S meter calibration and scale shape

Correspondents often raise the issue of S meter calibration and scale shape.

Though some people are dismissive of S meters, they do form a useful quantitative adjunct to the receiver, more so if their strengths and weaknesses are understood and can be worked around.

Calibration convention

There is not a true standard for S meter calibration, but a widely adopted convention is 50µV at the receiver input terminals for S9, and 6dB per S point. One manufacturer states 100µV emf (meaning unloaded generator voltage) which is the same thing when loaded with a matched load.

IARU captures this convention, but has a variation in Region 1 at VHF and above. For simplicity, lets stay with the simple 50µV at the receiver input terminals for S9, and 6dB per S point.

50µV at the receiver input terminals corresponds to -73dBm in a 50Ω load, so S9 is equivalent to -73dBm.

S meters in FM receivers are rarely calibrated to the convention for reasons related to FM receiver design, they will often read full scale on about 10µV.

The remainder of this article is about SSB communications receivers.

Typical HF SSB receiver configuration

A HF receiver is commonly configured with a selectable preamplifier, and an S meter that is 'calibrated' with the preamp OFF (which is the normal mode on the lower HF bands where external noise dwarfs internal noise, even with the preamp OFF).

Fig 1:

Lets assume that our model receiver is a conventional receiver (meaning not SDR) with 2kHz effective noise bandwidth that, without preamp, has a noise figure of 16dB or Te=11,200K (using Receiver sensitivity metric converter above).

Now lets put a preamp in front of that with a gain of 20dB and NF=5dB (Teq=630K).

Fig 2:

System Te=11200/10^(20/10)+630=742K, or system NF=5.5dB, sensitivity of 0.12uV or -125.5dBm for 10dB S/N. Noise floor is -135.5dBm.

Now, AGC is delayed, ie it does not operate on the weakest signals but is delayed until the signal reaches a reasonably high level compared to the internal noise before the receiver noise figure is degraded by reducing gain. The AGC threshold is typically 25dB or so above the noise floor, so it is set for say, -110dBm (~S3) preamp on, or -90dBm (~S6) preamp off in the model case.

The fact that the S meter zero is S0 and it creeps off the stop above -90dBm (~S6) with the preamp off (which is how they are commonly calibrated) means there is necessarily almost 6 S points of error at the bottom of the scale if the scale is marked S0 at the low stop! If the system is calibrated to the convention of -73dBm at S9, the scale below S9 must be compressed below 6dB per S point at least in part of the range.

This would be a pretty common configuration, the numbers will vary a little from receiver to receiver, but the example captures the issue that an AGC derived S meter will not lift off the stop until the signal reaches more than the convention of -127dBm for S0, and if the meter is scaled from S0, it is inconsistent with the convention. 

Receivers often have a facility to calibrate the S meter at one, two or three points on the scale, and so become accurate at those points and interpolate with some success between those points. It tends that S meters may be reasonably accurate from S7 or so to about S9+40dB, but are wildly inaccurate above and below those points.

Users can usually establish the S meter calibration configuration (eg preamp, attenuator etc settings), and depend on mid scale readings, but be suspicious of high end or low end readings until such time as they are checked by measurement.

Case study

S meter readings on a TS2000 with preamp OFF were converted into dBm using the convention that S9 is -73dBm and 6dB per S point, and those measurements were compared with the input signal required to indicate S0, S1, S3, S5, S7, S9, S9+20dB, S9+40dB, S9+60dB.

Fig 3:

Fig 3 shows that error is moderately low from -85dBm (S7) to -17dBm (S9+60dBm), but there is inevitable compression below as they try to shape the scale for S0 at zero deflection. Because AGC is delayed until about -103dBm (S4) there is inevitable error when the S meter is scaled down got S0 for zero deflection.

Different receivers will be a little different, but receivers that use delayed AGC and derive the S meter from AGC are likely to give similar effects.



Version Date Description
1.01 05/10/2012 Initial.

© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.