Measuring receiver sensitivity.

[Station X]

This query has arisen because I want to check the sensitivity of the Trio 9R-59DS receiver which is the subject of this thread:-

https://www.vintage-radio.net/forum/...d.php?t=127728

I'm used to sensitivity being specified in terms of X microvolts of modulated signal at the aerial socket resulting in Y milliwatts of audio at the speaker. I've never had any difficulty checking this using a signal generator and output meter.

However the manual for the 9R-59DS quotes the sensitivity for three of the wavebands as being:-

Less than 6 dB (for 10 dB S/N ratio).

Now I THINK I know what this means. With a Signal to Noise Ratio at the speaker of 10 dB, the input at the aerial socket should be 6 dB or less. However I don't know what the signal levels are referenced to and neither do I have a means of measuring S/N Ratio. Can someone enlighten me.

Is it possible to convert the figures given to the more common uV in and mW out format?

Thanks.

[Radio Wrangler]

It's not easy to convert directly because you need one extra piece of info... um, make that two.

What do they mean by 6dB? I'd guess 6dB above one microvolt.

For 10dB S/N ratio you need to measure the audio output power with the modulated signal on (And with the specified % depth of modulation oops, that's three missing items) and then again with the signal on at the same level, but unmodulated.

The idea is that that input with that modulation should give an audio level 10dB above the noise level. Because you're comparing a tone level with a random noise level, this is one place you do need a real true RMS meter.

The easy way is if you have a SINAD meter. With one of these, you just feed in modulated audio and the meter measures the level, then it notches out the tone and measures again. It displays the difference in dB. It really reads (Signal-plus-noise-and-distortion) to (Noise plus distortion)

There was a SINADDER sold on the forum a few months ago. They're just a reorganised audio distortion meter. I use an HP 8903 which automates things.

David

[Station X]

Unfortunately the alignment instructions are the most uninformative I've ever seen. Nothing about how to set the receiver's controls (other than waveband and frequency) and no modulation frequency or modulation depth specified.

It is suggested that the LO trimmer is adjusted so that the signal is received at the indicated dial setting and then the RF circuits are adjusted for maximum S Meter deflection. The latter test would not require modulation.

[ms660]

I wonder if it's meant to be microvolts for the sensitivity not db. Eg: 6uv for 10db S/N ratio?

Lawrence.

[Station X]

That's not what it says in either the service or operating manual. It's definitely 6 dB.

It might be an idea to just try it and see what input is needed to achieve the required S/N ratio. Unfortunately I have no way of measuring S/N ratio and neither do I know whether the receiver's sensitivity is up to spec.

[Radio Wrangler]

Sometimes you come across specs which were plainly written by someone who did not understand what they meant and was merely parroting bits of phrases seen elsewhere.

Sometimes the value got stripped out in translation.

It's not possible to be sure where the damage happened.

6dB signal/noise ratio gives intelligible speech, but the noise makes listening hard work. By 10dB the amount of work has diminished a lot, the noise is still very intrusive, but you're more certain of what you hear. On the HF bands, achieving this at a couple of microvolts (6dB over a microvolt) would be good with AM. Most sig gens seem to be set for 60-80% mod if they have no metering/calibration/depth control.

Some professional radios are specified for 6dB S/N with x microvolts of RF at 30% mod.

THen there's the matter of do they mean EMF or PD microvolts... there are sig gens calibrated in each system.

Let's ignore the book of words, it's broken. If you get 10dB S/N at 2uV (PD) with 1kHz mod at 80%, you have a useful and typical HF receiver.

To measure S/N at 10dB you can usually ignore the difference between SINAD, (S+N)/N and S/N.

All you need is an RMS AC voltmeter. A lot of better grade DMMs do this.

David

[ms660]

Originally Posted by Station X

That's not what it says in either the service or operating manual. It's definitely 6 dB.

It might be an idea to just try it and see what input is needed to achieve the required S/N ratio. Unfortunately I have no way of measuring S?N ratio and neither do I know whether the receiver's sensitivity is up to spec.

Yes, I understand that's not what it says, I was wondering if they've made a mistake.

The 6 db they quote is meaningless to me.

Lawrence.

