Jensen AN-004 HUM & BUZZ IN UNBALANCED INTERCONNECT SYSTEMS ‘by Bll Whitlock ORIGINS OF HUM Often sound systems exhibit strange and perplexing behavior such as hum that appears and disappears when power to other equipment, nt even partof the audio system, isswitched on or of “Traditional methods to eliminate hum often seem more lle voo-doo than engineering and, more often than not, are tial and enor exercises that end only when someone says "Ican lve with that This author has previously umtien about balanced lines in audio stems, so this paper will be strictly confined to unbalanced systems [2] In contrast fo a balanced system, an unbalanced system uses only to wires, one for signal and one for ground. lis uze still prevail in consumer audio, probably because it is cheap to male and it performs acceptably wellin very smallsystems such as typical home stereo setups. However, any unbalanced scheme has an inherent problem called common impedance coupling. From Ohm's lw ‘we Imow that when current love in a resistance, a voltage drop appears across that resistance With the exception of superconductors, any conductor {wite) has resistance. Iftuo diferent cizcuitschare the same conductor or wire, a curtent flowing in either cicuit will produce a voltage drop across the wire. As shown in Figure 1, a partial schematic of the simplest possible system, the shield conductor of the interconnecting cable becomes the offending common impedance. Since the cable shield is effectively connecting the grounds of the devices together, itcamies a cwrent derived from the powerline as well as the audio signal cuzent Although this fact is often overlooked orignored, itisfundamentaltothisdiscussion. Whatever voltage s present between its inputs, points A and Cin Figure 1 all ‘beampliied by Device B.Iteannottellthe difference between signal and hum, and will amplify both if they are present. To determine ‘what the input sees", we must trace the circuit bop from point Ato BioC. Since the voltage A to B (shield voltage drop) isin series with the voltage B to C (the signal, the voltages will diectly add. Cleat, it would be very desirable forthe shield voliage drop to be zero to avoid contaminating the signal Inthe realwwovld regardless field construction, material er gauge, we cannot make the shield ‘conductor resistance 2ero. Ouronly remaining choice sto somehow reduce the intechassis curent, in Figure 1, to anacceptable level INTERCHASSIS CURRENT: THEORY ‘This current is caused by the charge and discharge of capacitances betueen power line and chassis. An undesired but unavoidable primary to secondary capacitance exstsin the power transformer of every piece of AC operated equipment. Sometimes intentional capacitors and resistors are added from power line o chassis to suppress RFlandior meet safety regulations. Te predit the severity ofthe hum problem these capacitances create, we can analyze the cucuit using the steps of simplification shown in Figure 2 Oe abe smgemie 7 ‘ osiE_gg, semen = sue = CAPACITIVE REACTANCE 21 60 HE ¥, Figure 2 - ANALYZING THE EQUIVALENT CIRCUT @®- REFERENCE POINT AT RECEIVER GROUND @- Hun voLtaGe E AT ORTUER GRAUND = Tx O= HUN VOLTAGE @re® + stave @ v2 ©) DEVICE Al DEVICE EB (oe [amc oe 1 fe no 5 | mS ERRATA é | = [fetes tS Figure 1 - THE HUM GENERATING MECHANISM “The bas problem i tha the shield isa commeon path for both interchassis'ground” and signal caren Jensen AN008"Note the V, + V, term in the single capacitor equivalent circuit. The + accounts for the fact that most consumer equipment has a two- prong AC plug which can be inserted into an outlet either of two ‘ways. A special example illustrates how extreme the effects of plug reversalcan be. Consider the case wheze Croc Ja the value of Cpu and Cg i 1 the value of Cy, (this condition would be highly unlikely in the real world). Ifthe two plugs are connected to the AC line as shown in the diagram, each Tair of capacitors forms a volage divider with 3-1 division ratio, maling chasse voltages Vie and Uy each 40 volts AC with respect to ground. Since no current ‘will flow in a wire connecting two points of equal voltage, current I willbe 20, However, ifone of the AC plugs i reversed, the chassis voliages will no longer be equal and curent I will ow. Except for this special case, phig reversal simply cause a change in the interchassis current rather than the total cancellation seen in this example. For this reason, reversing AC plugs will almost always change the hum level in a system, Its also very unlilely that the two capacitances, Cp, and Cpa: OF Cy and Cay, Would be exactly matched in any piece of equipment ‘Mismatch ratios of two to one are common, Since allutlty 120 VAC ‘power in thie county is distnbated asymmetrically with respect 0 earth ground, one side called “neutral. is grounded. The other, called "hot" o "ine", i at 120 voli with respect to ground [4] Recently, proponents ofa scheme called "Balanced AC Power" have claimed that "[hum reduction] results