Measuring receiver sensitivity.

[G0HZU_JMR]

To show you the potential of the soundcard method I posted up a quick youtube clip showing how well the sig gen and the soundcard agree in terms of 1dB steps over about a 60dB range. I stepped the HP sig gen down in 1dB steps over about a 60dB range in the video.

There are various contributors to overall uncertainty here including the mixer compression/linearity at the top end and also mismatch uncertainty vs attenuator setting when looking into the mixer. So it is never going to be perfect but this sig gen has an electronic attenuator and it always seems to be remarkably accurate.

You can see that the agreement is within about 0.1dB over the whole range. I get similar results if I swap the soundcard for a modern spectrum analyser with a digital IF. You can also see that there is about a 90dB range available on the soundcard. But the mixer will compress slightly at the top end (you can see about 0.06dB compression in the first few steps) and the marker readout gets noisy as you drop to the bottom of the range. So I only showed the cleanest part of the range available. Note also that I'm just using a cheapo eeePC netbook with a very basic on board soundcard here.

But you can still check it over a 90dB range like this of you use some averaging or a different RBW setting and maybe use a stronger mixer. I just used an SRA3 mixer from Minicircuits here. I've got some stronger mixers than this but this one shows what a typical diode mixer can cope with.

Once you get the main fault cleared on the Racal attenuator you could maybe test it in a similar way. I think your Racal generator is capable of very good performance in terms of attenuator/level accuracy so it would be worth doing something like this final verification test once you think it is repaired

https://www.youtube.com/watch?v=lpl1...ature=youtu.be

[Station X]

Interesting. Thanks for putting that up.

I assume you used a second SG as a "local oscillator"? I'm thinking I could use the SG's frequency standard output as a LO and set the SG to 5kHz higher?

[Dickie]

I always have good ideas after the event, but did you try and find the faulty pad/relay before cleaning them?

[Station X]

That's not possible. The RF Amplifier and switched attenuator are contained within the same screened unit and the junction between the amp and the attenuator isn't accessible without some dismantling.

The unit isn't easy to get at, being buried deep inside the SG. so while it was out I decided to clean the contacts anyway. After that I made a careful inspection under a magnifier to make sure no debris from the alcohol soaked cloth had been left behind.

Dismantling the unit isn't a difficult process. Just lengthy on account of the sheer number of screws holding together the screened assembly.

[Radio Wrangler]

If you want to try the sound card method, I think I have some minicircuits Ray-3 mixers kicking about. These are their high level +23dBm parts.

At a pinch you can use the sig gen at high level and use an external atteunuator.

David

[Station X]

There was a fault with the attenuator. One of the remnant relays was only operating in one direction, so one of the 40dB pads never got switched in. This was due to a wire having broken off a relay coil tag. Depending on the orientation of the attenuator assembly, the wire would or would not make contact. That's why I didn't find it when I checked the continuity of the relay coils back to the plug. This may have been a self inflicted injury following a previous repair to the RF amp.

Having resoldered the wire I confirmed that the pads were being switched in in their correct sequence as the SG's output was decreased. When connected to a receiver the S meter now rose steadily and didn't suddenly take a downward leap.

However I still can't get sensible SNR readings. The only way I can get sensible readings with the SG set to the suggested 0.5uV pd is to introduce 40 dB of attenuation.

I've also found that with the SG's output set to -93dbm, the receiver's S Meter reads S9. It should require -73dBm to achieve this, so the SG's output may be 20dB high. I say may be, as I don't trust S Meter readings.

I'll have to accurately check the SG's output levels so that I know whether I can rely on it.

[Dickie]

If you're having problems at low levels it's also possible that the RF screening and gasketing in the SG is getting a bit "tired" and allowing a higher than normal leakage level out of the SG. Particularly if the relevant assemblies have been dismantled several times. It may not be very likely but it is something to be mindful of.

[Station X]

It occurred to me that I had another SG, namely an HP601A which can be used in CW mode and has 1Khz switchable modulation.

I compared its output levels with those of the Racal using the receiver's S Meter as an indicator. The values tied up exactly over the whole of the S Meter's range with S9 corresponding to -93dBm for both generators. I think it highly unlikely that both SG's have their output 20dBs high. However I will check them both using the mixer method as soon as I have the necessary hardware and software.

I did an SNR measurement using the HP and needless to say got the same spurious readings.

