Text and photos by Gary Koehler
[Editor] As Gary points out, while significant flow rate gains were achieved as the result of the extensive modifications he made to the Zenith 32NDIX carburetor, he has not yet done dyno testing of an engine with his modified carburetors and compared the results to the same engine with stock carburetors, so potential HP increases have not been measured. Also, he carefully notes that "Flow rate gains do not guarantee HP gains unless all other elements in the system are designed accordingly."
I have finished the rebuild and modify phase of the Zenith carburetors for my big bore rebuild, realized some nice flow and velocity gains, and wanted to share what I did with Registry members who are interested. While I don't claim to be a carburetor expert, and with due respect to those who have done carburetor, flow bench and dyno testing, I could find no published flow data or guidance for modified Zeniths, so I decided to flow test the modifications I did, and present the data here. Naturally, flow rate gains do not guarantee HP gains unless all other elements in the system are designed accordingly.
I looked at each section of the carburetor, starting with the carburetor base, the bowl section, the carburetor top and the air filter. Following are the major details of what was done; flow bench tests indicate that flow rates through a modified Zenith 32NDIX carburetor (29mm venturis) and air filter improved by 16% at full throttle opening and by the same amount at half throttle; this in comparison to a stock Super carburetor (28mm venturis), manifold and a 'B' Knecht filter canister with a new Mahle filter element. All testing was done at John's Fuel Systems and Carburetion in Hayward, CA, using a calibrated Superflow SF-600 flow bench (left). John has just about every type of carburetor there known to man, his shop is filled with carburetors and spare parts of every description: antique, aircraft, and even Pro Stock carburetors that are about a foot square, huge things when compared to a humble Zenith.
Starting with the throttle body, I removed the throttle shafts and cut flats on them parallel to the butterfly slot using a wire EDM to eliminate cutting stresses, which might distort the shafts. Flatting the shaft on each side decreases the area of the shaft exposed to flow, and at full throttle the shaft area presented to airflow is about 22% less. I did a basic calculation of what the stresses might be on the shaft from the reversion pulse, and considered it to be acceptable. Time will tell if this is true. The pictures following show a stock shaft compared to a flatted shaft, and the opposite side with similar flats.
While the shafts were removed, I counterbored the throttle shaft bores at each end of the throttle body to incorporate a miniature PTFE (Teflon) lip seal.
The lip seals prevent air intrusion along the throttle bore/shaft interface due to wear of the shafts, and more importantly prevent the entrance of grit that accelerates wear. My shafts were in good shape, so there was no need to re-bush the bores.
The central bowl section of the carburetor is an area where real gains can be made. If you look at the space adjacent to the venturi bores, there is a roughly triangular flat area of about .7 sq. in. normal to the incoming airflow. In fluid dynamics the term 'inefficient corner' is an area where in-rushing air abruptly has to make a 90° direction change when it collides with a solid surface that is perpendicular to its flow direction. When in-rushing air hits this shelf, some of it instantly reverses direction and collides with the air that is still entering, causing turbulence that affects both flow direction and velocity. This type of flow condition, an inefficient corner, is solved in industry by using turning vanes that smooth and redirect the flow around the corner to minimize losses. It's not practical to do that here, so I made deflectors that re-direct airflow away from the hard surfaces in the flow path and toward the venturis. The entrance area of the two venturis is about 2.1 sq. in., and the shelf area is about .7 sq. in., about 30% of the total entrance area at the level of the venturis, so it is significant. There are some other minor modifications shown in the pictures that are also meant to redirect and smooth the flow. By reducing the gross Mean Free Path (MFP) of the air molecules in the flow, a highly turbulent condition is made less so. In a 'perfect' carburetor, air molecules would flow into the carburetor opening and to the venturis without bouncing off anything — but this is never the case — reducing the number of so-called dirty bounces can increase flow, velocity is increased accordingly.
Next I looked at the crossbar; in another application it was used to secure an air cleaner through the center threaded hole, also acting in our case as a breather to the float bowl through internal passages in the carburetor top. I removed it as it does disturb the flow, and is a classic example of bluff body flow. Bluff body flow around a cylindrical surface (crossbar) is a form of flow separation that occurs where the incoming air collides with the crossbar, separates on either side and goes into a highly disturbed turbulent flow condition; at the rear of the cylinder there is actually a stagnant area created that creates a swirl vortex graphically shown in this animation:
Note how the 2 colors describing each path of the separated air indicate where it goes as it moves downstream. It looks like a blender in there, and you don't want it entering the venturi that way if you can avoid it. This type of flow occurs where the airflow separates at the crossbar, at both the mixing tube stalks, and to some degree the throttle shaft. To this end, I removed the crossbar, thinned and shaped the mixing tube stalks and flatted the throttle shafts. The flow is still far from perfect, but improved to the extent it can be, given that not much else can be removed.
