12
\$\begingroup\$

I am reading a datasheet of a motor and I keep seeing wires crossing each other and crossing them again. (see the picture below). Does anyone know what this means?

enter image description here

EDIT:

I also see this picture in this datasheet the following picture (below) which is more like the situation I have. I want to control the motor with my microcontroller. Does this change the situation on the crossed wires in the schematic (So is it now still differential)? And can I just simply control the motor with a transistor as shown in the control block in the schematic?

enter image description here

\$\endgroup\$
  • 3
    \$\begingroup\$ Twisted pair used in current loops. \$\endgroup\$ – Optionparty Nov 6 '14 at 14:43
  • 2
    \$\begingroup\$ Can you link to the datasheet? \$\endgroup\$ – ACD Nov 6 '14 at 15:42
  • 1
    \$\begingroup\$ yes of course. orientalmotor.com/products/pdfs/2012-2013/A/usa_st_ar_dc.pdf \$\endgroup\$ – T J Nov 6 '14 at 19:05
  • 1
    \$\begingroup\$ it's on page A-159, I can't seem to edit comments after 5 minutes :S \$\endgroup\$ – T J Nov 6 '14 at 19:14
  • \$\begingroup\$ Yep, twisted pair. The twist helps prevent noise pickup and crosstalk. Standard ethernet cable (the type with RJ45 connections) is twisted pairs inside, eg. \$\endgroup\$ – Hot Licks Nov 7 '14 at 20:46
10
\$\begingroup\$

Several sets of differential pairs routed on the red layer of this PCB

It's a differential pair. I'm surprised at the strange answers and arguments regarding twisted pairs and LVDS and things of that sort.

Differential pairs are notated symbolically as twisted pairs because, at a macroscopic level, that's how they're carried (think Ethernet, HDMI, USB, Firewire -- all use twisted-pair cables). And visually, it describes the signals as interconnected, which they are (they reference each other, not ground, after all!)

At the board level, differential pairs carrying high-speed digital or analog signals -- depending on their operating frequency -- are often routed as transmission lines with controlled impedances (both with respect to ground and with respect to each other). Regardless of operating frequency, differential pairs carrying digital signals should be routed with equal length, to prevent the positive and negative signals from arriving at different times.

Many low-speed analog signals are carried differentially as well; the most common example being audio, which is often carried over XLR cables, which use a twisted pair. For these signals, this transmission scheme is often called a balanced pair (or some variation on that). The balanced receiver uses the difference between the signals -- not their actual values -- to carry information. Because of this, noise rejection is extremely high, since any noise affecting one of the signals will affect the other in the same way (and will eventually be subtracted out altogether by the receiver).

In these low-speed signals, the matched lengths and transmission-line routing is less critical. Most importantly, keep the signals very close together, and always route them identically, so that any interference that affects one will also affect the other.

In the above image, the red traces are each differential pairs.

To interface with differential pairs, a line driver is generally used, which is a circuit that converts single-ended signals into differential, and vice-versa. One such circuit designed for digital differential signals is the SP3485 (an example in use would be on this SparkFun breakout board), though there are many others like it.

\$\endgroup\$
  • \$\begingroup\$ Ok this is the first time in my life that I hear about differential pairs. Can you tell me a bit more about how to connect the driver to a 5V micro controller? I'm a bit lost here.. \$\endgroup\$ – T J Nov 7 '14 at 7:07
  • \$\begingroup\$ is it just switching with an transistor (PNP) with the microcontroller on the + line and connect ground to the - line as shown in the datasheet or should I do more to get it working? \$\endgroup\$ – T J Nov 7 '14 at 8:03
  • 1
    \$\begingroup\$ I updated my answer with a paragraph to explain some examples. What you're looking for is a line driver, or a differential transceiver IC. You'll need to figure out what the voltage levels are (5V, 3.3, or 1.8?), and then you'll be able to find something out there. Here's one such driver-only chip you could use: datasheets.maximintegrated.com/en/ds/MAX3293-MAX3295.pdf \$\endgroup\$ – Jay Carlson Nov 7 '14 at 18:12
  • \$\begingroup\$ thank you for the reaction. will look into it. I also updated my question, if I wire my micro controller outputs on the base of the transistor will it then also work as a differential signal? \$\endgroup\$ – T J Nov 7 '14 at 18:51
  • 1
    \$\begingroup\$ Sorry, I was trying to answer your question in a more general sense (to make it more useful to the community). Ironically, in your case, you don't need to think about differential signals at all. I would connect 32 and 36 to ground, and drive 31 and 35 directly from MCU pins. They're just running into an opto-coupler, so your MCU should be able to drive the pins directly. The reason they show the differential traces is in case you're doing long cable runs -- like, 20 ft or more -- you should use separate grounds to prevent cross-talk. Sorry my answer wasn't more specific to your question! \$\endgroup\$ – Jay Carlson Nov 8 '14 at 4:40
12
\$\begingroup\$

To me it looks lite a twisted pair. It must be a LVDS transmitter on the left.

