Dear Brad, (name changed to protect confidentiality)

I wanted to capture our findings and thoughts from my recent site visit while they are still 'hot-off-the-press' so here goes:

Headsets are David Clark HD10-76 models, connected at various locations in and around the trainer to RIUs that are housed in an equipment rack adjacent to the DACS processor. The run from headset to RIU is estimated to be around 40 feet, worst case. The RIU output channel is fed to a Rane HC6 headphone amplifier, which then drives the headphones.

Problems are characterised as being affected by background noise, and sidetone having an echo-like "head-in-a-bucket" sound characteristic. Breaking this into two issues:

This is apparent--to a greater or lesser extent--at different positions, and is different in character at each location. This generally manifests itself as a 'stuck-car-horn'-type tone, or a messy hum-buzz. It was also noticed that the noise included some pickup of local AM radio stations (I particularly liked the observer position with the Mexican band music). Different positions were 'tuning' different stations.

This all seemed indicative of some grounding issues, and lack of shielding in certain parts of cable runs. This was indeed discussed, as the cockpit is GFE, and parts of the audio wiring are thought to be non-screened. It was also noted that the part that is screened is brought together at an I/O breakpoint that includes not only audio, but also general cockpit I/O.

It seems that in order to start addressing this noise issue, a systematic review of all audio cable runs, connections, and shielding arrangements should be done. I would also suggest that the headphone output signals may not need to be screened, while the mic signals certainly must be.

Another suggestion: Place the RIUs at or as near as possible to the crew members. Moving the RIU closer to the crew eliminates many feet of analog, low-signal-level cable. The RIU is more than capable of driving the David Clark 10-76 directly without needing additional headphone ampifiers.

The David Clark 10-76 is known to us as a poorly performing headset for crosstalk behavior, and I believe we are seeing this exemplified here.

To examine the exact nature of the problem, I built a new model containing only RIU5 in and out objects in the feeders (channel A thru D); and a Vox object per input, a comm panel per position, and a single intercom in Signals. So we had 4 operators talking to each other.

I then disabled all sidetone signals in the whole model by setting the sidetone control to 00 in the comm panel. With this configuration, each operator should hear the others clearly, but have NO sidetone at all. This was not what we heard. Sidetone was evident at a lower level than before, and somewhat delayed. This was true for all positions. The suggested cause for this was electrical crosstalk in the headset. We then disconnected all headsets except one (so we should have no source for any crosstalk) and there was no sidetone with this configuration (so the crosstalk was not in the cabling, it only happens when the headsets are plugged in).

I also tried a Telex headset plugged into the system (directly into an RIU) and this sounds good, clear and without any problems. But since we only had one headset, and could not plug this into the crew positions (due to connector differences), we could not proceed any further with this line of investigation.

My approach to optimizing this would have been to increase the input gain as much as possible, and to decrease the output gains as low as possible. With the level of noise (see problem 1), however, we could not increase the input gain as high as would be needed. My thought would be to lower the output such that the crosstalk-inducing signal is less, though it is possible that the increased input gain might offset this.

I'm sure we'll be in contact soon regarding your findings, particularly on the cable side.

Regards,
ASTI Project Engineer