S Meter Comments

I was interested to read this thread about the S-meter in the MP. For any of you who are interested, I thought I would pass along the results of some testing I did on my MP's S-meter recently. I have always suspected that my S-meter is quite generous. So I brought home a Hewlett Packard HP E4432B calibrated RF signal generator from work to check it out.

From what I understand, the traditionally accepted definition of S9 is a 50 microvolt rms signal at the 50 ohm antenna input to the radio. A little algebra shows this to be the same as a signal level of -73 dBm. And for the size of an S-unit, I used the common 6dB-per-S-unit assumption. So, in my calculations, S8 = -79dBm, S7 = -85dBm, 10dB over S9 = -63 dBm, etc.

My goal was to create a simple graph showing the indicated S-meter reading on one axis, and the correct S-meter reading on the other axis. I was a little disappointed to find that there are lots of things that make huge differences in the indicated versus true S-meter readings. The curve was very different for each band I tested (80, 40, 20, 15, 10), and it was different depending on whether the preamp (menu setting 8-4) was set to "tuned" or "flat." So I ended up with a family of curves on my graph.
Fortunately it didn't change much with the IF bandwidth setting, or that would have made it even worse.

I would also like to thank the author of the nice little program "S-Meter Lite" which was very helpful in capturing the S-meter readings for each condition. For example, I started out with a signal level of -33dBm (40 over S9), and then decreased in 5 dB steps until I got to S9. Then I kept going down, but in 6dB steps to hit S8, S7, etc. The S-Meter Lite program captured a little graph of the S-meter setting as I let it pause on each reading for a few seconds. Then I entered all the data into an Excel
spreadsheet and made some calculations, and finally the graph, from these recorded values.

Just for a teaser, here's what I got for a power input of -73dBm (which should have indicated S9 on all bands, I would assume).

Condition Indicated on MP

80M Flat S9+14dB
80M Tuned S8.2 (the best--actually quite good
for indicated levels above S9)
40M Flat S9+10dB
40 Tuned S9+2.5dB
20M Flat S9+15dB
20M Tuned (same as 20M flat)
15M Flat S9+12dB
15M Tuned (same as 15M flat)
10M Flat S9+15dB
10M Tuned S9+26dB (the worst--WOW, very generous indeed!)

I also found that the slope of the S-meter curves was not too bad for signals above about S8 or S9. What I mean by that is that a 5 dB change in the input signal level produces about the right amount of change on the S-meter, etc. In this upper region, the S-meter is just reading too hot by so-many dB, as the table above hints. But at about S7 or S8 the indicated S-meter readings drop like a rock with modest reductions in input power. Going from indicated S6 to S2 is only about 6 dB change in input signal level, but it should be 24 dB to cover 4 S-units. So the bottom half of the S-meter is pretty useless, if you ask me.

You know, I was kind of disappointed that the good folks at Yaesu didn't do a better job on the S-meter in such an expensive radio, and otherwise fabulous radio. Obviously the AGC-derived S-meter signal is digitized, since it can be read via the CAT port by the program S-Meter Lite. So it would have been a simple matter to run the digitized S-meter word through a lookup table and correct it for the various mode settings, so that it produced a more accurate result. I don't think it would have cost them a penny in recurring cost--just a little additional software (OK, maybe a few bytes of extra ROM) between the segments on the front panel, and the digitized S-meter value. But alas...Does anyone have any different experience? I only have one radio to test--mine. Perhaps yours is different. Tom / W7GT
____________________________________________________________________________________________

This message is being sent to those of you who requested my S-meter plot. So here it is. Download MP RX Tests.xls here.

Let me explain a few points about it, in case it isn't obvious. The horizontal axis is the "true" S meter reading you should get from a perfect radio, with S9 = -73dBm and 6 dB per S-unit. The vertical axis shows the readings I got from my MP. For example, if you look at the purple curve (the one labeled "80M Tuned") when I put in an S9 signal (-73dBm), the
S-meter on the MP actually read a little over S8 (-78 dBm or so). The right edge of the vertical scale shows the S-meter values read off the MP's meter (actually read using S-Meter Lite). The left edge of the vertical scale shows the power levels (dBm) which correspond to the S-meter readings on the right. In other words, it's just the MP's S-meter, but with the scale labeled in dBm, rather than S-units.

