AD8307 probe-How is it used?

[Radio1950]

Jeremy,

Thanks.
I read all that, hand on heart.
Very interesting.

I saw some of that third harmonic variation with a quick test setup here, but I cannot control the phase of my second sig gen.
You have the benefit of owning two special sig gens.
If I vary the frequency of my third harmonic by 10 Hz (not the same but a minimum avbl step), I get "accuracy wobbles" on my AD8307 Milliwattmeter.
My two sig gens are M2018 and M2022 within 2 Hz by GPSDO.

The reason for your result with IMD two tone test is that the signals are not exactly coherent?

rgds

[Radio1950]

David, OP,

whilst the academic discussion is interesting, you don't have to worry about it at all as it normally applies to special cases only.

Your AD8307 probe, if checked for reasonable accuracy, will be fine for 99.999999 % of tests that you might do in the home workshop.

Just keep in mind that if you see a result that looks odd, it may be true, and a good think is required.

And don't be put off too much by the "stories" as it gives us a chance to read about some historical stuff.

[Radio1950]

Stories,

Here's one, and even about Specans.

In about 1984, ominous year, our organisation installed about 48 Satellite Ground Stations, to support ATC HF (being converted to SSB) and VHF AGA Coms.

Japanese NEC 800 chan FM FDM, 12 and 14 GHZ; really beautifully built.

We needed a Specan for lineup, and as we were on an airport which was joint civil (me) and military, we asked (them) for assistance.
We helped them with their Thomson ER720 TAR radar, so they really had to.

"Here you go mate!"

They offered what looked like a WW2 Panadaptor, keeping their good Specans safe.

"I'll just go and see if it fits through our door!"

Stall.

The very next week, our new Anritsu MS710 turns up.
Drool.
Lovely thing to use.

Advance 30 yrs.

Our Bean Counters at work had a "Resource Rationalisation" and threw out about 20 MS710s, 15 Marconi 2018s, same M2018, and some Rhode and Schwartz SMHU sig gens.

I was on holidays.

The medicine helps, and on the good days, I mostly grab my Weller iron by the blue end.

(Jeremy, the NEC satellite stuff used 70 MHz IF.)
.

[Radio1950]

And the answers to my deceptively simple questions to Jeremy, with sometimes complicated side concepts?

As he says, the sum of two unrelated pure sine CW signals is the sum of each algebraic voltage vector, which allows for phase.

So eg 0.0 dBm and 0.1 dBm summed is +0.1 dBm, or 1.1 mW.


The level at which one interferes with the other in a AD8307 meter environment is a difference of 1.0 dB.

A bolometer (thermal sensor) will give identical results, it has to.

But the end effects of two unrelated pure sine CW signals when analysing a Device Under Test, depends entirely on the non-linearities of that particular DUT.





Sorry if I am preaching.



New question.

If I combine two 0.0 dBm CW signals, S1 is 5 MHz, S2 is 6 MHz in a 6 dB hybrid combiner (eg ARRL HBK 2007), and monitor the output of the combiner with my AD8307 Milliwattmeter, I see -6.0 dBm.

If I remove one of the input cables to the combiner, the AD8307 meter still shows -6 dBm.

I reconnect the cable and remove the other cable it shows -6 dBm.

If I remove both cables, the meter falls to nothing.

I did this last night.

What is going on?
Why don't I see -3 dBm, the power sum?.

Kewpie Doll available for cautious responder with correct answer, personal pickup.

With thanks to moderators for tolerance with slightly off topic stuff.
.

[Radio1950]

At the risk of being verbose and over gushy.

Reference my older question, the adding of two “unrelated” sine CW signals, the correct algebraic vector sum result will only be apparent on test equipment when the two signals are tending to be “coherent in frequency and phase”.

I know that’s somewhat tautological, but explains the situation.

This is probably what your instinct tells you, and is reproducible in the home workshop.

If I feed the 0 dBm 5 MHz signal, and 0 dBm 6 MHz signal, into an adder (not a mixer) and measure the added output on my AD8307 meter, I “measure” - 6.0 dBm, due to the loss in the adder (the hybrid combiner) and due to the two signals being so far apart in frequency.

The added power algebraic sum is actually occurring, but at a rate of the frequency difference, ie 1 MHz, and is not visible on the meter.

If I then adjust the frequency of the 6 MHz sig gen down, approaching 5 MHz, that rate decreases.

No change in the measurement is observable on the power meter until we get very close to 5 MHz, within a few Hz, when you notice the AD8307 meter starts to vary between a low figure (looks about -20 dBm) and about 0 dBm.

The “low” figure is theoretically much lower, but, visibly, meter and circuit response and hysteresis is at play; similar results with the ‘high”.

I cannot get exact coherence because I have two different sig gens.

If I could have it, and have a zero loss combiner, the power sum should be +3 dBm, 2mW, and if I could change the phase of one sig gen to be exactly 180 deg away, from the other, but on frequency, the power sum should be minus infinity, but in practice will be probably about -60 dBm.

I hope that explains it better.

Conversations tend to make assumptions with those directly involved, to the detriment of casual readers.
.

[Radio1950]

Reference that new question.
To elaborate.

It doesn’t matter if I use a three port 6 dB hybrid (transformer type) combiner, or a resistive three port (ie 3 x 17 ohm) star combiner, results are the same except the resistive type output is not – 6.0 dBm, but is about – 4.0 dBm or so, (due to the resistive type).

It also doesn’t matter to either combiner type if I turn off the sig gens by their internal Carrier ON/OFF, or by removal of cable.

And it doesn’t matter if I terminate the input port that has had its sig gen cable removed from, results are still the same.

All this is realisable in the home workshop with two sig gens, a combiner and AD8307 type Meter.

Who said they were not useful?


Pondering? ...

[G0HZU_JMR]

The AD8307 can certainly give really good results but I think it's important to understand its limitations when it is fed something other than a sine wave.

It definitely will fall over if given a signal with odd order harmonics like a square wave and theory suggests the error is 3dB in this case. That's a lot worse than other types of meter. I definitely wouldn't advise using an AD8307 to measure the power of audio test signals unless a pure sine wave was used.

Here's a screenshot of my old software after testing my AD8307 sensor back in 2019. Hopefully I used a lowpass filter after the sig gen although that sig gen does already have low harmonics. I can't remember what the test frequency was but you can see the AD8307 did quite well in this test. The detector linearity isn't that great but for most things this doesn't matter.

[G0HZU_JMR]

The AD8307 is a log detector and it is very sensitive to odd order harmonics and this can cause confusing results. A picture speaks a thousand words so here is a chart showing the measurement uncertainty 'window' for the AD8307 if there is a third harmonic present. This is based on carefully measured data using two vector signal generators made by Agilent. These allow the phase to be accurately adjusted in 0.1degree increments.

You can see that the uncertainty is fairly negligible if the harmonic is suppressed more than 30dB and so this would be fine for making absolute power measurements.

However, if like me you want to use the AD8307 to make relative measurements such as a low insertion loss measurement then it really is important to make sure the signal source has very low odd order harmonics. 50dB suppression would be nice in this case and I usually try for 60dB.

Otherwise, an attempt to measure a lowpass filter with an insertion loss of 0.3dB could give very strange results because the effect of the harmonic is present during the through calibration but the filter will then remove this error offset when it is placed inline. This means it is possible to falsely measure 'gain' from a passive device using the AD8307 log amp.