Spectrum Analyser

[Dr Wobble]

In a moment of madness I bought a spectrum analyser- a Marconi TF2370. It was sold as spares repairs, mainly I believe because the seller didnt have the means to test all functions.

The display lights up, and I think it sweeps and nothing appears faulty. I put a sinewave in from my sig gen and a spike appeared.

I realise this is a far out piece of kit and I havnt the nouse or other kit to test it properly. Could anyone guide me as to it's use and maybe test out a few functions etc. I've looked online at the basic "How to" videos,(one showed how to convert a P-P voltage into Dbm- but was too quick for me) but to be honest I'm a lost. I've also looked at the manuel, but it seems to deal mostly in RF;I bought it to test amps for THD.

Thanks, Andy.

[peterscott]

Hi Andy,

Maybe start by connecting the Std 10MHz Output to the Input. (bottom left).
On the lower unit

Select Range MHz/div
Horizontal Scale 10
Vertical Scale +10dB
Filter Bandwidth Normal (2)

On the upper unit

Select Reference Frequency LH (left Hand)
Centre Frequency Past Centre
Sweep Mode Auto
Vert Scale Range 10dB/div
Store Display High Def.
Counter on

Using the Reference Frequency control (the left hand of the four) set the digital read out to 50(MHz)

You should now see on the screen all the frequency components that make up a 10 MHz square wave.
Ideally these would be 10 MHz and all the odd harmonics (30MHz, 50MHz, 70MHz etc) but if the mark:space ratio is not
exactly 1:1 then you will see the even harmonics too (20MHz, 40MHz, 60MHz etc)
If you change the Vertical Scale (attenuator) you will see the whole display change corresponding to the attenuator setting you selected.

Now disconnect the cable and you should see a vertical line that represents 0Hz frequency.
Now connect the cable between the Tracking Gen Output and the Input and you should see a horizontal bright area covering
the whole screen at the -10dB level. The display is showing you all the frequencies (from 0 to 100MHz in this case) that the tracking
generator is sweeping through.

The Tracking Generator is very useful for looking at the frequency response of a circuit.
If you connect the Tracking Gen Output to an oscilloscope you can watch it sweep over the range covered by the Spectrum Analyser.

To see one in use take a look at https://www.youtube.com/watch?v=9pBrLPaI78k
In this case I'm looking at the RF frequency response of an old radio and observing it as I tune the various stages.
The tracking generator output is fed into the aerial connection and the spectrum analyser input is connected at the detector.

You mention that you are less interested in RF than AF and this analyser does cover AF too but you will find that everything is much
slower and you will need to use the finer Reference Frequency controls as well. Note that if you sweep too fast you will lose the amplitude accuracy of the display. (Try it and see.) Learning will be easier at RF.

Assessing distortion is certainly something you can do with this instrument. If you inject a pure sine wave you should only see one vertical line (plus the one at DC 0Hz). Odd and even harmonics will be easy to see to the right of your signal.

You should get the hang of the other controls just by experimenting.

HTH

Peter

[peterscott]

Two things that it can be helpful to know about when operating at low frequencies are that:

1. It is possible to tune lower than 0Hz into negative frequency. (Keep an eye on the direction that the digital display is incrementing or decrementing when trying to tune in to the frequency of interest.)

2. It can be helpful to increase the sweep rate (and ignore the compromised amplitude accuracy) just when searching for the frequency of interest. There are two spring loaded sweep speed-up buttons in the bottom right corner of the lower unit.

[Dr Wobble]

Thanks Peter, I will experiment later following your instructions.

Andy.

[Alistair D]

Regarding dBm, the m is for milliwatt dBs relative to 1 milliwatt, the online calculator below will do the conversion to volts for you.

http://www.coretechgroup.com/dBm_Calculator.php

It can also be done with the dB calculator in the Service Data Sources Spreadsheet but as that is for dB the answer will need to be multiplied by 0.224 to be correct.

Your analyser could give an answer that is either +ve or -ve, e.g. -10dBm, do not forget the -ve sign, if present, when entering the value in either calculator.

Edit: To prove this try the following. Enter 1Vp-p into the calculator, the result will be 3.97dBm. Now feed a 1Vp-p signal into the analyser and hopefully you get +4dBm.

Al

[Refugee]

Be careful not to exceed the maximum input voltage that should be marked on the front panel.
If you exceed the voltage and blow something the front end will be a very fiddly repair indeed.

[Cathovisor]

One other thing to add to Ref's warning, Andy - some spectrum analysers won't tolerate ANY DC on the input. The service manual on G1SLE's website shows the input to be DC coupled but it may just result in a trace shift.

