novak@tscs.UUCP (Albert Novak) (08/19/89)
In article <442@qtc.UUCP> (2 Aug 89 15:07:25 GMT), Jim Battan writes: > I'm in the process of converting our thin coax Ethernet from a mixture of > RG58/U and RG58A/U to something that is within the IEEE spec and > electrically consistent. I've heard numerous conflicting opinions about > what to do, from this newsgroup, computer manufacturers, and coax cable > makers. I know I want one of RG58A/U, RG58C/U, or IEEE 802.3 10BASE2 > (aka ThinWire) cable. I'll be using UTP when we move to our new building > in eight months (if the 10BASET spec is approved by then), but need some > new coax now. This is for a network of 35 Suns, a Sequent, several DEC > VAXen, and a GatorBox, all talking through a DEMPR and DELNI. > > My questions: > Are the technical advantages of fully IEEE-compliant cable (e.g. Belden > 9907) worth the extra cost (cable, new connectors, new crimper)? > (We're not up against any of the 10BASE2 limits (#nodes, length) yet.) The following inter-office memo from Belden to their sales force indicates that there is a reluctance on the part of most Ethernet users to opt for the more expensive cable. They even admit that problems may appear only when one reaches the limits of the IEEE 802.3 specification. The differences between RG-58A and RG-58C and Ethernet ThinWire coax are shown in the table made up by Belden. Differences in outside diameter and center conductor diameter only come into play if one uses Ethernet connectors on RG-58 or RG-58 connectors on Ethernet cable. If you use quality connectors made for your cable, reflection problems at the connector will be eliminated. Also, the IEEE 802.3 standard seems to match RG-58A and RG-58C dimensions better than the Belden Ethernet cable. Differences in impedance can be dealt with by using a cable that matches the IEEE 802.3 specification of 50 +\- 2 ohms. Belden 9262 (RG-58C) meets this specification per Belden's own table. The attenuation specification is where RG-58 does not match the specification. As Belden points out, this will limit the maximum length of each Ethernet segment. Using a good quality RG-58A or RG-58C coax that meets the Military specification MIL-C-17F (M17/155-00001) will keep this problem to a minimum. ############################## ## Beginning of Belden Memo ## ############################## INTER-OFFICE CORRESPONDENCE Belden Wire and Cable Richmond, Indiana April 25, 1989 TO: Field Sales SUBJECT: Differences Between ThinWire Ethernet (9907) and RG-58 Belden 9907 was designed specifically to meet the IEEE 802.3 10 Base 2 specification for ThinWire Ethernet systems. In general, the market perceives that generic RG-58 (e.g., Belden 8219) is a ThinWire equivalent. This is not so. Please notice the attachment, developed by John Staats, that details the different constructions. What all this means is a substantial decrease in performance. 10 Base 2 is a 10 Mbyte/sec system with a single segment length (between repeaters) of 185 meters. Five (5) of these segments may be placed in a row, but only three (3) lengths may contain work stations. Each available segment may contain up to thirty (30) work stations. If the system is heavily loaded, there is a good likelihood that there will be: 1) an increase in bit error rate, 2) a decrease in available distance, and 3) a much slower responcce time. In some cases, the system operates fine until one more terminal is added. The last terminal causes enough reflection to be diagnosed as a collision. At this point, the system will shut down. As it tries to re-transmit the signal, it continues to read it as a collision. This, in essence, shuts down the network. The point is, make sure the electricals are met if a system is to be fully utilized. Another point worth mentioning pertains to the Teflon (plenum) version. The industry standard for velocity of propagation is 77%. Cable manufacturers are intended to observe this upon their cable design. Some manufacturers will heavily foam the cable. While this reduces the cost (through reduced diameters, less material, etc.), it makes the standard connectors too large. This is another point that needs to be explained to our customers. Phil Dunn THINWIRE ETHERNET SPECIFICATIONS (by John Staats 3-15-89) Parameter IEEE 802.3 DEC BELDEN CABLES 10 Base 2 1701248 YW22392 9907 8219 8259 8262 8240 Standard ThinNet ThinWire RG-58A RG-58A RG-58C RG-58 ------------------------------------------------------------------------------ Conductor: Stranded 19x7.4 19x7.4 19x7.7 19x8.1 19x7.2 19x7.2 Solid OD 0.035" 0.0371 0.037 0.037 0.0375 0.035 0.035 0.0332 Tol (+/-) 0.002" 0.0005 0.0004 Nom Nom Nom Nom Nom Dielectric: Solid Cell Cell Cell Cell Solid Solid Solid Prefered HDPE HDPE HDPE PE PE PE PE Dielectric Diameter: OD 0.11687" 0.117 0.113 0.114 0.114 0.116 0.116 0.116 Tol (+/-) 0.006" Max* Nom* Max* 0.003 0.003 0.003 0.004 (* Over bonded tape.) Shields: Foil N/S Yes Yes Yes None None None None Braided TC TC TC TC TC TC TC TC 95% Min 93% Nom 93% Min 95% Nom 95% Nom 95% Nom 95% Nom 95% Nom 16x6x36 16x6x36 