Linear Microstrip Patch Array -- Need Guidance 1

[jediknight7]

I am new to antenna design and I am designing an 8x8 microstrip patch 4.9GHz array. Unfortunately, my first prototype results are dissapointing. I am only seeing about 8dBi of gain and my target is 23dBi in the 4.4-4.9GHz band.

Can anyone provide some insight on how they attack development? How much do you rely on simluation vs. testing/debugging? Do you fab out different versions of a starting design with slight variations?

I simulated one row 8x1 in Hfss and it looked really good -- about 15dBi and 1.5:1 VSWR

Also, I was wondering if anyone has experience with PCB prototype milling machines. Are they worth getting for speeding up development? Has anyone here used them before?

For antenna measurement, we are using a system from Diamond Engineering; seems to be very slick so far.

 

[MacGyverS2000]

So long as you get an endmill, proto machines are great if you do not know what the circuit should be like... antennas start to approach the black arts at those frequencies, so the ability to control conductor spacing/width is a huge help.

Your only other option is to lay down a number of small-change iterations and have a board house etch all of them. If none fit the desired profile, figure out what the next set of iterations should look like.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Comcokid]

Proto milling machines work very well. However, it takes time to learn how to use them mechanically (mill bit selection, setup, tricks and techniques). Also, it takes time learning the in-and-out of the Gerber conversion software. Don't expect immediate results.

There are some outside PCB houses that are focused on milling one or two PCBs as a design service. Don't have a name off-hand.

 

[MacGyverS2000]

SparkFun Electronics has a deal going with Gold Phoenix PCB.. I want to say it's $2/in², which is perfect for small antennas. There are others, but SparkFun was the first that popped into my head.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[jediknight7]

Thanks guys.

Another quesiton -- does anyone know a patch array expert? I am working as a temporary solution, but I am afraid of letting my team down and I am looking for some help.


Thanks.
Robert

 

[Higgler]

Yes. I work with one real patch antenna expert and I've had alot of experience in patch arrays. We're actually helping out a company right now with a similar problem in a 2x4 element array.

Normally in an array, as the number of patch elements increases, the smaller the patches has to become to maintain the correct operating frequency (due to mutual coupling). i.e. a single patch antenna is large, a 4 element array has mid sized patches and a 64 element array has even phsically smaller patches, all to maintain the same frequency. At around 100+ elements (depending on antenna to antenna spacings) your patch size settles down to a final smallest size.

If you pretested your power divider circuitry and it's good, then your antennas may have shifted low in frequency. If you did swept gain at boresight, and adjusted gain for known power divider losses, you may see the antenna frequency has shifted low when put in the array.

If you used thin dielectric and can handle thicker dielectric, thicker equals better bandwidth and it's easier to have good gain at your desired frequency, even if you're off a bit in dimensions.

You can almost get your gain with a 16 element patch. You can space the elements out towards one wavelength spacing and if the patches are made with air dielectric you'd minimize your transmission loss and get around 22 dB peak gain. Orienting the elements slightly tilted in a recent antennas and propagations article in a 16 element array showed you can optimize gain by cancelling crosspole at further out angles.

We have a high end milling machine ($12K) and use it alot for prototypes. Works ok, but cost of the bits is equal to the cost of having antennas etches we've found. For higher volume we usually use E-Fab and ProtoExpress.

khiggins at toyon dot com

kch

 

[SamuraiMike]

No offense meant, but that is really something to have that poor a performance. There is something crazy wrong.

Are you sure your antenna measurement system is working correctly? Sounds like you might be working in an outside environment? Maybe you're getting a multipath which is completely out of phase and destructive? Do the patterns look like your simulations? Do you have a network analyzer? What is your standard gain reference for dBi? I wonder if your connector transitions are working properly? What is your VSWR? Element spacing? Maybe a description of your substrate? How much have you tested the individual components? Antenna element, power divider, etc? Are you routing the power divider on the same layer as the elements? If yes, that could kill everything, if not, your transitions from the power divider to the elements could be the problem? That frequency is starting to get a bit more magic as was stated earlier. Is this only looking broadside or steered?

I think someone with a bit of experience could look at your design and see what the problem is pretty quick? We build an 8x8 array at a much lower frequency which produces that type of gain and is steered.



mikecanned at hotmail dot com

 

[biff44]

As suggested, I would probably test the power splitter circuit first, as it is easy to test if you put connectors on it and it has a 50-50 chance of being the problem.

 

[Russell67]

If you simulated in HFSS and it looks good how does the computer simulation compare to the range measurements of the same thing? Remember Engineering 101, make one change at a time. If you simulated 1 in HFSS why would not simulate the whole thing. Sounds like you may have some phasing issues with your feed network. How did you join the 8 linear arrays together? Are you sure you are not looking at the crosspole? What do your side lobes look like?