tl/f/11894 twisted pair fddi magnetics overview and recommendations AN-902 national semiconductor application note 902 todd vafiades july 1993 twisted pair fddi magnetics overview and recommendations 1.0 introduction the use of twisted pair cable for high speed lan signalling necessitates the inclusion of transmit and receive magnet- ics to couple the transmission signal to and from the copper media. the choice of magnetics in a given implementation can have a significant effect on the integrity of the transmis- sion signal. several important factors must be considered when choosing the magnetics for fddi twisted pair pmds. this application note discusses key performance parame- ters of magnetics suitable for use within a pmd designed for ansi x3t9.5 fddi twisted pair draft proposal compliance. although magnetics are required for both shielded and unshielded twisted pair media, this note focuses specifically on magnetics suitable for fddi signalling over category 5 unshielded twisted pair. this note includes layout recom- mendations for a typical pmd transceiver implementation employing the national semiconductor dp83223 twister transceiver and suggests the use of some readily available magnetics. 2.0 what are magnetics? in the case of twisted pair fddi signal transmission, the term ``magnetics'' refers to the one-to-one isolation trans- formers and common mode choke transformers which cou- ple the signal to and from the twisted pair media. these elements couple the serial data stream from one fddi node to the twisted pair media and again from the twisted pair media to another fddi node. it is also possible that these transformers may coexist with other filter elements, such as resistors or capacitors, which attempt to enhance the integ- rity of the transmitted and/or received fddi data stream. these additional filter elements may or may not be de- scribed as part of the magnetics depending on the individual vendor's perspective. as a point of clarification, ferrite beads or inductors, sometimes used to decouple sensitive power and ground pins from potential noise sources on transceiver ics, may also be referred to as magnetics. this application note is only intended to report on the media cou- pling magnetics. 3.0 why are magnetics required? in most electrical signal transmission systems, the data moving between two nodes is ac coupled in order to isolate potential system ground differences between the transmit- ter and receiver which could interfere with proper signal transfer. a one-to-one isolation transformer is a convenient component for use in signal transfer for several reasons: dc current blocking (system ground differences are isolated from one another), end stations protection from static charges that may build up on the cable, inherent differential signal coupling and common mode rejection. when using the dp83223 twisted pair transceiver, it is not necessary to employ complex multiple pole lc filters which are commonly found in many 10base-t, token ring and fddi implementations. due to the controlled output tran- sition times of the dp83223, simple networks which include only the termination resistors, isolation transformers and common mode chokes may be all that are required. some designers may choose to add simple filtering at the receive end of a system in order to reduce the susceptibility to tran- sient or continuous noise injected onto the media from out- side sources. this note contains example schematics detail- ing components and interconnection. 4.0 key parameters magnetics play an essential role in ensuring signal integrity within a transmission system. parameters such as insertion loss, crosstalk and transition time contribute greatly to the performance of the magnetics within a system. this ap- plication note briefly examines several important parame- ters which contribute to the effectiveness of a given mag- netics design. 4.1 insertion loss this is the loss introduced by the insertion of the magnetics and can be generally expressed as: i l (db) e 20 log v in v out where v in is the voltage across the input of the magnetics while v out represents the voltage across the output of the magnetics in an appropriately configured system. some fac- tors which may contribute to loss include: dc resistance of the windings, variation from a true one-to-one (primary to secondary) winding relationship resulting in a ``step-down'' effect, core loss as well as the inherent loss of additional filtering. it is important to consider insertion loss when set- ting specified transmit amplitudes for standard compliant twisted pair fddi signalling. 4.2 return loss this is a measure of the match between the two impedanc- es on either side of a junction point, defined by: r l (db) e 20 log z1 a z2 z1 b z2 where z1 and z2 are the complex impedances of the two halves of the circuit. if an impedance mismatch does exist, signal reflections will measurably decrease the performance of a given system. the effects of return loss are signifi- cantly reduced by the controlled output transition times of the dp83223. these controlled transition times basically eliminate the need for additional filtering which can increase the potential for a mismatch in transmit and receive imped- ances. c 1995 national semiconductor corporation rrd-b30m105/printed in u. s. a.