16x6x36 16x7x36 16x7x36 16x7x36 16x7x36 Jacket: PVC PVC PVC PVC PVC PVC NC PVC PVC NEC CL2 NEC CL2 NEC CL2 NEC CL2 Outside Diameter: OD 0.193" 0.183 0.183 0.182 0.198 0.193 0.195 0.193 Tol (+/-) 0.012" 0.007 0.003 0.003 0.004 0.004 0.003 0.003 Impedance: 50 Ohms 50 50 50 50 50 50 53.5 Tol (+/-) 2 Ohms 2 2 2 5 3 2 Nom Attenuation (dB/100 feet): @ 5MHz 1.00 0.99 0.90 0.90 0.91 1.05 1.05 ---- @ 10MHz 1.42 1.40 1.30 1.30 1.30 1.51 1.51 1.20 Max Max Max Max Nom Nom Nom Nom Velocity 65% 77% 80% 80% 77% 66% 66% 66% of Prop: Min Min Nom Nom Nom Nom Nom Nom DCR Loop: 50 mOhms/m 50 50 50 43 48 48 46 Max Max Max Max Nom Nom Nom Nom Edge Jitter: +/- 7 ns +/- 7 5 Nom 5 Nom 5.5 Nom 7 Nom 7 Nom ---- NOTES: 1. Out of specification conductor size could result in poor connector performance. (Belden 8219, 8259, 8262 and 8240). 2. Out of specification impedances cause large signal reflections which may cause false collision detection. (Belden 8219 and 8240). 3. High attenuation creates excessive signal loss which effectively shortens the usable cable length. (Belden 8259 and 8262). 4. High loop resistance can increase attenuation and edge jitter. These shorten the usable cable length and decrease the number of nodes per segment that can be used. (Belden 8259 and 8262). 5. Excessive edge jitter creates timing uncertainty which shortens the usable cable length and decreases the maximum number of nodes per segment that may be used. ABREVIATIONS: 1. Conductors are stranded where the listing is of the form "19x7.4". This denotes 19 strands of 7.4 mil (0.0074") wire. OD - Outside diameter. Tol - Tolerance expressed as a +/- specification or Nom - Nominal. 2. Dialectrics are cellular (Cell) or solid and made of polyethylene (PE) or HD polyethylene (HDPE). 3. Shields of foil are an aluminum/poly/aluminum sandwich material of 0.002"/0.001"/0.00035" thickness. Braid sheilds are made of a tinned copper (TC) briad (e.g. 16x6x36 is a 16 by 6 weave of 36 gauge tinned copper wire). 4. Jacket material is polyvinyl chloride (PVC) with optional National Electrical Code rating of CL2 (NEC CL2). EXPLANATION OF DIFFERENCES (by Phil Dunn): The column "IEEE 802.3" refers to the published standard. The column "DEC 1701248" refers to the DEC specification. The column "YW-22392" refers to the Belden number sold to DEC. Digital Equipment is the company that really pushed ThinNet into a major product line. The DEC 1701248, Belden YW-22392 and 9907 are really the same product. The only differences are in the way the specification spells out either nominal (Nom) or maximum (Max) tolerances. Digital Equipment, Belden and AMP all worked closely together to establish standards for cable and connector dimensions. The IEEE 802.3 standard cable is substantially larger. There often are problems getting connectors to fit this construction. The factors that affect performance with 9907 (and other 10 Base 2 cables) versus RG-58 are as follows: Dielectric: Should be foamed HDPE. The RG-58's have different electrical characteristics, this reduces performance. Bonded Tape: Specified on the 9907, not used on the RG-58's. This increases the attenuation on the RG-58's. Impedance: Make sure the impedance is held to a tolerance of +/- 2 ohms, otherwise, resulting reflections may be read as "collisions". Attenuation: Partly due to the bonded tape, but notice that the RG-58's have substantially higher losses. This limits the length of usable cable. Loop Resistance and Edge Jitter: Again, if not within tolerance, these can limit segment length. ######################## ## End of Belden Memo ## ######################## > What coax stripper is used with ThinWire cable? Xcelite technical support > says they don't make a compatible stripper. Some coax strippers have adjustments for each of the cutting blades used to make the cuts. If you get one of these types for RG-58 coax, it will probably work for Ethernet coax as well, as the differences in dimensions do not exceed 0.01 inches. > Can I use the same AMP or Amphenol connectors that are now in use on our > RG58/U cable? If not, what are the correct part numbers? AMP and Amphenol make connectors specifically for Ethernet cable as the differences in dimensions will cause problems with standard RG-58 connectors. The Amphenol part number is 36650-1003 for a BNC cable plug for Belden 9907 or DEC ThinWire cable. The AMP literature I have from 1988 does not mention Ethernet cable, but they should have a connector. > Can I use the same crimper (a Paladin hex crimper for RG58)? Probably not. Amphenol makes a crimper set for the Ethernet cable that is different than the one they recommend for RG-58. The part number for Amphenol's crimper die set is 227-1221-09 and for their tool handle is 227-944. ------------------------------------------------------------------------------ Albert Novak UUCP: uunet!pdn!tscs!novak Electronic Maintenance Supply Company PHONE: (813) 253-2684 1516 Cypress Street FAX: (813) 254-1739 Tampa, Florida 33606 ------------------------------------------------------------------------------