3
\$\begingroup\$

As part of a product upgrade we are looking at replacing looms of wires bound together by cable ties or spiral wrap with ribbon cable. We are doing this to try and cut cost and assembly times.

We are happy with the current noise performance of the system, but ribbon cable always seems to be referred to as "Bad" in terms of noise and interference.

The signals consist of a few medium speed digital signals (4 MHz SSI interface) and predominantly +-10V differential analogue signals with a 2 kHz filter on the receiving end. All up we are planning on using a 50 Core cable.

The environment is lab/light industrial, the cable will be approx 300mm long running parallel thin steel plate.

We are looking at using the standard 1.27mm Pitch cable (2 row 2.56mm plugs).

Ultimately, while rated as bad, can I expect at least similar performance from a ribbon cable as bundled wires.

\$\endgroup\$
  • 1
    \$\begingroup\$ Why don't you try and build one - you'll get far better feedback by prototyping it. \$\endgroup\$ – Andy aka Dec 15 '15 at 9:44
  • \$\begingroup\$ I am working on the design, I just thought it was worthwhile asking first as it is much easier to design things in rather than hack them in later. \$\endgroup\$ – Hugoagogo Dec 15 '15 at 10:29
  • \$\begingroup\$ Design = come up with idea, try it out (test prototype), proper implementation (test again). I don't think getting SE feedback on something as hard to define as this is going to help much. Just my opinion based on countless years in the industry! \$\endgroup\$ – Andy aka Dec 15 '15 at 10:54
  • \$\begingroup\$ I unfortunately don't have countless years in the industry, yet. We all need to start somewhere. And yes lots of testing is planned. \$\endgroup\$ – Hugoagogo Dec 15 '15 at 10:57
2
\$\begingroup\$

If you compare a ribbon cable with some random wires, laced into into a bundle (which I presume is being done at the moment) then the differences would be described as 'good'. What you get is a well defined impedance between adjacent lines. IIRC, a gnd-sig-gnd triple on 50thou pitch is around 100\$\Omega\$, but you had better check! You can use the well defined ordering to impose further ground lines between signals and pairs that you do not want to cross-talk.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I went with this answer as while it is not as complete as KyranF's answer it more directly answers what relative difference can be expected, and why. \$\endgroup\$ – Hugoagogo Jan 28 '16 at 0:56
3
\$\begingroup\$

If the ribbon itself is mostly shielded by steel plate from the major sources of industrial plant noise, this will help.

If you have enough spare wires in the ribbon, you should separate each differential pair on each side with ground wire signals, and keep one side dedicated to the serial signals, while the other side is all of your analog pairs (again, make sure to separate unrelated pairs with spare ground wires).

You can actually get a shielded ribbon cable with twisted pairs already organized for you, which would be great for differential pair serial comms and for the analog signal pairs. You can get some at Digikey for example: here and datasheet here which are "3M" brand and you can get 30 metres for $1050. I'm sure buying direct in larger amounts would make a huge saving, as Digikey only offers these as "value add" which can be grossly priced. Not including tax, this makes it only $1 of cable for your 300mm you want per unit, for the full service of shielded twisted pairs in an easy to assemble ribbon cable, which can be easily IDC terminated and grounded on your PCB/chassis ground.

The capacitance is just as bad as similar multi-core cables I've seen, between 15-30pf/foot, and I've run gigabit ethernet over similar capacitance in twisted pairs in custom cables over 5 metres so i'm sure this will do whatever communications you want in addition to your analog signals.

With your signals being +-10V and such a low frequency filter on the other end, i'm sure un-shielded will still be fine, provided you try doing the signal types as far away from each other as you can, using blank ground wires between each different pair.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I have seen the availability of shielded cables, but ruled them out due to needing to have in the range of 3-6 connectors per cable. I forgot to mention this in the question, this also makes it pretty hard to use the twisted ribbon cables. \$\endgroup\$ – Hugoagogo Dec 15 '15 at 6:12
  • \$\begingroup\$ So along the 300mm you will be having 3-6 IDC connectors attached along it?? wow. \$\endgroup\$ – KyranF Dec 15 '15 at 6:15
  • \$\begingroup\$ the twisted pairs are just a handy feature, but if it makes it too difficult then go for the normal ribbon and space signals accordingly \$\endgroup\$ – KyranF Dec 15 '15 at 6:16
  • \$\begingroup\$ Yea, our board has to hook into an external product with a number of modules. The ribbon connects our main board and number of adapter boards that plug directly into their device. \$\endgroup\$ – Hugoagogo Dec 15 '15 at 6:20
  • \$\begingroup\$ Are the blank ground wires for the purpose of protecting against contamination between signals for from external influence. \$\endgroup\$ – Hugoagogo Dec 15 '15 at 6:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.