CB and Ham Radio Techs Love Their Bird Wattmeters

[joeqsmith]

Wrote a quick program to sweep the board.   

[xrunner]

You may be interested in the following:
https://www.nutsvolts.com/magazine/article/a-discussion-on-dbs-and-rf-power-meters

First eval board appears to work, or at least the LED turns on.

That guy referred to DBS in the title and in the article "DBs" I kept trying to find out what the Hell was DBS. Then I realized he meant DBs like cookie"s" or beer"s". 

[xrunner]

Wrote a quick program to sweep the board.

That's good Joe you have confirmed that the red LED means it works correctly. 

[joeqsmith]

Wrote a quick program to sweep the board.

That's good Joe you have confirmed that the red LED means it works correctly. 

I was trying to following your lead. 

Same setup but sweeping a commercial diode detector from -20dBm to 12.   

[xrunner]

So I made up a little input compensation circuit that was in an article that bd had found a few years ago in a QST magazine (the article is 20 years old). The circuit is attached but all I needed was the parts at the input consisting of C2, R2, and L1 (C1 is already there on my board). I found the capacitor and resistor, and I made the one-turn coil myself. I assembled it all on a little perfboard sitting right next to the PCB traces of the main board and ran short wires to connect it. I removed the input cap on the board and put it on the perfboard so the input signal would go through the new components. When I turned on the system the output V which indicates power was way too high. Without any input power it is normally ~ 0.33 V. However it was up at 0.7 V and was really flaky, any movement of the perfboard would make the readings go crazy. So I thought I made some sort of mistake wiring it. I removed the perfboard and inspected it but found no problems. I put the AD8310 board back to original and it worked correctly. I then carefully put the new input circuit back on and the same weird responses happened again.

I put the board back to it's original state, I removed the coax from my Agilent and stuck a sewing pin in the SMA connector of the board - ooh nice signals coming like gangbusters! This thing is very sensitive to signals in the environment. I'll have to come back to this issue later because I can't test it the way I was. Or I might compensate for frequency in software with a menu. Anyway, I got bigger fish to fry now.

I started a shell program so I can get going on the microcontroller side of things. I looked at what displays I had available. I decided on the 128 x 64 Yellow over Blue (may change to something else on a whim though). I just put together a simple test at the moment. I set up to read an analog voltage which is referenced to the on-board 3.3 V reference (since we aren't reading anything over 3.3 V). I coded in the equation to give power in from Vo.

Pin = (Vo - 2.28) / 0.024 

I then connected the analog input to a power supply which is simulating the output of the AD8310 system Do we get the right answer given a simulated Vo input? Yes. So now those parts are out of the way.

In fact, I could connect it right now to the AD8310 and it should show the right input power (at the lower frequencies). But I don't have to.

Why?

Because nothing can go wrong ... wait let me re-read the first part of this post again. 

[joeqsmith]

So far I have been testing without any shield on the AD8307.  You may want to consider adding one.   

To automate the testing, I have the PC running the RF generator, DVM and SA.   The eval board is attached to the forward port of the coupler (20dB).  Output port goes to a 20dB attenuator and then to the SA.  I'm just swagging things for now so no need for the reflected signal just yet.   For now, no amplifier. 

I ran a sweep at 800kHz, 25MHz and 50MHz.   Coveroff is showing the signal without the cover on the coupler.   Adding the cover helps with the lower signal levels. 

Both showing with the cover and compared with the SA. 

[joeqsmith]

Set for about as much power as a child's CB walkie talkie (displaying 0dBm or 20dBm at the amplifier or 100mW).   

[xrunner]

So far I have been testing without any shield on the AD8307.  You may want to consider adding one.   

Yea I know. I ran into that issue last year working on an analog Boonton power meter. I had it apart and the sensor wasn't even connected. When I applied power to check another issue I noticed the meter was erratic on the lowest power settings even without the sensor connected. I eventually moved it to the kitchen to see if it was interference where my workbench was located. When I moved it the problem went away. Turns out it was my WiFi router. Its about 4 feet away and has enough power to cause the input circuits to respond. I could actually block it a little just by standing between the router and the Boonton. I'm sure the AD8310 is responding to RF fields a little even at those frequencies.

I did find something yesterday when I was hacking around the AD8310 board. In the AD data sheet they use 0.01 uF capacitors for C1 and C2 (in more than one example). However when I removed C2 to put it on my input compensation circuit, I found it tested as 0.1 uF. I then removed C1 and tested it. It also measured 0.1 uF. I changed both to what AD datasheet was recommending. For you to use or ignore at your pleasure.