4.3 common mode rejection this is the ability of the magnetics, either transmit or re- ceive, to reject common mode energy which may exist in the transmission signal. also, the ability of the magnetics to not impart any common mode energy to the signal. com- mon mode energy can be described as some potential ex- isting equally (in phase) on each side of a differential pair with respect to some fixed potential such as ground. as an example, some twisted pair conductors are routed through typical office locations which contain significant ambient energy. this can inject as much as 30v ac (in some cases even higher) of common mode potential to the twisted pair. if this common mode voltage is not blocked, the line receiv- er, which may be powered by a single 5v rail, will fail to receive a signal that is well outside of its specified operating range. 4.4 crosstalk this is the amount of energy coupled from the transmit channel to the receive channel within the magnetics. the effects of this type of crosstalk are virtually eliminated due to the physical isolation between transmit and receive mag- netics as shown in subsequent connection diagrams. 4.5 output transition time this is the standard ``rise and fall time'' as measured from 10% to 90% of full amplitude. with mlt-3, it is important to measure both rise times and both fall times of the three level signal. again, due to the controlled output transition time of the dp83223, additional wave shaping filters re- quired by some implementations are unnecessary. 4.6 overshoot given a square wave, overshoot may be defined as the amount of energy above or below the intended final high or low voltage level(s) as expressed in percent. overshoot may result from unintentional emphasis of some high fre- quency harmonics and or transitions coincident with reflec- tions. due to the controlled transmit transition times of the dp83223, the potential for overshoot is reduced by the in- herent decrease in high frequency energy of the transmitted transition times. 4.7 baseline wander in an ac coupled digital transmission system, baseline wan- der is the variation in the dc content of the transmitted datastream at any point in time. this phenomenon is depen- dent on the digital content of a given data stream and the low frequency cutoff of the magnetics. the scrambled fddi line code generated by a twisted pair fddi pmd can result in run lengths (no transitions) of up to 480 ns. if the magnet- ics low frequency pole is not sufficiently low to allow, with- out attenuation, a 480 ns static condition, then the attenua- tion at the critical frequencies will result in a ``droop'' or ``tilt'' of the waveform during the run length. this droop will effec- tively offset the baseline reference of the datastream result- ing in baseline wander. an increase in baseline wander con- tributes directly to increased jitter. in general, the higher the ocl (open circuit inductance), the lower the low frequency pole for the magnetics bandpass region and the less severe the baseline wander. 4.8 conducted power spectrum this is the power spectrum of a properly terminated pmd transmitter (including the magnetics) as measured by direct connection into a spectrum analyzer. this spectrum analy- sis is a convenient method of comparing the results of dif- ferent signalling techniques. the degree of randomness within the data stream as well as the differences between binary and mlt-3 are easily compared via conducted emis- sions. 4.9 radiated emissions this is the radiated power spectrum of a properly terminated pmd transmitter (including the magnetics) as measured by a near field antenna within a strictly controlled environment. although this application note does not report on the radiat- ed emissions results of the recommended magnetics it re- mains a very important parameter. it is the responsibility of the systems vendor to ensure that the performance lies within mandated limits set forth by the various and appropri- ate regulatory agencies. 4.10 emi susceptibility this is a measure of the tolerance of a working tp-pmd receiver to a controlled ambient field of radiation imposed on the twisted pair cable carrying the scrambled fddi line code. the receive-end magnetics can be supplemented with some degree of high frequency filtering to afford great- er immunity to susceptibility. 5.0 recommended magnetics this application note highlights specific magnetics from four vendors. it is important to understand that this note does not suggest preference to any one vendor or magnetics so- lution. the results herein are made available strictly as a means of objective comparison intended to assist the sys- tem designer in making the best possible choice for a given implementation. due to the relative immaturity of twisted pair fddi, this application note reports on only a limited number of magnetics solutions. future updates or adden- dums to this application note will include a larger selection of magnetics suggested for use with national semiconduc- tor pmd solutions. the four magnetics solutions are listed, in alphabetical order, by company name followed by product number. (contact information for each of the vendors is lo- cated at the end of this applications note.) bel fusee y 0556-3899-04 coilcrafte y q3950-c pulse engineeringe y pe-65620m valore y st6021 please contact each individual magnetics vendor for the lat- est product information and part numbers. 2
