Article First Published in Electronics World Magazine, March 2009
Users of unlicensed wireless modules are likely to be familiar with the EN300-220 specification, governing the use of such devices on the European ISM bands. Those who have looked a little closer (or who work in the alarm or security industries) will furthermore be aware of the receiver performance classifications detailed in this spec.
To quote the latest release, there are three categories (see section 4.1.1):
Highly reliable SRD communication media; e.g. serving human life inherent (may result in physical risk to a person.)
Medium reliable SRD communication media; e.g. causing inconvenience to persons, which cannot simply be overcome by other means.
Standard reliable SRD communication media; e.g. inconvenience to persons, which can simply be overcome by other means (e.g. manual)
According to 300-220 , the lowest class of receiver (class 3) only has to achieve a certain minimum level of rf sensitivity. Class 2 receivers are also required to meet a certain blocking performance level, while the highest performance class 3 parts must meet a more stringent blocking level, and an adjacent channel rejection test also (This corresponds to a receiver performance not much inferior to long ranged PMR or land mobile radios, governed by EN300-086).
Which is all very well written in the spec, but how do you relate this to the performance of your receiver ? How should these parameters be tested, and most particularly, is the receiver you’ve paid for actually compliant with the class claimed by it’s manufacturer ?
The receiver test methods stipulated in 300-220 are fairly straightforward, although for ‘old school’ RF engineers used to measuring absolute sensitivity and then relating rejection specs to that there is a surprise: EN300-220 specifies in terms of absolute interferer levels, not relative values:
A level of -107dBm must yield 20dB (or better) sinad when measured through a psophometric filter as detailed in ITU-T rec. 0-41.
For data-only links an 0.1% error rate, or 80% message decoding success, are given as equivalents to the 20dB sinad point.
The levels above relate to a 25KHz channel spacing (16KHz BW) radio, For other receiver bandwidths, the limit sensitivity level is given as 10 x log (BW/16) - 107 (dBm)
The basic method for all adjacent channel and blocking tests is the same. Remember that the wanted and interfering signal levels are at the combiner output / radio input. Not the level at the signal generator.
Two generators are used, feeding the test radio through a combiner or coupler.
A wanted signal is applied at a level 3dB higher than the limit sensitivity (-104dBm for a 25KHz unit).
An unwanted signal (un modulated) is then also applied, at a particular offset and level depending on the test, and the measured sinad must remain 20dB or better
This test applies only to class 1 radios.
The interferer has a level of -50dBm and tests are made with this signal one channel above and one channel below the wanted carrier.
The adjacent channel tests are repeated with the wanted level increased by 43dB (-64dBm for 25KHz) and the interferer at -20dBm
(For channel spacings >25KHz this interferer has a level of -44dBm in the first test, and -10dBm in the second)
For class 1 radios, the interferer has a level of -20dBm and tests are made with this signal 2MHz above and below the wanted carrier. A saturation test is then made, with the same level of interferer but the wanted signal increased by 40dB
Class 2 radios are tested with a -69dBm interferer at +/-2MHz, and with a -44dBm interferer at +/- 10MHz
No receiver is permitted to emit any spuri with levels exceeding:
2nW between 9KHz and 1GHz, or
20nW above 1GHz.
The methods used are fairly straightforward, but there are a few problems in these test techniques that aren’t immediately obvious:
1. The audio filter. A psophemetric filter of this type provides a noticeable improvement in sinad compared to the flat 0.3-3.4KHz ‘speech bandwidth’ filter provided by many sinad-measuring instruments, and a direct comparison is difficult to make. Ensure you’re using the right filter, or you could be throwing away valuable performance margin or failing your radio unfairly.
2. The combiner. Use a good, broadband part. To keep generator output power levels down, use a low loss directional combiner (a part such as the Mini-Circuits ZSC-2-4) with loss around 4dB in preference to a 12dB resistive type. Remember to check your actual signal levels at the combiner output, to allow for cable and coupler losses and other imperfections.
3. Signal generator noise. This is the biggest cause of test-error and wasted time in measurement regimes such as this.
The wanted signal generator is uncritical. It is a modulated low level, on-channel signal. Almost any RF source or communication test set will be good enough.
The interfering signal generator is another matter. This requires a good adjacent channel noise level or the measurements will be dominated by the noise sidebands of this generator directly impinging on the carrier. A noise level of around -115dBc/Hz at 25KHz offset or better is required.
When the blocking measurements are made, a different problem becomes evident, as at the high output levels required many signal generators exhibit excessively high wideband noise output. Even middle-range generators can be inadequate.
(To check for this generator noise effect, conduct a blocking test on a known good, high performance radio unit. Increase the interfering signal level from about 25dB below it’s eventual level in 1dB steps. A sudden degradation in receiver sinad, out of proportion to the signal level change, indicates the generator noise floor has risen. This will usually be co-incident with an internal attenuator switch over in the generator)
Inexpensive signal generators and comms test set RF sources are rarely adequate as off channel, interfering sources. Middle range units (such as IFR2023A or similar) are barely sufficient. Seriously consider (briefly) hiring a really high specification generator (such as an Agilent 8665, a R&S SMHU or an IFR2040. Or better) for your tests.
The EN300-220 class 1 radio specification is quite exacting, but there is no reason why a well designed low power module shouldn’t comply with it. On the other hand, there are plenty of designs which are nowhere near this performance level in the marketplace, and some of them are being sold as ‘class1’ units.
Good luck testing and, as always, “buyer beware”
by Myk dormer, Senior RF Design Engineer, Radiometrix Ltd
Note: EN 300 220-1 V2.2.1 (2008-04) Implementing 2009