[Synchrodyne]

The attached 1969 advertisement for the Trio 9R59DE quotes a sensitivity, at 10 MHz, of 2 microvolts for 10 dB S/N ratio.

2 microvolts is 6 dB above the commonly used 1 microvolt reference level. So that might explain the 6 dB number in the specification.

We might also glean something from the specifications for the early (1970s and 1980s) Trio and other Japanese solid state HF receivers, which seemed to follow a reasonably common approach to sensitivity specifications and which for which a signal level diagram was usually provided.

The 0 dB reference point was usually 1 microvolt of signal generator voltage, as measured with the SG loaded by the aerial input. Output S/N ratio was measured with the audio gain control set to provide 50 mW output, which was evidently typically achievable even at the 0 dB input level.

Thus HF receivers often had sufficient overall gain that measuring the signal input required for 50 mW output was of no value in determining usability, as it occurred well below usable S/N ratio. Rather it was S/N that was the determining factor, hence its widespread use in HF receiver specifications. On the other hand domestic receivers, particularly those without RF amplifiers, did not have abundant gain, and so usable sensitivity was better described as being that required for an output that provided adequate entertainment volume, namely 50 mW, by which time the S/N ratio would have been reasonably acceptable.

Buried somewhere in the alignment instructions for the above-mentioned solid-state HF receiver group was usually a note to the effect that where a modulated output was required from the SG level was required, the modulation level was 30%, which number had been something of a de facto norm anyway.

Of course, there is no assurance that Japanese OEM norms and practices of the 1970s do read back to the 9R-59 era, but in the absence of hard information, they provide a reasonable starting point.

So perhaps the best working assumption is that the stated sensitivity is 2 microvolts (6 dB above the 1 microvolt 0 dB level), measured as loaded SG output, for 10 dB S/N ratio, at 30% modulation, with the audio gain control set for 50 mW output (0.63 volts into 8 ohms). There does not appear to be any way though of avoiding the need to make S/N measurements in order to check receiver sensitivity. My guess is that the 9R-59 would reach 50 mW output before it reached 10 dB S/N ratio. Possibly its sensitivity curve would be similar to that shown in the second attachment for the single-RF-stage case.

Cheers,

[ms660]

Yes, 2uv would make sense, out of interest I looked at the manual for a Hammarlund HQ150 I used to own (also a single stage RF) for the sensitivity specs, couldn't find any

Used to check my comms receivers with Shannon and RAF Volmet as a rough guide to performance and on a good day (morning) transmissions from VK.

I've never tried the signal to noise measurement route properly, one tends to get a feel if the receiver is up to it's best.

Lawrence.

[Synchrodyne]

The numbers for the GEC BRT400D, attached, are I think illustrative.

Less than 7.0 µV for 20 dB S/N above 1.3 MHz and less than 10.0 µV below 1.3 MHz.

Very very roughly, one could lower those numbers by 10 dB to approximate the 10 dB S/N sensitivity, so around 2.0 µV above 1.3 MHz.

But also quoted is a sensitivity of less than 1.5 µV for 1.5 W output. Or in other words, within 2 dB of full output is available at signal levels that are likely to be below usable level in S/N terms. Thus the domestic receiver parameter, sensitivity for 50 mW output, would be of minimal utility for this kind of receiver, and for other HF receivers of comparable overall gain.

The BRT400D had two RF stages, but as the previously posted curve suggests, the second RF stage confers but minimal sensitivity advantage; rather it is there to improve image rejection and better protect the mixer from out-of-band signals.

Cheers,

[Radio Wrangler]

If you're designing a preselector filter and want to make it very narrow, then you are into very very small coupling factors between stages, and the filter's overall Q is getting painfully close to the best resonator Q you can achieve. This means that the corresponding insertion loss is very high. Having an amplifier stage to buffer resonator stages is very very helpful, especially something with a naturally high input and output impedance like a valve. The gain goes to overcome the loss as well.

Most of the old single downconversion (to 455kHz region) receivers with two RF stages did it to isolate RF tuned stages and to make up for the losses of narrow RF selectivity, all in the service of trying to get some image rejection.

You didn't want too much total gain ahead of the mixer, you'd just overload it. If you needed more overall gain in the receiver, it was more economical and a better fit with overall performance to stick in another IF stage.