6.0 parameter measurement this section summarizes pertinent data as measured from some of the key parameters mentioned previously. all tests were performed using the same specially designed evalua- tion platform. this platform consists of a multi-layer ``odl replacement'' emulation board fitted with a dp83223 trans- ceiver in order to duplicate, as closely as possible, the per- formance of a true tp-pmd application. each of the four magnetics solutions were tested against the same dp83223 transceiver in the same environment to ensure comparable conditions. all tests were performed identically on each of the magnetics solutions for both binary and mlt-3 encoded data transmission unless otherwise noted. again, it is very important to understand that any data reported herein is preliminary and is provided for reference. each magnetics vendor should be contacted for the latest performance in- formation. 6.1 insertion loss insertion loss is measured in two steps. first, the magnetics under test are replaced by shorting wires which dc couple the transmitted signal to the digitizing oscilloscope and the transmit waveform is calibrated to exactly 2v peak-peak dif- ferential. second, the magnetics under test are reinserted and a second peak-peak differential measurement is per- formed. the insertion loss resulting from scrambled fddi code is tabulated below. insertion loss bel fuse coilcraft pulse valor (db) scrambled fddi b 0.26 b 0.67 b 0.26 b 0.35 6.2 return loss although this parameter was not measured, the return loss due to the magnetics alone should be minimal because complex filtering is not required. potential return loss may be inferred by examining the magnetics vendor's manufac- turing tolerances. 6.3 common mode rejection this parameter was not tested. refer to each vendor's da- tasheet for performance specifications. 6.4 crosstalk the virtual absence of interchannel crosstalk between the transmit and receive magnetics is due to sufficient physical separation of the components as specified by national semiconductor. there will be some degree of crosstalk that occurs between the transmit and receive channel outside of the magnetics which will most likely occur within the media connector and within the media itself. this effect can be minimized by observing good high speed layout practices and will not be increased by the use of the magnetics solu- tions outlined herein. 6.5 output transition time the rise and fall times of a transmitted signal are a direct indication of the bandwidth of the transmit channel. the transition time specification depends somewhat on results of emi radiation testing and other performance tests. slow- er transition times can be achieved using different magnet- ics components. to test the rise and fall times of the mag- netics, the input of the magnetics were presented with the 2.0 ns transition times generated by the dp83223 twist- er. the output of each magnetics solution was then mea- sured to determine the transition time performance. transition (ns) bel fuse coilcraft pulse valor binary rise 2.41 3.25 2.37 2.34 binary fall 2.63 3.12 2.29 2.18 mlt-3 rise ( b 1 to 0) 2.40 3.16 2.37 2.38 mlt-3 rise (0 to 1) 2.67 3.32 2.59 2.43 mlt-3 fall (1 to 0) 2.48 3.08 2.25 2.19 mlt-3 fall (0 to b 1) 2.76 3.32 2.49 2.36 (refer to figures 1 and 2 .) 6.6 overshoot overshoot, especially in mlt-3 mode, will decrease the noise margin of the transmitted signal. serious overshoot may also contribute to unwanted bit errors in the received signal. the overshoot at the output of the magnetics is mini- mized because the input signal to the magnetics includes the controlled transition times generated by the dp83223 twister. overshoot of less than 2% can be considered negligible. overshoot (%) bel fuse coilcraft pulse valor binary k 2.0 k 2.0 k 2.0 k 2.0 mlt-3 k 2.0 k 2.0 k 2.0 k 2.0 (refer to figures 3 and 4 .) 6.7 baseline wander the effects of baseline wander can be directly inferred by measuring the magnetics droop characteristic over a worst case run length period of 480 ns for scrambled fddi code. the baseline wander is arrived at by doubling the percent- age droop exhibited by a given magnetics solution. baseline bel fuse coilcraft pulse valor wander (%) 480 ns width 5.6 12.1 5.6 6.9 (refer to figure 5 .) 3
