From: EPI Inc.       Contact:


  Race Engine Technology Magazine CURRENT ISSUE of Race Engine Technology Magazine

Race Engine Technology Magazine



Valid XHTML 1.0 Transitional Valid CSS!  

- EPI Service Bulletin #3 -


Document Number 260003
A06/24/02Initial Release.
B06/30/02Revised SUMMARY and COMPLIANCE sections.
C07/02/02Corrected date in COMPLIANCE section to 5/21/02.


This bulletin applies to all owners and operators of any aircraft using the EPI Mark-9 PSRU, as supplied on EngineAir V8 aircraft engines or from any other source.

EPI, Inc. is aware of two incidents in which the input shaft (part number 790048-C) of the EPI Mark-9 PSRU, installed on EngineAir V8 engines, has failed suddenly and catastrophically, causing a complete loss of power to the aircraft propeller.

Both failures occurred during ground runup operations. However, in view of the serious and potentially fatal nature of these failures, EPI must take the position that any further flight operations conducted with any PSRU containing the subject input shaft constitutes a severe hazard to life and property.

This bulletin constitutes a formal notification to all owners and operators of this powerplant that, based on these two failures, EPI considers the PSRU unairworthy until the subject input shaft is replaced with the new design (part number 790056).

The remainder of this document explains the cause of the two known failures, corrective actions taken by EPI, and provides information on compliance with this bulletin.


EPI Inc. is the designer of the PSRU which is currently being produced by EngineAir (currently known as Engine Power Systems). EPI Inc. manufactured the first five of these PSRU’s, as well as additional finished internal parts sufficient for the construction of approximately 15 more PSRU’s.

In mid-November of 2000, EngineAir purchased (a) the rights to manufacture the Mark-9 PSRU, (b) the Mark-9 engineering drawings and specifications for all the parts of that PSRU, and (c) EPI’s complete inventory of Mark-9 PSRU parts (finished gears, shafts, bearings, fasteners, etc. and the remaining supply of unmachined housing castings). In late 2001, EngineAir began delivering PSRU’s assembled from the inventory of EPI parts and housings finished by one of their vendors.

The Mark-9 PSRU contains a proprietary vibration absorber between the engine and the PSRU. Part of that mechanism consists of two concentric drive shafts. The inner shaft connects the engine to the outer shaft (part number 790048-C). The outer shaft connects the inner shaft to the PSRU drive gear.


Both of the known failures occurred on the ground, during runup operations. It is significant that both failures occurred immediately subsequent to the performance of maintenance operations involving the PSRU.

The first failure occurred in the prototype Mark-9 PSRU; the second occurred in an early production PSRU.

Both failures were sudden and complete breakage of the outer input shaft (EPI part number 790048), causing an immediate and total loss of power to the propeller.

That input shaft (790048) has a large outside diameter and a thin-wall cross section. It is made from a very high-strength steel alloy suitable for torsional loading in a fatigue environment, and incorporates special design features which enhance it’s resistance to fatigue.

At any given engine torque, the outer shaft operates at a lower stress level than the inner shaft which drives it. That fact, together with a conscious effort, by the owner of the aircraft, to hide the true circumstances preceding the first failure, added to the difficulty of finding the cause of these failures.

EPI recently received a few pieces of the broken shaft from the second failure (some of the other pieces were mysteriously "lost"). Examination of those pieces, together with some new information recently received about the first breakage, strongly indicate that both failures were due to mishandling and incorrect installation, which caused the application of impact and buckling loads which this shaft was never intended to absorb.

3.1 DETAILS: Failure #1

The first failure occurred in late October, 2001, on the prototype Mark-9 PSRU. EPI delivered this prototype unit to EngineAir in April of 1998, and according to the contract between EPI and EngineAir, that PSRU was never to be installed on any aircraft other than the R&D aircraft owned and operated by EngineAir.

After having flown this PSRU for several hundred hours, including several long cross-country races at very high power levels, EngineAir violated their agreement and delivered the prototype PSRU to one of their customers.

Although the engine delivered to that customer reportedly operated correctly during testing at the EngineAir facility, the customer claimed that he experienced a series of problems allegedly caused by malfunctions of the computers which control ignition and fuel delivery. (It was later discovered that the electrical system on the owner’s aircraft was the source of most of those problems.)

Subsequent to an owner-performed maintenance operation on an engine accessory, the owner incorrectly reassembled the parts, leading to a failure of the coolant pump drive during takeoff. That failure caused severe overheating which resulted in extensive damage to the internal components of the engine. The owner took it upon himself to disassemble the engine, and repaired it using parts other than those specified by the manufacturer.

The owner then reinstalled the PSRU on the engine (without having bothered to obtain a copy of EPI’s detailed installation instructions) and reinstalled the modified powerplant in the aircraft.

It was recently reported to EPI by a highly reliable (and, we believe, objective) witness that, during the run-in of this engine subsequent to the customer’s modifications, the engine operated for extended periods at high-load, while enduring severe, continuous backfiring and misfiring, presumably due to improper wiring of the ignition system.

