| ??? 03/15/07 02:28 Modified: 03/15/07 02:40 Read: times |
#135022 - Almost total agree... Responding to: ???'s previous message |
Richard said:
Lots of mistakes, e.g. bundling of wires, use of pre-cut wires, etc, have often given wire-wrap a bad reputation. Yes, especially this dumb bundling was the death of many fast digital circuits. "But I only wanted it to look nice..." Siemens sold this "Fädeltechnik" (thread-technic, looking like "diet" wire-wrapping) in the 80', you might remember. Bloody hell, what a crap!! They praised their special bundling combs... And, no, no solid ground plane... Richard said:
The fact that wires intersect randomly has an averaging effect on crosstalk, which is minimized. This is right, to some degree. But please take note, that fast analog chips don't like capacitive stray coupling from output to input at all. Here wire wrapping would result in a desaster, wherefore certain MAXIM datasheets warn against the use of wire-wrapping. Ok, the master prevents all trouble by avoiding any stray capacitance between input and output. But the best idea is to have all signal traces running directly over a solid ground plane. This heavily decreases the inter-signal stray capacitances, as can be estimated from this picture: 
Richard said:
Wire-wrap allows for higher density than two-layer boards, with shorter signal paths than 4-layer PCB's would allow. That results in better signal quality and reduced HF power-gnd noise. Forgive me, Richard, but I doubt this. If you analyze the EM radiation performance of 2-layer boards and compare to 4-layer boards, a big difference can be found, even if both contain a solid ground plane. The reason for this is, that the distance between copper traces and ground plane for the 2-layer is three times greater than with the 4-layer board. This underlines the importance of having undisturbed transmission lines formed by the copper traces and the underlaying ground plane. When using wire-wrapping just this is the problem: Most signal wires see a much bigger distance from the solid ground plane, and when routing them criss-cross over the board only the least of them "see" the ground pane at all. The consequence is, that most signal traces have an unfavourable low capacitance to the shielding ground plane and a much too high inductance, and, because SQRT(L/C) is the characteristic impedance of resulting transmission line, an impedance mismatch to the output impedance of signal driver outputs occur. Not routing signal traces over a solid ground plane, but routing them criss-cross over the board results in heavily increased transmission line impedances and makes it much harder to achieve series termination. Impedance mismatch results in ringing and increased ground bounce. So, by wire-wrapping Hf power-gnd noise will just increase, to my understanding. Of course, a master will prevent all this by inserting suited series resistors to certain critical lines. Richard said:
Most of the original 805x boards I bought back in the '70's and '80's worked just fine, despite the fact they were 2-layer boards. Of those, most are still in service, running machines and doing what they were intended to do. And to be honestly, many of them failed and must be redesigned, which was my job for some years... Richard, we both are capable to fabricate working boards even made of wood. I totally agree with you, that wire-wrapping has worked, does work and will work. But only in the hands of skilled people. I think an unskilled person will be more successful when using a 4-layer board than when using wire-wrapping. But that's my personal opinion. Kai |