\$\endgroup\$
  • \$\begingroup\$ My god, was it that simple? Thank you for the answer. Just a side question: What is a LVDS / what is it used for?? \$\endgroup\$ – T J Nov 6 '14 at 14:47
  • \$\begingroup\$ LVDS is low voltage differential signalling. Note that twisted pairs are used in a lot of places, like ethernet cat5e cable, telephone cable, field cabling in industrial instrumentation/control circuits, etc. Twisted pairs doesn't necessarily imply LVDS. \$\endgroup\$ – Li-aung Yip Nov 6 '14 at 14:49
  • 4
    \$\begingroup\$ It's electrical communication interface. One common usage is data ports of LCD panels where big amount of display data needs to be transmitted at very high speed. Big advantage of it is that data is transmitted through pair of conductors differentially. One conductor carries negative pulse, second conductor carries positive pulse. There's always some external noise sitting on top of those signals. At the receiving end you subtract negative signal from positive signal effectively eliminating noise and increasing signal amplitude. \$\endgroup\$ – Farid83 Nov 6 '14 at 14:59
5
\$\begingroup\$

In this context it means the output of the module on the left is differential (versus single ended output). In other words, the output level isn't referenced to a common ground but to each other, a complementary pair. To get the output value you subtract the two signals.

Differential pairs are less susceptible to noise. You can read more about it here.

I don't believe it means the wires are a twisted pair. Twisted pair is often used if transmitting complementary signals, but doesn't need to be, especially on a PCB. Also, twisted pair is often used to reduce magnetic noise on power supply cables.

EDIT with sources...I think the mob is wrong on this one.

In the OP's datasheet the signal is not twisted in equal parts.

enter image description here

If this was twisted pair you'd expected it to keep going in a uniform way. Googling "twisted pair schematic signal" and going to images confirms this.

enter image description here enter image description here

Notice they are equal groups. That's extended to the real world because when you make a twisted pair it's a constant and even twist. Notice the edges are also rounded. The top four hits in Google show this.

There are also no images in this Google search that show images like the OP's, implying that doesn't mean twisted pair. Generally if a wiring diagram wanted to show that twisted pairs were needed for interconnections it would be blatantly obvious and also labeled, just as the Google search and referenced images show.

Next, Google "differential signal" and go to images shows pictures like the following in the top results.

enter image description here enter image description here

If you follow those images to their sources, they are clearly using it to demonstrate differential signalling, NOT twisted pair. I'm a little disappointed in the community for just blindly downvoting an answer based on someone else's opinion with nothing to back it up, but oh well.

\$\endgroup\$
  • 5
    \$\begingroup\$ No, it's the twin outputs, one with inverting bubble that indicate differential outputs. The crossing wires would generally mean twisted pair. \$\endgroup\$ – Brian Drummond Nov 6 '14 at 14:53
  • 4
    \$\begingroup\$ @BrianDrummond Source on this? Otherwise it appears like a mob opinion to me. \$\endgroup\$ – ACD Nov 6 '14 at 15:12
  • 2
    \$\begingroup\$ There is no "mob" here, look carefully at how @BrianDrummond typically answers question, when you are calling one of the calm voices around here a mob leader it's probably time to take a break. \$\endgroup\$ – placeholder Nov 6 '14 at 16:45
  • 2
    \$\begingroup\$ @placeholder He states his opinion as a staunch fact, as if there's no possible chance he could be wrong. He is highly upvoted, I am highly downvoted. That is a mob mentality, synonymous with hivemind. I think you misunderstood me to be calling him rude, which I wasn't. \$\endgroup\$ – ACD Nov 6 '14 at 16:48
  • 4
    \$\begingroup\$ It's definitely a differential pair. The recommended connector for those signals is mouser.com/ProductDetail/3M-Electronic-Solutions-Division/… (according to the datasheet). The literature there says it can use ribbon cable, twisted pairs, or twin-ax, any of which could work depending on the application. Also in the data sheet, the comment: Use a multi-core, twisted-pair shielded wire of AWG28 to 24 for the control input/output signal line (CN5), and keep wiring as short as possible [within 2 m (6.6 ft.)]. \$\endgroup\$ – engineerC Nov 7 '14 at 4:05
4
\$\begingroup\$

I think it suggests "twisted pair", but does not positively imply it. Basically, the idea with twisting pairs of wires is that the amount of noise picked a wire loop will pick up from a given direction can be approximated by projecting the wires onto a plane which is perpendicular to that direction (imagine moving the wires perpendicular to the plane until they reach it), adding the total area of the loop in which a current flow in some particular direction would be clockwise, and subtract the total area in which the current flow would be counterclockwise.

If two wires in a cable are uniformly twisted many times over the length of each relatively-straight section, then projecting the cable onto a plane in any direction will result in clockwise and counter-clockwise areas that nearly cancel out. Twisted pair wiring won't eliminate 100% of noise pickup, but it will eliminate a lot.

My interpretation of the schematic is that because the wires are not shown as a twisted pair, twisted pair cable may not be essential, but if one happens to be using a cable that contains twisted pairs of wires, the indicated wires should be paired together within that cable.

\$\endgroup\$
0
\$\begingroup\$

My answer, too, is that it's a differential pair and that it carries no relevance how many "loops" you see in the schematic. If there were more "loops", it just would be more difficult to read - now you can trace e.g. the signal on pin 31 to the topmost pin of the differential driver on the left, so it's clear that pin 31 goes to the positive drive and pin 32 to the corresponding inverted drive.

My rationale is that when you generate a netlist from a schematic, the crossing of wires has no relevance - the same netlist is exported with or without the crossings. So the crossings are a visual courtesy of the person who draws the schematic to show that these signals form a differential pair. It makes it easier for the reader to see this "at a glance". Textual remarks in a description or in a datasheet would give you more precise information of how the signals should be carried physically. Schematic capture, when translated to a netlist, doesn't carry this information.

I would have posted this as a comment, but couldn't fit it nicely in the space available.

\$\endgroup\$

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.