What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

I think I have uneven airflow....

Up draft 540 sump

rv6ejguy said:
Here's what we're up against with typical vertical induction Lycomings:

540sump_zpsm7dxhs5t.jpg


Bad pairings, unequal shapes, small plenum volume and unequal runner lengths.


This one
Click to expand...
 
Interesting. And is there a only 1.3% flow difference between .0245 and .0255 nozzles even though there is an 8% area difference?
 
Last edited:
The division of flow between the injector nozzles with the interaction of the flow divider for this installation goes like this:

Cyl #1 16.8%, Cyl #2 17.5%, cyl #3 16.1%, cyl #4 16.1%, cyl #5 16.8%, cyl #6 16.8%
 
I guess I don't understand how this mechanical injection system works. On Kinsler types that I played with eons ago, going from a .024 to a .025 nozzle was around a 6.5% flow change for an 8% area change at a fixed 30 psi.

I am guessing to get the numbers above, you flow the whole setup on a test bench and measure the flow amounts from each injector individually?
 
rv6ejguy said:
I guess I don't understand how this mechanical injection system works. On Kinsler types that I played with eons ago, going from a .024 to a .025 nozzle was around a 6.5% flow change for an 8% area change at a fixed 30 psi.

I am guessing to get the numbers above, you flow the whole setup on a test bench and measure the flow amounts from each injector individually?
Click to expand...

DanH has a good article on fuel injection systems in this months Kitplane. I was surprised to learn the flow divider had that much influence on the operation too.
 
BillL said:
DanH has a good article on fuel injection systems in this months Kitplane. I was surprised to learn the flow divider had that much influence on the operation too.
Click to expand...

I guess there is more to it than meets the eye on the outside. I need to pick up a copy and learn some more then...
 
a few comments looking back at the whole thread

I've really enjoyed following this, and learning along the way.

I am really happy that the initial issue lead to an increased capability in the EFII system that solved the problem. I thought it interesting that Ross was open-minded enough to make the change after earlier saying that it was a mask for the root problem. He was likely correct, but was open to the 'symptomatic relief' anyway to address a common systematic problem.
Yes, it would be great if we all had manifolds that flowed well and evenly.

I feel pretty lucky to have the angle-valve horizontal induction manifold which does seem to flow very evenly, right out of the box. My GAMMI spread is less than 0.2 gph, right out of the box. No doubt the plenum volume, and runner design is a big part of that.

Interesting that the fuel divider plays much of a role at cruise fuel flows. It has a pretty symmetrical geometry.

Excuse the thread drift here for a funny story:

I got a huge chuckle out of DanH bringing up the Bosch D-jetronic. I spent a lot of time fussing and toying with the D-jetronic on my 1971 Volvo 1800E race car. I really hated it. Its fundamental problem was relying on manifold pressure, rather than a direct measurement of airflow. A real challenge for the old analog computer box was that airflow is a non-linear function of manifold pressure. The manifold pressure sensor was pretty low quality by todays standards. A small error in manifold pressure near WOT would give a very incorrect mixture. But the bigger issue for me was that it is completely un-adaptable to changes in engine breathing. Improved breathing means more airflow at the same MAP, and so it got leaner as I improved the engine. Eventually, it had a horrible lean spot at the RPM where the induction tuning gave the best flow, to the point where it would miss. I tried adding a trim pot in the computer box to bias the mixture, and I tried fudging the calibration of the MAP sensor. Marginally solved the issue, but never perfect. And then, trying to get it to pass California smog tests. I will spare you all the onslaught of swear words I've directed at the D-Jetronic over the years.

My final solution was to find a Bosch K-Jetronic system off of a 1975 Volvo. What a wonderful system that was. It used a direct mass-flow sensor, and a constant-flow injection. (sound familiar?) That made the system much more amenable to engine changes. My engine finally ran right, delivering about 170 hp on 94 (R+M)/2 gas from an old B-20 overhead-valve motor. Not much by todays standards, of course.

That system was so versatile that it was the backbone of the first (or one of the first) production FI systems with a closed-loop mixture trim from an O2 sensor. It was called Lambda-sond. I think it was the first, but could be wrong.
Anyway, just thought I'd share that -- The old CIS, whether it be Bosch or Bendix, is not a bad way to go.
 
dave17_zps4kjvz0qz.jpg


Interesting photo. Plenum volume and runner taper isn't the only thing the Murrays are doing differently.

Check out the injector port location (inside the bend, not outboard), alignment (pointed toward the intake port, not away from it), and port style (positive injector retention).

This set of runners are not likely to be for Dave Anders, as there are six of them.

Photos shamelessly borrowed from the SDS website:

 
Last edited:
DanH said:
...This set of runners are not likely to be for Dave Anders, as there are six of them...
Click to expand...

That's what I thought too, but upon closer inspection it looks like 4 induction tubes and one exhaust (showing the O2 sensor bung).
 
DanH said:
dave17_zps4kjvz0qz.jpg


Interesting photo. Plenum volume and runner taper isn't the only thing the Murrays are doing differently.

Check out the injector port location (inside the bend, not outboard), alignment (pointed toward the intake port, not away from it), and port style (positive injector retention).

