EZ30R AVLS Cam Specs deep dive

[donkeytits1]

Knowing the EZ30R has an extra 20kW on the EZ30D in my car, I’ve always been interested in the cam specs. Especially with the AVLS system. I wasn't sure if the low lift profiles support the majority of the driving range, allowing a 'hot' high lift profile, or if there was some other strategy going on. 

After picking up some cams some years ago, I have finally I have had the opportunity to measure them.

I also measured the cams for an EZ30D and an EJ255 WRX (2006) for a comparison.

All cams are from Australian delivered models, EZ30D cams are from a junk engine I got years ago, probably 2002. The EJ255 are from a 2006 WRX with 'V25B' heads, we just happened to have some, and the EZ30R cams I have since found are 'Phase 1' or 'Pre facelift'.

MEASUREMENTS

Engine	Cam	Clearance	Gross Lift	Net Lift	Factory Spec				At 1 mm Lift			At 0.050" Lift
					Open	Close	Duration	At Lift	Open	Close	Duration	Open	Close	Duration
EJ255	Inlet	0.20	9.60	9.40	5 BTDC	55 ABDC	240	0.218	-102.23	103.80	206.03	-99.93	101.25	201.18
	Exhaust	0.35	9.80	9.45	55 BBDC	5 ATDC	240	0.172	-104.64	103.80	208.44	-102.28	101.36	203.64
EZ30D	Inlet	0.20	9.78	9.58	5 BTDC	55 ABDC	240	0.239	-105.30	105.96	211.26	-102.99	103.50	206.49
	Exhaust	0.25	9.29	9.04	52 BBDC	0 ATDC	232	0.171	-101.42	101.17	202.59	-99.10	98.79	197.89
EZ30R	Inlet HL	0.43	10.12	9.69	22 BTDC	48 ABDC	250	0.255	-107.18	108.18	215.36	-104.11	104.83	208.93
	Inlet LL L	0.20	3.13	2.93	N/A	N/A	147	0.246	-54.07	53.72	107.79	-49.76	49.35	99.12
	Inlet LL H	0.20	6.32	6.12	N/A	N/A	218	0.245	-89.26	88.61	177.88	-85.55	84.65	170.20
	Exhaust	0.35	9.70	9.35	60 BBDC	6 ATDC	246	0.177	-107.82	107.47	215.29	-105.44	105.03	210.47

 

AVLS

https://commons.wikimedia.org/wiki/File:Switching_tappet_for_SUBARU_EZ30_(INA).jpg
The AVLS system in the EZ30R engine is a direct acting system with an innovative concentric split follower. For each valve there are a set out outer high lift lobes and an inner low lift lobe on the cam shaft. The inner lobe runs on the inner section of the follower and the high lift lobes run on the outer section of the follower. The inner section of the fowler acts directly on the valve stem and when the system is not engaged the outer section is free to slide along the axis of the valve. When engaged, oil pressure moves a pin across that locks the outer section of the follower to the inner section and the valve then follows the high lift profile.
The follower has a curved surface, with the flat axis parallel to the cam centreline. The cross section of the surface was measured to be an arc, and there is a different radius used for the inner and outer sections. The inner section was measured to have a 43mm radius and the outer section was measured to have a 52mm radius.
The inner section of the follower protrudes slightly. The lobes seem like they have a slightly different base circle diameter as well. 32mm vs 32.5mm. This still leads to the high lift lobe having more clearance. During an engine build, the valve clearance is set on the low lobe, and the high lift clearance is not adjustable.

 
To measure valve motion for the EZ30R ALVS lobes a flat follower was used and then a correction was applied to account for the curved follower.

SERVICE MANUAL SPECIFIED DURATION
The valve timing specifications given in the service manuals appear to be with respect to the corners of the profile ramps. The ramps end at very low lift levels and so in a practical sense, the factory open and close angles can be described as the start and end of any visible movement of the valve when viewed from above. It also roughly aligns with the peak acceleration and deceleration events. When specifying the factory specs in terms of a check lift, the non symmetrical size of the ramps makes it a little misleading. The lead-out ramp is longer that the take-up ramp. None the less, the check lift for intake cams appears to be around 0.25mm and the exhaust cams 0.17mm.
The lower peak acceleration of the EZ30R cams (discussed later) seems to exaggerate the duration stated in the service manual.

