• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 23 subscribers
  • Views 595 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPIC1801 Integration cycle

Bob Workman
Prodigy 30 points
Other Parts Discussed in Thread: TPIC8101

I am working on a replacement for a legacy knock system. For it I need the flexibility available in the TPIC1801.
I do however have some questions:

1. How quickly can the integration cycles occur and what is a reasonable max integration cycle?
   i.e how quickly after a int to hold transition can the next hold to int transition occur with out effecting output
       and how long can the integration window be without affecting knock detection?

2. What is the state of the OUT line during hold state? If I feed it to an alalog input will I get a garbage signal during hold state?
     Since I original wrote this I saw Innocentio's post and it looks like the OUT signal returns to zero during hold.

The legacy system I want to replace is for muscle car era EFI. These engines tend to be highly modified so the
OEM systems seldom perform as desired and offers no flexibility. These setups offer very little in the way of timing signals,
only an Ignition Reference Pulse fixed at 6 BTDC and no indication of which cylinder it is.

One of the options I am looking at is a constant poll to sample the knock signal level and pass this on to the ECU,
ideally via the analog OUT line. Reading the knock level off the SPI port and processing it to make a knock/no knock
determination and then outputing this as an appropriate signal to ECU is another more processor heavy option.

I first considered just a single constant fixed length poll triggered by the Ignition Reference Pulse but as a 50 degree knock window for a V8 varies
from 1.4 msec to 20 msec depending on speed so chances of missing the knock event are high if poll is limited to smallest window and overlap problems if larger.
This then suggests an untriggered constant poll of a fixed size window meaning that depending on integration window size and engine speed
I either integrate several knock windows into one reading or take several readings for one knock window.
Here the gap between integration cycles must be small or there is a chance that the knock event which itself is relatively short will be missed.

Any information or recommendation would be greatly appreciated.
Bob

  • 0
    TI__Mastermind 23810 points

    Bob,

    1.  The integration time can be programmed from 0.5ms to 10ms.  I will have to look into the minimum amount of time between HOLD to INT, but ultimately the length of the integration window will be set by your system.

    2. When in the hold state, the OUT line will be held constant.  You should be able to read to read the value of the analog OUT line from an external ADC.

    My initial reaction is that you should probably focus on catching the shorter knock events, since the longer knock events could be detected by combining several integration cycles.  I'll give you an update when I learn more information.

    -Clancy

  • 0 in reply to Clancy Soehren
    Prodigy 30 points

    I am still waiting on the HOLD to INT interval value.

    i.e. if I run a 1.4ms integration window how long until I can start the next 1.4ms integration window?

    You had replied " I will have to look into the minimum amount of time between HOLD to INT" but have never responded.

    My goal is to issue a constant stream of 1.4ms integration windows as close together as possible and then feed the OUT signal into the ECU

    knock in input.  Original input was a non tunable low-pass filter. I hope the TPIC will allow me to tune the signal to match altered engine conditions.

  • 0 in reply to Bob Workman
    TI__Mastermind 23810 points

    Bob,

    Thanks for the reminder, and my apologies for the delay.  I've been looking at the part some more, and initially it looks like the limit is essentially how long it takes to transmit the necessary information over SPI.  I'm planning on taking this to the lab and running some experiments myself to see if you will run into problems if you try and push the HOLD time down as much as possible.

    Have you had the chance to try this yourself?

    -Clancy

  • 0 in reply to Clancy Soehren
    TI__Mastermind 23810 points

    Bob,

    It doesn't look like the minimum hold time will impact you too much.  I don't start seeing problems until reaching a hold time of less than 10us.  Below this point, the hold state isn't always recognized by the TPIC8101, and so it continues to integrate.  Will this minimum hold time work for your system?

    -Clancy

  • 0 in reply to Clancy Soehren
    Prodigy 30 points

    This should not be a problem. Knowing the lower limit means I can ensure that this delay is always met. There may be some problem with extremely high RPMs where the ignition event is very small but as the legacy analog system begins to give false results at around 4500 RPM it is a moot issue.

     

    Thank you for your reply