RF power meter calibration

[Steve G4WCS]

Not sure where to put this but I know there’s a large collective of expertise on here. (Moderators, I wonder whether there would be any mileage in having a measurement/calibration/useful standards section ?)

I have one of those cross meter swr/power meter things that I need to check the calibration of.

Now even back in my calibration days, RF Power was easy, we had a signal generator, 1kw solid state power amp, and Bird thrulines that we sent off to an NPL lab to use as standards.

Not so easy at home unfortunately.

I do have an all singing all dancing transceiver that will do up to 100 watts, and can be easily adjusted, and a calibrated TEK 463 scope. Is it as simple as running the transceiver into a dummy load through the wattmeter, and tee’ing into the coax to measure the voltage waveform on the scope, or do I make a diode detector, and use a dvm to measure the dc output, or are there any other tips I can use please ?

[MrBungle]

Scope is probably the most accurate without spending lots. I wouldn’t tee it though. It needs to sit behind the termination so you’d need a through terminator and attenuators.

I went the opposite way and bought an AWG (Rigol DG1022Z) which does 0-25Mhz dead flat with a reference output to a 50 ohm line in dBm. This is more than good enough for me. While acting and a signal reference this also has the advantage of being able to do log sweeps and marker output so you can use it either with a through terminator or a probe on a termination to do extremely accurate filter sweeps as well. Not satisfied with that I built a W7ZOI power meter which gives log power output so you can get a log-log display on the scope. Basically it’s then a scalar network analyser. You can even use it with a return loss bridge to sweep antennas.

[trh01uk]

Steve,

you don't say what frequency you want to do this calibration at. I presume we are talking HF - up to 30MHz in other words?

I think the method you are proposing of teeing into the coax to measure the voltage will be fine, provided the mismatch introduced by the scope probe isn't excessive. If we assume a standard 10:1 probe is essentially a capacitor - typically 10 - 15pF or so - then its effect is going to be a minimum at the lowest frequency your power meter/transceiver combo can run at. At 1.8MHz, 10pF will have a reactance of 8.8kohms, which falls to 530 ohms at 30MHz. I'd certainly be reasonably confident of a good forward power measurement up to a few MHz, with that setup.

One way of getting a calibrated RF power meter is to buy one of the older thermal types, which can be calibrated with DC. The Marconi TF1152A - as shown on this webpage - would be a typical one that I think can be picked up for modest sums in the usual places.

I suspect the easiest way to check the reverse direction power calibration on your SWR meter, is just to reverse the meter, feed RF into its output and take its input to a dummy load. Any forward power then will read on the reverse power scale. I think this is a lot easier than trying to make up known high power RF loads to give a defined mismatch.

Richard

[Radio Wrangler]

Bird 43s aren't precision calibration lab instruments. They are handy robust meters intended for setup and repair operations in the field.Errors can get down into the 5% region.
The HP 435A, 436A 437A with 848x series heads are more accurate power meters (non directional) reading to 100mW and operate thermally.

The HP432A with 478A series heads is a DC-substitution thermal power meter and can be used to transfer precision DC measurements to RF. NPL use these and the more sophisticated NIST type IV power meter, again with 478A heads.

The analogue power meters sometimes turn up cheaply, and I bought my pair of 478A heads, new unused in sealed NATO packaging for 50p.

Er, I've just looked.. a 478A with a buy it now price of over £400. Wow! a reminder to only trust completed prices!

Anyway. Use a decent scope. do not use a scope probe, use an attenuator and 50 Ohm load and the result should be as good as with a Bird 43.

David

[Bazz4CQJ]

With a 100W HF tx, does temperature measurement of the dummy load offer a cheap and cheerful option? Could it be set up using a DC supply (or supplies)?

B

[Terry_VK5TM]

Originally Posted by Radio Wrangler

Er, I've just looked.. a 478A with a buy it now price of over £400. Wow! a reminder to only trust completed prices!
David

Yes, people have cottoned on to them and they are attracting significant prices now.