SPECIFICATIONS AT 1mm LIFT

Aftermarket cam manufacturers tend to spec cam duration at a higher check lift, 0.050” or 1mm being common. At these levels of lift the valve has been past the ramps and through most of the acceleration phase of the profile, leaving what I'll call the bulk lift phase within the duration measurement. While the ramps and the acceleration phase can vary between cams, the bulk lift phase tends to be close to harmonic motion, which minimises required valve spring force and high frequency vibrations.

SPECIFICATIONS AT 0.050" LIFT

Edited Sunday at 11:51 PM by donkeytits1

[donkeytits1]

VELOCITY AND ACCELERATION PROFILES
The measurements were detailed enough to give reasonable insight into the higher derivatives.
The EZ30R cams appear use a lower peak acceleration, ie they lift off the seat slower. This could be called 'less aggressive'
Peak negative acceleration is the biggest determinant for required valve spring rate, and it can be seen that all three engines are quite similar. (Actual valve spring rate is also proportional to valve mass)

Edited Sunday at 11:53 PM by donkeytits1

[donkeytits1]

CAM DOCTOR SHEET

There have been a few cam measurement sheets and specs getting around. Like in the service manual, the duration specs seem suspiciously large.
To check them, I plotted the information on a measurement sheet I found on facebook over the top of the corrected results. It appears this measurement was accurately taken with a flat face follower but not corrected for the curved shape of the AVLS follower.
This exaggerates the measured duration of the cam. In reality, the EZ30R has remarkably similar cam design to the preceding EZ30D and the EJ255.
Given the file extension and layout of the sheet, the measurement would have been taken on a 'Cam Doctor' machine.

 

Edited Sunday at 11:59 PM by donkeytits1

[donkeytits1]

MEASUREMENT METHOD
For my measurement, he cams were mounted in a lathe with a 2000 point optical encoder attached to the back of the headstock. A digital indicator with micron accuracy was mounted to the carriage as rigidly as possible. The encoder and the digital indicator were then logged against time, data transfer being achieved via RS232 serial. The indicator outputs its screen reading in its own format at roughly 10Hz, and the encoder at around 50Hz in plain ASCII. Measurements were merged with respect to time after the measurement was completed. The cam was rotated as slowly as possible, approx 1.5 RPM and number of rotations were logged and averaged together to improve signal to noise.

Measurements were taken with a flat face follower. This has a number of advantages:

1)      The inlet cams for the EZ30D and EJ255 engines as well as all the exhaust cams run in the engine with flat faced buckets. By measurement the cam with a flat follower, you are directly measuring the valve motion. 

2)      Side loads on the dial indicator is minimized. This minimizes friction and deflection in the indicator.

3)      The measurement is theoretically unaffected by an offset between the centreline of the indicator and the centreline of the cam. This removes one source of inaccuracy and makes setup easier and faster.

The main disadvantage is that a flat follower cannot be used on a cam with concave sections, ie undercut.

This is not a problem for the EZ30R cams, as viewing the reflection of a point light source against the ground surface while the cam is rotated shows that all sections of the cam are convex.

The curvature of the EZ30R bucket was measured by mounting the bucket in a milling machine and dragging an indicator across the follower surface. Essentially, I constructed a makeshift gantry style Coordinate Measuring Machine (CMM). Two micron accuracy indicators were used, one for height and one for translation, for this giving XY coordinates that could be analysed. Once it was clear the curve was an arc, an arc was fitted using least squares.

I did the bucket measurement a long time ago, and do not have images.

THE MATHS

To transform the translating flat face follower motion to that of a translating roller follower, the actual polar lobe profile needs to be calculated and then a geometric offset applied equal to the radius of the roller follower. The lift for a given angle of cam rotation is then that of a translating point follower, which is simply r – base circle diameter/2  - roller radius for any given angle of theta.

I wrote quite a bit of MATLAB code to perform the data merging, measurement delay estimation, de-glitching, runout correction, and smoothing on top of this, but that is the main gist.

Below is a graphical representation of the 52mm AVLS follower running on the EZ30R High lift cam at the point of maximum valve velocity. It is showing that the higher the valve velocity, the further to the edge the contact point is on the follower/bucket. I think this is why the AVLS follower is curved. Being off to the side, of the circular shaped follower, the available running area is less. The curved surface reduces the distance relative to the valve velocity.

