RF TESTING OF MODEMS EMBEDDED IN RADIOS

1
RF TESTING OF MODEMS EMBEDDED IN RADIOS

• J. W. Nieto
• Senior Scientist
• Harris Corporation
• RF Communications Division

2
Presentation Overview

• Overview
• Motivation
• Baseband Testing
• RF Testing
• RF Test Setups
• Recommendations
• Summary

3
Motivation

• Most HF waveform standards define a set of
  minimum performance requirements which must be
  met in order for a modem to be compliant with the
  standard
• These performance requirements have been
  determined using a baseband HF channel simulator
• STANAG 4539 3200-12800 bps adds baseband radio
  filters (digital filter taps provided in STANAG)
• If minimum performance numbers have no margin,
  demodulation process may meet performance specs
  when tested at baseband but not when embedded in
  radios

4
Motivation

• Differences between baseband testing and RF
  testing
• Transmit Side
• Transmit radio filter
• Transmit Gain Control (TGC)
• Automatic Level Control (ALC)
• Receive Side
• Receive radio filter
• Automatic Gain Control (AGC)
• To complicate things further, radio manufacturers
  will likely implement all of the above differently

5
Motivation

• Example US MIL-STD-188-110B Appendix C 12800 bps
  waveform utilizing radios with poor AGC
  characteristics (e.g. too fast)
• 12800 bps is 64-QAM constellation with no
  coding/interleaving
• Waveform will run with errors even in a noise
  free environment
• BER will never be less than 1e-5 at any SNR !
• Will fail certification !
• At baseband, same demodulation process likely
  exceeded performance requirement of standard

6
Motivation

• Why did modem fail at RF ?
• AGC causes sudden changes in amplitude of
  waveform. Since amplitude contains information
  for M-QAM constellations, errors have been
  introduced
• TGC and ALC can drive signal too high into Power
  amplifier, distorting signal and possibly
  modulating signal. This effectively limits the
  maximum SNR measured at the demodulator and may
  cause bit errors
• TX and RX radio filters spread impulse response
  of system. Demodulator process has to use more
  taps for channel estimate which allows more noise
  into system. Typical radio filter losses between
  1-3 dB

7
Baseband Testing

• Goal of testing Repeatable and comparable
  long-term BER modem performance measurement
• HF Channel Simulator
• Watterson Model
• Mechanism for simulating fading characteristics
  very important for achieving repeatable and
  comparable results
• No radio filters or known (specified) digital
  radio filters
• No TGC, ALC, or AGC
• Signal to Noise Ratio (SNR) is average long-term
  SNR of waveform in 3 KHz
• For 110B Appendix C and F Waveforms, SNR
  calibration may require several minutes
• Test should ignore first few minutes of results
  in order to avoid Acquisition related bit
  errors (specially in fading channels)

8
RF Testing

• RF setup extremely important
• Improper setup can limit SNR of system with no
  noise (RX SNR must be checked !)
• If possible, TGC and ALC should be turned off at
  transmit end
• Test should be performed at low power with enough
  attenuation between radios
• Slowest AGC setting should be selected for RX
  radio
• If AGC is turned off for RX radio, RF level into
  radio must be carefully selected

9
RF Test Setup 1
Baseband
RF
MODEM
X
RADIO 1
BER TESTER
Synthesizer
HF_SIM
X
RADIO 2
RF
10
RF Test Setup 2
Baseband
MODEM
HF_SIM
BER TESTER
Radio 1
RF
11
RF Test Setup

• RF Test Setup 1 (Traditional)
• Using low-cost mixers, maximum SNR achievable
  close to 30 dB but very sensitive to input/output
  signal levels at both mixers
• Reliable BER measurements up to 20 dB unless SNR
  in channel simulator is selected based on
  combined channel simulator and setup SNR (For
  example, if max SNR 30 dB and the channel
  simulator is set to 30 dB SNR, effective SNR is
  27 dB, 3 dB off from desired point)
• RF Test Setup 2 (courtesy of JITC _at_ Fort
  Huachuca)
• Maximum SNR achievable gt 40 dB
• Radio 1 should have TGC and ALC disabled (or
  preferably, an exciter with no PA is used)
• Reliable BER measurements gt 30 dB

12
Recommendations

• Overall testing strategy needs to be defined
• Standards should
• Have performance specifications for baseband
  testing and for RF testing
• Provide more details on how testing should be
  done for baseband testing and RF testing
• Specify characteristics of HF channel simulators
• Provide details on how HF channel simulators
  characteristics can be verified
• Provide details on how calibration for proper SNR
  should be done

13
Summary

• Modems which are embedded in radios suffer more
  performance losses than modems tested at baseband
• HF waveform standards currently dont address
  this issue which can result in embedded modems
  not meeting performance requirements with same
  demodulation process that meets performance at
  baseband
• HF waveform standards should provide more detail
  as to how testing should be done
• HF waveform standards should provide more detail
  about channel simulator calibration, generation
  of fading characteristics, noise, filtering, etc