1
\$\begingroup\$

I'm doing pre-certification testing to ETSI EN 303 372-2 V1.1.1 (2016-04), specifically the Adjacent Signal Selectivity test in Section 6.1.

I am a Systems engineer with lots of experience designing and performing equipment testing to DOD, DOE, FAA and FDA standards. Though I'm not an RF engineer, the test mentioned above makes no sense whatsoever. Here's the relevant section:

6.1 Adjacent signal selectivity

a) Two test signal generators shall be used. Each signal generator shall generate a modulated signal in the IDU input frequency range and thermal noise.

b) The signal generators shall be connected to the IDU input through a splitter (combiner).

c) The symbol rate shall be set to the low end of the IDU's range.

d) The test signal generators shall be set to the frequencies and levels according to table 1. For each row of the table:

a) The IDU shall be set to receiving the signal of the first test signal generator.

b) The second test signal generator shall be set to output signal off.

c) The noise level (or signal to noise ratio) of the first test signal generator shall be varied in order to determine the threshold for quasi error free reception.

d) The second signal generator shall be set to output signal on.

e) The noise level (or signal to noise ratio) of the first test signal generator shall be varied in order to determine the threshold for quasi error free reception.

f) The degradation is equal to the noise level (or signal to noise ratio) determined in step e minus that determined in step c.

e) Repeat from d) with symbol rate set to the high end of the IDU's range.

f) The result is the highest degradation found.

What makes no sense to me is step e) under d): There is no limit to how much the SNR of the desired signal can be increased to counter the influence of the interfering signal!

What I believe is needed is to quantify the effect of the interfering signal on the desired signal, and this step actively cancels that out.

What am I missing here? What aspect of selectivity is this test seeking to measure?


Update:

Based on comments below, my understanding is shifting:

For now, I've decided to take an approach that splits the issue down the middle:

  1. For the "wanted" signal, determine QEF at the receiver, then record both the generator and receiver SNR (they will likely differ by a small amount).
  2. Add 1 dB SNR to the "wanted" generator, and ensure the same step change is seen at the receiver.
  3. Add the interferer signal, and ensure the receiver SNR is reduced by less than 0.4 dB

My hope is this test procedure meets both the letter and spirit of Section 6.1 of ETSI EN 303 372-2 V1.1.1 (2016-04). It keeps the "wanted" signal very close to the receiver's QEF while making the effects of the interferer signal clear and unambiguous.

One minor concern is to not push the receiver below QEF during this test, because SNR measurements below QEF may not be completely reliable. That is, I don't want to have to fully characterize the receiver SNR value outside of the fully functional state, as it tends to deviate from the generator SNR at both high (>30 dB) and low (below QEF) values when directly connected.

Also, the spec says nothing about the SNR of the interferer signal: I initially chose to make it equal to the SNR of the "wanted" signal, though I am now revisiting that decision as degraded SNR looks like increased roll-off (alpha), so doing any SNR reduction to the interferer would make it differ from what the spec requires.

\$\endgroup\$
  • \$\begingroup\$ Ran the test as described in the update above, and it was both sensitive and accurate, in the sense that I have very high confidence that the actual certification testing will yield no surprises. Thanks for the discussion! \$\endgroup\$ – BobC Jun 6 at 15:20
1
\$\begingroup\$

The test measures the DUT's sensitivity while in the presence of an in-band, adjacent channel interferer. It measures a ratio, the source level (6.1 C) vs the interferer (6.1 D) Since it is a ratio, the absolute level does not matter for linear systems. The desired signal is set to the quasi free error level, that is usually 3 to 6dB above the sensitivity level.

\$\endgroup\$
  • \$\begingroup\$ Yes, but there is no limit on how high the SNR of the desired signal can be set to overcome the interferer. The ratio measured is at the tuner, not the signal source (a DekTec 2115B-GOLD in this case). In other words, this test can't be failed unless a really crappy receiver is used. \$\endgroup\$ – BobC Jun 3 at 19:42
  • \$\begingroup\$ To clarify: I'm looking at test results from a UK testing firm for the prior spin of our product, and they make no sense to me. It appears they didn't record any generator values, and looked only at tuner values. Hence my confusion. \$\endgroup\$ – BobC Jun 3 at 19:45
  • \$\begingroup\$ The limit to the SNR in this case is your own system performance. Personally, I don't know why ETSI tests RX, but this is their house, their rules. They also want you to demonstrate that your system doesn't start to behave erratically in such conditions. \$\endgroup\$ – Lior Bilia Jun 3 at 23:51
  • \$\begingroup\$ As I read this spec (over and over), I'm beginning to suspect it is written to minimize assumptions about the equipment being used, both the unit under test and the test equipment. That is, it does not literally describe the actual test to be performed, but more the requirements for that test, the goal. See my comment to Sunnyskyguy's answer. \$\endgroup\$ – BobC Jun 5 at 14:23
0
\$\begingroup\$

Final draft ETSI EN 303 372-1 V1.1.1 (2016-06). Seems to have changed the criteria to 1dB of Rx compression must tolerate 20dB rejection of adjacent channel.

https://www.etsi.org/deliver/etsi_en/303300_303399/30337201/01.01.01_30/en_30337201v010101v.pdf

\$\endgroup\$
  • \$\begingroup\$ We must be very careful about "adjacent" channel (with normal spacing) and an "interfering" channel (lacking normal spacing). This test specifically concerns interference. For now, I've decided to take an approach that splits the issue down the middle: 1. For the "wanted" signal, determine QEF at the receiver, then record both the generator and receiver SNR (they will likely differ by a small amount). 2. Add 1 dB SNR to the generator, and ensure the same step change is also seen on the receiver. 3. Add the interferer signal, and ensure the receiver SNR is reduced by less than 0.4 dB. \$\endgroup\$ – BobC Jun 5 at 14:19
  • \$\begingroup\$ I have only done co-channel interference CNR vs SNR and optimal Rx IDC BER for ISM band at 1GHz to our in-house requirements and not to this std. Sorry I could not help more. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jun 5 at 15:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.