If you can find the heads for 50p a time, grab as many as you can, you could have a very comfortable retirement.

If/when they become available (getting hard to get now apparently), they are attracting even higher prices.

[Radio Wrangler]

I paid 50p for two!
They were just anonymous cardboard boxes sealed in multiple thick polythene bags but I recognised them. I worked in the same building as the 478A repair facility. Drove home with a bit of a smile on my face.

David

[MrBungle]

Those power sensor heads are a bit fragile. I killed mine.

Thus I refer to this thread for a thoroughly modern and cheap solution https://www.vintage-radio.net/forum/....php?p=1018915

I’ve actually designed and manufactured a board for this. I will post it once I’ve got it in an enclosure.

[G0HZU_JMR]

It's worth estimating the various contributors to overall uncertainty especially if you go down the route of using an analogue scope and a dummy load. Otherwise you could end up with some odd results especially if the dummy load isn't very precise at RF or if the transmitter has harmonics up around -30 to -40dBc or if the scope is only accurate to maybe +/-3% at best.

I'd expect that even with some care it will be difficult to get reliable power measurements that are better than +/-10% if using a typical low cost dummy load and an old 75MHz analogue scope and this also assumes the transmitter harmonics are low enough to be negligible in terms of their contribution to overall uncertainty.

Also, I'd be nervous to apply more than maybe 10Vrms to a scope input once up into the HF bands. I have put 28MHz at about 25Vrms into a scope (plenty of times) in the past with my old Tek 585 scope without any ill effects but I don't think I would do it again. I think it is normal to lower the voltage limits for a scope quite a bit once you get above a few MHz.

At power levels up to about 10W it is possible to use a decent 50R dummy load and a simple (schottky) peak detector diode and a DVM to estimate power based on the peak detected RF voltage. But the 50R load ideally needs to be within about 1% and the voltage drop in the diode needs to be factored in to the DC voltage measured at the DVM. But the accuracy should improve with this method and it could be as good as +/-2%. But it is more likely to be +/-5% and also the harmonics also need to be very low. I would prefer all harmonics to be -50dBc to make sure they are negligible with this method.

Above about 10W the diode will be stressed too much so one way round this is to tap the diode down into the dummy load and correct the reading using a spreadsheet/calculator. This is best done with a homebrew dummy load made from lots of resistors as it is then easy to tap the diode detector in place. This would be OK up to about 30MHz if the load was made with decent resistors.

The other way to do it at high power would be to use a calibrated attenuator or a calibrated homebrew directional coupler based on the classic Sontheimer design.

[Steve G4WCS]

Thanks for the replies. Ill give it some thought. Will certainly have a go at your board mr bungle as we have discussed.

[Bazz4CQJ]

I just spent about 15 minutes playing with my dummy load and running DC through it and results attached.

The 50R dummy load is a commercial piece of equipment picked up at a junk sale, but has no info on it other than a sticker warning it contains beryllium. Temperature was measured on an IR camera. I'll need to strap another PSU in series to get to 100W.

I'm not going to claim that this is "highly" accurate, but doesn't it at least give you some indication about what the power meter says?

B

[G0HZU_JMR]

I've used a similar method with tiny chip resistors when measuring a balanced load but I didn't expect great accuracy from it. In my case I wanted to measure frequency response without touching (unbalancing?) the balanced load and I used something like 100mW power into the circuit and used a thermal camera to look at the temperature of the chip resistor. I crudely calibrated the temperature vs power with a known +/-1dB change in drive level and then measured the frequency range that stayed within this window.

The chip resistor changed temperature quite quickly and settled quite quickly too. However, you will be stressing that load a bit to get it up at 100degC on the outside when you run it at higher power. How long does it take to settle each time you change the power and does it settle the same at each power level on the way back down?

[Radio1950]

Hi Steve

If your Power Meter with crossed needles is a Bird type or similar, it will have known characteristics on its spec sheet.
If it is an amateur radio type it may have a spec sheet with dubious characteristics.

