apache-labs.com

Is there a way to calibrate the S-Meter such that it matches reasonably to what my yaesu 3000 ft-dx and kenwood 870 exhibit? I primarily use 40 meters and have an average of an s-4 noise floor. I notice using my 7000 mk ii the average noise floor sits around s8 to s9. This is on a fully reset database and fresh install of thetis. Is some sort of pre-amp turned on? Any way to match to the other 2 radios? THx kindly.

Signal Generator.
I use an XG3.
https://elecraft.com/products/xg3

You're more likely to find your other two rigs are wrong...as most all rigs are notoriously bad on smeters adcuracy.
You'll also proabably find your other rigs smeters depend on the RF gain which is the wrong way to design an smeter.

Mike W9MDB

Hello,

I use the Elecraft XG3 Signal Generator too, like Mike. Perfect tool! And the calibration in Thetis works excellent too.

Main difference between Icom/Kenwood/Yaesu rigs and (ANAN) SDR's is that the Japanese rigs use 3dB/S-unit until S9, and 6db/S-unit above S9. You have to take this into account when you read the S-meter of these rigs.

73, Hans PA0Q.

Yes, you will find that the big three, Kenwood, Yaesu and Icom, all cheat very badly when it comes to S-meter accuracy and linearity. They all want to claim they have the "quietest receiver". What they all really have is the "quietest S-meter"

This is a particularly good link to review: https://vu2nsb.com/radio-systems/amateur-radio-station-ham-shack/radio-transceiver-s-meter/

As you can see from the data at that link, a true signal or, more typically, noise level of S4 is reported by these radios as an S1. Hence your noise level can be quite high but you placidly think how wonderful your receiver is because it is reporting an S1.

The only receivers that have anywhere near accurate S meter readings are those from Flex, Apache and Elecraft. It is not unusual for new owners of those makes to complain about how "noisy" their receivers are because the S-meter readings are true and correct (dare I say honest?), when in fact those are some of the best receivers on the market.

Thx for the replies, I went to the calibration menu and changed the default -70db level reference to -80db reference at 5.000 mhz WWV and at least have a closer representation to what I'm used to. I'm not after exact perfect values, but enough to match the two other radios. Thanks !

I'm curious - why would you want to match what is known to be inaccurate - i.e. not correct? The factory setting was I'm sure very close to correct from the get-go. It is calibrated such that S9 = -73 dbm. Note all other manufacturers - ok, the Japanese, match this as well as seen in their reviews in QST. So they ARE the same at that level - the level you changed. From there going up S9+ values they continue to be quite similar.... but below S9 it falls apart, the further it is below S9 the more inaccurate the other radios are. For example, S8 on the Anan is S7 on the other radios... S7 is S5.... S4.5 (between S4 and S5) is S0 on the other radios! So with them, you THINK you are really low noise - but the reality is it's S4.5 or -100dbm.

By changing the calibration point you only changed the single point where they agreed and now BOTH would be inaccurate, the slope is different for each and thus the inaccuracy will still be there.

What I will do is I'll tell people that their report is an "accurate S meter reading on a calibrated meter".

Also: be sure you have the RX Meter set to "signal" and not "signal avg".

Gary
K9RX

What good is the S-meter if most of my signal contacts, even weak ones, are S9 + 10 over? This is why... I can't respond to my contacts telling each and everyone of them you're 10 over lol. I'll reset the database yet again and see if the S-meter is more reasonable upon reset.

S meters are not supposed to be "reasonable", they are supposed to be accurate.

Resetting your database will not make it any more "reasonable", but it may make it more accurate depending on what you did to the calibration.

If you have access to an accurately calibrated signal source, you can further improve the accuracy of the S meters. Based on my own experiences, and some anecdotal ones I have noted in the past, people have typically seen around a 3dB difference after calibration. Note that this is an offset only and has nothing to do with linearity. Our S meters seems to be very linear, i.e. a 6dB (S unit) change in signal strength is reported properly at all signal levels.

However, at the end of the day, providing S meter based signal reports really isn't the best method of describing signal strength and quality. It would be better to provide everyone with a signal-to-noise (SNR) report. SNR is really what matters. If somebody is S9 and you neglect to tell them you have an S8 noise floor they may think they are perfectly readable when in fact they are not.

I see the conundrum. Call it old habits, Im simply not used to having ALL signals generally above S9, especially when biased with 2 other radios.

I just reset the database and will try it out again for a day. The dbm and microvolt readouts appear within rational scope as before. Guess I'll get one of those elecraft signal gens you fellas spoke of.

There's another significant difference between the "big" three and SDR's like our ANAN.

The S/dBm-meter in our SDR software is completely independent from the adjustment of any attenuation, pre-amps, RF (AGC) gain, or whatever. The readout of the meter will remain exactly the same, no matter what you adjust between the ADC's and the final stage of your SDR. Our S-meter just shows the value of the signal received from the antenna lead into the ADC, nothing more, nothing less.

Now do this with an I, K or Y rig: When the signalstrength of the other station is not S9+20dB (what he wants to hear from you) you just add 20dB gain with the pre-amps and tell him the result. Friends forever!
Same you can do with your noiselevel. When it is'nt S1 you just give 20dB attenuation.

Who is fooling who?

73, Hans PA0Q.

Good point, Hans! I believe the Flex also references all S meter measurements to the antenna jack, i.e. takes out any attenuation or preamp gain. Do the Elecraft radios work that way as well?

It's worth noting that this is how laboratory and technical test equipment works. When you look at a power measurement on an Agilent spectrum analyzer (for example), the power is referenced to the input connector and is not affected by any internal attenuation that might be switched in.

The S meter used to bug me too. It took a couple of years to get over it and I still leave it on average. The Anan agrees with my HP signal generator across a wide range but it just didn't feel right.

You can use the panadapter and peak blobs to give you the readings you want, but you'll have to convert dbm to S units on the fly.

You cannot use readings from the panadapter to determine S-units except for CW:

viewtopic.php?f=6&t=2463

My old Flex 3000, and now the ANAN, finally made me "comfortable", because RF system design is my day job. I despise the so-called metering on KenYaeCom equipment.

w-u-2-o wrote:You cannot use readings from the panadapter to determine S-units except for CW:

viewtopic.php?f=6&t=2463

My old Flex 3000, and now the ANAN, finally made me "comfortable", because RF system design is my day job. I despise the so-called metering on KenYaeCom equipment.

Sure I can. The beautiful thing about doing something wrong is making up your own rules.

The noise floor on the panadapter seems to fall more inline with the S meter on the KenYeaCom rigs.

AG5CK wrote:The noise floor on the panadapter seems to fall more inline with the S meter on the KenYeaCom rigs.

Under average conditions (e.g. a passband of 2900Hz, and a panadapter bin width of 2.93Hz), the displayed average noise level (DANL) on the panadapter will be 30dB less than the S-meter (in Sig Avg mode). That's 5 S-units difference. Given that many of the Japanese radios show an S1 on the meter when the noise floor is really S5, that's not far off! And a sad indictment of how the marketing department probably put pressure on the engineering department