ADF4351 signal generator

stephenkearney

New Member
I received the ADF4351 signal generator board yesterday and have been playing around with it. Doing antenna return loss measurements works okay but I haven't figured out a good way to sweep frequencies. The Analog Devices app does have a sweep function but it appears the 4351 only outputs a very narrow bandwidth signal and it needs ~1 sec. to step to the next frequency increment. I'm still looking for a good way to get that slow sweep to sync with the RTLSDR spectrum analyzer. For now it seems like a better to use SDR# to check return loss at the desired frequency. If it's acceptable (e.g. -10dB) then do some spot checking of the return loss at a few points above and below the target frequency.

I did find that the 1090 MHz monopole I made from music wire is terrible: -1.5 dB return loss or a 71% loss of power :(

BTW the 4351 board came with a bad USB cable. If you have trouble getting Windows to recognize the USB device that's likely the problem. I wasted an hour fiddling with it before trying a new cable.
 
Last edited:
Sounds like you are starting to have some fun..
I haven't got mine yet. But, I've been thinking of some settings.

upload_2014-7-6_20-45-51.png

This should sweep 2 MHz pretty quick.. I was going to use SDR# to look at it, but it runs so SLOW on my old XP systems..
Found HDSDR and that seems to run full speed.

Cheers,
Rich

PS:
When using a spectrum analyzer that's looking at the band you want, you don't really need
the sweep generator to be in Sync with the display.
You just need a fast scanning spectrum analyzer, and set the pause or dwell time of the
sweep generator, so you can see the scan.
A max-peak-hold function is a good thing to have in a spectrum analyzer.
If you have that, you don't need a fast sweeper speed. Slow works too.
 
Last edited:
This should sweep 2 MHz pretty quick.

That's the thing... it doesn't sweep quickly. Each step (0.1 MHz in your settings) takes a few seconds to do. Like you, I expected it to be fast.... but it's not.

When using a spectrum analyzer that's looking at the band you want, you don't really need
the sweep generator to be in Sync with the display.
You just need a fast scanning spectrum analyzer, and set the pause or dwell time of the
sweep generator, so you can see the scan.
A max-peak-hold function is a good thing to have in a spectrum analyzer.
If you have that, you don't need a fast sweeper speed. Slow works too.

I should have explained better. I understand how it's supposed to work, as seen in the How to measure return loss of a 978 MHz antenna video. The problem is both the generator's sweep and the spectrum analyzer's scan are slow. Since both are so slow it takes a long time for all the points in the sweep to be scanned. It's nothing like the fast scan a "real" spectrum analyzer does.
 
Humm, sounds like something isn't working correctly.. I wonder if the gen board has NV memory, and can be re-booted without the USB sending commands?
I've read about that feature in some other 4351 eval boards..

Is that Delay Time (ms) in Mil-secs or micro-secs? I'm sure there must be a min-max limitation to the delay time..
If you exceed those times, perhaps it defaults to slow-mo?

Or, could it be running so fast, you can't see it with your SDR display?

Maybe try just sweeping a small segment of spectrum around a station you can hear on a radio?
Weather channel (162.5mhz FM) or maybe an FM radio station? 93.7 maybe? then sweep from 93.6 to 93.8 Mhz..?.
Using small steps, like 0.01 Mhz..

If mine comes in and seems slow too, I'll set it for 50 to 53 Mhz and plug it into my 100Mhz analog O-scope..
That way, I can tweak the settings and see the results instantly and clearly..

I just looked at the tracking.. This might take a while longer, since I'm on the east coast..
Processed Through Sort Facility
Jul-03-14, 07:19 AM, GUANGZHOU EMS
Acceptance
Jul-01-14, 19:37 PM
 
Last edited:
This morning I did some testing with HDSDR centered on 435 MHz.

I think the HDSDR sample rate is set to Max. I believe it's ~ 2 MHz.
It was only using about 30% of my old single core Pentium processor power. (running XP).
(Unlike SDR# which runs the CPU at 100%).

The spectrum plot HDSDR is real fast and doesn't freeze up at all.
I used a handheld radio and transmitted very short bursts on (approx) 434, 435 & 436 Mhz.
Each time I hit the x-mit button, for a 1/10 of a second, I would see a big peak shoot up on the display..
I'm pretty sure that a 2 to 5 second sig-gen sweep will be very visible on the the display.

I know that HDSDR is only about 3 MHz wide, but I think that might be wide enough to do some useful work.
Just need to get my new generator!
 
I have no problems with SDR# except the ~3 MHz spectrum display (which is actually a limitation of the RTL dongle). A return loss scan for a 1 GHz antenna would typically cover 50 MHz of bandwidth and with a 3 MHz window it's hard to do :)

That's why I was trying to use RTLSDR Scanner -- it attempts to get around the dongle's 3 Mhz bandwidth by doing sequential scans. It works pretty well except for the speed.

At this point I'm using SDR# almost exclusively for return loss testing and only using RTLSDR Scanner for very limited scans. In SDR# I check the return loss at the target frequency (e.g. 1090 MHz) and then spot check at 10 and 20 MHz above and below the target.

