measuring RF in a circuit

[Flump]

I have an old ham radio (transistor) that has no RF output power
so I need to measure the RF to see where it is stopping, it's likely
to be in the milliwatt range, I have a scope but guess I cant just
start probing around as I would get false readings ?
frequency can be anywhere from 3.5-30Mhz

[T3sl4co1l]

A 10x scope probe will work okay there, as long as the impedances are low and the tuning is not strict.  If this is not the case, try an E-field or H-field probe (i.e., wave the 10x probe tip near something without clipping it on; or use a couple turns of wire as a magnetic pickup coil near the tuning coils).

Tim

[w2aew]

In addition to the advice above for looking for the RF signals, you'd likely be more successful starting off with more of the basics.  Work your way through the schematic checking power supply voltages, and bias voltages of the various stages.  You'll likely find the fault much faster than by simply chasing low level RF signals...

[Teledog]

..used to use a 1000pF ~0805 cap at the end of a skinny (rg174?) cable - connected to the spectrum analyzer.
Bring the tip near your suspected dead section (LO or whatever) and the SA set to approx the same freq.
If you actually touch the end to the point in question with zero level..then there's your problem.

Could also use the same technique (sans SA) with a low buck homebrew RF probe (diode, cap resistor) to a multimeter(scope?).

..big ham HF rigs are a c***sucker to work on (been there, done that)

Ask the MFR (or Google) for a service manual..manuals for the obsolete units are usually free now.

G'luck!

[German_EE]

Take the ground lead of a scope probe and wind it into a small coil, then clip the ground lead to the probe tip. This will be enough to pick up RF at transmitter levels without connecting to the circuit.

Do you have a licence for this transmitter? If not then you will need to connect it to a dummy load whilst you do your tests.

[KJDS]

The art of probing any circuit is to do so in a manner that doesn't affect the circuit.

The most common way it to use a 470R resistor on the end of a piece of thin coax connected to a spectrum analyzer. If the analyzer isn't AC coupled then a DC block should also be used. This, with the 50R input impendance of the analyzer gives a 20dB probe and the 470R resistor doesn't have much of an affect on a 50R RF circuit.

Another way is to use an RF probe with a multimeter, either a commercially made Fluke 85RF or similar or a home made one.

[Flump]

In addition to the advice above for looking for the RF signals, you'd likely be more successful starting off with more of the basics.  Work your way through the schematic checking power supply voltages, and bias voltages of the various stages.  You'll likely find the fault much faster than by simply chasing low level RF signals...

I will try this but it will also be interesting to see if i can measure what's going in and out of the pre-drive / drive / to
ascertain in what section the fault is.

connected to the spectrum analyzer.

Ask the MFR (or Google) for a service manual..manuals for the obsolete units are usually free now.

I dont have a spectrum analyzer but there are loads of info on the web abvout the radio so a schematic
should be easy to  find

Do you have a licence for this transmitter? If not then you will need to connect it to a dummy load whilst you do your tests.

I do! , but it will be on a dummy load anyway 
.

Another way is to use an RF probe with a multimeter, either a commercially made Fluke 85RF or similar or a home made one.

RF Probe sounds very useful so I will see if I can pickup a SH one from ebay, thanks!


Also I just won this off ebay which might be useful ?

[KJDS]

Another way is to use an RF probe with a multimeter, either a commercially made Fluke 85RF or similar or a home made one.

RF Probe sounds very useful so I will see if I can pickup a SH one from ebay, thanks!


Also I just won this off ebay which might be useful ?

It's possibly useful, try it with the 470R on the end of a bit of coax and see if it's sensitive enough, if not then it's an RF probe and a multimeter. PM me if you need one, there's a pile of them sitting in my storage unit looking for new owners.

[G7PSK]

I have one of those lake power meters that I purchased new in 1993, I think I still have the instruction sheet and circuit diagram that came with the unit. They were sold as kits or complete.

[Flump]

It's possibly useful, try it with the 470R on the end of a bit of coax and see if it's sensitive enough, if not then it's an RF probe and a multimeter. PM me if you need one, there's a pile of them sitting in my storage unit looking for new owners.

I have no idea how much these probes go for but ill try the meter first and see how that goes,
whats the reason for the 470R resistor ?

I have one of those lake power meters that I purchased new in 1993, I think I still have the instruction sheet and circuit diagram that came with the unit. They were sold as kits or complete.

If you ever come across any paperwork for this meter I would love a copy,
never see RF meters that measure this low so its definately a keeper.

[KJDS]

It's possibly useful, try it with the 470R on the end of a bit of coax and see if it's sensitive enough, if not then it's an RF probe and a multimeter. PM me if you need one, there's a pile of them sitting in my storage unit looking for new owners.

I have no idea how much these probes go for but ill try the meter first and see how that goes,
whats the reason for the 470R resistor ?

Without it you're adding a 50 ohm load onto whatever part of the circuit you probe. You may get away with it, but it's also quite possible that it will have a significant effect on the circuit, rendering meaningless the results. If you're probing round an amplifier it is also possible that you'll kick it into oscillation which could kill the active device. The 470 ohm resistor means that you have much less affect on the circuit, but put a DC block in series with it if you're using that power meter, otherwise you could be introducing a big change to the bias of a device, which could cause it to oscillate, or in the case of a FET gate, could drastically change the bias. ...

[SoundTech-LG]

Recently testing output on the 5nS, and 2nS range on a Tek TG501. The output was weak and almost impossible to see. "Loading" the output with the proper 50 ohm termination brought the signal right up to spec. I don't see any 1nS output at all. But using a Spectrum Analzer is a good idea.

[Flump]

Without it you're adding a 50 ohm load onto whatever part of the circuit you probe. You may get away with it, but it's also quite possible that it will have a significant effect on the circuit, rendering meaningless the results. If you're probing round an amplifier it is also possible that you'll kick it into oscillation which could kill the active device. The 470 ohm resistor means that you have much less affect on the circuit, but put a DC block in series with it if you're using that power meter, otherwise you could be introducing a big change to the bias of a device, which could cause it to oscillate, or in the case of a FET gate, could drastically change the bias.

Thanks KJDS it makes sense now.

A few posts ago I made attached a picture of a low RF power meter, well it turned up today and it doesn't work! 2nd thing I have bought this week thats broken.

Anyway bit of a strange problem on this, the resistance measures near enough 50R but when you the SWR goes through the roof instead of being a 1.1 match, also the meter does not read anything.

It has to be something to do with the resisters, but what ?

[Howardlong]

Euww... well that's mutton dressed as lamb if ever I saw it! What frequency are you testing this on? Should be just about OK at HF I'd have thought, but it ain't pretty.

What is the return loss/SWR like on top band or 80m? Hopefully the construction techniques won't matter for much down there. That doesn't look like it'll cope with too much power, I wouldn't want to put 20W CW through it for much more than a few seconds.

Check your patch cables too, I've had far too many failures in that department for my own good.

[w2aew]

Here's my video on how to make a simple RF detector probe:

[Flump]

hi Howard !
I tried it on 7mhz and 28mhz

Alan I will look at your video in a bit and maybe try some of your component values,
as I don't think this would ever work in the condition i received it but i might be wrong,
and I don't know what diode is used in this meter, i might end up having to rebuild it all
but i bought it from an older ham radio guy and he has no reason to lie about it so i am
intrigued why it wont work in its current form.

Okay had another quick look at it and disconnected one end of the diode and the dummy load is perfect now but the meter will not read anything,
connected the diode back up and then disconnected the 10nF capacitor that goes down to ground and same thing, perfect SWR but no
meter reading, adjusting the VR on the back of the meter to full and trying various different values of capacitors (including the original) the highest reading I can get is 500mW with 5w in to it, so I am not sure if there is another problem because all the caps gave the same reading on the loads meter.

[G0HZU]

It looks like a very basic peak reading voltmeter but scaled in Watts for a 50R load and a pure sine wave signal. Provided that the dummy load resistors are suitable types and reasonable quality then it should be fairly accurate across the HF bands once adjusted via its internal trimmer pots.