I had a similar cal situation a while back and needed a 200 Watt (level switchable) HF power meter with reverse capabilities, useable down to 2.0 MHz (2182 KHz actually).

I tried and tested a few amateur SWR meters and found the response below 5 MHz unuseable for my task (HF Marine TRX and ATU test and adjustment).
I ended up constructing a dual directional power meter using the classic two ferrite cores and two diodes circuit.
Works very well.

1 - I tested the frequency response of my reflectometer using a known sig gen at +10 dBm observing the DC output (10 K load) on a VTVM.
It helps to know if the Zo of your sig gen and dummy load are matched.
My reflectometer was within 1.0 dB from 500 KHz to 50 MHz.

2 - I then calibrated my power meters using a newish commercial ICOM HF transceiver in FM mode, no mod, and varying the carrier power, observing the RF voltage with a CRO probe across a good dummy load. You must use a 10x CRO probe, or other low capacitance probe and RF millivoltmeter etc.
You can also read the Po on most transceivers, for correlation.

Some problems you encounter in practice may be -
The frequency response of some power meters is dreadfully non linear, and that the amplitude response is poor at low power, and especially low frequency.
Then there is the cal accuracy of your test equipment and loads.
Your cal power source (eg transceiver) may not be the same impedance as your power meter.
Your power meter may not be low SWR in itself.
Testing at three power levels per power meter meter range is required to check overall cal. Use say 5 watts, 50 watts, 100 watts, to check the 100 watt range. More important for home made meters; your commercial meter scale should reflect the specific non-linearity of your actual diode detector.

Don't expect any accuracy more than 5%, even for normal commercial power meters.
A practical result of 10% measured accuracy is fine for most RF jobs
(However, SWR, a ratio, can be determined with more accuracy).



Even my Telewave HF VHF power meter has a frequency cal chart on the back to overcome freq response, and as you probably know, Birds use separate slugs for various frequencies and power levels (except for the rack mounted types).
Both use a type of coupled transmission line for sampling.

If you don't have to measure powers at one watt or less, the detector diode sensitivity issue with these power meters does not matter so much.

So, as per your original post, I would say that your own suggested method of using a CRO and probe is OK, provided normal accuracy restraints are taken into account.
If you use your suggested method of diode and DC meter, you will have calibration issues with your detector as well, even before you calibrate your meter. and then you would have to then use a calibrated attenuator.
Better to use a CRO.

To check that the Zo of your sig gen is the standard 50 ohms, monitor the output at say 0 dBm with no load on a CRO with 10x probe at a frequency well withing the rating of your CRO, eg 1.0 MHz.
Attach a known good 50 ohm load, and verify that the RF voltage across the sig gen output falls to 50% of the previous O/C reading.

To check that the Zo of your transceiver is the standard 50 ohms, monitor the output voltage with a CRO and 10x probe across a known 50 ohm load with say 10 watts RMS at a low frequency eg 3.5 MHz.
(Attach the 50 ohm load with a coaxial "T").
Then double terminate the output with another known 50 ohm load, and verify that the RF voltage has fallen to 66% of the first reading.
The transceiver should easily cope with the 2:1 Z mismatch.

Good luck with your task.

[Bazz4CQJ]

Originally Posted by G0HZU_JMR

However, you will be stressing that load a bit to get it up at 100degC on the outside when you run it at higher power. How long does it take to settle each time you change the power and does it settle the same at each power level on the way back down?

I really did just a quick and dirty test; that unit has no ratings info on it and I've never run so much power in to it before, so I was just looking for a 'ROM' sort of response. The practicality of measuring temperature to indicate anything meaningful is going vary enormously from one dummy load to another depending on the physical construction. That unit does look as though it might be giving viable reading, at least to 50W. Obviously, working with a dummy load that's mounted in an oil-cooled arrangement would be much more difficult.

In Steve's OP, he doesn't say specifically what equipment he has, but I assume it's some fairly standard HF transceiver and maybe the cross-meter power/swr is part of an antenna tuner?