BTW I'm using 40 dB of attenuation on the output of the ADF4351 before it goes into the dongle. Using the level settings in the Analog Devices app and SDR# I can set the reference level to close to zero in SDR#.
 
Some interesting results from my return loss testing:
  1. To get a lower physical profile I originally soldered the coax to the base of the antenna instead of using a BNC connector. That break in the coax shield resulted in a significant return loss. Now I'm making all connections with proper connectors.
  2. A ground plane made of radials angled at 30-40 degrees down has a slightly better return loss (3-5 dB) than horizontal ground plane radials.
  3. A cone ground plane had a much better return loss compared to the ground plane made of four radials.
This 4 radial ground plane resulted in a ~9 dB return loss (87% of the power radiated):
monopole with 4 radials GP.JPG


This cone ground plane resulted in a ~19 dB return loss (98% of the power radiated):
monopole with cone GP.JPG
 
Last edited:
I have no problems with SDR# except the ~3 MHz spectrum display (which is actually a limitation of the RTL dongle). A return loss scan for a 1 GHz antenna would typically cover 50 MHz of bandwidth and with a 3 MHz window it's hard to do :)

That's why I was trying to use RTLSDR Scanner -- it attempts to get around the dongle's 3 Mhz bandwidth by doing sequential scans. It works pretty well except for the speed.

At this point I'm using SDR# almost exclusively for return loss testing and only using RTLSDR Scanner for very limited scans. In SDR# I check the return loss at the target frequency (e.g. 1090 MHz) and then spot check at 10 and 20 MHz above and below the target.

BTW I'm using 40 dB of attenuation on the output of the ADF4351 before it goes into the dongle. Using the level settings in the Analog Devices app and SDR# I can set the reference level to close to zero in SDR#.

I too would like a wider BW, like 100 MHz for doing 1090 antenna work. But, doing fine tuning on resonate components, 3 MHz might be good enough.
 
Wow, -19 dB of return loss is a professional grade antenna.. That cone design is good at 1090, since it's going to have a small wind-loading area.. Be too big at VHF.. Maybe okay at UHF..

You seem to be getting into testing nicely. Might be a good idea to make a video
or post a step-by-step procedure for doing return-loss on an antenna, for novice DIYers..
Or people like me, who don't want to learn the theory, but end up with a great antenna.. ;)


Did you see this yet? Is this what you are running?
 
I think that scanner plug-in is neat but it doesn't have the spectrum analysis features that RTLSDR Scanner has.

The return loss measurements are pretty easy. Basically I followed the directions in the How to measure return loss of a 978 MHz antenna video. Except I used the ADF4351 for the signal generator and the dongle + SDR# for the spectrum analyzer. Of course it doesn't work quite as well and requires more effort compared to a commercial unit like the $1500 Rigol DSA815 Spectrum Analyser. But it's good enough for basic measurements :p

Here's a return loss scan I did with RTLSDR Scanner before I decided to do simple (and faster) measurements in SDR#. The signal generator was set to sweep with 10 MHz steps. Note the dip at 1090 MHz to around -12 dB.
Return loss.png
 
I've been madly testing different ground plane designs; it's pretty easy to construct them out of paper with a layer of aluminum foil. Or thin aluminum sheet from cut up beer cans ;). However I've had trouble getting consistent results. I suppose it's due to the tiny values we're dealing with when the return loss is down around -15 dB. I'll get a measurement of -25 dB but when I reset everything and repeat the test it's -15 dB :mad:. Finally I've come to the conclusion that the wire radial GP as seen in How to measure return loss of a 978 MHz antenna works very well, is simple to make, and reasonably sturdy. One thing I noticed is that the radials' downward angle seems to work better than orienting them perpendicular to the radiating element.
 
Last edited:
You will get inconsistent results when working at microwave frequencies.
The higher you go, the (shorter wave-length) makes for more problems.
At 20 Ghz, you should to use a calibrated torque wrench on your SMA connections. :eek:

If you can consistently get -15 dB return loss (or better), your antenna (VSWR 1.433 : 1) is working a lot better than many factory made $$$ antennas.

I've seen a lot of down-angled radials on smaller GP antennas. Must mean someone had some test equipment.. :p
The VHF antennas that I've seen with perpendicular radials weren't 1/4 wave antennas.
I think 5/8th wave verticals are the most common.. They are base-loaded with a coil for impedance matching.
I've used 5/8th UHF & VHF mobile antennas, where the car's roof is the perpendicular ground plane.

http://vk6ysf.com/5-8_Ground_Plane.htm Looks like a mobile on a simulated car roof.. :D

5-8_Ground_Plane_image10.jpg
 
Hey! my 4351 sig-gen was in the mailbox early this morning!

I used a quick and dirty return loss test on the mystery 915 antenna that I want to hack.
It's around -18 to -20 dB down between 917 & 918 MHz.
I always wondered why it never worked very well at 911 MHz when it was used to transmit Ham TV (ATV).

My simple test rig used a diode detector on the directional coupler tap. I piped the detector DC right to a DVM.

