Advice
Beginners
BIOS Guide
CPUs
Links
Mainboards
Memory
Network
Storage
Video/Sound Cards
Contact
Forum
SiteMap
Sponsors
Home
PC Hardware from Made-in-China.com
Prices:
|
Advice Contact Forum SiteMap Sponsors Home PC Hardware from Made-in-China.com |
. | |
|
Prices: |
LOSTCIRCUITS |
||
| ATA Not-So-FAQs Or: Why Ribbon Cables are unsuitable for RF transmission of data |
 |
|
|
(Article by snn47, March 17, 2003) -.. ..-. ..... .--. -.-- | ||
Real System Problems
The general problems of a signal bus have not changed since Z80 and 6502 CPU (1-2MHz) systems.
1. Any pulsed signal requires a defined signal-form, otherwise it can't be detected e.g. defined min-./max- signal levels as well as a minimum-signal, -length, -rise- and -decay-time.
2. A constant impedance is a must for any RF wirebound Duplex RF-transmission-system. Such a system consists of a transmitter (controller), transmission-line (flat-ribbon-cable) and receiver (drive-electronics). Any part from start to end should have identical impedance. Aside from transmission-line problems of the cable the driver-/receiver-termination-impedance normally different between high/low state so please don't ask about adding more slave drives in between.
If the impedance is not constant across the entire bus or if termination resistance is not correct, part of the signal may be reflected on the bus. These reflected signals will add/cancel the wanted pulses and degrade signal quality and detectability in almost all cases.
A passive termination will require a lot of power for the termination resistors, while an active termination has impedance problems.
Since a system bus will most of the times be designed for the best cost benefit (= it will just work in mass production) but may not provide a high safety margin.
3. If more then one signal line is run unshielded in parallel, another negative effect that degrades signal quality is coupling between parallel signal lines. Like for the reflected pulses, coupled signals from parallel lines will add/cancel the wanted pulses and degrade signal quality and detectability.
Unshielded signal lines cannot be run in parallel without adding at least a ground wire to "isolate" both.
A simple solution is the use of twisted pairs of wires as they are used for SCSI. It improves cross-coupling, but lower impedance and increase signal delay. SCSI systems require the termination by the last device on a bus and that any powerless device will not load the bus.
4. The introduction of the 80wire flat-ribbon-cable required a switch in the connectors to provide ground contact for the added 40 lines. While the 40-pin-connector remains at the front we have a 80-knife solution at the back, where we have a 80 wire flat-ribbon-cable pressed onto 80-knives.
5. By adding the third connector for the slave drive to the cable we connect all ground additional wires in parallel which will interrupt the impedance at that point. I don't know how good/bad the controller/HDD system terminates this contraption if the slave drive connector is added and a slave drive connected as well.
6. If the cable is run above metal the electronic properties change we'll have another jump/change in impedance.
7. Different dimensions/materials of the flat-ribbon-cable will change the impedance and signal velocity.
8. Impedance of controller and drive-electronic will change from model to model and have different transmitter and terminations values.
9. Drive, data and signal lines have different data rates.
10. If a drive bay is used then another jump in impedance is added by the connector/cable between the HDD and the cable to the back of the slide-in-bay and the connector at the fixed drive-bay which will have to connected to the ATA-cable somehow.
The faster the system get's, the more sensitive it will be to any change in cross coupling /impedance/ propagation velocity.
Since not to many parameter are defined it could be many things that make your system (un-)stable.
next page: => Conclusion =>
