Wireless Systems for Measurement – Part 1

by Steve Liddle

Wireless Systems for Measurement: Steve Liddle explores the use of wireless systems for audio system measurement.

Introduction

A recent thread on the SynAudCon forum generated an interesting discussion on the use of wireless systems for audio system measurement.

This was sparked by my experience of seeing a well reputed German doctor of physics performing a sound system setup in a theatre using a Sennhesier Evolution wireless system in combination with a NTI MiniSPL microphone.

I asked the usual question about the compatibility of the compander with audio measurement and was assured that extensive testing revealed that it was actually surprisingly linear.

So 2 years later I finally got round to purchasing my own Sennhesier EW100 G1 system and and was pleased with the transfer function magnitude response. Happy days, or so I thought. The thread on the forum resulted in several suggestions for more revealing test signals than the pink noise and log sweep I had already tried.

Why use a compander?

Before we examine the measurements let’s look at some background regarding the need for a compander in a wireless system.

In simple terms, the dynamic range of a microphone can easily span 100dB or more from the noise floor to the maximum SPL before distortion occurs. An analogue wireless transmission link can have a dynamic range of less than 60dB from the RF noise floor in the link and the maximum transmitted signal level. How do we transmit that 100dB over a link that only has 60dB dynamic range? We compress it. However, at the receiving end we would like to be able to appreciate the full dynamic range of the performer, from intimate whispers to howling banshee wails. We therefore need to reverse the compression, a process called expansion. A compressor followed by an expander is known as a compander.

Not all companders are made equal. Ideally a compander would compress and expand in a completely linear fashion. However, many have varying ratios of compression and expansion which means that the incoming signal levels  would determine the level of companding. With regards to measuring the signal chain this could mean, for example, that our carefully executed equalisation could be ineffective at a higher or lower signal levels.

Wireless systems are not the only use for companders. You’ve all heard of Dolby Noise Reduction, right? (Am I showing my age here? I can remember the days of Chrome Bias vs. Normal Bias cassette tapes as if it were yesterday). Dolby use companders to increase the levels recorded to the magnetic storage media so the signal to noise ratio is improved. The expander restores the dynamic range on playback. They also use pre-emphasis before recording to tape, increasing the high frequencies with a shelf filter.

Companders are also used in sound systems at inserts in the mixing desk. In this case they are used to increase dynamic range by attenuating signals that fall below a threshold and compressing above another threshold. This serves to lower the noise floor when there is no signal present.

What do we need in a measurement system signal path?

1. Linearity – what goes in as X comes out the other end as X. Frequency shifting or gain effects are not desirable here. There is a case for adjusting the output level of the wireless receiver to suit the sound card but there should be no uncontrollable effects.

2. Time Invariance – any time variant phenomena must be avoided. For example, opening and closing a door during testing, or a latency delay that changes with each average.

The Measurements

Ok, so now we know what is required of our wireless link lets compare 3 systems and see how they perform. I borrowed the Lectrosonics kit from my esteemed colleague, Ingemar Ohlsson, of Audio Data Lab in Stockholm.

3 products, 3 price ranges:-

1. Lectrosonics TM400 – regarded as the industry standard. Highest cost of the 3 systems up for testing. Compander free. Uses a proprietary algorithm to encode digital audio information into an analog format. The audio is transmitted as a digital signal over an analogue RF link.

2. Sennheiser EW100 G1 – Still a significant investment required for the newer versions but used kit can be obtained for roughly 50% list price.  High quality analogue transmission using proprietary HDX linear compander.

3. Behringer ULM2000 – Budget analogue transmission with compander (and interesting EQ).

I chose a number of test signals, some at the suggestion of the heard, and performed transfer function measurements using EASERA Systune.

The following colours apply in the charts below:-

Green – Behringer ULM2000

Blue – Lectrosonics TM400

Red – Sennheiser EW100 G1

Edit: I have added a pink line showing the reference responses for a loop back cable connection after the comments from Dan Bavholm. These are marked M (for Motu)

The wireless systems in these tests basically replaced a loop back cable in a dual-channel FFT setup. When making the measurements I was careful to set the delay offset in Systune to avoid the high frequency response being affected. This showed that the Lectrsonics kit with it’s digital conversion and algorithms had a latency of several miliseconds (3.72 to be precise) whereas the analogue systems had almost negligible latencies (0.062ms)  – something worth bearing in mind when trying to determine an absolute propogation time in a signal path. The measurements we  generally perform with these systems would be relative delay times which would not be affected by these latencies anyway.

Pink Noise

All 3 systems handled this signal well with no surprises other than the Behringer ‘Ultra Musical’ frequency response. Pink line shows the flat reference response.

Again, no surprises here – the ripples in the response are artifacts of the display driver in Systune, only 2 averages being taken and no smoothing being applied.

 

Linear Sweep 1.37s

Things start to fall apart below 1kHz for the Sennheiser and Behringer systems with the linear sweep. Even the Lectrosonics system shows some deviation from the normally flat response.

 

Linear Sweep 0.09ms

I was forced to take 32 measurement averages to be able to see this set of responses. Interestingly the Lectrosonics system showed the largest deviation. Without knowing anthing about the proprietary digital processing algorithms it is difficult to draw any conclusions as to why this happens.

 

Tone Burst – from the Don Keele CD, 100Hz centre

This signal was taken from the CD that Don Keele kindly made available last year. It is a composite tone burst with content covering the octave centered on 100Hz. I chose a signal with 2 bursts per second and set the FFT size to 1.37s. The measurement system didn’t show any readable response until I got got up to 32 averages. Each wireless system more or less handles the burst with the Lectrosonics showing less artfacts in the out of band regions.

Edit: I applied the coherence masking after the suggestion from Dan Bavholm. You can see the traces fading above 250 to 300 Hz. I only used 2 averages to obtain the pink reference line in this case.

 

Tone Burst – from the Don Keele CD,  50Hz centre

Just for fun I also tried a composite burst centred around 50Hz. I wasn’t expecting anything spectacular as this is the region where the reponses of all 3 systems have their high pass filters. Nevertheless, the results weren’t too bad.

Edit: again the application of the coherence filter shows the valid response data within the range of interest and only 2 averages required to obtain the reference.

Summary

In the results above we see how signals with different time structures and various frequency ranges affect the performance of the wireless transmission path. The compander dependent systems generally show the greatest deviation from linear but aren’t as bad as I expected. They handle broadband pink weighted signals quite well.

In part 2 of this 3 part blog I will present the results of varying the stimulus level in the test setup.

Thanks to Pat Brown for reviewing this article.