So, if the rig has a nominal 100W RF out, I wonder what is the meter actually reading and what sort of accuracy would he'd like to have? If we were talking about the older rigs with a pair of 6146's in the PA, depending on their condition, the RF out could be anything, whereas with something that is all solid-state, I suppose you'd expect it to be within x% of the spec (where I will let someone else suggest what x is)?

B

[Radio Wrangler]

The output of the transmitter probably doesn't look like a 50 Ohm source, nor does it need to. It could do, but it doesn't need to. We only want power transmission in one direction but we do want optimum efficiency.

There has been a long running argument on this matter in the US. The chief proponent of the "it must be fully matched" camp was called Walt Maxwell. He'd written on it for the ARRL and the ARRL hadn't baulked. The "It doesn't have to be" camp wasn't so individual, but perhaps the most prominent was Warren Bruene, senior engineer at Collins and designer of the VOA's 250kW transmitters. The full matching camp just kept saying the maximum power transfer theorem demanded it.

Most high power systems do not run in a fully matched situation. The maximum power transfer theorem tells you how to extract the theoretical maximum amount of power that could, in theory, be taken from that Thevenin or Norton model of source. It fails to take into account that this may be significantly higher than the source will survive, or than you want.

Apply the max power transfer theorem to the light bulb in your room: The source looks like 240V and 0.5 Ohms, so you need a 0.5 Ohm light bulb. It will therefore get 120V at 240 Amps = 28.8kW which might be a bit brighter than you want and the heat in your wiring will be equal. It'll probably set your house on fire. If instead you choose an old fashioned 100W bulb, you breach the max power transfer theorem. It takes 100W at very close to 240v, = 0.42A. Your wiring drops 0.21V and dissipates about 87mW. Your efficiency is about 99.91%

OK, Zout of a transmitter isn't all resistive. But blindly assuming the max power transfer theorem without doing the exact analysis each time is daft.

Back on topic, The 8481 faily of power meter heads ARE resistors with thermocouple sensors and work open loop. Like a dummy load and a thermometer.

Stick a dummy load in a polystyrene foam-insulated bucket of water and performing calorimetry is a good way of measuring power.

David

[Radio1950]

Wrangler,

I thought we were suggesting practical non-lab arrangements for a calibration procedure for the OP, where it might be comforting to the user to know that the source impedance was at least close to the impedances of both the Power Meter and the calibration load.

For the OP and everyone's interest, the Zo of my ICOM 7600 TRX at 3.5 MHz transmitting 10 watts, is close to 50 ohms measured, and ideal for some DIY calibration uses, as it has front panel adjustable power output.


I will leave it to others to ponder on the relevance of your Mains Power Transmission "analogy", and whether it is an actual valid comparison, and also whether it actually considers all factors.

Thevenin applies to ALL electrical real power transfer situations, no exceptions.
So does the maximum power transfer theory, no exceptions.

[Radio Wrangler]

I earlier said that a dummy load and a good scope ought to be sufficient.

I was also careful to not close off the possibility that some transmitters under some circumstances could look like a 50 Ohm source. The point is that they do not have to.

All sources at all times can be modelled by Thevenin or Norton equivalents.

When I developed the "Dual Directional Power Meter" for the G-QRP club, I used a little Trio TS430S as the signal source, into the directional power meter and into a dummy load of a standards lab calibrated power attenuator and an HP precision RF power meter (8482A and 435A). The attenuator was also known to be quite an accurate 50 Ohms at the input.

That directional power meter design has no adjustment for directivity. It relies on transformer turns ratios and the accuracy of two low power 50 Ohm resistors on the simulated coupled line. I marked meter scales for both 100W and 10W fsd with two different meter resistors. The reflected meter showed no power. A check with a network analyser showed the directivity of the two-transformer hybrid coupler to have over 35dB directivity, and my power attenuator was in that league too.

To calibrate the reflected power meter scale was simple. I turned the directional power meter around and drove RF through it in the reverse direction. The Forwards power meter now read zero, and I could calibrate the scale on the Reverse power meter in the same way I'd just done the Forwards meter.