The black line is the curve I would have expected from a perfect S-meter in the radio. I put a heavy orange line at S9 on each axis. Note, as I said in my first email, that the SLOPE of the curves isn't bad for larger readings on the S-meter. But they all drop off quickly when the meter reading on the MP gets below about S9 or so. Note how few dB it takes to get from S7 down to S2!

The curve for "80M Tuned" tracks reality pretty well from about S9+5 on up. I wish they were all that way. Other than 80M tuned, and 40M tuned, all the others are really hot, leading to giving very generous S-meter readings. Look at how bad the 10M Tuned is!

FYI, I have the Inrad mod installed, and I have turned down the Menu 9 IF gain setting from 12 to about 10, as the instructions indicate. I haven't looked, but it might even be lower than 10 Earl, you suggested I might turn down the IF gain to make it right, but as you can see from the graph, it would require a different IF gain setting for each band. And there's no way you could turn the IF gain down 25 dB (as you would need to for 10M Tuned) and still have the radio operate well.

As an editorial comment, I'll bet there are some out there who would say, "Well, of course you would expect the S-meter readings to change when you switch bands, or when you turn the preamp on or off, etc." But I don't think this is a correct idea, since as I understand it, the definition of S9 is "50 uV rms signal level at the antenna input to the radio," which has nothing to do with what goes on inside the radio. (But S-meter reading can certainly have a lot to do with your antenna, of course.) It just tells you how strong the guy's signal is at the input to your radio, not how good your radio is. All that your radio can do is to try to optimize the SNR and to provide appropriate filtering and processing to dig the signal out of the noise and the QRM.

As a bonus for you guys, who are clearly my soul-mates, I made some MDS (minimum discernable signal) measurements too. MDS measurements are always a little subjective. So what I did is to run my PSK waterfall program. This displays a histogram of amplitude versus time, with signal amplitude mapped to brightness of the trace. I 'm sure you all know what a PSK waterfall program looks like, such as Zakanaka, WinPSK, or Digipan. Anyway, that made it a little more objective, since I could turn down the input signal level until the little white line I was looking at began to break up. That's what I called MDS. And I could clearly hear the signal at the MDS level, although it was very weak. You will see that the MDS level changes quite a bit from band to band, and with the preamp set to Tuned or Flat. I measured the MDS at all filter bandwidths. FYI, I have the stock Yaesu filters in my SSB positions (since I don't ever operate SSB!). I have the stock Yaesu 500 in the 8.2 IF, and an Inrad 400 in the 455 IF. I have a pair of Inrad 250s in the 250 position. I just made one measurement in the CW portion of each band. I don't have any particular comments. I just wondered what the noise floor on my radio was, and how it varied with band and IF bandwidth.

I would love to get any rebuttal or reaction from you about my S-meter plots or the MDS measurements.

Earl, FYI, we exchanged a little email about switching COM ports to the radio recently. If I could get a program that allowed me to simultaneously run my Logger program with the radio, and at the same time run another radio control program, then I could write a program like S-Meter Lite that reads the S-meter, reads the radio's mode (band and tuned/flat), and then does an
inverse lookup table to then display PROPER S-meter readings on the computer. It wouldn't be hard, but the trick is having two programs talking to the radio at once. But that's a project for another day. I have never written a Windows driver, so I've got a lot to learn before tackling that.

And Greg (Mr. S-Meter Lite), thanks for the great little program. I would be happy to have you include whatever you think is useful from my data in your program. And I would also appreciate the opportunity to "approve" what you end up saying, as you offered to do. Of course one complication is that you can't just include a single lookup table for the MP, since there are
really a whole bunch of curves. And I only did the five non-WARC bands. If I included the WARC bands, it would only make it worse, cuz I assume they would be different from the ones I did do. I don't know if the MP is the only radio that has such a poor S meter. I did a single check of my trusty old TS-520S, and it was really pretty good on the one band I checked.

