How to Measure Receiver Sensitivity (MDS) & Noise Figure (NF)

[@hervefaro4450]

Thanks for that very interesting tutorial.

[@w8bya]

Thank you for the feedback Herve.

[@N4HAY]

Many thanks for your video. It is very instructive to me to actually see the test being performed. This augments the texts on the topic.

[@w8bya]

You are very welcome and glad to have helped out.

[@davideisenberger6197]

It's been over 30 years now but if memory serves me, to measure our FM 2 way units, we used a sig gen that was Calibrated in uV. The best I can recall, spec levels were about .5 uV for 20 db of quieting. Certainly some units were better than others in sensitivity i.e., .35 uV. Unfortunately since I have been away from it so long, I am positive that RX sensitivities have increased along with high levels of Intermodulation degradation. When I retired from the company 100 db intermod rejection was common. Enjoyed your video very much! 73's David. K8KEM

[@w2aew]

Couldn't you start from the other direction? In other words, start with the signal generator OFF, then start at a VERY low power level, and increase it until you get a 3dB increase on the analog meter? Seems like this would be quicker (only adjust the volume control once). This is similar to how you demonstrated the digital meter.

[@w8bya]

Hi there and thanks for the question. Yes, there is no reason you could not measure the signal level that way as the ultimate goal is simply to see that 3 dB delta. I guess after doing MDS tests for over 30 years for tempest work one way it kind of grows on you hi-hi.

[@israelhershkovits5675]

He is talking as a real amateure child!!!!

[@billturini8533]

Thanks a lot for this video. I appreciate your pulling it all together in one place.

[@w8bya]

Hi Bill. Thank you for taking a moment to provide feedback and very glad to have helped out. 73

[@laurentdrozin812]

Thank you for the presentation. Very useful. I would like to understand better the requirement on the bandwidth of the RMS voltmeter. In your case, you have a carrier at 144kHz, and a signal bandwidth at 500 Hz, so a voltmeter bandwidth at 10MHz is ample enough. But what happens if you work at another signal carrier, for example 150Mhz, and signal bandwidth at 500Hz still? Is the instrument bandwidth related to the signal bandwidth or to the signal carrier frequency?

[@w8bya]

Glad you found the video useful. The BW of the RMS voltmeter does not depend on the carrier frequency of this receiver since it uses an IF and the meter is sampling the post-detected output. The same meter has been used to measure the MDS at 10 GHz using a 500 Hz BW. If we tuned our receiver to 144 MHz and wanted to know the MDS of the receiver when using a much wider BW, say 15 MHz wide IF, then you must make sure the meter has at least 15 MHz BW also. In general you want to make sure that your meter has at least the same BW of the passband you are listening to.

[@laurentdrozin812]

@@w8bya thank you! I managed to figure it out when I found the standard you refered to in the video.

[@i82996]

Thanks much for sharing. I am restoring tube sets and I can directly use this method, which helped me very much. - Albert

[@w8bya]

Hi Albert that is really great to hear. GL

[@oh6cn]

Great Video Thanks

[@iz7fmm]

My R.M.S. voltmeter has scale for dbm & dbv, which one is the correct to use? Thanks

[@w8bya]

The key is that your meter can display differences in dB. That is what is key. You can actually do this measurement with a meter that does not have a dB scale but it would require you to make calculations to determine dB differences. I would start by measuring in dBV and see what you get. I am thinking you may get the same value even if you measure in dBm, let me know. 73

[@allthegearnoidea6752]

Great video you have a lovely collection of test equipment. I have done a few videos on RF test equipment myself but I really enjoyed your video. Thanks for sharing. Most of my gear is Marconi but I do have a hp 8903. Do you really need to use the step attenuator I am thinking that the output would be well calibrated on your test gear. I’m wanting to do a test on an old AM domestic radio I thinking I should do the test at 1khz 30% modulation could you give me some advice. Thanks Regards Chris

[@w8bya]

Thank you very much. It is nice to be able to make some measurements to finish a home project or to help someone else out who does not yet have the gear.