As a check, I ran the dual directional power meter in the normal direction, checked for no reflected power reading, and then Tee'd two good 50 Ohm loads together and checked that the readings corresponded to a 2:1 VSWR. (and out of simple cussedness turned the directional power meter around to check the meter readings were the same, but on opposite meters).

There is a later version of this design, using the same hybrid coupler but Analog Devices logarithmic detectors (The old AD6440) - ADI had given me some samples. The biggest addition was a phase meter comparing the limiter outputs of the two logger chips. This reads the phase of the reflection. The difference in voltage of the two detector outputs reads the return loss (in dB!) of the reflection. This arrangement does a vector impedance measurement on the load - at full operating power. This is particularly useful when measuring the input impedance presented by linear amplifiers. It gave me polar format coordinates ready to plot onto a Smith chart.

Power meters can be good fun.

But for a transmitting setup, you want to tune the load to get its reflection down to a tolerable level. You don't need to know or even care about the source impedance of the transmitter. It doesn't matter. This is important because on some transmitters like Class-C stages, it varies wildly through the cycle of the RF waveform and you aren't going to match that with any linear load.

The 240V light bulb analogy is amusing, but it's only an illustration that in a perfect world, not everything would be matched source to load. Applied to a transmitter, it has a gaping omission. The output Z of the transmitter is not made out of simple ohmic losses. Thevenin and Norton models are "black-box models" which means that though they are useful for modelling what goes on outside the model box, they are NOT representative of behaviour inside. In fact they can be very misleading and the Thevenin and Norton models disagree with each other especially if yo try to use them on the power dissipation in the source impedance.

This is all far too much for the question in the original post. A decent scope and a reasonable dummy load is sufficient for HF. 10% change in power is hardly noticeable and 99.999% of reports are 5&9 anyway

I think my grasp of RF theory and practice is OK. I spent 30 years designing RF test equipment at HP, and someone would have noticed if I was below par. The little dual directional power meter was published in Sprat (issue 61) and copied in various other QRP clubs' magazines. Doug deMaw put it in his design notes book and used the coupler, with more elaborate detectors in the Oak Hills Research power meter. (I got given one). Since my time at HP, I've been designing avionics, and there are tens of thousands of transmitters I've designed now in aircraft. All approved for use in certified aircraft. They passed all their tests and they have a reputation for efficiency, which is particularly important to balloonists and glider pilots who have to run off only a battery. Aviation things are very closely monitored in use so we would get to hear if there are any problems.

David

[Bazz4CQJ]

I decided to throw caution to the wind and remove the two screws to open up my dummy load. Inside is just a simple flange mount resistor, which is rated at 150W and with a max temp of 100'C; see attached PDF.

Price for 1 is $23. Not yet identified a UK supplier. My unit has the centre lead going straight to the pin of a N-type connector, and the flange on an aluminium heat sink.

Which leads me to a little bit of grief, since running it up to 50W yesterday, the N-type connector is firmly resisting removal. I wonder if the connector on the end of the jump cable did not appreciate being warmed up?



Hold the press; similar item from China, 99 pence, post free.


B

[Radio Wrangler]

THere are some N-connectors from a large SE asian country that seem to be slightly lacking accuracy, threadwise as well. Some are fine but some bits of gear can be very devils to get connectors off.

David

[Bazz4CQJ]

Yes, time and again it seems that the folks out east can make IC's at wonderfully low prices but really struggle when it comes to doing the same with anything that has an internal thread!

I think I might risk buying a fivers worth of their flange resistors and perhaps re-build my dummy load. I suspect that, as is, it will be hard pushed to run 100W for more than the briefest time. I could re-build it using 4 off 50R, two in series, in parallel with another two in series. Done 'neatly' that should still be a reasonable dummy load at frequencies at least to 30MHz? Mounting a thermocouple on the heatsink close(ish) the the resistors could prove useful.

B