[Top Cap]

Just came across this whilst looking for some info on repairing my TS940.
"Clif Holland of Avvid, a respected repairer of Kenwood radios, emailed me to note that the Japanese specification for the standard signal generator used in alignment is different from the US signal generator calibration. The 930 service manual refers to signal levels in dBuV, so I had assumed 0dBuV was 1 uV and 40dBuV was 100uV.
But not so. Cliff is right and I'm off by 6 dB. I checked it out, and although I see no mention of the issue in the TS-930 or TS-950 manuals, I found a table in the TS-850 service manual, pg. 96, that confirms this. It has two columns:
Japanese "SG" American "SG"
-6dB 0.25uV
0dB 0.5uV
6dB 1uV... etc.
40dB 50uV... etc.
Apparently the Japanese generator defines output in terms of open circuit voltage rather than voltage into a matched load. "
So confirmed, 0dB on a Japanese SG is 0.5uV and not 1uV

[Station X]

Thanks Top Cap.

That table appears in several Trio manuals, I found it in the R1000 manual.

The Japanese reference level (0dBuV) is defined as 1uV emf applied to 50R which will result in a 0.5uV pd across 50R?

So if I want to set the output of my SG to 0dBuV (Japanese reference level) I set it to read 0.5uV pd which is 1uV emf?

That level is about -113dBm, so to convert Japanese dBuV to dBm I subtract 113?

As far as the actual testing goes I'm waiting for a Micro Circuits mixer to arrive so I can check the calibration of my SG's

[Top Cap]

Yep, you have to visualise the source impedance of the generator as 50 Ohms. Therefor if you set it to say 1V with no load then applying a 50 Ohm will cause the output to be divided by 2 and an output of 0.5V will be present. We used to check the source impedance of sig gens in this way, making sure they were 50 Ohms. I don't have a RF power meter here at home but I do have a HP3400 rms meter which I use at 10MHz for doing a basic 'works or not works' attenuator check. It's hopeless of course for looking for anomalies at higher frequencies, but good for a basic test.

[Station X]

I now have a Leader LDM-170 Distortion Meter which is also capable of measuring S/N ratio.

I set about measuring the sensitivity of a Trio-Kenwood R-1000 Comms Receiver. The sensitivity figure quoted in the service manual is better than 10dB with a 0.5uV pd input signal on SSB.

Following the level diagram in the manual, I applied a 14.25MHz, 0.5uV pd unmodulated signal to the 50R antenna socket, switched the receiver to LSB, tuned for max audio output as shown on a power meter and then used the volume control to set the audio output to 50mW as indicated by the power meter. With the Leader set to read RMS volts the reading was 0.63V across 8R, which gives 50mW. So I reckon the calibration of the Leader is good.

I then did a S/N measurement and got a figure of 20dB. To get the figure down to 10dB I had to reduce the RF input to 0.15uV. I then did the same test using a different signal generator and got the same result.

Do these results sound reasonable or not?

[Radio Wrangler]

They're about what I'd expect. The 'Big-Three' tended to make their receivers somewhat more sensitive than is really needed on the MF/HF bands.

When using an S/N meter, remember that it cannot tell the difference between noise and distortion, so play with the AF gain control to find what output level gives the best indicated S/N.

David

[Station X]

Thanks David, that's reassuring.

I made the test with 50mW of audio output, as that's the figure given in the manual. I will see what happens when I change the output level.

One of the advantages of using a "manual" SNR Meter, as opposed to a PC sound card one is that it's easier for me to understand how the measurement is made.

I'll try testing some other receivers before getting back to the original task, namely measuring the sensitivity of a Trio 9R59-DS receiver.

[Station X]

I used the Leader to check the SNR of my JRC-NRD525 Comms Receiver. The advantage here is that all control settings, RF levels, mod depth etc. are given in the receiver's manual.

Quoted figure for HF SSB is 0.5uV for 10dB SNR. I measured 20dB.
Typical figure for HF SSB is 0.25uV for 10dB SNR. I measured 15dB.
Advertising blurb suggests 0.2uV can be achieved. At that level I measured 10dB.

Quoted figure for HF AM is 2uV for 10dB SNR. I measured 26dB.
Typical figure for HF AM is 0.5uV for 10dB SNR. I measured 19dB.
With 0.3uV input I measured 10dB.

I tried repeating the test using the PC based SINAD/SNR Meter, but the results were hopeless.

Now I have faith in the test equipment I'll try testing the Trio 9R59DS which is where I came in 74 posts ago!