This is the best explanation of flat face follower kinematics that I’ve found to date. Props Dr Starr. This is the foundation for generating the lobe profile from the flat face follower lift coordinates

http://www.me.unm.edu/~starr/teaching/me314/camprofile.pdf

The dependency of the cam profile on the derivative of the lift measurement makes noise rejection / smoothing very important. Derivatives greatly amplify noise, and if too much noise is present the lobe curve begins to self intersect.

Smoothing however rounds off features like the ramp transitions, so there is a balance. Obtaining low noise measurements in the first place gives the best chance of getting a clean lobe profile with an acceptably low amount of smoothing.

To generate the offset lobe curve, you can offset each point. Naively you could do this from the lobe coordinates, but this involves calculating the normal angle and hence another set of derivatives. An easier way is simply adding the offset distance to Ro in the original equations and re-generating the lobe.

Other considerations:

1)      Run out correction. With no run out, the measurement of the base circle should be constant. If there is run out, the measurement will be a sinusoid at the frequency of once per revolution. To correct for run out, fit a sine curve,, to the measurement of the base circle and subtract it from the entire measurement.

2)      The indicators have a small delay between taking a measurement and beginning a transmission of the reading, which when merged with the encoder measurement with respect to time, leads to noise in the measurement. The arduino begins transmitting the encoder reading almost instantly.  By experimentally adjusting the time offset over a long measurement to minimise noise, the delay in the indicators was estimated to be around 22ms.

 

Edited Wednesday at 01:27 AM by donkeytits1

[donkeytits1]

EZ36 Cams

A facebook user was kind enough to share an EZ36 measurement. It appears to be measured using the same machine as the EZ30R 'Cam Doctor' measurements. So, it is reasonable to assume that it was measured using the same flat face follower. It certainly appears legitimate. As the EZ36 uses flat buckets, not correction is needed.

Now, we can plot all EZ engines on the one chart.

Inlet cams are not very different.

The measurement appears to take into account clearance but based on the FSM check lift, there might be a 0.05mm difference. This is possibly doe to using the lower limit for the clearance value rather than mid range. 

The EZ36 has the smallest exhaust cams.

Service manual check lift

1mm check lift

0.050" check lift

Engine	Cam	Clearance	Gross Lift	Net Lift	Factory Spec				At 1 mm Lift			At 0.050" Lift
					Open	Close	Duration	At Lift	Open	Close	Duration	Open	Close	Duration
EZ36	Inlet	0.20	9.90	9.70	15 BTDC	49 ABDC	244	0.243	-108.99	107.43	216.42	-106.64	105.00	211.63
	Exhaust	0.35	8.66	8.31	24 BBDC	24 ATDC	228	0.192	-100.16	99.79	199.95	-97.78	97.31	195.09

Edited Monday at 10:49 AM by donkeytits1

[donkeytits1]

QUICK NOTE: On service manual open/close times

The lead out ramp on all cams measured tends to be larger than the lead in ramp, and when calculating the check height for the factory specs it tends to fall below the start of the lead out ramp. This tends to skew the 'close' point further away from the maximum lift point of the lobe.

I don't have any confirmation on this, but Subaru may be using the start/end of the ramps or the start and end of the max acceleration events as their definition of 'open' and 'close'. Quickly checking the EZ30R, which has about 5 degrees offset between open and close and max lift when using check height, gives the correct duration and is close to symmetrical about the point of maximum lift when using the edge of the acceleration curve.

The importance of this, if it is correct, is that you can calculate the timing of max lift by calculating (close - open) / 2 from the service manual.

You get no insight into the lobe timing when the cam is on the lathe, obviously.

I did measure the sensor chopper disk that the ECM uses to sense cam timing, so I might post that up later.

 

Edited Tuesday at 02:53 AM by donkeytits1

[donkeytits1]

QUICK NOTE: AVLS Follower witness marks

Just found a blurry image of the AVLS follower that was measured. It is good enough to see the areas of actual lobe contact.

You can see how pushing the high lift lobes outwards limits the maximum contact distance away from the cam axis. This directly limits the maximum velocity of the valve. for the same velocity, the distance would be larger for a flat follower.

Here. the AVLS cam designer is clearly using all the available velocity, for both the high lift and low lift lobes.

Edited Wednesday at 02:17 AM by donkeytits1