joltes@husc9.harvard.edu (Richard Joltes) (04/05/91)
We've been looking at ways of measuring dB loss along UTP wiring using a method we found in our Tektronix 1503B TDR manual, and considering our current findings I'm wondering if anyone uses this method successfully. What's making me even more skeptical is that 2 different engineers at Tek have given us two different methods and they are both convinced they're right! Personally, I'm skeptical whether it's a valid instrument for testing Freq. related variables anyway, but if anyone has tried this out please let me know via e-mail. If anyone's interested in the results of this survey let me know and I'll summarize if there's interest. Dick Joltes joltes@husc9.harvard.edu Hardware & Networking Manager, Computer Services joltes@husc9.bitnet Harvard University Science Center
jrd@cc.usu.edu (04/07/91)
In article <1991Apr4.163415.420@husc3.harvard.edu>, joltes@husc9.harvard.edu (Richard Joltes) writes: > We've been looking at ways of measuring dB loss along UTP wiring using a > method we found in our Tektronix 1503B TDR manual, and considering our current > findings I'm wondering if anyone uses this method successfully. What's making > me even more skeptical is that 2 different engineers at Tek have given us two > different methods and they are both convinced they're right! > > Personally, I'm skeptical whether it's a valid instrument for testing Freq. > related variables anyway, but if anyone has tried this out please let me know > via e-mail. If anyone's interested in the results of this survey let me know > and I'll summarize if there's interest. > > Dick Joltes joltes@husc9.harvard.edu > Hardware & Networking Manager, Computer Services joltes@husc9.bitnet > Harvard University Science Center ------------- Dick, Since you are not letting any cats out of the bag then I wouldn't expect many responses. Here is the simple, clear, foolproof solution: put a scope across one end of the twisted pair system, right next to a real load. Observe waveform quality and strength. Step two is a little tougher but incisive: put an attenuator between the real load and the wire. You should expect an attenuation margin of up to 10dB on short wires. But this is twisted pair wiring where pulse risetimes can be horribly messed up by coupling and impedance variations and so on (hence the scope). Coax is pretty much attenuation bound (assuming good connectors and same-reel wiring). TDRs, especially the really nifty ones, use extremely short pulses and consequently very wide bandwidths. The Ethernet board is not such a good receiver, but it's the only one which matters. That makes pure TDR observations something of an art. If in doubt consult a competent EE for some insight. Btw, when applying scope to twisted pair one wire goes to each channel of the scope and run the scope in differential mode to avoid grounding either wire. Attenuators are tougher to obtain because twisted pair wiring has an impedance in the neighborhood (a BIG neighborhood) of 120 ohms; best to make these from a handfull of resistors on the spot. Your problems are very likely not attenuation at all (twisted pair is naturally low loss) but rather from diffuse reflections caused by objects adjacent to the pairs. Joe D.