The primary reason that I went with this simulation package, is that I don't have the luxury of large spaces. I have described these applications as being "like building a pipe organ in a submarine". Like most products the antenna is hated by marketing and forced into an "afterthought" position. My job is to make it as good as it can be given the constraints. I don't (and likely never will) have the luxury of a quarter wave of free space. Some of my designs aren't that big in total. The antenna that I am having trouble with is several forms of dipole (This is an unusually large board for us), with varying balun structures to match them back to the 50 ohm feedpoint. The baluns are working in that the appropriate load reflects a good match back to the feed point, but when I add the antenna elements things get wild. I will usually calculate the lengths from free space numbers adjusted for the PCB dielectric, then set up a simulation to vary the lengths across a range that covers the calculated value. That will usually get me to resonance, then I adjust the feed structure to match the impedance, and add harmonic stub filters, then add in the enclosure and mechanicals and re-tune. This process takes about a full day of simulation time at 0.05 mm resolution using the Nvidia Titan card as the calculation engine. There is no feedline as such, there is a track between the balun and active element that is also part of the balun structure. Generally I can solve these match problems down to about 25dB return loss. That has yet to be verified by VNA, but informal testing shows more than double the range of the antenna designed by the company that produces the chips. I cant' give too much detail, it's all proprietary. I was just looking for anyone who has had similar issues to deal with. In this particular PCB, antennas that are usually easy to tune up, are falling apart as I adjust them through resonance. The return loss will drop from 25dB to 6 or so, then re form on the other side. I am theorizing that there is some other structure coupling in that is resonant in some destructive way. As I have time, I will simplify the structure on the PCB and re-sim. On Fri, Aug 1, 2014 at 1:27 AM, Vasile Surducan wrote: > David, > > Simulation will not help too much. A good simulation can be 50% away from > the crude reality and that one is good simulation! Workaround: > > - separate the antenna area to the rest of your circuit with an air gap i= n > the PCB. However this will not help if your antenna is too close from you= r > circuit, preserve at least a lambda/4 (better more) open space around you= r > antenna. You didn't say what type of antenna do you have... a 0dBi or a > -xdBi SMD antenna is actually not an antenna... > > - check if your ground area is below or intersect some active path of the > antenna, this usually will not help. > > - check the delay of you feeding line, on the antenna you should usually > have the max. RF voltage > > - is your antenna impedance matched with the PCB feeder and with the > amplifier output? > > > > > On Thu, Jul 31, 2014 at 6:44 PM, David VanHorn > wrote: > > > I am doing a lot of PCB trace antenna design lately, and I am seeing > > another occurence of something that I had seen before, where a given > > antenna design that should be workable, simply won't tune in a given > > design. > > > > Previously all I had to address this was renting some time on a VNA and > > hacking a physical prototype. > > > > Today I am working with XFdtd simulating the entire PCB, plus all the > > surrounding materials. That makes the tuning process one of time on a G= PU > > instead of rental lab time and X-acto knives. > > > > It seems that there is some other structure on the PCB that is resonati= ng > > in some way that interferes with the antenna. I've done some > > investigation with stitching vias and definitely things are a lot more > > stable with those in place, but it's not affecting the main problem tha= t > I > > am seeing. > > > > I am wondering if others have hit this, and if so what they did to > isolate > > and get rid of the problem? > > -- > > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > > View/change your membership options at > > http://mailman.mit.edu/mailman/listinfo/piclist > > > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .