Custom Tester Speeds Design, Production

PC-driven test equipment cuts PCV valve production in half

—Edited by Larry Olson

Automotive positive crankcase ventilation valves — small, simple devices, with only one moving poppet and a spring — must meet rigorous performance specifications. "Manufacturing PCV valves is not difficult," says manufacturing engineer Brent Veeneman, CMfgT, of Novo Products Inc, Holland, MI,. "The hard part is the design and prototyping stage. It’s the quality of the testing that separates the premium product from the commodity."

Each new valve must prove that it will admit specific airflow amounts at various vacuum levels. To satisfy such specs, Novo found that experimental testing had to become an integral part of the design process. "It’s largely trial and error," Veeneman notes, "and that can be very time consuming. We start with a rough design that experience suggests should come close to meeting the requirements, test it against the customer’s specs, then make changes and test again."

PCV valves must relieve engine crankcase pressure properly to prevent backfiring or blown gaskets and seals. They do this by regulating pressure outflow from the engine with a tapered poppet moving in and out of a tapered orifice. The orifice outlet connects to the engine’s air/fuel intake, allowing the vacuum generated by the air/fuel intake to modulate the poppet. Spring tension holds the poppet open when vacuum is low, while increasing vacuum gradually pulls the poppet closed. When the poppet is open, engine gases are vented back into the air/fuel intake for re-burning, rather than venting to atmosphere.

Each engine’s design determines a different relationship between the amount of vacuum and the amount of crankcase pressure allowed to vent. That relationship requires each PCV valve design to have a unique balance among the internal poppet and orifice geometry, the spring force, and the vacuum level. Customer specs define the valve’s external size and shape, along with flow values desired at designated vacuum levels, called "delta points." Vacuum levels typically range from 1 to 20 in. hg, and flow values can be from 0.2 to 8 cfm. Each valve design must be verified by flow-test data collected at up to 20 delta points.

While Novo engineers tweak the internal geometry to achieve the flow spec, they also try different spring and poppet designs to shorten the stabilizing time, which in turn allows faster part-to-part cycles later in production testing. Veeneman explains that one major complication in testing is the time needed to let airflow stabilize at each vacuum point before the flow is measured.

Other complicating factors are the airflow’s absolute pressure and temperature, since these conditions will cause vacuum pressure and air density to vary. "Some buyers base their vacuum and flow specs on standard atmospheric pressure of 14.7 psia (29.92 in. hg) at 20°C," he points out. "Others prefer different ambient pressure and temperature baselines, so our design-test system must be able to adjust test data to show flow under whatever standard conditions the buyer has specified."

Homemade tester slow, accuracy doubtful

Novo first tried to satisfy design-test requirements by combining a commercial flow meter with a homemade valve system. "It worked," Veeneman recalls, "but manual data gathering was time-consuming, and we weren’t really sure about its accuracy." Consequently, they contacted InterTech Development Co, Skokie, IL, that had earlier provided several turnkey assembly systems with integrated leak-testing functions, as well as leak-testing instruments and consultation on several other leak-test stands.

Collaborating with InterTech, Novo engineers built a two-station production QC tester. One station tested valves at flows of 0.2 to 2 cfm, and the other tested valves at flows of 2 to 6 cfm. Separate low-flow and high-flow fixtures at each station were connected alternately to two InterTech M-1035 mass-flow instruments (one each for lower and higher flow ranges), to cover the full 1 to 20 in. hg range.

In the typical production QC test at three delta points, the PCV valve remained in the same fixture while the M-1035 sequenced through three programs, one for each specified vacuum point. The total part-to-part test cycle time was 36 seconds. Each M-1035 displayed the flow rate at each vacuum, compared the flow rate and vacuum against pre-set limits, and signaled accept/reject. This dual-fixture, dual-station, dual-instrument setup made it possible to test in one station while the fixture in the other station was manually unloaded and reloaded. A second fixture in each station allowed a different part with a different set of programs to be tested at the same time.

The M-1035’s ability to program precise vacuum points and flow limits, and recall test data from its last 1000 tests (333 valves), made the production tester the instrument-of-choice for design testing as well. "When new valve designs began calling for full 20-point curves spanning both low and high flow ranges," Veeneman reports, "we would test the low-flow and high-flow delta points on whichever M-1035 had the appropriate range. Then we’d collect the data from both and manually plot the curve by patching the segments together."

He admits that was not the best way to do it, since the curve would have a void in it when the data jumped from one instrument to the other, as well as having possible calibration discrepancies between instruments. However, that was the best available at the time. Using the production testers for design work also wasted time with tooling and program changeovers for the test parts, then changing everything back to resume production testing. This also meant doing more extensive calibration.

Six months after production start-up, the high-flow production instrument was refitted with a low-flow transducer to provide more testing capacity for the increasing volume of valves in that range. "With that," he says, "the only way we could do full-curve prototype testing was to fill in the missing-range points with our old home-made tester, which made the task even more difficult and less reliable. We needed a full-range tester and a better way to graph the data."

Overall design time reduced

InterTech proposed replacing the tester with a custom version of their vacuum-regulated M-1035 production test stand. Instead of the mass-flow transducer, this tester featured an external laminar flow element with a master lab calibration accuracy of ±0.6%-of-reading across the 0.2 to 8 cfm flow range. This was coupled with a differential-pressure transducer (certified at 0.15%-of-reading), an absolute-pressure transducer (certified at 0.3%-of-reading), and an airflow temperature sensor.

While the M-1035s in production testing provide a mass-flow transducer accuracy of ±0.5%-of-range for Novo’s high-accuracy requirements over relatively short cfm ranges, percent-of-reading capabilities are essential in the lab to ensure consistently high accuracy at all points along the extended cfm range. Adding high-accuracy transducers, a temperature sensor, and a custom program enables the M-1035 to measure flow values at ambient temperature and pressure, and display and record data corrected to buyer-specified standard conditions.

The test program runs on a laptop PC using a custom version of InterTech’s S-3085 monitor software. This provides for storing test parameters for up to 20 delta points for each PCV valve design, along with correction to atmospheric pressure and temperature per customer specs. It also allows the PC to combine real-time results for all delta point tests into a single continuous graph, showing changes in vacuum level and airflow over time against each point’s upper and lower (accept/reject) limits. Full-range curves can be displayed, exported to Microsoft EXCEL format, printed out, or even e-mailed directly to customers for approval.

While a standard M-1035 can store and recall test programs for up to 33 three-point production tests, the lab tester needed greater storage capacity. Since each delta point test requires a separate test program, the 20-point tests for each of seven PCV valve models meant a total of 140 programs — a number that is certain to grow as more valve models come into Novo’s product mix. The S-3085 software enables the lab tester’s PC to store much more test data (2 million test records per 100 MB, by date) in Microsoft Access database format.

The PC also makes possible recall of stored test data, either in total or by date, accept/reject results, or causes of rejects (e.g. high or low pressure). Data tables and curves for multiple samples of the same design can be merged to compare average flow performance vs. high / low extremes for the sample population. The custom program enables the M-1035 to receive the multi-point programming for a specific PCV valve design, sequence through the test for each delta point, and send real-time test data back to the PC for display and storage.

"We see it cutting our actual testing time by 40 – 50 percent," Veeneman concludes. "Because it’s a dedicated lab instrument, we no longer need to wait for a break in production — or worse, interrupt it." Better design testing eventually meant an overall improvement in productivity. "That’s because we have more accurate flow data in 20-point curves that make the data easier to interpret. Real-time graphic display lets us see and document exactly how a design is behaving at all times during its cycle. We no longer have to plot curves manually or patch curves together, which gives us better direction and helps us respond more quickly with design, tooling, and equipment adjustments."