As you've probably already discovered, these things are big, and heavy (I have its successor model). But they're made from pretty much standard off the shelf components so they'll probably long outlive their modern counterparts

A theory someone had at work years ago was to feed white noise into an amplifier and then do a sweep on the output - and in theory, it should show you the frequency response. What it does in practice though...

[peterscott]

A theory someone had at work years ago was to feed white noise into an amplifier and then do a sweep on the output - and in theory, it should show you the frequency response. What it does in practice though...

With the TF2370 having a built in tracking generator it's very easy to get the overall response of the back to back pair.

Peter

[Dr Wobble]

Thanks for all your replys. I will tread carefully before I start mucking about. I'm trying to find some test leads in order to use this beasty as my test lead collection is woefully deficiant.

I have brain freeze re decibels Alistair. I know a dB/mv/V etc is a logarythmic expression of power. I really need to get my head round this by doing some calculations. Like a lot of us, I tend to shy away from maths, that said I do enjoy the "fizz" I get when doing calculations. My problem is I'm starting from such a low base as I don't know how to use a scientific calculator or what 90% of the buttons do. I've learnt a lot in five years about electronics, I now need to" up" a level or two. Not sure my calculator can do letters Alistair.

I have to suss out what some of the controls do. Most seem to be to control the display parameters. I'm puzzled however what the 5 turn pots do; they are labelled Reference Frequency.

Andy,

[peterscott]

The four pots labeled Reference Frequency all do the same thing but the range from course on the left to fine on the right.
The spectrum analyser is just a big superhet radio and the reference frequency is just the tuning control to take you to the frequencies you want to look at.

The push buttons below these "tuning knobs" control how selective you want your radio to be. The more selective the easier it is to separate adjacent frequencies but the narrower bandwidths (the buttons towards the left) will slow down the sweep rate of the display.

You can actually use the spectrum analyser as a radio receiver. There is an output on the back that you can connect to an amplifier/speaker.

Peter

[Herald1360]

Bels express a logarithmic power ratio. A Bel is log10(P1/P2) so a power ratio of 10 = 1Bel. A decibel is 1/10 of a Bel, so a power ratio of 10 = 10dB. 100 = 20dB 1/10 = -10dB etc.

It gets more confusing when Volts come in- provided that the impedance is the same for P1 and P2 then since P is proportional to volts squared and to square a number you multiply its log by 2, then 20log10(V1/V2) gives the right power ratio in dB.

Trouble is, dB are often used carelessly to express just voltage ratios- a typical example being to say that a cathode follower (voltage gain of 1) has 0dB gain. It doesn't. It has quite a high power gain since the input impedance is high and the output impedance is low, much more power can be taken from the output than is needed todrive the input.

More fun again with dBm- unless the impedance is carefully noted. 0dBm in 50R is about 0.225V, but in a 600R sytem it's about 0.775V.

Eeeek!

[Terrykc]

Some might query why RF signal levels tend to be measured in dB. Also, there are two ways are doing this using either dBµV or dBmV, which can confuse the uninitiated.

The important thing to note is that, as a dB measurement is always a ratio, as pointed out in previous posts, there has to be an absolute reference point if they are to be used for level measurement. Also, the impedance must be a constant.

Thus the references are: 0dBµV = 1µV in a 75Ω load whilst 0dBmV = 1mV in a 75Ω load. The scale used should be that which fits the range of levels best, thus someone concentrating on very low levels - for reception measurements, say - would use dBµV whereas someone like myself who is always concerned with the larger levels found in RF distribution is happier with the dBmV scale.

There is a further complication here in that people who deal with transmission parameters work with 50Ω loads rather than 75Ω but we will assume here that 75Ω is used throughout.

Sometimes it is necessary to convert between the two scales which is easy as the ratio between the two is 1,000 - which equates to 60dB for voltage measurements. Thus 0dBmV = +60dBµV. The specification for the output of the Aurora converter is specified as +76dBµV but I automatically convert this in my head to +16dBmV ...

But "why?", I hear you ask - what's the matter with good old fashioned volts?

Good question! Let's try it!

A UHF modulator set to Channel 68 (847.25MHz) has an output of 6.3mV and feeds the input of an amplifier with a gain of 50. The output of the amplifier feeds a 100m length of CT100 cable which attenuates the signal @850MHz by 90%. This feeds a two way splitter which loses a further 36% of the signal. This then feeds an 8-way splitter and, finally, an 8-way tap, both of which have losses of 72%.

What is the output level?

