Page 2 of 2 FirstFirst 12
Results 11 to 20 of 20

Thread: LV direct burial splices

  1. #11
    Join Date
    Apr 2008
    Location
    Ann Arbor, Michigan
    Posts
    7,054
    180918-0723 EDT

    Open Neutral:

    Your Dallas reference from post #1 is good, but it covers many aspects in a general way with some specifics mixed in.

    You have not told us your maximum radial distance. Is it 100 or 200 feet, or 10,000 ft. Nor your baud rate.

    Is the application at some home, or at a research site?

    How deep in the earth are the sensors located? Can the tap points be located in a JIC oil tight box above earth? Does the CAT 5 cable really have to be buried vs conduit above earth?

    As Dallas mentioned energy propagation in a transmission line, like CAT 5, is about 0.7 the velocity of light in free space, or 0.7*186,000*5280 feet/second. Divide by 1,000,000 for feet/microsecond which is 687. For a radial distance of 345 feet the reflected pulse will arrive at the origin in 1 microsecond. In my plots that I referenced you see 0.4 microsecond for a 150 foot open cable. Good correlation with the 0.7 factor.

    If you operate at 9600 baud, sample at the bit midpoint, and have a 350 foot cable, then the sample point is at 50 microseconds, or 50 times away from the first reflected signal. One sees very little disturbance beyond one reflection. See my waveforms at P4 and P5.

    If you operate in the 3 wire mode you get rid of the problem of switching between power and data on the same line. Also with slew rate control you reduce the reflection problem.

    Note that maximum reflection occurs when the far end of a transmission line is shorted or open. If the far end has a load resistor equal to the characteristic impedance of the transmission line, then there is no reflection.

    I can not link to your master information. Dallas makes a master driver chip that has good characteristics.

    Why three channels?

    At each well are you using two temperature sensors, one for input and one for output?

    Are the sensors inserted into the pipe for direct contact with the liquid, or on the pipe outside surface?

    If the Dallas driver is a very long distance from the wells, then consider putting the driver near the wells, and use RS 422 back to the computer.

    .

  2. #12
    Join Date
    Apr 2008
    Location
    Ann Arbor, Michigan
    Posts
    7,054
    180918-0921 EDT

    Open Neutral:

    You mentioned POTS and DSL. These probably have little to do with your application. DSL is a much higher baud rate, distance is possibly in the 1 to 3 mile range (we don't know your distance), and cable characteristics may be somewhat different.

    The major things you need to consider are:
    1. Total cable capacitance. What Dallas links into the term "weight".
    2. Baud rate.
    3. Distance to farthest point. Reflection time.
    4. Loading. Also in Dallas "weight".

    Both cable capacitance and reflection time suggest using a low baud rate. The 1-Wire bus is not a low impedance balanced bus. Thus, this bus limits baud rate, and is not very immune to capacitively coupled noise. But still a good bus for its purpose.

    .

  3. #13
    Join Date
    May 2010
    Location
    Inside the Beltway
    Posts
    642
    Your response is confusing. First off, this is not RS232 or RS-185, etc. It is "1-Wire" -- a Maxim (previously Dallas Semi) product. It's a bidirectional serial protocol with data rates of 16.3kbps & an upgraded "overdrive" of ~140kbs [I see several values mentioned.] Its selling point was/is that it took one wire (and ahem, a ground...) to move data and power. I have never seen the term "baud rate" mentioned in 1-Wire documentation, except re: a 1-Wire to RS-232 bridge.

    My reference to DSL is this. A twisted pair spur (T) is in effect, a linear capacitor across the pair. This causes HF rolloff grief. In telco-speak, it's called a "half-tap" & such plagued telcos when they were rolling out DSL because it uses higher frequency across the pair than the 300-3000 Hz of POTS; what worked for POTS did not for DSL. They had to find & remove them from the pair. This is a similar issue. Maxim warns against star and stub configurations in the AN-148 "Guidelines for Reliable Long Line 1-Wire Networks" that I mentioned.

    I am using a Embedded Data Systems OW-SERVER
    https://www.embeddeddatasystems.com/...n-2_p_152.html
    as the master. It helpfully has 3 switched output channels so loads on one don't affect the others. This avoids the star problem for some installations. I'll have less than 22 DS18b20 based probes off of it; the nearest to the master will be ~130 ft, the furthest ~400 ft; all in a semi-straight line. I will likely spread the probes across the 3 channels, reducing the "weight" on each; and if one channel's pair fails, the others will still function. [They are separate at this layer, but not in the software, they say.] If I ever need more than 22, I can get another OW-SERVER.

    The cable will be in the trench with the GSHP HDPE tubing back to the compressor. The probes will be about 4+ ft down, strapped to the well's tubing & then both (thermally) insulated from the backfill. (Other probles will be in the earth between the wells.) To minimize the stub issue, the splice will be near the probe. It will NOT be in above-ground conduit with accessible junction boxes; the homeowner would not like that, and there's the stub issue.

    Hence my interest in waterproof splices. I've seen the wirenut+gel-filled "testtube" scheme, & the 3M clamp-on. In the past I can recall seeing some sealant you mixed up and poured into the mold with the splices inside. Anyone familiar with same?

  4. #14
    Join Date
    Apr 2008
    Location
    Ann Arbor, Michigan
    Posts
    7,054
    180917-1317 EDT

    Open Neutral:

    I did not say it was RS232 or 185 (whatever that is).