The PSRU's outer shaft (p/n 790048) failed during a ground runup just subsequent to these backfiring and misfiring episodes, reportedly just after the flaws in the wiring system had been discovered and corrected, which elinated the backfiring and misfiring.

The owner reported the failure to EPI the day it happened, but he denied that anything unusual had occurred prior to the failure. EPI attempted, on several occasions, to obtain the broken pieces from this shaft for examination, and offered to replace the shaft (a $1,000 part) for free in order to obtain the pieces for examination.

When EPI’s efforts to obtain the broken shaft for analysis were unsuccessful, EPI appealed to the owner, on the basis of concern for the safety of others using this product, to assist with the discovery of the problem and it’s correction. The owner explicitly refused to cooperate. Neither the failed shaft nor any of the critical diagnostic information detailed above was ever obtained from the owner.

It is EPI’s understanding that this prototype Mark-9 PSRU had accumulated over 700 hours of flight time at the time of the failure, which represents well over one hundred fifty million fatigue cycles. With the exception of an oil pump drive tang, no replacements to any internal parts of that PSRU had occurred after it’s delivery in April of 1998, and EngineAir had refused to periodically return that prototype PSRU to EPI, as previously agreed, for inspections during it’s use as a prototype.

NOTE: Contrary to the information being given out by EngineAir, the prototype PSRU is identical to the delivered production versions in every structural and operational sense except for the different reduction ratio. EngineAir's argument that "…there is no concern regarding production PSRU’s because this failure was due to a prototype anomaly…” is PATENTLY FALSE.

3.2 DETAILS: Failure #2

The second failure occurred on the Mark-9 production unit number 3, during Sun n’Fun 2002.

After arriving at Sun n’Fun following an uneventful flight of several hours, the owner of the subject aircraft detected a small oil leak at the front of the powerplant. The leak was perceived to be coming from the oil pump at the front of the PSRU.

Instead of inspecting the pump which was on the PSRU, EngineAir personnel decided to install a new pump from EngineAir’s parts inventory. Within minutes after startup with this new pump installed, the PSRU outer input shaft (p/n 790048) failed completely.

Post-failure inspection of the PSRU revealed that the new oil pump had jammed the outer input shaft. Inspection of that oil pump revealed that the housing (manufactured by a vendor selected by EngineAir) had not been machined in accordance with the engineering drawings, and therefore interfered with the outer shaft by a significant amount. That interference caused a huge axial compression load on the outer input shaft, which buckled the walls of the shaft, leading to it’s immediate failure.

The broken shaft was replaced with one manufactured by EPI, and the original oil pump was reinstalled. The aircraft was flown extensively during and subsequent to Sun n’Fun.

The total time on the PSRU at that time (and on the failed part) was approximately 75 hours. Within 20 hours prior to the failure, all critical parts (including the failed shaft) had been inspected and found airworthy in accordance with standard aircraft NDT procedures (specified in EPI Service Bulletin 260002).


EPI has re-examined the design data for both the PSRU input shafts, and verified the calculations of stress levels for both shafts. These calculations confirmed that the shafts (identical in the prototype and production units) are suitable for the intended environment.

However, it is a known fact, which is explicitly detailed in the EPI Mark-9 Installation Manual, that both input shafts must be handled with great care. This is no different than the level of care required with many parts in certificated aircraft engines.

In point of fact, neither input shaft was designed to withstand the abuse to which the failed items (part number 790048) were subjected.

In the interest of safety, EPI has designed a replacement outer shaft (part number 790056) which incorporates features making it more tolerant of the severe impact loads encountered during backfiring and misfiring incidents. No amount of redesign can compensate for ham-fisted mechanics or the installation of incorrect parts.

The new part (790056) is visually different from the original part, making it unlikely that the original part will be confused with the replacement part.


It is clear that a sudden failure of the PSRU input shaft in flight will cause an immediate and complete loss of power and a forced landing. The success rate of forced landings with high performance, high wing-loading experimental aircraft has been less than encouraging. Such a failure while operating close to the ground, especially during takeoff, would very likely result in fatalities.

Although both documented failures occurred on the ground during runup, and both occurred immediately subsequent to the performance of improper maintenance operations, nevertheless, EPI takes the position that any flight conducted prior to removing the subject input shaft (part number 790048-C) and replacing it with the improved product (part number 790056-A) constitutes a severe hazard to life and property.

On May 21, 2002, EPI Inc. notified EngineAir, in writing, about these concerns and safety issues. EPI, Inc. recommended that EngineAir acquire the design of this new shaft (790056) and implement an immediate recall program which would take all 790048 input shafts out of service and replace them with the new design.

In view of the serious nature of the situation, on July 2, 2002 EPI voluntarily provided EngineAir with the engineering drawings and process sheets necessary for the manufacture of the replacement shaft (p/n 790056).

<< Return to: Contents   Go to: Top of Page