For now I'm moving ahead with my software. I have more to add and the display has to be prettied-up. I am going to add at least one button for the user (don't they need something to press?). It's going to select an external attenuation value. External att. is needed if the user wants to measure higher powers than the board can handle on it's own. I'm going to have these options (dB):

0
10
20
30
40

Of course it doesn't physically add it - that's what the human has to do. But it adds the value back to the measurement so the readout is corrected. I also want to show both watts and dBm on two different display lines ... and the Ext. Att. value selected. I need to get all that on the display, and sometimes playing with the formatting on these little OLED displays can take more time than writing the entire program.

I want to make some more graphs to look at something else, perhaps for a second button. That should keep me busy for a while.

Good luck on your side Joe. 

[joeqsmith]

It also measured 0.1 uF. I changed both to what AD datasheet was recommending. For you to use or ignore at your pleasure.

The data sheets have a lot of pretty pictures but also explain how to determine the values based on your requirements.   They even explain how to DC couple it.   Before   checking the LED,  I read the datasheet and made a few tweaks.   

https://www.eevblog.com/forum/rf-microwave/cb-and-ham-radio-techs-love-their-bird-wattmeters/msg4244893/#msg4244893

I think my wife bought me the Signal Hound back in 2016.  I wrote some software using their API to allow it to decode a simple FSK modulation.   They didn't offer much as far using LabView with it and it was a bit of a pain to get it worked out.   I thought I would try their new library when I started working on this test program.   Sadly, they only offer support for 2016.   After NI's change to a subscription model which would have cost me $3000 to rent their software for a year,  I am back to using LabView 2011.   Watching the Titanic slowly sink where their top management seems to think if they can just ram the iceberg a little harder, it will somehow stay afloat.      Any long term vision is gone but I'm sure everyone receives their bonus for a job well done.  It's just my warped perspective.   My first rant for the day.    

I do like the Signal Hound. Outside of a problem with the USB connection on startup, I have not had any problems with mine.  That problem appears to be something on their end and not the OS and it looks like in my case, they have a fix for it. 
 

[xrunner]

I think my wife bought me the Signal Hound back in 2016.  I wrote some software using their API to allow it to decode a simple FSK modulation.   They didn't offer much as far using LabView with it and it was a bit of a pain to get it worked out.   I thought I would try their new library when I started working on this test program.   Sadly, they only offer support for 2016.   After NI's change to a subscription model which would have cost me $3000 to rent their software for a year,  I am back to using LabView 2011.   Watching the Titanic slowly sink where their top management seems to think if they can just ram the iceberg a little harder, it will somehow stay afloat.      Any long term vision is gone but I'm sure everyone receives their bonus for a job well done.  It's just my warped perspective.   My first rant for the day.    

Oh yea LabVIEW. I used that extensively in the lab I worked in during the period say from 1990 to 1998. National Inst. came to the company and gave a demo to my boss and me because he thought we could use it. I agreed and from that point on I was the guy who wrote all the applications and data acquisition stuff using their DAQ cards (in 486 and Pentium PCs) and GPIB instruments mostly Tektronix plug-ins . Eventually I went to another group and some poor schmuck had to take over my job. I don't know who did but nobody else at the time knew how to use it. Lets just say I got a few phone calls over the next year. 

[bd139]

LabVIEW is what caused me to learn Python 

[xrunner]

LabVIEW is what caused me to learn Python 

 

Yea some of those LabVIEW diagrams can get - shall we say - fairly busy on the screen. 

[joeqsmith]

Yea some of those LabVIEW diagrams can get - shall we say - fairly busy on the screen. 

I saw a lot of people asking for a Zoom feature but don't understand its use.  If I showed you the source for the LiteVNA, you would laugh at how small it is.   There are a lot of other features I would have liked to have seen but instead their management/marketing/sales squandered all their time on NXG.  Like I said, my warped perspective as I am sure that was a good business decision and everyone involved received a bonus.     

One thing I am pretty sure of, LabView doesn't make good programmers out of bad ones and analog Bird Watt meters that read high don't magically produce more power.   

[bd139]

LabVIEW makes terrible artists out of terrible programmers. It’s like painting with dung 

[joeqsmith]

Without a calibrated reference, I'm at the mercy of what ever my out of cal vintage equipment does.   If I connect the RF generator directly to the spectrum analyzer and sweep from 1MHz to 100MHz and -120dBm to 10dBm we get something like the attached.   If the two instruments and cable were perfect it should be obvious that we would see a perfect line at -120,-120  to 10.10.     The hams and CBers would need to chime in how their Birds would fit into this...    Oh right, the gold standards....

