The joys of engine monitor diagnosis

[Bob_Belville]

16 minutes ago, kortopates said:

I had to download your CSV data file, add an elapsed time column as =B15-TIMEVALUE("14:51:30") in order to get us on the same time line and then I could see what I believe you are referring too.

Notice that on Savvy's elapsed time line, my Gami spread data was between 1:01:24 to 1:01:53. The 1:01:53 elapsed time corresponds to to 15:53:23. 

After 1:01:53 window of the gami spread, indeed the FF does go back up to 9.4 for a few seconds, the last data point its at 9.4 is 1:02:14 elapsed or 1:53:44 actual time, the next 2 following seconds and data point is where Eztrends reported 3 & 4 peaking (again). But this is really invalid to go by that because. As you said, the leaning past peak the first time was done earlier as reflected by the gami spread reported above ending at 1:01:53  elapsed (or 15:53:23 actual). But only since the monitor is in Lean Find mode it continues to see it come back to peak in reverse with FF being enriched for a second peak. But if I simply expand the gami spread measuring window to include both peaks  it will report the same way EZTrends did (incorrectly).

Probably not saying anything differently than your last post but I was pretty confused till I dug into the details to resolve the time line differences and take a closer look.

I think a graphic does a much better job of illustrating it.

Thanks Paul. After my posts I noticed that the Savvy time line starts at 0 rather than Zulu time from the EZTrends, I'm sorry you had to convert in order to compare. After studying the results I'm pretty happy with the balance of the stock injectors. (I just replaced the fuel injection servo so I'm glad to see it seems to be operating as expected.) 

Is there a user accessible Savvy utility to generate a lean find graph and summary like the one you've posted here?

Thanks,

Bob

  

[kortopates]

49 minutes ago, gsxrpilot said:

I'm paying for the Savvy Analysis as well. How can I determine GAMI spread from the Savvy graph? Or should I just request analysis? 

 

39 minutes ago, Bob_Belville said:

Thanks Paul. After my posts I noticed that the Savvy time line starts at 0 rather than Zulu time from the EZTrends, I'm sorry you had to convert in order to compare. After studying the results I'm pretty happy with the balance of the stock injectors. (I just replaced the fuel injection servo so I'm glad to see it seems to be operating as expected.) 

Is there a user accessible Savvy utility to generate a lean find graph and summary like the one you've posted here?

Thanks,

Bob

  

Bob, No problem.

Bob & Paul - yes, you have access to all the tools I use. To measure the Gami spread first zoom in on series of gami spreads that you want to measure. I prefer to add FF to the graph as I pre-examine the data for quality. You don't have to incorporate the following check I do first, but its especially good for turbo's and any NA engines not running at WOT. So I first have a graph of MAP and RPM and check to verify MAP and RPM were reasonably constant the entire time. 0.5" MAP variation is acceptable multiple inches of MAP variations is not and very common on turbo with either fixed or pneumatic waste gates. But that's another tangent. Once satisfied only FF is changing, then click on the G for Gami icon which is highlighted in my screen shot above. That will eliminate all but the one Gami mode chart and why I pre-screen with MAP and RPM before entering Gami mode. Then in Gami mode click at the start of the time line - and see a cursor line labeled "Start" and then click again when FF is at the opposite range (high vs Low or Low vs High). Just be sure unlike the screen shot above to only bracket one peak with FF either going up from LOP to ROP or down from ROP to LOP. After both start and end cursors have been clicked you'll be able to see your Gami spread data in the data box - upper right corner. To get out of Gami mode and back to multiple charts just click on the X at the upper corner of the data box. There is also help published on all the tools available by clicking Help in the upper right corner.

Paul - feel free to request analysis at any time. But we ask you to fly the Savvy Test profile along with a full power takeoff and climb for analysis request so that we have multiple gami sweeps and LOP Mag test data to report on. With full Savvy test profile data you'll get a full report back. Cruise data alone doesn't give us much to discuss unless were looking at an anomaly. We especially recommend requesting analysis of your Savvy test profile data before and again right after annual to ensure all is good.  

Edited October 8, 2017 by kortopates

[gsxrpilot]

Thanks @kortopates!

Is the Savvy Test profile the document I found under Help > Flight Test Profiles? It looks like it suggests three GAMI sweeps and an inflight Mag check along with setting the JPI to 2 second sampling.

Additionally, since you fly the same airplane I fly, can you suggest a MP/RPM and Altitude you'd recommend for this flight test profile?

[kortopates]

Hi Paul  @gsxrpilot that's the profile under Help or http://content.savvyanalysis.com/static/pdf/SavvyAnalysisFlightTestProfiles.pdf 

We suggest no more than 65% power, I suggest 24"/2500 and 6K'. That equates to near 60% power in ours  and 24" at 6K puts ambient atmospheric pressure near the same as MAP which is advantageous because it prevents an induction leak for significantly affecting the results.   (But if we see evidence of an induction leak we'll recommend a follow up test - we just want to ensure we can get good gami results). Be sure to go slowly as each cylinder peaks and well into the LOP side (about 50 LOP if able).

With the JPI 900, I suggest putting it on 1 second sampling rate for the highest resolution - then re-boot the monitor before collecting the data. Then you need only do 30 sec per mag doing the LOP Mag test including Both in the middle to give time for EGTs to re-stabilize. But shoot for around 50 LOP when you do the mag test and I suggest doing it right after your last gami sweep from ROP to LOP so I can see exactly how LOP the test was done. That way if we do see missing we can tell if it was premature or acceptable.  

I don't mean to overly complicate the procedure, its just that the above helps to give the best recommendations for maintenance if we do see any anomaly's. 

[Bob_Belville]

25 minutes ago, kortopates said:

Hi Paul  @gsxrpilot that's the profile under Help or http://content.savvyanalysis.com/static/pdf/SavvyAnalysisFlightTestProfiles.pdf 

We suggest no more than 65% power, I suggest 24"/2500 and 6K'. That equates to near 60% power in ours  and 24" at 6K puts ambient atmospheric pressure near the same as MAP which is advantageous because it prevents an induction leak for significantly affecting the results.   (But if we see evidence of an induction leak we'll recommend a follow up test - we just want to ensure we can get good gami results). Be sure to go slowly as each cylinder peaks and well into the LOP side (about 50 LOP if able).

With the JPI 900, I suggest putting it on 1 second sampling rate for the highest resolution - then re-boot the monitor before collecting the data. Then you need only do 30 sec per mag doing the LOP Mag test including Both in the middle to give time for EGTs to re-stabilize. But shoot for around 50 LOP when you do the mag test and I suggest doing it right after your last gami sweep from ROP to LOP so I can see exactly how LOP the test was done. That way if we do see missing we can tell if it was premature or acceptable.  

I don't mean to overly complicate the procedure, its just that the above helps to give the best recommendations for maintenance if we do see any anomaly's. 

Paul @kortopates, I notice that my 930 which was set at 3 second sampling rate on the last flight reverted to 1 second sampling during the JPI lean find mode! After about 4-1/2 minutes it returned to 3 sec.

[kortopates]

Just now, Bob_Belville said:

Paul @kortopates, I notice that my 930 which was set at 3 second sampling rate on the last flight reverted to 1 second sampling during the JPI lean find mode! After about 4-1/2 minutes it returned to 3 sec.

Yes, the 900 series monitors automatically go into 1 sec sampling rate while in LeanFind mode so that  you'll get better resolution using LeanFind. We prefer to always be looking at the better resolution data and suggest setting it for its fastest rate - which 1 sec for these. But 2 sec (such as the EDM 830) work okay too. 

[gsxrpilot]

Hmmm.... might have to go fly tomorrow. Evidently I need to fly a Savvy Flight Test Profile. 

[carusoam]

Paul,

Nice job of taking a complex subject and making it easier to follow/understand...

Best regards,

-a-