are often quite dramatic’ [3] Balanced power uses center-fapped transformer to male each side of the line 60 volts with respect to ground. Although intuitively attractive, thisapproach can completely cancelinterchasss currents ina system of three or more devices only in the caze where each of, the devices had such matched capacitances. This would be an extremely rare occurence. Although 10 to 15 dB hum reductions, which would be more routinely achieved, might be considered ‘dramatic’ in a video system, this author cannot recommend thisor any other ine conditioning’ method asa cost effective olution for audio system hum problems INTERCHASSIS CURRENT: MEASUREMENT ‘When designingor troubleshooting a system, a highly recommended first step isto measure actual ground cunents of the system devices. This can be done quite simply using an AC volimeter adapted, 25, shown in Figure 3, fo measure AC curent. This same setupcan ako rmearure chassis cuntent between devices, The 1 10 resstor converts current to vollage at 1 millvolt per ricroamp while the capacitorlimits the measurementto frequencies ‘under about 1H One lead of this current meter is connected 10 the shield of an input or output jack on the device under test. An IHF/RCA plug is handy for this and t generally won't matter which jack you choose, since all shield grounds are usually ied together inside the device. The other lead of the curent meter i connected “SEY conte cone Figue 3- MEASURING THE CHASSIS CURRENT to a ground equivalent to power line neutral. The safety ground of ‘any modem outlet i convenient for this. It i required by safety codes to be tied to earth ground, as i the neutral The cuzent ‘measwrement should be talen under four conditions: device "of device "on", then repeated with the device's AC plug reversed. Tabing the highest reading willgive usa "worst case” number which can then be used, along with Table 1, to estimate the system hum levels produced when this device is connected to others via cables, ‘The author has tested a variety of consumer devices, including CD player, cacrettedecle, tuners, receivers, and power amplifiers, for chassis cunent fo ground. The broad categories of typical ground cunents developed from the testing were used in Table 1. Ground cunent 5 generally related to AC power consumption of the device, since this dictates the sizeof its power transformer and, to some ‘extent, its interwinding capacitances. Ground current L 5 wA RMS, Js typical of "low power’ consumer gear drawing under 20 watts Ths includes most CD players, cassetie decls, and tumtables, (Current M, 100 A RMS, is typical of "medium power" consumer gear drawing 20 to 100 watts. This includes most tuners, low to medium power receivers or power amplifiers, and some small TV receivers. Curent H, 1 mA RMS, i typical of "high power consumer gear drawing, orcapable of drawing, wellover 100 watts ‘This includes most high powered amplifiers or powered subwoofers ‘and kage screen or projection TV receivers. Table 1 shous the calculated effect of these currents when they flow in interconnect ‘able shields in an unbalanced audio system. Acontact resistance of ‘50 mf per connection was used and the 0 dB reference levels 300, mV FMS or about -10 dBV. All renults have been rounded to the nearest dB, Please note tht this charactercation of chassiseurrentapplicsonly to devices wth two-prong AC plugs. Three-prong plugs effectively connect the device chassis to safety ground, making the chassis a voltage source. System effects of this will be discussed later. esbie Lg ions 3m 20% 26m Sof = Sm 100% = 303m osm Curent ct[u[el[il[mule{[i[uJel[i[u[«l[i [ule fics cagimapeny | -109 | -23 | -6 | -101 | -r5 | -s5 J -96 | -70 | -s0 | -a | -03 | -33 | -s3 [ -7 | -37 sd Contactor Jens ca@smapany [-mn | -ss [es | -105 [=r | -s0 [-i00 [7 [= | -23 [-e7 [-a [-a [-a | -2 sd Contactor fio2 cacsmapen) [-12 | -26 | -e6 | -207 | -21 | -o1 | -103 [77 | -s7 | -26 [-70 ~or | -65 | 1s sd Contactor ‘Table 1 -CALCULATED HUM LEVEL, aB xe 300 mV, vs GROUND CURRENT, CABLE LENGTH, and SHIELD GAUGE 2 Jensen AN008AUDIBILITY OF HUM & BUZZ ‘Just what level of hum or buzz is audible depends on many factors A recent AES paper indicates that noise artifacts should be under 7120 €B to be inaudible for senous lstening in residential environments [5] The experience ofthis author indicates that levels higher than about -80 dB are annoying o mostlisteners. The noises onginating with the power line are generally described as either Thum’, which is predominantly 60 Hl or"buzz’, which consists of mixture of igh-order harmonics of 60 Fz. These harmonics are the zezul of power ine waveform distortion, which commonly reaches 5% THD and is caused by many types of non-linear power line loads. Because the human ear is much more sensitive to frequencies inthe? lb to § Lee range at these very bw levels, buzz is usually more audible than hum, even though the hum level may be electrically larger. BREAKING THE INTERCHASSIS CURRENT