This set of runners are not likely to be for Dave Anders, as there are six of them.

Photos shamelessly borrowed from the SDS website:

Click to expand...

The photos are indeed from Dave Anders and TooBuilder is correct. Dave is doing this work himself, following his usual lightweight, well thought out approach. We're lucky to be associated with fine people like Dave, Mike, Rusty and Clark who give us valuable feedback so we can improve the product.
 
Last edited:
scsmith said:
I've really enjoyed following this, and learning along the way.

I am really happy that the initial issue lead to an increased capability in the EFII system that solved the problem. I thought it interesting that Ross was open-minded enough to make the change after earlier saying that it was a mask for the root problem. He was likely correct, but was open to the 'symptomatic relief' anyway to address a common systematic problem.
Yes, it would be great if we all had manifolds that flowed well and evenly.

I feel pretty lucky to have the angle-valve horizontal induction manifold which does seem to flow very evenly, right out of the box. My GAMMI spread is less than 0.2 gph, right out of the box. No doubt the plenum volume, and runner design is a big part of that.

Interesting that the fuel divider plays much of a role at cruise fuel flows. It has a pretty symmetrical geometry.

Excuse the thread drift here for a funny story:

I got a huge chuckle out of DanH bringing up the Bosch D-jetronic. I spent a lot of time fussing and toying with the D-jetronic on my 1971 Volvo 1800E race car. I really hated it. Its fundamental problem was relying on manifold pressure, rather than a direct measurement of airflow. A real challenge for the old analog computer box was that airflow is a non-linear function of manifold pressure. The manifold pressure sensor was pretty low quality by todays standards. A small error in manifold pressure near WOT would give a very incorrect mixture. But the bigger issue for me was that it is completely un-adaptable to changes in engine breathing. Improved breathing means more airflow at the same MAP, and so it got leaner as I improved the engine. Eventually, it had a horrible lean spot at the RPM where the induction tuning gave the best flow, to the point where it would miss. I tried adding a trim pot in the computer box to bias the mixture, and I tried fudging the calibration of the MAP sensor. Marginally solved the issue, but never perfect. And then, trying to get it to pass California smog tests. I will spare you all the onslaught of swear words I've directed at the D-Jetronic over the years.

My final solution was to find a Bosch K-Jetronic system off of a 1975 Volvo. What a wonderful system that was. It used a direct mass-flow sensor, and a constant-flow injection. (sound familiar?) That made the system much more amenable to engine changes. My engine finally ran right, delivering about 170 hp on 94 (R+M)/2 gas from an old B-20 overhead-valve motor. Not much by todays standards, of course.

That system was so versatile that it was the backbone of the first (or one of the first) production FI systems with a closed-loop mixture trim from an O2 sensor. It was called Lambda-sond. I think it was the first, but could be wrong.
Anyway, just thought I'd share that -- The old CIS, whether it be Bosch or Bendix, is not a bad way to go.
Click to expand...

The D-Jet actually worked very well as designed- for stock engines. I have customers finally switching over to our EFI now from D-Jet because they can't get parts for them any more. D, K, L, Motronic all suffer from the same problem when you put them on modified engines- they are not calibrated for the new airflow characteristics and so have lean and/or rich spots- even with airflow meters as in the K and L systems. We see similar issues on Continental engines at Reno with stock Bendix setups over about 600hp.

Programmable systems solve this problem by allowing you to fix any of these spots, even using speed/ density. We sell a lot of systems for Porsche Turbos which had CIS and people are amazed at the increased driveability, power and fuel economy on modified engines. Mechanical injection would never cut it today in production cars due to emissions, driveability and fuel economy demands but it works ok in aircraft in a steady state, non-emission environment.

We're just giving people more market choices. You can buy whatever suits your fancy and mission.

In this thread, the inability to be able to correct AFRs in each cylinder was seen as a drawback to our EFI. Now we've fixed that and done the mechanical systems one better by being able to do it at any power setting and with more precision and far less work. Combine this with our ignitions where we can advance with rpm and MAP and even with mixture strength and now you can squeeze every drop of energy out of your fuel that the Lycoming is capable of. That's what lots of folks want these days according to what we hear from feedback.
 
vertical sump flow balance idea

I am gratefull for access to such experience and knowledge. I have a simple O320 with pmags and carb. If I eliminate the carb and use Don's controlled leakage device, I still have numerous sources of flow imbalace potential(Dan's point). If my only objective is smooth running and good fuel utilization, flow balancing for EGT match is suggested because it would accont for all variances in composite .

So, what would be wrong with a clever, robust ie safe restrictor system in the tubes between the sump and the head ? It might not be perfect from idle to WOT, but surely better than "stock" ?
 
Thanks Ross, more good stuff.

Although, I'm unclear why a mass-flow based system would have rich and lean spots, since there is presumably a fixed relationship between airflow and fuel flow. In the case of the K-jetronic, the flapper valve in the air stream operates a fuel metering valve. I suppose there are non-linearities in the characteristics of both the flapper valve and the fuel metering valve? I always assumed those canceled out to maintain fixed AFR.