[@allthegearnoidea6752]

w8bya Sorry we double posted. Do you know how to do the test for an AM radio that does not have CW

[@w8bya]

@@allthegearnoidea6752 Sure. But you need to be careful as there are many different definitions for AM sensitivity. In my past life, in the world of TEMPEST engineering, we used specific standards that could have different criteria than what is commonly used for say, ham radio gear etc. To avoid confusion let's just assume that we wanted to measure the AM sensitivity the way the ARRL would measure it. That standard is commonly accepted. By their definition the goal of the AM sensitivity test is to determine the RF level of an AM signal, 30% modulated at 1 kHz, that will produce a tone 10 dB above the noise level (MDS) of the receiver. Fortunately the test setup is nearly identical to the one used in this video except we enable AM modulation as stipulated above. In a nutshell you set the signal generator to the frequency of the receiver and adjust the RF signal level such that when only the modulation is turned on and off you notice a 10 dB difference on the AC RMS voltmeter. Hope that helps. If not I could probably make a quick video.

[@PapasDino]

Nice lay down! I did have to pop a couple of dramamine though as I was watching the focus move around there for a while hi hi. Have noted your procedure down in my lab notebook so I don't have to keep reinventing the wheel anymore every time I do this measurement. I have a old HP400EL on my bench just for these kinds of voltage measurements, can't beat the old stuff! One way to adjust for cable/connector losses is to mock it up and simply measure the output at the final connector that goes to the radio on your analyzer and adjust the RF generator's output upwards to whatever you need to overcome those losses; that gives you a more accurate starting point vs. trying to estimate the cumulative loss. 73 - Dino KL0S

[@w8bya]

That made me chuckle....sorry about that Dino, did not mean to make you dizzy. Yes measuring the I.L of the RF cables involved will tell us exactly what the loss is but IMO it is not required here. I am sure we were well within 1 dB. But if I were forced to use longer or unknown cables I would certainly characterize them 1st. Thanks for the view & comments. 73

[@PapasDino]

I had to break down and order some RG-142 jumpers to have on hand for these kinds of measurements, thanks for reminding us how important it is for maximizing the shielding to keep junk out of the circuit.

[@w8bya]

My pleasure Dino. Another cable that is even better for everyday bench use is RG-223. It is the low power but slightly higher loss version of true mil-spec RG-214. RG-223 is much more flexible and has approx the same shielding effectiveness of RG-142. RG-142 has a breakdown voltage in the thousands of volts and great for high-power applications while RG-223 is wonderful for low-level signals that need to be isolated from inadvertent coupling effects like cross-talk & RFI etc. Just be careful when buying from overseas....so many cheap knock-offs that are not worth a crap. True mil-spec RG-223, RG-142, and RG-214 ALL use very high-density, silver coated, double shielded braids along with a silver coated center conductor.

[@PapasDino]

Yeah, I know I'm taking a chance getting them from over there and will probably sacrifice one connector to see what the cable looks like inside. I guess there's always Pasternak (or whatever that company is called nowadays). You can see the workbench/radio desk pix on my QRZ page.

[@PapasDino]

Well, my jumpers arrived today from overseas...got some with BNCs on both ends and a couple with N/BNC (RG-142 cable). Thought it might be interesting to actually try to measure, at least qualitatively what kind of leakage might be exiting the cable along its length. So set the generator to 20dBM, 100MHz for grins and used my large H-field probe with the analyzer to see. First I just checked the frequency response between 1MHz to 3GHz with the tracking generator and found about 1+ dB losses along the way, highest toward 3GHz as you might expect; slightly less loss with the N connector. Then I checked along the length of the cable between the generator and a dummy load with the H-probe. Nothing really along the cable but when you got to the BNCs there was noticeable leakage; none really around the N. So lesson learned is if you really want to minimize ingress/egress to the system use N connectors! These should probably be sufficient for my level of hobby work. 73 - Dino KL0S

[@Igortruntrek]