The idea is that power = (Vpeak*Vpeak)/100 for a clean unmodulated sine wave signal source. i.e. a transmitter with very low harmonic content.

So with 5W going in you should see 22.36V DC (less one diode drop) = about 22V DC on a DVM at the cathode of the diode assuming it has the correct diode and there is a proper RF decoupling capacitor fitted here. eg 1nF ceramic disc or a 1nF feedthrough would be ideal. If the capacitor is no good then the DMM will give crazy readings if any RF gets into it. The one that is in the image doesn't look like a suitable cap.

Also note that the diode will be operating way up in the linear region here rather than square law and it will need to be rated to cope with a high peak inverse voltage. Something like a HP2800 or 1N5711 Schottky diode would work well here although I'd be happier if the diode was tapped at the mid point of one of the resistors to reduced the PIV stress on the diode. This would mean recalibrating the meter adjustment though...

[Flump]

with 5w going in and cap connected I get 1.3v DC at the cap end of the diode,
SWR is always off the scale unless the cap is disconnected,
and thats the same for all the caps I tried, they all made the SWR very high.

With cap disconnected I get about 2v DC at the diode and perfect SWR.

[G0HZU]

Is the diode damaged and leaking/shorted out?

[Flump]

Nope the diode checks out okay.

cant really makeout what is wrote on it too well but it looks like

55c
3v3
6zy or gzy

[G0HZU]

Hopefully it's not a 3V3 zener diode... 

[KJDS]

Hopefully it's not a 3V3 zener diode... 

A BZX55C3V3 is indeed the mostly likely candidate from the markings, and the least likely that any competent designer would use.

[Flump]

LOL it looks that way

I just replaced it with a BAV 21 as thats all I have
and now I get 18.77 VDC with cap d/c
and with cap origonal cap connected I get
19.30 and perfect SWR 

Will it be okay th leave the BAV 21 in there ?

[Alexei.Polkhanov]

Hmm I always assumed that best way to probe RF signals on board is with Active FET Probe.
There was some earlier discussion on a subject https://www.eevblog.com/forum/projects/good-design-for-a-diy-active-probe/15/. Also look at some DIY FET probe designs like this one:
http://elektrotanya.com/files/forum/2009/10/e04a036.pdf
and this
http://welecw2000a.sourceforge.net/docs/Hardware/Aktiver_Tastkopf_mit_OPA659.pdf
and here
http://www.mikrocontroller.net/topic/188227

[Alex Eisenhut]

Don't forget tubes...



850+ MHz aint too bad.

[G0HZU]

I did a quick simulation to help you understand further how the meter is designed to work.

If we assume the meter movement is about 50uA then for 5W we need 25uA through the meter to get a halfway reading on the dial because this is where is it marked as 5W.

We know that 5W is 22.36V peak...

Also, the diode is a peak voltage detector so we should get 22.36V less 1 diode drop at the diode cathode = 22V approx with a Schottky diode. We then need to fit a calibration (current limiting) resistor inline such that this 22V will give 25uA indicated on the meter for 5W.

So the total calibration resistance (trimpot+ series resistor?)  needs to be about 860k ohm according to my quick and dirty simulation.
There are graphs for the RF AC voltage across the load and you can see the green DC voltage output from the peak detecting diode is just under the peak at 22V.

you can also see a graph for current through the meter movement at 25uA. But your meter might not have a 50uA movement so this is just a guide. If it had a 100uA movement then the 860k resistor would be a lot lower in value. But it may well use a regular 3 pin potentiometer here instead of an inline two pin trimmer anyway. But the idea is the same... to adjust the pot to produce enough current from the 22V to just drive enough microamps into the display meter to give a mid scale deflection for a 5W test signal.

 Hope this helps...

[Howardlong]

LOL it looks that way

I just replaced it with a BAV 21 as thats all I have
and now I get 18.77 VDC with cap d/c
and with cap origonal cap connected I get
19.30 and perfect SWR 

Will it be okay th leave the BAV 21 in there ?

Loving this thread!

