate. This is faster than an LC-1 in "instant mode"

We say 20mS but you are correct, the sensor response is much closer to 10mS. The quoted 20mS time is the time our entire test rig takes from solenoid initiation to sensor T63 response. The test rig response time has been measured to be between 7mS and 11mS but explaining that was thought to be to cumbersome for the average user so we left that part of the response time in with the total. Better to be conservative than quote unobtainable values.

Thanks.
5. There does not seem to be a provision to increase the CAN data rate. It is obvious, looking at the DAC line, that there is more resolution to be had there, time-wise.

Here is a shot of the DAC output when the sensor is hit with a blast of CO2. You can see the individual 2ms (500Hz) steps. It covers the entire span in 3 steps ( 6ms ). Allot of that detail is lost in the CAN output updating at 10ms per sample ( 100Hz ). It, obviously, knows more than it's telling Hopefully, there will be a firmware update. I'd rather have the CAN at 500Hz and the DAC at 100Hz. Although, that may flood the bus.

Yeah, the decision to limit the can bus transmit rate to 100hz was made to keep from flooding the bus. We have to be careful since a lot of our customers use multiple units and some even use 1/cylinder so we could kill the bus in a hurry if we are not careful. If you really want to see it, I can change yours on a one off basis to something faster (It can be reset using an in-house tool via the CAN bus). But you don't gain as much as you might expect. The analog output is updated at 500hz and gives a t63 time of 19-20mS and the 100hz CAN data from the same test has a t63 time of 25mS. It is slower but IMO the signal is better. I take the CAN output any day over the analog.

BTW, I cant believe you buy so many O2 controllers just to test them. Major props! I have 21 different current AFR/UEGO/Lambda controllers under my desk right now.

DISCLAIMER!!!!
YES: I work for AEM but I'm not speaking for AEM nor am I here to sell AEM stuff. I just love this stuff like you guys do.

[DelSolid]

I notice lots of references to CO2 or test gas or lack of test gas used for O2 sensor testing.

I always wanted a desktop test rig that I could access whenever I wanted while using my main computer, without going down to the shop to get to the huge custom-mix bottles we have down there.

Turns out there is a really good source of certified test gas available over the internet in small bottles that is specifically designed for exhaust gas calibration use. The California Air Resources Board (CARB) requires all cars here to be smog tested bi-annually, so the test stations are all over the place. The regulations require the test machines to be calibrated using specific test gasses and they are commonly available from lots of sources on the internet and can be easily shipped.

http://www.emissionsdepot.com/bar-97-calibration-gas/

They are much cheaper and easier than getting a full bottle of custom mix test gas and are small enough to be easily used. They come with the actual mix percentage on the bottle. While you don't have a lot of options on mix compositions but it is really good for some type of testing. for example, these are what I use for response testing.

Of course you still have to build a rig to deliver the gas but that's not hard.

John

[Fred]

Sorry to revive a thread from a couple months ago but this thread is actually pretty good so it wouldn't have stayed down for long anyway.

Amen, I have nothing against bumping old threads with a valid purpose. I bump 5 year old threads on other boards all the time :-D This one has to be continually updated to stay current, really. I've not updated the early post(s) for ages, but anyone coming in here should read the whole thing anyway. Thanks for your valuable additions! :-)

[bmotorsports]

I notice lots of references to CO2 or test gas or lack of test gas used for O2 sensor testing.

I always wanted a desktop test rig that I could access whenever I wanted while using my main computer, without going down to the shop to get to the huge custom-mix bottles we have down there.

Turns out there is a really good source of certified test gas available over the internet in small bottles that is specifically designed for exhaust gas calibration use. The California Air Resources Board (CARB) requires all cars here to be smog tested bi-annually, so the test stations are all over the place. The regulations require the test machines to be calibrated using specific test gasses and they are commonly available from lots of sources on the internet and can be easily shipped.

http://www.emissionsdepot.com/bar-97-calibration-gas/

They are much cheaper and easier than getting a full bottle of custom mix test gas and are small enough to be easily used. They come with the actual mix percentage on the bottle. While you don't have a lot of options on mix compositions but it is really good for some type of testing. for example, these are what I use for response testing.

IMAG0951.jpg

Of course you still have to build a rig to deliver the gas but that's not hard.

John

Neat idea for validation and testing. Do you happen to have the expected lambda values for some of the common mixes they offer?

I'm interested in what your test rig looks like, care to share?

On a similar disclosure note, I sell AEM & ECM widebands .

[DelSolid]

I'm interested in what your test rig looks like, care to share?

The goal for this tester was to try to come up with some method of getting repeatable system response measurements. I had built a few before but they were overly complicated and really just a PITA to work with. My latest version is a desktop setup that is almost fully automated and seems to give extremely repeatable results.

The basic design is two gas bottles (call them A & B) and the mechanism to quickly switch from one to the other, as closely to a step function as possible. And monitor the output to see how fast the sensor output reaches the new value.

Test gas A is flowed steady "on" at about 2 L/min through a solenoid. A signal is sent to Solenoid A to switch it off and to Solenoid B to turn it on. Solenoid B is a high current Nitrous solenoid with a solid state driver to minimize on time delay. The A gas stops flowing and the B gas starts flowing (also at 2L/min).

Both gasses are at room temperature and have the same flow so there should not be any thermal shock associated with the switchover.

The volume of gas after the solenoid and before the sensor itself has been minimized to pretty much as little as possible.

There is a pressure sensor screwed into the O2 sensor holder as well as exhaust ports for the test gas that has passed by the O2 sensor.

The solenoid trigger line, the pressure sensor output and the O2 sensor controller output is logged at either 500hz when I am just screwing around or 10Khz if we are collecting for real.

In early testing, the two gasses were run at slightly different flow rates to allow the pressure sensor to try to detect the arrival of the pressure wave from the second gas. Doing this netted a measured difference between the signal being sent to the solenoid and the pressure changing of between 8ms and 11ms, time and again. But when the gasses were flowing at the same rate this was not measurable. This time was the arrival time for the pressure wave, which should theoretically be faster or arrive earlier than the test gas itself so it should be considered a conservative estimate only. Ultimately, we decided to ignore this number since it could not be measured during the actual A/B testing with the same flow rate so just keep that number in mind and remember that about 10mS is spent just physically switching the solenoid and replacing the gas volume with the new gas.

We got the flowmeters from McMaster and the majority of the fittings at the same gas company we got the test gasses at.

For control I used an ECU I had and set up a program in it like an industrial controller. It would just flow gas A for a couple seconds to stabilize then do a series of A=>B, B=>A, A=>B step tests, 10 each direction. So each test generated about 30 seconds of data.

I determined the lambda value the controller stabilized on for each gas and normalized the results to those numbers. Then I used the T63 time (the time it takes the value to change to 63% of the new value) to determine the response time.

There is also a T90 time but many of the controllers could not hold their final reading to stay within the T90 threshold value so it could not be used.

We used a full size car battery inline with a 20A power supply to try to get the cleanest possible analog signal out of the devices and followed the mfg instructions but many still had incredibly noisy signals.

The test would be better if we could have heated test gas (say 500C) but that's not going to happen on my desktop.

[Hentai]

where is the gas going after it hits the o2? Is it venting to atmo though the holes that the o2 sensor is connected to?