This equipment is going to make building antennas a lot easier! :D
 
I didn't pay much attention to 911, I started off stepping 1 MHz from 900 to 930, just to getting into the ball park.
When I saw the DVM blinking zero and .1 mV, between 917 & 918 MHz, I knew that was the sweet spot.
This is a pretty narrow banded antenna*. I did take readings 1 MHz above and below the sweet spot.
IIRC, those readings were about -12 dB.

I did an open connector test, using an open N-to-BNC adapter on the end of my 15' test cable to see full return voltage
while stepping 900 to 930, just to make sure the sig-gen had a stable output level. It's not perfect, but pretty good.

To finish up, I scanned 900 to 930 with 0.20 MHz steps and watched for any strange peaks or nulls in the antenna.
There were none.. Just the deep valley between 917 & 918 where I expected to see it.

*narrow banded antenna
This antenna has a LOT of coco elements. It's about 7 feet tall !
Plus, it has resonate filter sections at the first element joint.
http://forum.planefinder.net/threads/915-mhz-coaxial-collinear-conversion-to-1090-mhz-project.160/
 
The signal generator is working okay. Not as fast as I would like, but it does a pretty good job.

I decided to mount the parts on a PVC bread-board. Just to keep anything from getting pulled around on the bench (or falling on the floor) and shorting out.
There's a 50 ohm SMA terminator installed on the SigGen's un-used output. (Per recommendation from a review I read somewhere).

breadboard_zps0d77ae93.jpg


http://www.bellnw.com/manufacturer/Narda/3002.htm

There is also an SMA 3dB attenuator pad installed in the SigGen output line. I will be tweaking that pad to find the best setting that's usable.
I wanted an LNA in to give more drive to the detector diode, so I can reduce the amount of power going to the antenna being tested.
Bottom line, I don't want the SigGen to cause any RFI.
During testing, (in the basement) I monitor the 1090 Puck and use VRS with ModeSdeco to make sure I'm not jamming my own ADS-B radios.

The UTO-520 is sold as a 500 MHz amp, but can be used up to 1100 MHz without much lost in gain.
I'm not too worried about the VSWR, being a bit high at 1090, since it's a constant and will
cause the same loss into the diode in both antenna test mode and full reflect mode (short or open).

UTO-520
upload_2014-7-19_14-54-43.png
 
I am interested in buying one. I found four (4) different types. Dont know which one to purchase.

Any advice?

Which one is better, and why?


Type 1:
ADF4351 development board × 1
CY7C68013A base board× 1

http://m.ebay.com/itm/231266464162

Double board ADF4351 + CY7C68013A.jpg




Type 2:
1PC ADF4351 development board
1PC High Speed USB2.0 Dashboard

http://m.ebay.com/itm/261532585416

Single board ADF4351.jpg



Type 3:
1 X 35M-4.4G signal source signal generator/simple spectrum.
1 x USB Cable.
2 X SMA cable.

http://m.ebay.com/itm/121372341978

Signal Generator Enclosed.jpg



Type 4:
ADF4350 development board 0.137-4.4G 137M-4.4G signal generator


http://m.ebay.com/itm/111460880624


Type 4 signal generator.jpg


 
Last edited:
Mine is the Type 1: I liked it because it's a newer version chip.. https://ez.analog.com/thread/9581
But, for antenna testing at one frequency, or a small range of frequencies, the ADF4351 and ADF4350 there isn't likely to be much difference.
The improvements in signal purity in the ADF4351 means it could be better if you used it as an LO in a receiver..

Type 2: Seems to be the same package as Type 1.. They just don't show the USB board, which I assume is listed as the High Speed USB2.0 Dashboard..?.
include:
1PC ADF4351 development board
1PC High Speed USB2.0 Dashboard
1PC Software

I can't tell what kind of chip is inside Type 3, but I like the metal case and the USB socket (any old USB printer will work).
I do not understand what that SMA Input is. All of these Eval boards have an RF input, in case you don't want to us the on-board X1 Crystal.
IIRC, a 25 mHz clock signal can be fed into that Input, if you need an extra stable signal. (Assuming your 25 mHz source is super stable).
It seems the Type 3 only has one RF output. If you look at the open boards in the pics, you will see the all have two RF outputs. (They are 180 degrees out of phase).
So, you have an extra output, if you ever want to drive two devices... Or build a receiver that uses this type of LO (180 out of phase).
That extra output is not used on my test rig.. I put a 5o ohm terminator on it. (See my pic above).

The Type 4 is nice, because of it's single PCB configuration. USB right there, ready to go. But it's a 4350.
It has a 5vdc input socket, but that's not going to be needed, since the USB will provide power.


For an antenna test rig like mine, I think any of these would work.
I do like the ADF4351 because it's output is pretty clean. I might use it in a radio someday.
Type 3 looks like it's less prone to breakage or ESD death.
Plus, it's processor noise isn't going to radiated all over the area. Maybe interfere with your FM radio or wifi..?.

If I was replacing my unit.. I would try a Type 3..
 
Thank you Richard for the prompt & detailed reply. I think I will go for type 3, though type 1 is very tempting due to ADF4351.

What other piece of equipment I will need for antenna testing? A directional Coupler? Anythig else?
 
Back
Top