6.8 conducted power spectrum the conducted power spectrum offers a convenient meth- od of understanding and comparing the differential power spectrum of a given set of magnetics. due to the similarities between the conducted spectra of each of the magnetics tested herein, only typical measurements are presented. of specific interest are the differences between the binary and mlt-3 encoded conducted power spectrum for scrambled line code. although mlt-3 suffers from 6 db lower noise immunity than binary given equal transmit amplitudes, mlt-3 does exhibit an improvement in the reduction of dif- ferential conducted power at key frequencies. conducted power binary mlt-3 (dbm) @ 31.25 mhz b 47.0 b 53.0 @ 62.50 mhz b 53.0 b 62.0 (refer to figures 6 and 7 .) 6.9 radiated emissions currently, no data is provided for this parameter. however, preliminary data will be available soon. please contact na- tional semiconductor for information pertaining to the radi- ated emissions of suggested pmd implementations. 6.10 emi susceptibility currently, no data is provided for this parameter. however, preliminary data will be available soon. please contact na- tional semiconductor for information pertaining to the emi susceptibility of suggested pmd implementations. 7.0 additional parameters ultimately, the most important performance factor of twist- ed pair fddi signaling is long term, error free data transmis- sion. to ensure that each of the four magnetics solutions tested herein will support error free transmission, 16 sepa- rate bit error rate (ber) tests were performed. each of the four magnetics were tested against themselves and each other at both the transmit and receive ends of the transmis- sion system. specifically, each test was performed using 130 meters of category 5 cable with scrambled code set at 2.0v peak-to-peak differential transmit voltage. these tests were performed for both binary and mlt-3 signal encoding. each of the 16 ber tests passed proving acceptable inter- operability in terms of the magnetics to the tp-pmd stan- dard ber limit of k 10e-12. several additional electrical parameters exist for each of the magnetics solutions presented herein. although these pa- rameters are not included in this analysis, they are nonethe- less important and may help to further inform the system designer regarding performance. each of the magnetics vendors publish a list of these specifications, tolerances and test conditions included where applicable, to accompany their solutions. it is best to refer to these figures for a more comprehensive understanding of performance. some of the standard parameters associated with magnetics include: e turns ratio e ocl (open circuit inductance) e ll (leakage inductance) e cw/w (interwinding capacitance) e dcr (dc resistance) e hi pot (high voltage tolerance) e cmr (common mode rejection) tl/f/11894 1 figure 1. typical binary transitions tl/f/11894 2 figure 2. typical mlt-3 transitions 4
tl/f/11894 3 figure 3. typical binary overshoot tl/f/11894 4 figure 4. typical mlt-3 overshoot tl/f/11894 5 figure 5. typical binary droop tl/f/11894 6 figure 6. typical binary conducted power spectrum tl/f/11894 7 figure 7. typical mlt-3 conducted power spectrum 5
8.0 utp-pmd magnetics connection this section focuses on suggested interconnection and lay- out of the magnetics solution within the pmd. due to the high speed nature of twisted pair fddi, careful layout prac- tices are advised. maintaining a 50 x signal impedance and keeping high speed signal traces as short as possible are important design factors. the following design example highlights several key areas of concern and also suggests possibilities for improved overall system performance. figure 8 illustrates a typical magnetics layout using the na- tional semiconductor dp83223 twisted pair transceiver. this layout example assumes the use of four planes to ac- commodate the required power and signal routing as de- scribed in the cross sectional view provided in the legend ( figure 9 ). additionally, the legend provides component type and values as well as identification of various signal paths and power planes. circuit details of the layout follow: capacitor c1 optionally helps to ensure that high frequency energy outside of the intended passband across r3 will be attenuated. capacitors c2, c3 and c4 provide power supply decou- pling for each of the designated power planes. c2 decou- ples noise from txv cc to txgnd. c3 decouples noise from rxv cc to rxgnd. finally, c4 decouples noise from eclv cc to eclgnd. ferrite beads fb1 through fb4 provide good isolation be- tween unique supply islands and planes. fb1 isolates the rxgnd (receive ground) island from the eclgnd plane. fb2 isolates the rxv cc (receive power) island from the eclv cc plane. fb3 isolates the txgnd (transmit ground) island from the eclgnd plane. and f4 isolates the txv cc (transmit power) island from the eclv cc plane. while many implementations employ standard inductors of various values for power supply isolation, national semiconductor recommends the use of ferrite beads for improved isolation and enhanced performance. ferrite beads provide damping of high frequency noise while not creating problems caused by high q inductors. resistors r1, r2 and r3 form a voltage divider in which the receive signal, as presented to the dp83223, is attenu- ated relative to the full receive amplitude. this amplitude reduction is a good method of ensuring maximum operation- al headroom of the embedded adaptive equalizer and asso- ciated circuitry within the dp83223. in addition, this attenua- tion can be adjusted to accommodate for magnetics inser- tion loss. resistors r4 and r5 form the back termination for the transmit signal path. these resistors are terminated directly to the txgnd (transmit ground) plane. since the dp83223 twister