Hi Thanx for a nice video. But this method (NF) is relevant only if you have got FLAT bandwith filter. Your rig has DSP so its OK. But question is how to measure non DSP rig with analog filter. Analog filter is not FLAT and has for example on 2200Hz BW -6dB signal decrease. I cannot calculate with 2200Hz BW because only on center it has 0dB but on the sides there is attenuation. What do you think about this ? How to compensate ? I try to measure FT736R on 6m and MDS is -140,2dBm. Rig has only original filter so there is 2200Hz bandwith/6dB. But if I calculating NF with 2200Hz it is not so real (NF = 0,24dB). Do you have any idea how to calculate or compensate nonlinearity of filter on calculating ? Thanx 73 OM4AIK

[@w8bya]

Hello Igor. Sorry but not true. Measuring the MDS/NF of analog receivers does not create any special problems as long as you realize that the filters may not be perfect. Most every standard for MDS understands that "real" filters will be used and that a fall off in MDS will occur as you move off the absolute peak. If you must know what the MDS is away from the PEAK signal, within a bandpass, then you will need to carefully characterize the ripple and shape factor (skirts) of your filters and apply a correction factor to your MDS level taken at the peak frequency. Or you can repeat the MDS measurement at different frequencies away from the peak freq while staying within the bandpass. As far as getting such a low value 0.24 dB NF (very unlikely) remember that even TINY errors in your MDS measurements can mean big differences in the resulting NF. For example, if your MDS value was only 3/4 dB off (easy to be off this much) you would have a NF of 1 dB instead of 0.24 dB which is much more realistic. Making MDS measurements with absolute certainty AND repeatability, within 1 dB is usually difficult esp when you are looking at very sensitive receivers, front-ends, or LNA's etc. As you know the noise makes the measurements bounce around making it hard to determine an exact value. 73

[@Igortruntrek]

w8bya Thanx for reply my friend. I try to measure at every 100Hz from one side to other side of bandwith (2200Hz mean 22 measuring) and the cenrer (peak) I set VU meter to 0dB and on left side and right side it was about -6dB signal on VU meter down. I take 22 measuring with 100hz step and make arithmetic average (sum of every value divided by 22). The final number was 2.1dB. Can I use this value to a correction factor for a NF ? (0.24+2 1=2.34dB) Or its wrong method ? Do you think MDS must be corrected too ? (Isn't MDS a minimum detectable on peak value?) Thank you for your time. 73

[@w8bya]

Hello Igor. Years ago I used to own two FT-736R radios but I no longer have my MDS data for them (I always measure my S-meter to know what each s-unit is (how many dB) and I take MDS measurements on all bands to ensure the receiver is ok. I honestly do not believe your reading of 0.24 dB NF is accurate. Are you positive your 50-ohm signal generator is calibrated and accurate to within 1 dB esp with extremely low RF signal levels like -140 dBm? It is possible you are getting RF blow by like I demonstrated in the video. This is very, very important. Also you must make sure any attenuators etc you use are exactly 50-ohm. You should be getting a NF of between 2-7 dB on 6m (if I had to guess). But getting back to your last question, when I use the term "standard" I mean the organization or agency or company who wrote the specification & requirements for what MDS means to them and how it is to be measured. In my example I used the "standard" that is defined & used by the ARRL since it is very common and available to everyone. In my past life (TEMPEST engineer) we used different "standards" for how to define & measure the MDS. There can be MANY different ways to define & then to measure the MDS. You can certainly use the method that you described but you may be the only person in the world who does so hi-hi. If it is acceptable & useful to you then that is what is important but if you want to be like everyone else (right or wrong, useful or not) then you must follow their "standard" methodology. I hope this makes sense. 73

[@Igortruntrek]

Thank you for reply So at first its generally question which is very important. How do you measuring MDS/NF if there is not 500Hz filter installed and there is no DSP feature to set BW to 500Hz? Is it possible to measure in standart filter bandwith (2200Hz/-6dB) ? Which bandwith can I calculating on formula where -174dBm is compensated to 500Hz BW ? Can I calculating with 2200Hz BW ? Because this values are very different (on 500Hz it is -174dBm+27dBm but on 2200Hz BW it is -174dBm+33,4dBm). If I calculating with 500Hz BW NF is 6,67dB (looks real) but I am not sure if I can calculating with 500Hz because my TRX hasnt 500Hz filter. How you measuring old radios with no DSP BW features and no CW filter installed please ? Thank you, your answers are very helpful for me. 73 My measuring equipment is RF generator Marconi Instrument 2018A with 30dB external attenuator and VU meter I use SW for Windows: Orban Loudness Meter and SB Audigy 2 soundcard is HW.