The same thing about narrow filters and increased loss also applies, pro rata to IF selectivity. It was hard work before quartz crystals became affordable.

The 9R59DS was a good example of its era, with the budget of an AR88 or BRT400 thrown at it, it could have been better, but the shortfall was smaller than you might expect considering the difference in new prices. Trio done a good job!

David

[Station X]

Thanks for all the info everyone. I get it now.

I had hoped to carry out a simple sensitivity test to check whether realignment was needed, but it's not as simple as I thought.

I have a Marconi TF893, alas without the optional SINAD filter. Could I use this to take two measurements, one with modulation on and one with modulation off and then calculate the result?

An alternative might be to use a PC sound card with SINAD software.

[Radio Wrangler]

It should do and be near enough. For formal specs, it wouldn't be acceptable without true RMS measurement, because diode detectors mis-read noise. SINAD software on a sound card could be quite good, and would make life easier.

David

If a radio gets quite a bit noisier when you connect the aerial it is good enough. Not much help in you quest for the right figures but sometimes we have to be pragmatic.

[Synchrodyne]

Atmospheric noise is likely to be greater than receiver noise at frequencies up to the 15 to 20 MHz range, as shown in the attached graph. So receiver sensitivity is not a big issue in terms of actual usability for the most part. Still, sensitivity performance relative to specification could be used as a proxy for checking that the receiver is performing as intended; if the sensitivity is sub-par, performance elsewhere might also be suffering.

Above about 20 MHz receiver noise could become a limiting factor, which is probably why, in the later valve era, some designers used 1st RF amplifiers that were a bit quieter than regular remote cutoff pentodes, such as high-slope pentodes and cascodes.

Cheers,

[G8HQP Dave]

Using quieter valves makes little difference when most of the front-end noise comes from the first tuned circuit, made worse by the loose coupling needed for good tracking of a general coverage receiver coupled to a random antenna.

Tight coupling is one solution, but then you need either a carefully tuned or buffered antenna.

Broadband tuning is another solution, but then you lose front-end selectivity.

A high gain RF stage looks good in advertising, but makes little difference to noise performance and can significantly worsen dynamic range by overloading the mixer.

[G0HZU_JMR]

What you ideally need is an accurate signal generator (as in low level signal accuracy and lack of leakage) if you want to make a fairly reliable measurement of sensitivity. This normally means something fairly decent from the likes of Marconi or HP. Back in the 1980s I repaired a LOT of CB and ham radios and I measured receiver sensitivity on several radios a week across various modes like SSB, AM, FM and CW.

I didn't have a true rms meter back in those days and I mainly used a Tek 585scope to look at the AF output and I could judge 10dB S/N ratio by eye fairly well because I was doing this stuff on a daily basis. It didn't need to be any more accurate than this because none of my customers ever asked for sensitivity data. It was up to me to decide if the radio was healthy enough in terms of sensitivity

I found that an important thing to look (as a health check) for was how the AGC behaved as the signal level was reduced. Looking at the scope on AM or SSB and watching the noise and the signal change as the level fell was a good marker as to how healthy the receiver strip was. i.e. looking at the point where the AGC dropped out was useful. Also, on AM the detector performance can be monitored best if you use a scope because you can see distortion or unwanted spurious terms from the radio.

In terms of accuracy, I don't think the lack of rms meter was the weak link in my setup. It was the small signal accuracy of my old Marconi 801B sig gen and my little Marconi TF2015 sig gen. The TF2015 was obviously the better performer here and I used it with a fixed 10 or 20dB attenuator on the output to get the signal level right down when needed.

I used an Altai 360 analogue AC voltmeter to 'calibrate' my S/N measurements. At the time it was my only multimeter although I did buy a DVM back in the 1980s. I recall that I used the Altai meter to find the SSB receiver noise floor by connecting the sig gen to the radio and then I turned up the sig gen level until the meter showed an increase of 3dB in AF level from the receiver. The 10dB S/N ratio would be at a sig gen level 10dB higher than this.

Although the Altai meter can't measure Vrms I was just measuring a ratio of a 'noise' signal to a noise plus tiny sine wave. So the error wrt the 3dB change is quite small and much less significant than the absolute accuracy of the sig gen.

But I only did this a few times because once I saw on the scope what a 10dB S/N ratio looked like I could just use the scope and not bother with the Altai meter. Note that you have to boost the AF level to several volts in order to use a meter like the Altai. I used an old AF transformer back then but I could have used an opamp or similar. In later years I used a Racal 9300 true rms meter for stuff like this.

Today, you can either use a PC soundcard or a modern DMM. I only listed my 'vintage' method above as it was the best I could do back in those days on a limited budget

[Station X]

Thanks everyone for the additional information since my last post.

I had a bit of a play this afternoon.

Receiver's AF and RF gain controls set to max. Receiver's antenna trimming control set for maximum audio output at the frequency being tested.

Signal generator set for 2uV output, 1 kHz modulation frequency and 75% modulation depth.

On the lowest frequency band (Band A) the alignment points are at 600kHz and 1.4Mhz.

At 600kHz output level with mod on was 22dBm and with mod off 12dBm, a difference of 10dB.

At 1.4MHz output level with mod on was 17dBm and with mod off 7dBm, a difference of 10dB.

On the next waveband (Band B) the alignment points are at 1.7Mhz and 4Mhz.

At 1.7Mhz output level with mod on was 8dBm and with mod off 5dBm a difference of 3dB

At 4Mhz the difference between the signals was so small I couldn't make a useful measurement.

One problem is that with the output meter connected it's not possible to hear the audio output and use it a as a tuning aid. I'll have to arrange a changeover switch to switch between speaker and output meter.

Another annoying thing is that the tuning of the receiver tended to creep as the drive cord contracted after tuning in a signal.

So unless my methods are totally flawed, which is quite likely, I'll have to carry out a full realignment. I'll look out for the wax which has been mentioned in another thread.

My next step is going to be to try a PC sound card based SINAD Meter and see how the results compare.

[G0HZU_JMR]

I have a Marconi TF893, alas without the optional SINAD filter. Could I use this to take two measurements, one with modulation on and one with modulation off and then calculate the result?

I had a look at the schematic of the TF893A and it basically looks to be the equivalent of what I was trying to do back in my student days with the little Altai meter. i.e. an AF transformer feeding to a bridge of diodes that drive a moving coil meter. I've not investigated the detector circuit of my Altai meter but I suspect it will be the same kind of thing.

So you can expect the meter indication to be the result of rectifying and averaging the waveform. On top of this I presume it will include a fudge factor of x1.11 to convert average to rms (for a sine wave) and the dial calibration will include this fudge factor.

If you feed it noise then I think you have to add another fudge factor of about 1.13 (from memory) as the meter will under read noise. Also, my Altai really needs to be driven to a couple of volts in order to get reasonable accuracy on the AC ranges or it will read low on small AC waveforms due to finite performance of the diodes in the detector. This small signal deafness applies to all waveforms including sine waves.

I used a step up AF transformer on mine to get the meter to 'read' the low level noise from the CB/Ham radio with no mod/signal. All I had to do then was adjust the sig gen such that the on/off ratio was 3dB on the Altai meter to get the noise floor. I think the error here wrt a true rms meter is <0.5dB. Because I was just measuring a 3dB (S+N)/N ratio, I didn't have to worry about the absolute accuracy of either the noise measurement or the S+N measurement. So I didn't bother with any fudge factors.

The level for 10dB S/N ratio will be the sig gen level plus 10dB and this can be done as a simple +10dB calculation in your head without any need to actually change the sig gen.

However, if you try and measure S/N or (S+N)/N up in the 10dB region with a meter like the Altai then the fudge factor of 1.13 listed above will begin to dominate because the meter will under read the noise contribution wrt the larger signal and you could see an error of well over 1dB in the measurement. Eg you might see well over 11dB indicated on the test gear instead of 10dB. This may well apply in the case of the TF893 as well but there may be different detector types in the newer versions with the SINAD option?

20*log(1.13) is just over 1dB and this is part of the reason for the >1dB error.

[Oldcodger]

RE- #18, so as not to quote whole post- just the bit about wax- Commercial stuff ( military Mobiles) used a setting compound , like wax on the IF coils ,which we used to soften with a hairdryer to make alignment a lot easier- a tip not mentioned in a lot of alignment guides, but only found from life on the bench. Hopefully of some use.