It will be simpler to say that the output from the cable is 10% of the input, from the 2-way splitter it is 64% and from each of the two 8-ways it is 28%.

While you are sharpening your pencil, I'll do it my way!

The modulator output is +16dB, the amplifier gain is 34dB and the cable loss is 20dB. The 2-way splitter has a 4dB loss and the 8-ways each have a loss of 11dB So we have 16 + 34 - 20 - 4 - 11 - 11 = 4dBmv. Simple mental arithmetic!

So, how are you getting on? 6.3 x 50 x 0.1 x0.64 x 0.28 x 0.28 = ??? - and put that calculator away!

Eventually you should get 1.58mV which, by sheer coincidence, happens to be +4dBmV ...

Stick to voltages if you wish but I'm sticking with dBmV!

[johntheboffin]

If it helps, there is a manual with detailed instructions on my TF2370 web page:

https://sites.google.com/site/marconiin ... m-analyser

John

[Dr Wobble]

Thanks John for the link. I,d downloaded it but the manual presumes the owner to have a certain level of knowledge. Hence my thread.

Someone over on UKVRR has sent me a couple of test leads so I'm hoping to tackle the beasty later this afternoon.

I understand that dB* is an expression of a ratio and the last letter, IE v/V/u/m may indicate an impedance. However like impedance which I struggled with for years, dB remains fuzzy and tantelisingly just out of reach for the old noggin. I can tell you what it is but when I try to apply it, say on my mixing desk or in reference to an amplifier the concept illudes me. It drives me bonkers; sometimes when reading up on it I think I have it........then like a ghost in the night, it slips away. Hope that makes sense. However with help from your good selves, like when I got stuck on impedance, I hope to get there in the end.

Check this vid out. I'm no wiser at the end of it. https://www.youtube.com/watch?v=mLMfUi2yVu8
Dave may say, "Its bloody simple you peanut!" Well ain't for me cobber! It's just the way my brain works. God, how I struggled with ohms law, it took me 3 years. Then one day whilst out with the dogs in a field I got it. I think it has something to do with not excepting something like V and I are linear in respect to R. I have to mull the whole thing over for a while, so I know what I is doing when R = x and V is xx and what about W.

Thanks again for all your help,and reading the waffle, Andy.

[Refugee]

You need to look out for something like this for doing audio with a spectrum analyzer.
viewtopic.php?f=15&t=1915&hilit=
It should display a graph just like the ones given in data sheets scaled in dbs directly showing the frequency response of an amplifier.
You use the external trigger input to operate the scan and off you go.

[Alistair D]

Andy, for the present forget about all of the "waffle" Use the online calculator to convert the screen values to the volts you understand. Use your time getting to know the instrument. DBs will fall into place later.

Al

[johntheboffin]

I don't envy Dr Wobble's struggle with the dB concept. I was the same at college when I first encountered it. One day it just dropped into place and I wondered why I had trouble with it - I am sure the same will happen to you!

Maybe try a different approach. Have a look at the scale and switch markings on this valve voltmeter (click on the image to blow it up a bit):

https://sites.google.com/site/marconiin ... -voltmeter

In this instrument, 0 dBm (dB relative to 1 mW) is for a 600 Ω impedance. The voltage required to dissipate 1 mW in 600 Ω is 0.7746 V (V = Square root (power x resistance)). You can see on the scale that the 0 dBm point corresponds with 0.7746 V on the 1 V range.

In voltage terms, a given dB change is 20 log (V2/V1), where V1 is the "starting point" and V2 is the "finishing point". So, for example, if V1 is 0.7746 V and V2 is 0.07746 V then we have a change of 20 log (0.07746 / 0.7746) = 20 log (0.1) = -20 dB. You can see from the red markings around the switch that this is indeed the case.

You will also see that the "3" series scales have a slightly different full-scale point; in fact they are really multiples of 3.162 (square root of 10). This is to account for the fact that V = Square root (power x resistance), as noted above. This way the same dB scale can be used for all ranges, the switch indicating how many dB have to be added.

As has been mentioned earlier, the dB scale is always a ratio, any last letter (m, V etc.) defines a "starting point" for that ratio and therefore makes the scale absolute.

Why bother with all this? One good reason in audiometrics is that our hearing is not linear, it approximates to logarithmic. Therefore a sound level scale is better expressed in dB as the scale corresponds better to our perception. Going back to spectrum analysers, they are expected to display a huge dynamic range and so a logarithmic scale is the only practical solution.

Keep at it - the penny will drop one day! And, by the way, you have an excellent spectrum analyser!

John