[@w8bya]

First you must be 100% certain that you are able to accurately measure the true RMS value of an arbitrary (noisy) signal from your receiver. Some setups can measure the RMS level of easy & clean signals but may not be able to give you accurate "true RMS" readings when noise is involved. I am not sure if your setup can do this or not so be aware.....Also, using 500 Hz as the BW for MDS is just a common number used by some standards but not all. You do not have to use 500 Hz. You can use any BW you want or need to. You can easily convert the MDS from one BW to another BW by taking 10xLog (BW1/BW2). This will be your conversion factor. Once you have your converted MDS go ahead and calculate the effective NF. To convert the MDS you measured at say 2200 Hz subtract 6.4 dB to obtain the equivalent MDS for 500 Hz. At this point pse feel free to contact me directly at my e-mail address which you may find on QRZ.COM if you have any additional questions. TU es 73

[@billturini8533]

When you pick off the audio from the radio, do you have it terminated by an 8 ohm resistor and parallel the input to the HP meter from it, or is the audio fed directly into the HP meter?

[@w8bya]

The audio goes directly into the RMS voltmeter. As long as that is not changed throughout the entire measurement process matching is irrelevant.

[@paolonervi2208]

Nice and useful video.just a hint..next time please use a tripod..It's really difficult to view such a jumping video..

[@markvanwijk8153]

Although your methodology of measuring MDS is correct, it is not fully relevant with today's receiver -technology. Try modulating the generator signal (use s/n test-signal which produces the dit-dit-dit-dit Rf signal) and find out some big differences between DSP receivers and true analogue receivers. With some DSP the will drop in readability on a modulated signal by minus 3- 8 dB (-2 being the K3 and TTORION vs -8 on IC7400/756proII) compared to a pure analog receiver (K2) what I measured. You can hear it too; as soon as the signal is modulated, some DSP's make it disappear in their own AF noise. In the end MDS is not about static benchmark performance but communication performance. Which is recovered readability of the message. see details at: pa5mw.blogspot.com/2009/12/upgrade-current-vhf-station-iv-mds.html

[@w8bya]

I politely disagree with your 1st paragraph. Your 2nd paragraph can at times involve the MDS of a system using demodulation and is a whole different subject specifically not covered in this video (and is therefore irrelevant). Your 3rd paragraph is simply incorrect. Let me explain: Contrary to what you say, I can assure you that the MDS method illustrated here can and is used everyday with both pure analog & DSP based receivers (and not just by hams). Pse pay close attention to what I said in my video as I chose my words carefully. This video illustrates how to perform a simple MDS measurement using an unmodulated signal source similar to the method used by the ARRL. I never said that this is the only or best way to measure MDS esp if modulation or data was involved. There are other formal ways to measure the MDS (that the ARRL does not address) if analog or digital modulation/demodulation is employed. Also this video is not intended to pit one receiver type over another. Also contrary to what you say, MDS is NOT a benchmark for just communication performance. I.e., sometimes it is important to just know if a signal is present. There are many different industries (defense, intelligence, etc) that use different standards for determining MDS for reasons that you may not be aware of. Depending on who's standard you use and if a modulation source is employed or not MDS will necessarily be measured differently. Again, the reason I chose this method of MDS measurement is because it is very simplistic & can be performed by most people with very modest equipment. I also mentioned that there are many different standards that define how the MDS should be measured. For example you may want to study the applicable unclassified portions of NSTISSAM TEMPEST/1-92 or NATO AMSG-720B as references.

[@israelhershkovits5675]

Talking to much bla bla bla real amateure

[@w8bya]

Yup you are right.....I am an amateur & I do not make instructional videos for a living. I suggest for you to watch at 2X speed or find a different video from someone else that does not go into so much detail. Other people actually appreciate & find value in the details I provide. Maybe I can get some pointers by looking at some of your videos.

[@chrisbart5316]

@@w8bya I liked the video, and yes, I listen at 2x. :)

[@w8bya]

@@chrisbart5316 Smart man ! :-)