Germaniums seem to be preferred in this application, lower voltage drop. That BAV21 will be OK for now, not sure how it'll behave at higher frequencies though.

[G0HZU]

IMPORTANT!  I wouldn't recommend trying to measure the meter movement current if it isn't stamped on it somewhere. This is because it will be very easy to fry the meter movement if you make a mistake. Then once the meter movement is damaged you will have nothing...

You don't really need to know the meter movement current to calibrate it. I would recommend the old school HP2800 diode here or the 1N5711. They are very good quality Schottky diodes with a 70V PIV rating and low capacitance and decent frequency response.

If the layout was improved and tightened up a bit you could calibrate this meter reasonably well with a DC power supply if you used one of the above diodes because their detection performance is very predictable. eg apply +22V DC to the diode cathode (check it for 22V DC with a DVM) and then adjust the high power range trimpot to show 5W on the dial. And that's the meter calibrated on the high power range without even needing an RF signal source.

For the lower power range apply 9.64V DC to the diode cathode and adjust the other trimpot for full scale reading of 1W because 10V peak is 1W but you have to allow maybe 0.36V for the diode drop.

But make sure you don't touch the direct connection to the meter dial with your power supply wires or you will fry the meter movement with the DC power supply.

[TimFox]

Yes, never apply a DC voltage or an ohmmeter directly to an unknown meter movement.
To check the full-scale, you need to apply a reasonable voltage (maybe only 1 volt) through fixed resistors, starting with a high value (maybe 100k) until you get full-scale deflection by adjusting the voltage.
Then, to determine the meter's resistance, keep the voltage and large resistor in place and put a variable resistor in parallel with the movement.  Start at high resistance, and reduce the resistance until the meter reads half-scale.  Remove the variable resistor and measure its resistance, which is very close to the meter resistance (so long as the series resistor above is much higher than both).

[Flump]

G0HZU Thank you for the explanation and Drawings 
I am going to leave this meter as it is now as its all working
as it should do, the BAV21 I put in has a voltage drop of .600mV
so I am using that in my calculations to calibrate the scale with.
I also changed that Cap for a 1nF ceramic .

Volts to watts
+ 0.600 (diode)
/ 1.414
square it
/ 50

watts to volts
x50
square root
x 1.414
-0600

And the scale seems nice and linear across its range.
If calibrated at 5W it is spot on anywhere from 1-20

I do have a nice enclosure with quite a large analog meter
if i can find it i want to build a new RF meter from scratch
Possibly following your design above if its suitable ?
My thoughts on the new one is to be similar to this one
but measure up to 100w, so SO239 with big internal Load
but also a BNC for a flylead (scope probe?) and to have a scale which
reads 1 - 10 and make 3 switchable ranges,
0-1 , 1-10 , 10-100 .

And I have ordered some of these to keep in stock anyway
http://www.ebay.co.uk/itm/121518325327?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I also bought an RF probe and Blew it up within an hour of unpacking it,
apparently they dont like 100w on the pointy end
so I need to learn more about RMS voltages 

[G0HZU]

If you try and make a similar meter for 100W operation then I'd advise arranging the resistor network such that you can feed the diode detector from a tap point on the resistors.
This way the diode sees less RF voltage and so you could use the 1N5711 with its 70V PIV rating.

Obviously, this means the calibration theory will change a bit but this just means a bit more care is needing in working out the detector output for a given RF power.

Note that this type of power meter needs a clean sine wave to give reliable performance in terms of accuracy because of the detection system it uses. If there are any harmonics present then the uncertainty of the meter reading will go up a lot.

eg if you had a -30dBc harmonic the meter uncertainty could be +/-6% (approx) in the power reading. A -20dBc harmonic gives approx +/-20% uncertainty in power reading. Ideally, all harmonics should be something like -50dBc to make them insignificant with this type of meter if you want to get the most from it in terms of measurement uncertainty.

However, if you build this type of meter well then it should totally outclass something like a Bird 43 wattmeter in terms of accuracy across the HF bands. The Bird 43 is a very overrated power meter and I wouldn't want to own one. I'd be too embarrased to have one because they are so poor and unreliable both mechanically and technically.