Removing the crossbar still requires venting to the gasoline surface in the float bowl, so I made small vent risers attaching on each side that seal with o-rings to the holes in the carburetor body vent passages, with a provision for securing them at the top.
The holding screw securing the 2 mixing tubes can be removed as shown in the following picture, and the abutted flow in this area can be directed toward the main venturis. The mixing tubes are held in compression with a split stainless steel spring finger attached to the intake bore, it preloads the mixing tubes to hold them in place, and is shaped with the intent of re-directing the abutted flow toward the venturis. With all of the top modifications, the assembled carburetor top and bowl look like this.
Lastly was the air filter and canister. There are many design approaches, but I used the stock Mahle element because of its rigid construction and efficient paper element. The flow data following show that the Mahle element when attached to the filter mount will flow 210 CFM, 64 per cent more through the filter than required by the modified Zenith through a single runner, yet the flow rate through the modified Zenith at full throttle was reduced by only 1 CFM with the filter element mounted. This speaks volumes about the need to go any farther down the road using a K&N element, or similar unit that may not offer the same degree of filtration. The stock canister and Mahle filter are about a 4 CFM flow penalty.
To mount the filter, I fabricated a conical filter mount to match the inside diameter of the carburetor opening at the downstream end, and the inside diameter of the Mahle filter at the upstream end. The length of the filter mount compared to a C canister with this design adds an additional 3/4" of length to the intake. This design does not use a canister, but rather attaches the filter base to the filter mount. Made of Delrin, it presents a slippery surface as well as eliminating all the internal flow entry loss due to the mismatch in diameters that exist in the stock configuration. If water entry through the engine grill becomes a problem a rain hat might be needed.
Half way through the testing, gains of about 12 per cent were seen, and based on these gains it seemed to follow that with the already increased flow capability, a larger venturi could be used that would still provide the signal strength needed at the mixing tube venturi. I increased the diameter to 29 mm, and also increased the outside diameter. Making the o.d. of the venturi larger results in an increase in the surface path of the inlet radius (it's not really a radius), increasing the collecting area at the entrance of the venturi. The 29 mm venturi change added about 4 percent to total flow, with the end result being 16 per cent, or about 128 CFM as compared to the stock Zenith fitted with super venturis at 110 CFM.
About the Testing
Testing was done at JFC in Hayward, CA on a Superflow-600 Flowbench using 28" depression, which I believe to be the SAE standard for testing of this type. All testing was on a single manifold runner with the adjacent passage blocked. There was a series of 5 tests on different days where the temperature and humidity were different, so I standardized the flow data by comparing each test to a known standard (in this case, a stock Zenith carburetor with Super venturis, new Mahle filter element and stock canister mounted on the test manifold. I then compared the increased flow from the modified Zenith against the stock number, arriving at a percent value increase for flow. When CFM is quoted in a test, it is a number compared to that measured a few minutes earlier referencing the stock setup. The CFM numbers, while close, are mostly for quantitative reference, while the percent value better represents the actual performance.
The flow data is shown in the graph (right), measured data points are at 45° and 90° for both setups; throttle openings less than 45 were too hard to measure, as the flow regime here is not really a performance issue. The flow rate from half to full throttle is uniformly increased by 16% for the modified Zenith. It is reasonable to infer that there are also flow improvements at less than 45 but there are no data points to support it.
So, why all the effort, why not just slap on a pair of Solex carbs and be done with it? The attempt here was to increase the intake tract velocity (since energy = mass times velocity squared); so for a comparison I flow tested a Solex PII-4 solid shaft carburetor with its manifold and the stock Knecht mesh air filter, then baselined the values against the stock Zenith as before. The graph below uses the CFM data obtained to show full throttle velocity in feet/second for the three carburetors at the manifold-head interface, using entrance area elliptical values of 30 x 35 mm for the Zenith manifold and 35 x 40 mm for the Solex. CFM values are also shown for each carburetor; the modified Zenith surprised me with its velocity, despite the Solex flowing considerably more CFM. Velocity was determined by dividing the flow rate by the area of the port entrance at the head; this could vary somewhat based on individual head and manifold dimensions.
Thanks to John's Fuel Systems and Carburetion for all the help, tooling and advice.