Author Topic: Adjustable slew rate?  (Read 22136 times)

ruutboy

  • Member
  • ***
  • Posts: 19
Adjustable slew rate?
« on: November 24, 2013, 11:01:41 am »
I work as a Controls Engineer for GM, and we have a ton of ill-behaving DeviceNet networks. If it isn't the voltage giving us trouble, it's a "hi-z" connection (high being relative here), or in a recent episode, water being run *through* the panel. (Hard to blame DeviceNet on that, but getting the new ADN set up in RSNetworx was  a bear).

But one of design flaws of the system is the set slew rate of the RS485 chips. At 250k baud it doesn't need to be so stiff.

So why can't I put a tiny RC network at each transmitter to relax things a bit? Any thoughts?

Thanks!

Jan Axelson

  • Administrator
  • Frequent Contributor
  • *****
  • Posts: 3033
    • Lakeview Research
Re: Adjustable slew rate?
« Reply #1 on: November 24, 2013, 01:18:04 pm »
Maxim has chips with slower slew rates. See:

http://www.maximintegrated.com/app-notes/index.mvp/id/367

ruutboy

  • Member
  • ***
  • Posts: 19
Re: Adjustable slew rate?
« Reply #2 on: November 24, 2013, 04:57:41 pm »
Hi Jan,

Thanks for the response. Unfortunately these are already in equipment such as I/O Safety blocks (no modding these, the Safety Police will be after me), Turck (yes, spelled correctly. And embedded in epoxy, so likewise...), ADN modules (could be modded I suppose, but it would have to be a pin for pin match), Siemens SST blocks (could be done as well I suppose), and various VFDs.

According to my partner, we have over 300 DeviceNet networks. Each having anywhere from 2 to 20 blocks on them. A lot of chips... :)

DeviceNet, is on its way out (no doubt its incredibly poor performance and reliability had something to do with it). But we are stuck with them for the foreseeable future. We have already installed split resistor terminators with a cap tied to ground. This helped, but there is still a long ways to go. Coupled with that is less and less companies are supporting DeviceNet, so "store-bought" solutions are becoming rarer as well.

BTW, I've been a fan of yours since your "Circuit Craft" days. To every new Controls Engineer, I recommend two books. Yours, and "Bebop to the Boolean Boogie". :)

Jan Axelson

  • Administrator
  • Frequent Contributor
  • *****
  • Posts: 3033
    • Lakeview Research
Re: Adjustable slew rate?
« Reply #3 on: November 24, 2013, 05:45:38 pm »
Cool!

As you probably know, impedance-matched terminations should help if the problem is noise.

ruutboy

  • Member
  • ***
  • Posts: 19
Re: Adjustable slew rate?
« Reply #4 on: November 24, 2013, 08:10:00 pm »
Cool!

Oh yeah, your articles were the first ones that I read every month (I probably still have a few "Circuit Craft" magazines lying around). You have a knack for making electronics easy to understand (try "The Art of Electronics" for a taste of the other end of the spectrum). LadyAda of Adafruit.com (Limor Fried), Jeremy Blum (jeremyblum.com), and of course Jeri Ellsworth (all over Youtube) are the current generation's best (IMHO) at making electronics accessible.

As you probably know, impedance-matched terminations should help if the problem is noise.

They do. DeviceNet specs a 121 ohm terminating resistor, I found some precision 60.5 ohm resistors on Digikey and got a few k of them. To go with the resistors, I also bought 22 nf caps that we wired to the center point of the resistors (the resistors being wired in series to get us the original 121 ohms). The other end of the cap is tied to ground (not shield).

This should give us 3db in noise reduction. That's definitely helpful, and not to be sneezed at. But the cables that our benevolent corporate overloads foisted upon us contains not only the DeviceNet conductors, but 110VAC power conductors as well. You might ask "what were they thinking?" The simple answer is, they weren't. You should see the scope traces... Ugly. :O

And the cables that we use are the top of the line "Control Boss" made by Sine/Amphenol (no longer being made by all appearances), which despite their gold-plated contacts regularly build up a few ohms of resistance (and a few is all that it takes), which takes the network down.

And since DeviceNet carries its power in the same cable as its data lines (although it doesn't have to), voltage drop becomes a serious problem long before we run up to the max length of the network. We are currently trying out these: http://www.ebay.com/itm/251379658717?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649 to boost the voltage at the far ends of the network. Initial results have been very good, and at $6 you aren't going to lose much. You can actually get them cheaper, but they appear to be a cheap knockoff of a product that is already coming out of China (notice how the wires wound around the ferrite core are touching their neighbors. I'll pass on these).

S, anyways, the other day I was (re)reading a bit of "Serial Port Complete" and lamenting the zero slew rate of the transceiver chips. I had toyed with the idea of popping out the ones that I could get to and putting in newer/better ones. Then I thought, why not just add a RC network to each transmit and receive line and create my own slew?

Am I missing something obvious?

Jan Axelson

  • Administrator
  • Frequent Contributor
  • *****
  • Posts: 3033
    • Lakeview Research
Re: Adjustable slew rate?
« Reply #5 on: November 24, 2013, 08:31:35 pm »
Yes, sure, give it a try.

ruutboy

  • Member
  • ***
  • Posts: 19
Re: Adjustable slew rate?
« Reply #6 on: November 24, 2013, 08:54:07 pm »
Yes, sure, give it a try.

LOL! Not exactly spoken with confidence, but what do I have to lose? :)

I should be able to see with the scope if it is having any affect at all, and I'll let you know if it does anything for/to me.


Thanks!