You would think an S-meter would be an easy thing to get right in these days of digital electronics. Tom / W7GT


_____________________________________________________________________________________________

I think I will try the same experiment that you did Tom when I get back to Idaho where the MP is. Here in Minnesota I currently have a TS-870 and by coincidence had run essentially the same experiment on it as you ran on the MP.

Interestingly, my results were somewhat like yours in that the meter was very non-linear below S8 and the S9 level was approximately -73 dBm. On the low readings remember that this is a test on the bench and the signal to noise ratio is quite good!

Readings made at 3787 kHz, CW mode, 600 Hz bandwidth, 50 ohm load. WaveTek 3000 Generator

(-137 dBm)0.03 microvolt = detectable signal
(-133 dBm)0.07 microvolt = (S7 signal for 160 meter DX !)
(-123 dBm)0.24 microvolt = quite audible, but no S-meter reading
(-113 dBm)0.70 microvolt = very audible, but no S-Meter reading
(-109 dBm)0.84 microvolt = 1 bar on the S-meter (signal loud)
(-102 dBm)= S1 (3 dB to get to S3)
( -99 dBm)= S3 (4 dB to get to S5)
( -95 dBm)= S5 (8 dB to get to S7)
( -87 dBm)= S7 (10 dB to get to S9)
( -77 dBm)= S9
( -61 dBm)= S9+20
( -43 dBm)= S9+40
( -23 dBm)= S9+60

___________________________________________________________________________

I ran a test of my MP's S-meter a couple of years ago and noted similar trends to Tod's & Tom's test results -
reasonable linearity above S9 and lousy below S9. I also used the somewhat arbitrary 6 dB per S unit convention &
expected to see a 48 dB RF input change to go from S9 down to S1 indicated, but only saw about half of that.

Furthermore, I found the AGC threshold on my MP on 20m to be about -108 dBm or about 0.9 uV. In other words, below that
level, the audio output from my MP drops in direct proportion to its RF input. That's not very good considering the MP's minimum detectable signal level is some 20 to 30 dB below that, depending on the Pre-amp setting.

My MP's S-meter, at the time, read S9 on 20m for an RF input signal level of -73 dBm, or 50 uV signal. Lots of
granularity though given the number of bars on the S-meter. That puts my MP's total AGC range at about 95 dB, assuming
it doesn't lose control at S9+60, which I did not test.

Since that test, I read "A High-Performance AGC/IF Subsystem" in QST, May 1996, by Bill Carver, then K6OLG &
now W7AAZ. After reading Bill's article, at least part of the reason for the MP's S-meter compression below S9 became
clear.

W7AAZ's IF has an AGC threshold of -131 dBm, when referred to the RF input. That's about 23 dB below my MP's AGC
threshold. So there just isn't anything available to drive my MP's S-meter below -108 dBm. In an ideal world where each

S unit is 6 dB and S9 is 50 uV or -73 dBm, then S1 would be -121 dBm.

And there's the rub - the MP's AGC system simply runs out of gas before getting down to a 'real' S1.

Strangely though my MP's S-meter then read S1 at about -97 dBm, so there's a gray area of about 11 dB from -108 to -97
dBm where the AGC voltage is apparently changing, could affect the S-meter, but does not.

By the way, my S1 measurement is consistent with Doug's comment on the MP's Technical Supplement setup for 1 segment on the S-meter. So Yaesu fully intends for S1 to be some 24 dB shy of the S1 ideal.

W7AAZ's IF, with its nearly perfect linear-in-dB AGC voltage over a 120 dB range, easily displays S1 correctly, as well
as S9+60, and anything in between. I have not heard Bill's IF yet, but have been told that listening to an AGC system
that handles this wide an input signal range, with good attack and hang characteristics, quickly becomes addictive.

So, given the state of the art in IF amplifier & AGC technology, there just doesn't appear to be any technical
reasons for having poorly calibrated S-meters in modern transceivers.

By the way, Tom, Bill made a comment in his article similar to yours regarding the apparent inability of high priced
transceivers to properly account for changes in band and preamps in indicated S-meter readings.
73, Paul N2PK