PDA

View Full Version : Flow sensor response


MethoD
21-07-2016, 09:47 PM
I have the Aquamist injector "input" connected directly to an ECU Auxiliary output, rather than to one of the injectors. This way I have precise control over when it starts injecting and the rate. I have the flow meter 0-5v output fed back into the ECU so I can monitor flow and adjust ECU parameters based on the value.

I can view all this activity in the ECU logs and I have noticed that there is a small bit of lag between the FAV opening and the flow detected by the flow meter. Perhaps about 1/2 second.

While I expect a bit of lag when ramping up, I have also noticed that when the FAV duty cycle is quickly shut down completely, such as when lifting off the throttle (thus no flow past the FAV), it takes that same bit of lag time for the flow meter gradually taper down to 0.

What I am wondering is -- is the behavior I am seeing because of inertia of the flow meter impeller? I assume in the case of the taper down the answer would be "yes" since the flow meter is before the FAV thus fluid flow should stop instantly.

Thanks
M

Richard L
21-07-2016, 09:56 PM
This is the low pass filter doing its job. If you want fast response, you need to use the white wire where the raw signal from the sensor is taped. The output is a 5V square wave. You will need to create your own decoder. Fliw is proportional to frequency, not voltage.

The lag time is about 200mS:
Magenta = flow Cyan = signal

http://www.aquamist.co.uk/NEW/FAV-log.jpg

MethoD
22-07-2016, 02:13 AM
Ok, I think I understand. Since the 0-5v signal is conditioned there would have to be a slight delay between the actual and conditioned output.

I do have a spare digital input on my ECU that I could probably use the raw pulses, but I don't think I need to. Because of the way I am using the flow signal, if the signal is slightly behind the flow, then that should be ok.

If fact, because of the distance between the meth nozzles and the throttle body, there is no doubt a slight lag between the time the injection starts and the time the meth really starts hitting the cylinders. So in effect, the conditioned output from the flow meter may be more reflective of what's actually going on...

Thanks
M

Richard L
22-07-2016, 07:34 AM
Don't forget there is an offset of 0.5V and clips at 4.5V at the top. UPur actual flow range is 0.5-4.5V. The 0.5V offset is intended for monitoring the presence of the flow sensor, >4.5V is an indication of "loss of ground" supply to the sensor.

MethoD
22-07-2016, 11:50 AM
Yes, I have noted that in my testing. On mine it seems to sit at .55v when not flowing at all. Probably a slight difference in grounds at the Aquamist and ECU. At max flow it's hitting about 3.32v.

Can an actual flow rate be roughly calculated by the voltage output using a formula? So if
.5v = 0 flow
4.5v = 1000cc/min

and I want to solve for 3.32v (x)

first subtract the actual offset (.55v) from the known values, so

0 cc = -.05
1000cc = 3.95

use a simple formula:

x= 3.32 * (1000/3.95) = 840.5cc

Does that formula look right and does it give me a decent approximation of the actual flow?

I'm using the 0.9 and 1.0 jets and according to the flow matrix in the manual, at 20psi boost they'd be flowing about 797cc/min. I'm running about 17~18psi right now, thus my flow should be slightly higher. So 840cc/min seems pretty darn close to what it should actually be.

MethoD
22-07-2016, 12:34 PM
Or wait, maybe I should be subtracting .55 from the 3.32 value also?

With that, the formula would indicate that it's flowing about 701.2cc

Richard L
22-07-2016, 07:12 PM
formula:
Vflow = (Vout-0.55)x 1.25

MethoD
23-07-2016, 12:53 PM
What is your formula calculating? When I plug my Vout into it (3.32volts), VFlow is 3.4625. Do I use that number in another formula to calculate fluid flow?

Richard L
23-07-2016, 04:20 PM
Vflow = 0 to 5V (full span), instead of 0.55v to 4.55v. Once this is done, you can then calibrate your sensor. Make sure the SC trimmer is at 12pm. After this you can create the cc/v table.