allows these back termination resistors to be ref- erenced to ground, the noise coupled to the transmitted sig- nal is less than those implementations which reference the output to v cc . resistors r6 through r9 provide the two unused twisted pairs within the 4-pair bundle with 100 x differential termina- tion. resistors r10 through r13 provide good common mode termination for each of the four twisted pairs within the bun- dle. more specifically, r10 and r11 terminate the two un- used twisted pairs while r12 and r13 terminate the two active twisted pairs. r12 is connected between the primary center tap of the receive transformer and the common mode common point while r13 is connected between the secondary center tap of the transmit transformer and the common mode common point. within some magnetics the transmit channel isolation transformer primary and receive channel isolation transformer secondary center taps are pinned out. the example shown in figure 8 assumes these pins float. although the common mode termination design presented here is a viable option, other designs may potentially pro- vide improved performance as well. an additional point of clarification: to date, the ansi subcommittee on twisted pair fddi has not yet defined common mode termination of any kind. however, data has been presented that indicates a significant enhancement in emc performance for catego- ry 5 cable fitted with common mode termination. 6
tl/f/11894 8 figure 8. typical magnetics layout using dp83223 transceiver tl/f/11894 9 figure 9. legend 7
9.0 magnetics packaging package type information includes: package encasement, footprint and pinout for each of the three vendor's products. for precise mechanical information on each of the magnet- ics, please refer to the appropriate vendor's datasheet. the order of description is alphabetical by vendor name. please contact each vendor for the latest package information. bel fuse: product y 0556-3899-04 one 0556-3899-04 required for transmit channel one 0556-3899-04 required for receive channel through-hole/6-pin sip/100 mil pin spacing tl/f/11894 11 tl/f/11894 10 undesignated pins are no-connects coilcraft: product y q3950-d one q3950-c required for transmit channel one q3950-c required for receive channel through-hole/6-pin sip/100 mil pin spacing tl/f/11894 13 tl/f/11894 12 pulse engineering: product y pe-65620m one pe-65620m required for transmit channel one pe-65620m required for receive channel plastic/surface mount/16-pin dip/50 mil pin spacing/ 300 mil device width tl/f/11894 15 tl/f/11894 14 undesignated pins are no-connects valor: product y st6021 one st6021 required for transmit channel one st6021 required for receive channel plastic/surface mount/16-pin dip/50 mil pin spacing/ 300 mil device width tl/f/11894 16 tl/f/11894 17 undesignated pins are no-connects 8
vendor information bel fuse, inc. 5362 w. 78th st. indianapolis, in 46268-4147 (317) 876-0044 coilcraft, inc. 1102 silver lake rd. cary, illinois 60013 (708) 639-6400 pulse engineering, inc. p.o. box 12235 san diego, ca 92112 (619) 674-8100 valor electronics, inc. 9715 business park ave. san diego, ca 92131 (619) 537-2619 references 1. national semiconductor dp83223 device specification. 2. fddi twisted pair physical layer medium de- pendent (tp-pmd) working draft proposed american national standard rev. 0.3 dated 2/17/93. 3. bell, david a., solid state pulse circuits , reston publish- ing, (1981). 4. various, reference data for radio engineers , sams, (1981). 9
AN-902 twisted pair fddi magnetics overview and recommendations life support policy national's products are not authorized for use as critical components in life support devices or systems without the express written approval of the president of national semiconductor corporation. as used herein: 1. life support devices or systems are devices or 2. a critical component is any component of a life systems which, (a) are intended for surgical implant support device or system whose failure to perform can into the body, or (b) support or sustain life, and whose be reasonably expected to cause the failure of the life failure to perform, when properly used in accordance support device or system, or to affect its safety or with instructions for use provided in the labeling, can effectiveness. be reasonably expected to result in a significant injury to the user. national semiconductor national semiconductor national semiconductor national semiconductor corporation europe hong kong ltd. japan ltd. 1111 west bardin road fax: ( a 49) 0-180-530 85 86 13th floor, straight block, tel: 81-043-299-2309 arlington, tx 76017 email: cnjwge @ tevm2.nsc.com ocean centre, 5 canton rd. fax: 81-043-299-2408 tel: 1(800) 272-9959 deutsch tel: ( a 49) 0-180-530 85 85 tsimshatsui, kowloon fax: 1(800) 737-7018 english tel: ( a 49) 0-180-532 78 32 hong kong fran 3 ais tel: ( a 49) 0-180-532 93 58 tel: (852) 2737-1600 italiano tel: ( a 49) 0-180-534 16 80 fax: (852) 2736-9960 national does not assume any responsibility for use of any circuitry described, no circuit patent licenses are implied and national reserves the right at any time without notice to change said circuitry and specifications.
|
