Why is cylinder #4 so lean?

[Moonbat]

The plane is a 1967 M20E. The engine is a Lycoming IO360 A1A. It is turbo-normalized. The engine, prop and all major components have recently been completely overhauled to be "0" time. This overhaul process included the fuel servo, turbo, prop governor, all 4 cylinders, valves, cam shaft, fuel pump, spider, addition of Surefly ignition. For good measure, the GAMI injectors were all replaced with new ones giving us a clean starting point. The jets are, in order from 1 to 4: A A D J. We came up with this selection of injectors based on the GAMI spread seen shortly after the engine was overhauled / broken in. That GAMI sweep showed #4 was STILL the leanest cylinder, just as it had been prior to all the overhaul work.

A "baby jar" test was run and showed no appreciable variation in the  fuel delivered to the cylinders. (Photo of the jars with fuel in them is attached.) After all of this, the GAMI sweeps are still showing that cylinder #4 is too lean relative to the others. The spreads on the sweeps are .8 - 1.2. The data from the EDM930 was sent to Savvy and they concur that cylinder #4 is STILL TOO LEAN! (See Savvy's report attached.) I checked for induction leaks by doing a GAMI sweep at MP = atmospheric pressure and then another at MP = 10" Hg lower in an attempt to really force the induction leak occur so we could see the difference in the leaking cylinder - no luck. There did not appear to be an induction leak. 

The only thing I can think of at this time is that the turbo-normalizer setup in my plane "jams" too much air into cylinder#4. Do any of you have a t-n setup in this same engine type and have this same problem? 

Short story... I am befuddled and I don't know where to go with this investigation. Any suggestions? 

(On the positive side, the plane is 20-30 knots faster now that it is really running on all 4 cylinders instead of 2.5 cylinders and a bad cam lobe! Maybe I should stop whining about this, fly ROP and enjoy the ride. But what's the fun in that?) 

2020-11-23 Scott Winick N3215F.pdf

Edited November 27, 2020 by Moonbat
RICH should have been LEAN

[bfreelove]

If the lines from the spider to the fuel injectors haven't been changed I would check for a restriction there.  Based on your jar test I think you'd expect to see more fuel in the number 4 jar with it having the largest injector nozzle.  1-3 seem to correlate pretty well for the nozzle sizes.  

[Moonbat]

The baby jar test was done about 6 months ago when we were first digging into the problem of a too lean cylinder#4. At that time, the nozzles were E D G H, which is how they were when I acquired the plane. As is the case now, the nozzle in cylinder #4 was the largest. The previous owner had been messing with the GAMI injectors, too, but with a different set of criteria. He was not trying to set things up to fly LOP, he was trying to equalize the EGT's. Don't ask!!!

In view of that, and the fact I was worried about the nozzles having been tampered with (reamed out, or damaged in some other way) I asked GAMI to allow me to start over again. And that is how we got started with the current, brand spanking new, roster of nozzles: A A D J. I am going to re-run the baby jar test this week.

Two things have been suggested to me by Savvy. 1) the GAMI sweep is an accurate test, and, 2) the baby jar test is a crude test. Perhaps there is an obstruction, a kink, or an extra bend in the fuel line to cylinder# #4. Another carefully administered baby jar test may reveal something along those lines (no pun intended). 

[RLCarter]

Check the intake runners. Not enough fuel OR too much air, both will cause a lean condition. 

[carusoam]

The Gami spread is a real science...

The baby food jar is a visual representation of how the fuel system is working, today... it also limits the problem to only fuel flow issues, separating out the unknown air flow challenge... (if there is one)

Starting fresh... would include the JPI data shared here...

and re-run the baby jar test as well... to demonstrate the knew Gamis connected to the fuel lines...

Mixing old data and old pics is going to cause some confusion to where you are trying to go...

 

In the end... you are trying to get the same air/fuel ratio.... to all of the cylinders...

You started with a set of Gamis... and updated to a new set of Gamis...

Tuning a set of Gamis is often a two step process... no, not including the first set of Gamis... 

There are methods for testing intake systems for leaks... it looks like you have solved this too...

Want to post some data?

JPI data from a few different cruise altitudes would be interesting... Some really good run-up data, done slowly, can be extra helpful too...

If you are seeing different Gami spreads at different altitudes... the JPI data won’t be shy about showing it...

Load to Savvy, click the share button, copy link here...

Then add the details to each of the runs...

See if we are seeing the same thing you are...

Getting all of this organized takes a lot of effort...

The results are worth it... 

PP thoughts only, not a mechanic...

Best regards,

-a-

[corn_flake]

I'm being trying to the troubleshoot a similar issue as well.  The injector size in my IO-360-A3B6D is HJFB (1-4).  My cylinder 2 is the leanest cylinder but running the largest injector from Gami (J).  Cylinder 3 and 4 do not seem to peak at all.  I'm only willing lean until the engine starting to run with a hint of roughness.  Perhaps I could get a CFI with me so I may actually pull that mixture lever way back... .   

By the way, what's the actual cursing speed you are getting now at LOP?  20 to 30 knots faster is impressive.  How about ROP?  

[carusoam]

Flake,

Got any JPI data you want to post?

Best regards,

-a-

[corn_flake]

No JPI data, but I did post the Savvy graph here in the past.  John at GAMI ask me to do a lean test in flight.  I was brave enough to pull way back at least two times...  you can see those at 23 minutes into the flight.  It sure look like cylinder 2 went on strike.  

https://apps.savvyaviation.com/flights/3937839/1dfdc7f2-5f20-406b-853c-767e1d8c29bc

[carusoam]

Flake,

Looks like I grabbed one from your set of recent data...

-a-

[carusoam]

Then I reorganized it to really focus on just EGT and CHT...

There are a couple of things going on that we could discuss...

Start a new thread for this...

You have great instrumentation collecting data...

Essentially, you have four cylinders that should be showing near identical data... within reason...

near the 30minute mark... it looks like an attempt at leaning didn’t go very well...

Enriching, returned things to being close together... the leaning again the odd spread of EGTs occurs again...

If things are really well balanced... an engine will lean smoothly until it just stops running...

Are you familiar with the Gami spread test?

If yes, what is your Gami spread?

If you mentioned two cylinders don’t reach peak... I get it... now... 

In that case... cleaning fuel injectors is typically the next step...

Baby bottle test before and after... cleaning...

Best regards,

-a-

[corn_flake]

I didn't mean to high jack the thread.  @carusoam I also posted the wrong link.  I meant to post a link to a flight from last year where cylinder 2 EGT goes way down if I lean too much.  I'll start a new thread.  

[Moonbat]

Here is the EDM930 data from the most recent flight, including 3 GAMI sweeps, all done with the new set of injectors, at an altitude of 5K. (Special note: Please don't ask for a faster sampling rate for the data. This is as good as it gets for my particular model and firmware. In an attempt to partially compensate for the slower sample rate, I made the adjustments to the mixture knob precisely as follows: 1/2 turn CCW (leaner) every 30 seconds. On my particular unit, it appears to sample 1 time every 6 seconds in "normal" mode, but when I enter LF (Lean Find mode) it changes to 1 sample per 1 second. I believe in newer units, it can sample at 6sps. Not the case with my unit.) 

U201120.JPI

Edited November 25, 2020 by Moonbat

[carusoam]

8 hours ago, Moonbat said:

Here is the EDM930 data from the most recent flight, including 3 GAMI sweeps, all done with the new set of injectors, at an altitude of 5K. (Special note: Please don't ask for a faster sampling rate for the data. This is as good as it gets for my particular model and firmware. In an attempt to partially compensate for the slower sample rate, I made the adjustments to the mixture knob precisely as follows: 1/2 turn CCW (leaner) every 30 seconds. On my particular unit, it appears to sample 1 time every 6 seconds in "normal" mode, but when I enter LF (Lean Find mode) it changes to 1 sample per 1 second. I believe in newer units, it can sample at 6sps. Not the case with my unit.) 

U201120.JPI 24.99 kB · 3 downloads

Getting closer...

My iPad didn’t know how to handle the .JPI file...

What worked above, in Mr. Flake’s post... 

I believe he uploaded the JPI files to Savvy.com...

At Savvy, there is a button to click called ‘share’...

Then post the link here...

Its really a cool way to share what you were seeing... and free... (for that part)

Does JPI have an IPad app? I use a dos/windows laptop to download data...

Best regards,

-a-

[Moonbat]

I’ve shared the data via Savvy’s share feature.

https://apps.savvyaviation.com/flights/4533779/a716f153-fd4f-4750-8faa-7dfd77d5add1 --- Flight 2020-11-20 20:12 Z.    

   GAMI sweep #1 from about 22 - 25 minutes into the flight (JPI data time 20:34:00 - 20:38:00) yields a SPREAD of ~.8 gph

   GAMI sweep #2 from about 29 - 33 minutes into the flight (JPI data time 20:42:30 - 20:45:40) yields a SPREAD of ~.9 gph

   GAMI sweep #3 from about 38 - 42 minutes into the flight (JPI data time 20:50:00 - 20:54:30) yields a SPREAD of ~.9 gph

The data file (*.jpi) is the native format of JPI’s EDM 930. The tool to analyze it in its native format is JPI’s data analysis tool, EZ Trends. Windows only, afaik. 

Edited November 26, 2020 by Moonbat

[carusoam]

Oops...

Almost there....

Not sure what happened here.

Looks like the share button isn’t pushed..?