PATH. To eliminate hum, we must effectively eliminate interchassis round cunent. We could eliminate it by simply brealing the chassis to chassis chield connection. Of course, this alone would not solve our problem. We must break the signal line as well and insert a device ‘which willeence the vollge atthe outputofdevice A and regenerate itinto the input of device B, while ignoring the voltage that exists ‘between the now disconnected device grounds. These properties generally describe a differential responding deve with high ‘common-mode rejection, usually called a ground isolator. See reference [2] for more information on this subject ‘Two basic types of differential responding devices, active differential amplifies and audio transformers are available af reasonable cost We uon't consider active optical or carrier modulated isolation amplifiers here because such devices which alo have acceptable audio performance are stil quite expensive Active differential amplifier cicuits are used in a number of commercially available devices. To agreaterorlesserextent, they all share several disadvantages: they can further complicate the ground system by contributing interchassis cunents of their own, since they require AC power they cannot handle ground voltage differences over about 10 volts FMS: they use semiconductors or integrated circuits which are prone to degradation or failure caused by power Ine orlightning induction voliage transients; and, worst ofall they are exquisitely sensitive fo source impedance. This sensitivity limits ‘hum rection, even in a balanced system (for which they are intended), but it males them nearly useless in an unbalanced system [2]A typical example of such devicesis the popularSonance AGHI (which uses the Analg Devices SSM2141). Lab ‘measurements on this unt, shown in Figure 4, reveal that over the 200 0 to 1 10 range of source (output) impedances typical in consumer equipment, its hum rejection is only 15 to 30 dB. High quelity audio transformers are, by thei nature, relatively insensitive to source impedance and exhibit excellent hum rejection performance in ether talanced or unbalanced systems, Under the same conditions and same range of source impedances, the passive transformer based ISO-MAX® model CL2RR mearures 90 to 110dB. Asshown inthe Figure 4 graph, its measured hum rejection ‘2 over 70 dB better than the active device. Itrequizes no power of any lind and can handle ground voltage differences up to 250, volts RMS without malfunction, degradation, or damage. There is a widespread belief that all audio transformers have inherent limitations such as high distortion, mediocre transient ova Fatty (6) Sue tata (2) 10 1 {Ht TUTTI 0 100 tk Figue 4 - HUM REJECTION vs SOURCE IMPEDANCE, response, and lage phase enors. Unfortunately, many such transformers do exist and not all of them are cheap. The vast majority of available audio transformers, even when used as directed, do not achieve professional performance levels. As Cal Perkins wrote "With transformers, you get what you pay for. Cheap ‘transformers create a hostof interface problems, most of which are clearly audible "6] If well designed and properly used, however ‘audio transformers qualify as true high fidelity devices. They are -Fassive, stable, reliable, and require neither trimming, twealang, nor DIAGNOSIS OF A LARGER SYSTEM Most sjstems consst of more than two devices and often consis of ‘2 mivtue of floating 2-prong AC plug) and safety grounded (2- prong AC pli). In addition, devices may be connected fo external sources of ound curents such azcable TV. Our previous analves of a generalized tuo device system allows ws fo apply the same principles to analyze and treat hum problems in lager systems. Our ‘example system, shou in Figue 5, consists of a luge screen TV recetteruath audio outputs, a teeo preamp contol center a sub- ‘woofer with intemal power amplifiers and a stereo power ampliier {or the satelite spealers. Alldevices have 2-prong AC phigr, except thesub-woofer, which has 3-prong plug. niall, we wllnot male the cable TV connection shown by the dotted ine. The interconnect ‘able ia fol shiedded ype with a #24 gauge damn wie having a resstance of25 md perfoot Now. lefego through the process step. byte. Step 1 is to measure or estimate the worst case ground cutent of ‘each device having a 2-prong AC plug (as described under heading 43), To keepour analysis process a2 easy ae poseble, we willbe using some simplifying cssumptions and approximatons throughout Therefore mecsurements need not be made wth laboratory precision. Our calculated hum levele wil generally be pessimistic by ‘several dB. Step2 isto measure orcalculate the interchassis resistance foreach cable run. Vendor data usually provides either resistance per unit length or equivalent wire gauge information for the cable's shield Remember to include some shield contact resistance at each ‘connector (normally one at each end) as part ofthe total. Ifthe Jensen AN008