As you can see, things start to fall apart around 90dBm.   The crappy cables and adapters had to be removed to get this good.   No taped up red coax with wrong connectors on them.. 

[joeqsmith]

If I plot the difference between the Signal Hound and my Marconi generator, in the -80dBm to 10dBm range we are interested in, they seem to agree within a half dBm.   Just a sanity check.   

[joeqsmith]

If I run the Marconi up to 2.4GHz (max) it's still within 1.5dBm above -80. 

I wonder what my investment in vintage slugs would be to pull something like this off cause that's a lot of slugs... 

[xrunner]

If I run the Marconi up to 2.4GHz (max) it's still within 1.5dBm above -80. 

I wonder what my investment in vintage slugs would be to pull something like this off cause that's a lot of slugs...

Make sure you buy new just so you can trust 'em. 

[joeqsmith]

Shown running multiple sweeps from -80 to 10dBm at 30MHz.  It's repeatable enough that we could easily compensate for this error.   The problem is not having a known reference.   I could make the data look very good but it could be much worse.   Kinda like guys taking their screw drivers to their Bird meters....

[joeqsmith]

If I run the Marconi up to 2.4GHz (max) it's still within 1.5dBm above -80. 

I wonder what my investment in vintage slugs would be to pull something like this off cause that's a lot of slugs...

Make sure you buy new just so you can trust 'em. 

Do they even offer such slugs.  Google time:

https://birdrf.zendesk.com/hc/en-us/articles/4414502024727-What-Slugs-are-available-for-Bird-Model-43-

I'll bet some smart ham or CBer has hacked slugs.     

[xrunner]

Up to 1260 MHz, didn't know they made any that go that high.

Forgot Bird stuff Joe, we'll put 'em out of business with the power meters we're making. 

[bd139]

Just put "American made" on a pre-cut Hammond box, then pay some Mexicans to stick a Chinese board in it. Job done. Bird gone.

[joeqsmith]

Here are some photos of what they look like inside.  Looks like a lot of labor.
http://www.repeater-builder.com/projects/bird-element-tour/bird-element-tour.html

Moving beyond the 100mW walkie talkie test, I swapped out the 2W attenuator for the 100W.  40dB total on both legs. 0dBm is roughly 10Watts.   I'm just doing a point slope correction.  Using a polynomial would be a better choice.   Still, without some sort of standards, I don't see a point.   We've all fitted data before...   

When looking at the error,  again I am looking at the difference of the RF generator and the measured signals.  There are two amplifier gain stages and roughly 40dB of attenuation between them.  The 11dB offset is to be expected.  More of interest is how the two measurements track. 

[joeqsmith]

I suspect this is far too complex for the ham group.   There's software that would need to be installed and manuals that would need to be read.   The mechanical Watt meter is much better suited.  Even if the owner has no clue about accuracy, percent of full scale, load mismatches.....  They should be able to see the needle bounce. 

https://birdrf.com/Products/Sensors/RF-Power-Sensors/RF-Power-Sensors/5017D-Wideband-Power-Sensor.aspx

[xrunner]

I re-took some data because I had worked on the board and also replaced some capacitors. The plots still look very nice and linear. Then I did a little data analysis to help me decide a few things about what I really want to do with this power meter.

First I'm defining my project to go from -70 dBm to +13 dBm (without an external attenuator) but of course can be used with one. Next what I wanted to look at was the measured power errors and how the errors are distributed over the frequency range from 200 to 500 MHz. I plotted curves for dBm error over the defined power range for 200, 300, 400, and 500 MHz. I also calculated the mean error over each power range for each of these frequencies.

It's easy to see how the errors creep up higher as the frequency is increased. For example if you added the mean error back to the power calculation for a given band of frequencies centered around where the mean was taken, how big could that band be before you would want to define the next correction band? Say you wanted an accuracy of +/- 1 dB, how big could the frequency band be using a mean error correction before you'd have to define the next band with another correction factor?

I tried this method for the range 300 to 400 MHz, the correction factor came out to be 6.1 added to the regular equation :

Pin = ((Vo - 2.28) / 0.024) + 6.1

The actual input power was -30 dBm, calculated power plus correction factor shown below

300 MHz  -28.1 dBm
350 MHz  -29.3 dBm
400 MHz  -30.6 dBm

Of course in that range there were certainly corrected errors that were less than 0.5 dB. That's not too bad. I'd say +/- 1 dB error for this thing is pretty good. The Agilent 8648A signal generator doesn't have an output accuracy better than +/- 1 dB. Haven't decided on anything yet just thinking, that's all right now.