For your systems that you install in cars, do you put EGT sensors in the header pipes?

The mention of an O2 sensor in Dave Anders' system leads me to assume he is going to run mogas? Those tapered runners are beautiful, how were those made?
 
scsmith said:
Thanks Ross, more good stuff.

Although, I'm unclear why a mass-flow based system would have rich and lean spots, since there is presumably a fixed relationship between airflow and fuel flow. In the case of the K-jetronic, the flapper valve in the air stream operates a fuel metering valve. I suppose there are non-linearities in the characteristics of both the flapper valve and the fuel metering valve? I always assumed those canceled out to maintain fixed AFR.

For your systems that you install in cars, do you put EGT sensors in the header pipes?

The mention of an O2 sensor in Dave Anders' system leads me to assume he is going to run mogas? Those tapered runners are beautiful, how were those made?
Click to expand...

I think many of the issues with air flow meters come when the airflow of the modified engine exceeds the design parameters of the meter design. At X airflow, max power in stock trim, we might have the system delivering 11ish AFR (common with K-Jet). The new engine has 10% more displacement, more cam duration, ported heads, better exhaust and a 15% higher redline. It make 50% more power and flows 50% more air. Two things happen, the AFM can physically bottom out, being unable to meter any more fuel or two, at the old 100% airflow rate, the engine is well down in rpm and load but the metering system is still delivering a max power AFR of 11 when you might want it closer to 13. You can diddle with stuff but you'll never have the precise, automatic mixture control of EFI across the entire operational range. Certainly L-Jet systems don't take kindly to even a mild camshaft change and run like garbage. Non-programmable speed/density EFI systems also do not like cam changes, being perhaps worse in this regard.

Programmable EFI gets rid of all these compromises and allows you to get any AFR you want at any rpm and load point without leaving the vehicle and getting your hands dirty.

We rarely use EGTs in the automotive world for tuning except on engine dynos. Not particularly useful for every day use since we either run at best power AFRs or Stoich for lowest emissions.

We've seen people have good and bad luck with wideband O2 sensors and leaded fuels. The later 4.9 Bosch sensor seems much better in this regard than the older 4.2 sensor and Bosch publishes some general data for life expectancy with leaded fuels. They will certainly last long enough to accurately tune the engine with EFI. A couple users have over 350 hours on the same 4.9 sensor now. The sensor is not needed for general running in aircraft, we usually use just for initial tuning and monitoring.

The Sky Dynamics folks are wizards with their parts and induction systems. I have no idea how they create such art with these curved and tapered runners. I'd be very interested to see their process!
 
Last edited:
scsmith said:
Thanks Ross, more good stuff.

Although, I'm unclear why a mass-flow based system would have rich and lean spots, since there is presumably a fixed relationship between airflow and fuel flow. In the case of the K-jetronic, the flapper valve in the air stream operates a fuel metering valve. I suppose there are non-linearities in the characteristics of both the flapper valve and the fuel metering valve? I always assumed those canceled out to maintain fixed AFR.

For your systems that you install in cars, do you put EGT sensors in the header pipes?

The mention of an O2 sensor in Dave Anders' system leads me to assume he is going to run mogas? Those tapered runners are beautiful, how were those made?
Click to expand...

I think the system to which Ross refers (D-jetronic) was manifold pressure control. It had an aneroid reference. If the mass flow per rpm and manifold pressure was higher then the mixture would go lean. It was affected by exhaust pressure and airfilter plugging. I worked on hundreds of these as a green ignorant mechanic in early '70's. It inferred the mass flow. If volumetric efficiency changed so did the A/F ratio. It also fired the injectors via two sets of breaker points in the distributor, they fired in pairs, one on each side of an opposed engine. It had a terrible throttle response lag. Owners and mechanics nightmare.
 
Last edited:
BillL said:
I think the system to which Ross refers (D-jetronic) was manifold pressure control. It had an aneroid reference. If the mass flow per rpm and manifold pressure was higher then the mixture would go lean. It was affected by exhaust pressure and airfilter plugging. I worked on hundreds of these as a green ignorant mechanic in early '70's. It inferred the mass flow. If volumetric efficiency changed so did the A/F ratio. It also fired the injectors via two sets of breaker points in the distributor, they fired in pairs, one on each side of an opposed engine. It had a terrible throttle response lag. Owners and mechanics nightmare.
Click to expand...

The D-Jet was introduced in late 1966 I believe and was pretty cool by the days standards. They used a moving magnet/ coil attached to a vacuum diaphragm to sense MAP along with rpm so it was a speed/density type system but without the quick response of solid state MAP sensors like we have today. Also used a series of contacts inside the TPS instead of a potentiometer like we do today and an analog computer instead of digital.

It worked ok if maintained and the components such as pumps, regulator and injectors were bulletproof. The computers were even pretty good. As Bill stated, it suffered when people did not maintain certain wearing parts and the throttle response was nothing to write home about because of the mechanical sensors. It was the first really successful and widely used EFI system, fitted to many Volvo, VW and Posche engines in the late '60s and '70s.

Things have come a long ways in those 50 years.
 
You must log in or register to reply here.
Back
Top