-a-

[Moonbat]

The folks at Savvy are working on it. The only "share" that was possible was one that shared (made public) about 305 flights. Seemed like a bad idea. 

[takair]

Moonbat, in your first post, in caps you say STILL TOO RICH.  I assume you mean too LEAN?  
 

The injection system in our planes is simple, yet can be confusing, add the turbo and it can get complicated....so my thinking may be off.  The fuel is drawn from the injector on the intake stroke.  So, even with the turbo, an intake leak might have the effect of drawing air into the cylinder but  not adequate fuel from the injector (drawing air from the leak vs fuel from the injector).  I admit not looking at all of your data, but.....are the symptoms the same when you are “under boost” vs not?  In other words, does it behave the same down low and up high?  Any chance you have a standby vacuum hooked up to that cylinder that may be leaking?  Manifold pressure is usually on the other side...so not likely, but there are plugs for those on the jugs where MP line would be, be sure they are tight.  Of course check the riser at both ends for leaks.  Anyway, haven’t thought this through entirely, but maybe worth a thought.

[Moonbat]

I did mean "too lean". Thanks. Corrected now. I will check all your suggestions out soon and report back. As for with and without turbo, I did GAMI sweeps at 6K and 8K at high MP and 10" lower just to eliminate that possibility. Same results. 

Edited November 27, 2020 by Moonbat

[carusoam]

Sharing 305 flights may not be the most focussed way to discuss one particular flight... 

But, I’m not sure if this is as critical as sharing financial data accidentally...

Sometimes it can be helpful to invite @kortopates to the conversation...  Paul has a lot of insight on how Savvy things work... and he knows engine ops really well... and is a CFII... and flys an M20K.... and... and...   

Go MS!

Best regards,

-a-

 

[corn_flake]

Uploaded Moonbat's JPI data using my account.  

https://apps.savvyaviation.com/flights/4539366/d0c05c1a-03ee-41a5-9251-8d9b96a61a8c

[carusoam]

Way cool, Flake!  

It takes a community to make this happen...

Go MS!

Best regards,

-a-

[carusoam]

Situation

1) 6k’ 22 - 25 minutes

2) 8k’ 29 - 33

3) 10” lower 38 - 42

 

And generally the Savvy graph looks like this...

4 EGTs and the TIT...  where all the three situations are pretty visible and their peaks pretty well distributed from each other...

 

Then a close up of situation one... if I can get it tonight...

Where we see #4 peaking a bit early and lower than the other three...

 

So... let’s call this an early look at the graph, and what we can see to start with...

Then we can look a bit closer at the data...

 

Overall....

 

Situation 1:

FF at peak...

1) 8.3 gph

2) 8.7 gph

3) 8.4 gph

4) 9.1 gph (out of the range of the other three)

a 0.5 GAMI spread is kinda good...  you have three with 0.4...

When using GAMI, and going through the effort... expect much closer...

My IO550 was running about a 0.1 with six cylinders...all done before i owned her...  

 

 

Situation 2:

EGT 4 (purple line) is peaking two minutes before the other three...

At the same time, the rpm data is getting really noisy...

 

 

Situation 3 does the same thing...

The purple line is leading the others to peak by a significant margin...

We can come back and put some numbers on it...

 

Moonbat,

How is that for data mining...

Kind of showing what is possible...

We can compare the FF at each peak... the Gami spread of the three that are close together... isn’t bad... but the one to peak early is similar behavior to a dirty fuel injector...

Any questions so far?

 

PP thoughts only, not a mechanic...

Best regards,

-a-

[corn_flake]

Moonbat,

Since the injectors were new from GAMI, and you also have new Cam Shaft, the only items left to check are the fuel line to cylinder 4 and the fuel spider it self.  Where those replaced as well?  

[Moonbat]

Spider, yes. Fuel lines, no. The fuel lines will be the focus of the next phase of investigation. First by performing a baby jar test. If no significant variation found (again), then mechanical cleaning, reaming and close inspecting for obstructions and/or kinks. A critical piece of the story is that cylinders 1-3 have gotten a little tighter after replacing the motor, injectors, et al. Cylinder #4 was and still is the outlier going back to data collected prior to the change to the new motor. That makes me think it has to be something that is still in common, i.e., the fuel lines from the spider. Or, perhaps, as I initially asked in this post, does this particular turbo-normalized set-up somehow jam more air into cylinder #4, making it the leanest? 
 

As for Anthony’s suggestion about contacting Paul K, he is the one at Savvy who is watching over all of this, and he’s looking into the quirk I think I detected with the “share” feature. 
 

Thanks for your help to this MS place!

[Yetti]

I would check the placement of the probes.  Maybe switch some.  Also use a hand held to take temps and verify.  It's always good to verify the instrumentation before ripping into the mechanical.   Also pull spark plugs and verify colors.  And rotate the spark plugs.

Edited November 27, 2020 by Yetti