    The Dallas 1-wire system at a minimum can work with 2 wires, common and data/power. Overall I think it is a better system when it is 3 wire with data and power separated.

    The Dallas 1-wire communication is asynchronous serial data in a half duplex mode on a single unbalanced bidirectional bus at some predefined baud rate. Because a single bus is used only one communication path in one direction can exist at one time. When data and power are combined, then only one of power or data can exist at a time.

    RS232 is an unbalanced interface standard, but usually carries asynchronous serial communication. Either full or half duplex. Same for RS 422, except 422 is a balanced bus providing longer distance and greater noise immunity.

    The 1-wire system is similar to RS 232 in that it is unbalanced, and thus suffers some of the same problems. However, an RS 232 is better in that it has active drive for both 0 and 1. But RS 232 is not good for long cable lengths, nor is 1-wire. RS 422 is much better on long cables.

    A three bus system is good for reliability.

    You definitely want to stick with the base bit rate (to some extent bit and baud can be treated the same, but when doing so bit can not be considered as data bit rate).

    If a bus cable, connection to the bus, or the sensor fails how will you fix the failure?

    .

  5. #15
    Join Date
    Jun 2003
    Location
    Hawthorne, New York NEC: 2014
    Posts
    3,897
    In the past I can recall seeing some sealant you mixed up and poured into the mold with the splices inside. Anyone familiar with same?
    Yes. 3M. Not re-enterable but better than the gel. Actually an electrical item.

    http://www.hillas.com/Categories/3M-...iABEgKNqfD_BwE



    -Hal

  6. #16
    Join Date
    Dec 2006
    Location
    Eastern Oregon
    Posts
    3,615
    We do a lot of control work in various capacities, but I still can't get over the fact this sounds like terrible design using a direct buried control cable with direct buried splices.

    I always ask myself when something gets way out in left field, is this standard procedure or am I trying to reinvent the wheel?

    The only way I could even feel a little good about something like this, is if it was in conduit and spliced in a vault/handhole with DB or submersible rated splices.

    But to each their own, hopefully a few years down the road you aren't wondering why you're not getting any signals from your devices.

  7. #17
    Join Date
    May 2010
    Location
    Inside the Beltway
    Posts
    642
    We do a lot of control work in various capacities, but I still can't get over the fact this sounds like terrible design using a direct buried control cable with direct buried splices.

    I went for 1-Wire because of the variety of inexpensive sensors available, and I need the "reach" given lengths from the building. (I am not going to even consider mounting the electronics out in the woods near the wells, and getting power and telemetry to/from same.)

    The constraints/issues I see are:

    a) The sensors are going to be buried because the well lines will be.
    b) The wells & their lines are buried to limit the thermal losses from the piping back to the compressors.
    c) The issue of the stub lengths and the signal degradation they cause.

    IF we were to put buried conduit and junction boxes near the surface, then we could dig down to each splice box for repairs. But getting down to the wells themselves to replace failed sensors is another issue. One solution there is to over-deploy sensors, so you can lose a few and not suffer that badly. I do plan to split them between the three available loops with such losses in mind.

  8. #18
    Join Date
    May 2010
    Location
    Inside the Beltway
    Posts
    642
    Open Neutral:
    I did not say it was RS232 or 185 (whatever that is).
    Sorry, fatfingered RS-485 there. But you started talking 1200 vs. 9600 Baud, and I don't know how they relate to 1-Wire performance. It has one standard speed, and an optional "overdrive" of ~10X that.
    The Dallas 1-wire system at a minimum can work with 2 wires, common and data/power. Overall I think it is a better system when it is 3 wire with data and power separated.

    The Dallas 1-wire communication is asynchronous serial data in a half duplex mode on a single unbalanced bidirectional bus at some predefined baud rate. Because a single bus is used only one communication path in one direction can exist at one time. When data and power are combined, then only one of power or data can exist at a time.
    I think I can use one common power for all three channels; I'll look into that. (i.e. 3 pairs of CAT5 for 3 channels; power on both of 4th pair.) As I said, there's no need for speed on this tasking, an update every 5-10 minutes would be more than enough.
    But RS 232 is not good for long cable lengths, nor is 1-wire. RS 422 is much better on long cables.
    But the supply of direct RS 422 temperature probes is far more constrained; I've not seen any. And re: noise, this is not inside a steel mill with arc furnaces; it is out in the woods. The bridge manufacturer talks of 1000 ft. runs. (BTW, as far as I know, Dallas sold everything 1-Wire to Maxim some time ago.)
    You definitely want to stick with the base bit rate (to some extent bit and baud can be treated the same, but when doing so bit can not be considered as data bit rate).
    Turns out the Embedded bridge *can* do overdrive with some low-level commands but I neither need nor want same.
    If a bus cable, connection to the bus, or the sensor fails how will you fix the failure?
    Do without or dig it up.

  9. #19
    Join Date
    Dec 2012
    Location
    Placerville, CA, USA
    Posts
    19,647
    Quote Originally Posted by Open Neutral View Post

    Do without or dig it up.
    Assuming it fails in a way that does not disable the data bus!

  10. #20
    Join Date
    May 2010
    Location
    Inside the Beltway
    Posts
    642
    Quote Originally Posted by GoldDigger View Post
    Assuming it fails in a way that does not disable the data bus!
    Hence three buses.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •