Dear Si-List, dear Jan, I would like to comment on the issue on radiation from bends. Pleas find my statements inserted in your text: > I have done TDR measurements and the conclusion is, as everybody elses > conclusion, that the bends are very hard to see, except at very high frequencies. > The radiation from bends is harder but it has been studied numerically and > has been measured! Even if "high frequencies" is very relative I agree. I think it can be safely stated that radiation is a third order effect, whereas the discontinuity in ch. impedance is second order (i.e. setting in at much lower frequencies). For designers this means that it will be quite safe to approximate bend discontinuities by purely reactive networks (e.g. the L-C-L T-network) for all frequencies of interest. > results have been reported in the IEICE transactions (Japanese IEEE) and > later in the IEEE trans on EMC. The > paper in the IEICE is publically available: > http://search.ieice.org/2001/pdf/e84-b_9_2604.pdf I have not read the paper (I will though) but I have done some FDTD simulations on bends which might be interesting for you: Christian Schuster and Wolfgang Fichtner, "Parasitic Modes on Printed Circuit Boards and Their Effects on EMC and Signal Integrity", IEEE Tr. EMC, vol. 43, no. 4, pp. 416-425, November 2001. The paper discusses bends only in the context of surface wave excitation so it will not really cover all aspects of the phenomenon. However, what I found was the following: TM0 surface waves have zero cutoff on grounded dielectric planes and may be therefore present for all frequencies. However, for "low frequencies" surface waves behave as a "quasi plane wave" with most of their energy in the air above the substrate. For "high frequencies" the waves, i.e. their energy becomes "trapped" in the substrate and their behavior is changed (essentially they "slow down" and they are sometimes called slow-wave modes then). In this regime bends can radiate very effectively into surface waves increasing radiation losses drastically. For a 0.6 mm microstrip on 0.6 mm alumina this becomes true for frequencies >= 50 GHz (well beyond what usually is considered ). The bad thing about is that the energy once injected into surface waves remains in the substrate (which should then be better considered as a sort of waveguide) and can couple nicely into nearby traces. The good thing about is that the critical frequencies are usually "relatively high" and that you can push them even higher by using lower DC's or thinner subtrates. > Because most microstrip lines have very small h/lambda the radiation from bends is very small. Actually the > increase for practical tracks on PCBs is negligible and likely less than 1% (or even smaller), even for high > frequencies. I can confirm that h/lambda is an important parameter (h = substrate height). In terms of h/lambda I found a factor of about 1/3 or 1/4 critical, i.e. in my simulations an h >= lambda/4 gave me radiation into surface waves. > To make things easier assume that the radiated power is emitted isotropically than the > far-field electric field is proportional to square root of the radiated power. So a bend produces an > electric field that is about 5% higher or only 0.4dB. Again, for most practical stuff the 10% used here > is much to high. Although my FDTD simulations included free-space radiation I cannot make any conclusions about their importance or behavior. Likewise I have no idea how the publication mentioned deals with surface-wave radiation. However, there are publications which sort free-space and surface-wave radiation out.. I hope this is useful for you. Regards, Christian Schuster IBM ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu