Elektor Electronics Is Proud To Present An Unprecedented .

GENERALINTERESTyCartesian Co-ordinates(location x/y)circuitboardyxxorigin (reference)revolvingpoint0 Polar ed010024- 11Figure 1. Cartesian and polar coordinates.affordable. Closed systems keep the unpleasant chemicals to a defined area.Do not forg et tha t a ny rema ining etcha ntmus t b e w a s hed from the new ly-etchedboard first w ith w ater and then any still clinging to the b oa rd ca n b e trea ted w ith d eveloper.Ca us tic s od a is chea p a nd in a ny ca s ed oes not keep w ell. Pla in w a ter w ill not d othe job: it does not clean the board properly.After d rying , the cop p er s id e s hould b es p ra yed w ith a la cq uer to ma ke s old eringeasier and to prevent the copper from oxidising.A Boring StoryNext, holes a re b ored in the circuit b oa rdus ing drills of various diameters . In general,three d rills a re us ed : the commones t is0.7 mm or 0.8 mm, us ed for a lmos t a ll s ma llcomp onents s uch a s res is tors , ca p a citors ,ICs ; it is even s uita b le for the lea d s of s omeelectrolytic capacitors. A 0.9 mm or 1 mm drillis used for most connectors, square pins andthe la rg er d iod es . La rg er connectors , s old erp ins a nd s imila r comp onents req uire a1.2 mm or 1.5 mm drill.Usually a mini-drill is used in a simple pillar s tand . Great concentration is req uired toens ure tha t the holes a re d rilled centra lly in14the pads, so that the holes line up ins tra ig ht row s a nd columns . Otherw ise, fitting components such as Dconnectors a nd 40-p in ICs w ill p resent a major problem.La rg er holes , p erha p s 3 mm formounting b olts or s crew s , a re b es tnot d rilled w ith the mini-d rill. Morepow erful machines, how ever, tend topull aw ay at the circuit board materia l, w hich g ives a n uns a tis fa ctoryappearance to the holes. It is betterto d rill firs t w ith a 1 mm d rill a ndthen use a hand reamer.Hig h s p eed s teel (HSS) d rills ca nb e us ed for a ll excep t fib reg la s sb oa rd s . Ha rd meta l d rills a re nots uita b le for fib reg la s s b eca us e theglass embedded in the plastic w earsaw ay the cutting edges of the drill sofa s t tha t it s trug g les throug h thema teria l a nd d oes not p rod uce aclea n hole. Ha rd meta l d rills ha ve a3 mm or 1/ 8” thick s p ind le a nd as ma ll cutter. Thes e d rills , cos tingup w a rd s of tw o p ound s ea ch, a reb rittle a nd b rea k a t the s lig htes tp rovoca tion. It is a n exp ens ive w a yto make holes.Drilling under Softw areControlThe best thing about PCB layout prog ra ms is tha t they a utoma tica llyg enera te lis ts of coord ina tes ofpoints w here holes are to be drilled.If a comp uter-controlled d rillingma chine is us ed , the circuit b oa rdsimply needs to be fixed to the tablea nd the ma chine d oes the res t. Theonly intervention req uired is tochange drills, w hich can how ever bequite tiresome. The program must bes top p ed , the d rilling hea d mus t b emoved to a s p ecific p os ition, or a tleast raised aw ay from the board, inord er to g a in a cces s to the chuck.The heig ht of the d rill mus t b e correct, to ensure that the new bit is atthe s a me heig ht a s the old one: forthis purpose rings can be marked onthe drill spindles. Alternatively, drillsrea d y-ma rked w ith ring s ca n b eb oug ht — b ut a ll from the s a memanufacturer, s ince different manufacturers p ut their ring s at d ifferentheig hts ! And if, a fter a ll this effort,the new d rill is only us ed for a fewholes, the w hole process w ill have tobe repeated just a few seconds later.Manufacturers s oon realis ed thatthis w a s a p rob lem, a nd d evelop edchucks w hich ca n b e op ened fromabove using a small handle. This is ag rea t imp rovement over chucks ,s uch a s thos e freq uently found onmilling ma chines a nd mini-d rills ,w hich require a special tool to openthem.Such chucks make a rather expens ive a d d ition to a s ma ll d rillingmachine, but for the perfect solutionyou can expect to pay as much as athous a nd p ound s : you w ill need a na utoma tic comp res s ed -a ir toolcha ng er w ith tool ma g a zine a nd acomp res s or, a nd everything w illw ork as if by magic. The control program w ill run a little slow er, becausethe drilling head has to move over tothe tool magazine to drop off the oldd rill a nd p ick up the new one; evenhere, comfort has its price.But there is s till a b ig s tep fromthe p erfect ma chine to a s a tis fa ctorily d rilled circuit b oa rd . Ha ve youever cons id ered how reg is tra tion ispreserved through the process of circuit board manufacture? In the PCBla yout p rog ra m the d rilling coord ina tes a re know n exa ctly. The p os itions appear on the film and are thentra ns ferred to the circuit b oa rd —but w here is the reference point?Getting the film cleanly and accurately aligned w ith the edges of thecircuit board is made impos s ible bythe frequently dirty and roughly-cutb a s e ma teria l. The circuit b oa rd isgenerally not rectangular, the materia l b eing cut a fter p rod uction us inga guillotine.A comp uter-controlled d rillingmachine requires a device that guarantees that the drill lands repeatablya t the s p ecified p oint. This d eviceshould be firmly fixed to the machinea nd in a ny ca s e d es ig ned to ma keprecise alignment easy.There a re ma ny a p p roa ches toreg is tration, us ing try-s q uares , p red rilled holes (a rea l p rob lem w henmodifying a circuit), optical registration devices, sticking the film to thecircuit b oa rd , or ‘intellig ent’ motiona na lys is s ys tems us ing a ca mera ,TeachIN and coordinate transformations a ccord ing to reference ma rks ,and many other w onderful methods.In tw enty yea rs of exp eriencedeveloping printed circuit boards ins ma ll q ua ntities , the a uthor ha sfound a much simpler solution:Elektor Electronics3/2001

GENERALINTERESTPolar Coordinates!Here a p oint on a s urfa ce ha s itsp os ition d efined not b y X- a nd Ycoordinates as in the Cartesian system, b ut b y a leng th (d is ta nce froma fixed point) and an angle. Polar andCa rtes ia n coord ina tes ca n b e interconverted w ithout los s of informa tion.In the PCB layout, draw a circularp a d w ith d ia meter exa ctly 3 mm inan unus ed area; alternatively, us e amounting hole. In the circuit b oa rdits elf, b efore exp os ure, d rill a holew ith d ia meter 3.1 mm, in the corresponding place to the pad on the layout. Pos ition the film over the boardso that the pad is over the hole. The0.05 mm w id e cra ck of lig ht a roundthe p a d a llow s exa ct centring b yeye. The centre of the p a d is ca lledthe reference point. If the unexposedb oa rd a lrea d y ha s the rig ht d imens ions , the film ca rrying the p rintedlayout can be rotated until its edgesa re a ls o in a lig nment w ith thos e ofthe board.In the la yout, a s fa r a s p os s ib lefrom the reference p oint, a s econd3 mm d ia meter p a d is d ra w n. Thegreatest distance is across the diagonal of the circuit board, but if there isno s uita b le s p a ce, a d ifferent p la cecan be chosen. The important thingis tha t it is fa r from the referencepoint. This second point is called therota tion p oint. Ca rtes ia n a nd p ola rcoord ina tes a re comp a red in Fig ure 1.After exposure and etching of thecircuit board the reference point w illhave b een etched aw ay, b ecaus e ofthe hole p revious ly d rilled there.Ea ch other hole to b e d rilled is a t aknow n d is ta nce from the referencehole. This g ives us one of the p ola rcoordinates. All w e need now is thea ng le, w hich is d erived from therotation point. At the rotation point,d rill a nother 3.1 mm d ia meter hole.This hole mus t b e ma d e extremelya ccura tely, b eca us e the p os ition ofall the other holes depends on it. It isb es t to d rill firs t a 1 mm hole a ndenla rg e w ith a s ma ll round file orreamer. The copper of the pad gives agood indication of how w ell centredthe hole is.All w e require now on the table ofthe comp uter-controlled d rillingmachine is a pair of small pegs that3/2001Elektor ElectronicsCartesian coordinates: a heresyHas anyone every seriously wondered why CNC machines always work using the Cartesian coordinate system? Why are X-, Y- and Z-axes always used? Why, when such machinesare so difficult to build? The linear guides must be absolutely parallel, because otherwisethe carriage will jam. The axes must be at exactly 90 degrees to one another, or else everything goes askew. The table must be absolutely true and the whole machine must be solidlyfixed to a base.These are all disadvantages. But the greatest disadvantage is the linkage between theaxes. Consider the X-Y table, the original form of hand-operated milling machine. This hastwo handwheels, one to move the table in the X-direction, the other to move the table inthe Y-direction. There are thus two linear guides, one fixed at an angle of 90 degrees to theother, supporting one another. If the lower mechanism has play in it, this is transferred tothe upper one, even if the upper one is absolutely precise. And the lower guide also has tobear the weight of the upper one.This traditional mechanism stems from a time before computers, when positioning alongthe axes was controlled manually by a technician using handwheels. Technical drawings arenormally marked up with XY coordinates so that the successively required positions caneasily be reached by use of the handwheels. In the age of automation the technician is nolonger employed and the handwheels are replaced by motors under computer control. Butthe coordinate system has not changed: in the human imagination everything has a length, abreadth and a height. Curious, when most human actions are polar: ‘take three steps in thisdirection and then turn right’!Imagine now how you would drill a circuit board by hand without a pillar drill. With thefingers of one hand you would hold the board steady and with the other hand you wouldhold the drill. Your drilling arm would be rather higher at the elbow than the other arm,since the mini-drill has a certain height. But you do not move your arms in the X- and Ydirections — no, you turn your drilling arm about the pivot at the elbow and turn and slidethe circuit board to suit. You optimise your movements using your visual system — notperfectly, however, as sometimes you might miss a hole hidden by swarf. You do not need afirmly fixed base on which to work; your drilling arm is fixed at the elbow pivot, and what isbetween this pivot and the circuit board does not matter. Even a small tool between thetwo makes no difference. Your arms and your sitting position need not be absolutely parallel, or even anywhere near, and there are no 90 degree angles to be seen: two pivots areenough!fit smoothly and w ithout play in thetw o holes w e ha ve d rilled . The p egfor the reference point is fixed, w hilethe peg for the rotation point can beslid along a line allow ing for variousd is ta nces b etw een the tw o p oints .The coordinates of the fixed peg areknow n, as is the angle of the slidingrotation-point peg. The drilling datacomprise the X- and Y-coordinates ofthe tw o p oints , a nd s o the d is ta ncebetw een the tw o and the angle theymake w ith the Cartesian axes can beca lcula ted us ing a s imp le p rog ra mon the PC. After a trans lation and arota tion the p os ition of a ll the otherpoints can be determined, no matterhow a s kew the film w a s p la ced onthe circuit board during exposure orhow the board lies on the CNC table.This method of registration is suit-a b le for the ma nufa cture of one-off circuitb oa rd s , or for a numb er of d ifferent b oa rd s .For small-quantity production runs a slightlydifferent procedure can be follow ed.The ConceptThe s ys tem us es tw o p ivots , one for thew orkpiece (the circuit board) and one for thedrilling arm. This allow s any desired point onthe circuit board to be brought into range onthe turntable. This system has the big advantage over a linear construction that only tw obearing points are needed w hose exact separation is the only quantity that needs to beknow n. This requires no expensive specialistcomp onents : the b ea ring s s imp ly ha ve torema in vertica l a nd free of p la y. To a lig n a na xle to p rofes s iona l s ta nd a rd s of a ccura cy,tw o so-called taper bearings are used. Theseca n w iths ta nd enormous forces a nd a re15

GENERALINTERESTFigure 2. The well-known principle of the ‘axle drive’.exp ertly ma d e to rema in s olid a nd p erma nently free of p la y. This is the ma in rea s onw hy our machine is so economical to build.A rather significant disadvantage ought tobe pointed out. Normally, in conventional linea r ma chines the a xles a re long threa d edrods supported by bearings and turned by amotor. A nut is fixed to the moving part andmoves b a ckw a rd s or forw a rd s a ccord ing tothe d irection of rota tion of the s crew. Thismecha nis m na tura lly g ives a hig h effectivegear ratio. Suppose that the screw gives a linear movement of 4 mm per rotation and is driven by a s tepper motor w ith an angular res olution of 200 steps per revolution. The linearmotion corres p ond ing to one s tep is4/200 0.02 mm. That is ideal for this kind ofma chine. Gea ring is therefore comp letelyunnecessary.Our ma chine is not d riven b y a threa d edrod , b ut ra ther w orks d irectly w ith a ng ula rmotion. The 240 mm long tool a rm ha s atravel of240 mm 2 3.14 1510 mm(circumference of circle w ith a rm leng th a sradius)A stepper motor driving this directly w ouldta ke 200 s tep s to ma ke one revolution; thedis tance corres ponding to one s tep is therefore 1510 mm/200 7.55 mm, rather too greatfor a CNC ma chine. For a d es ired res olutionof les s tha n 0.04 mm w e need to g ea r themotor d ow n b y a fa ctor of a t lea s t7.55 mm/0.04 mm 190:1. That is not exactlystraightforw ard.16Gearing MechanismIn the opinion of experts in the fieldthere is no s imp le g ea ring mecha nis m tha t g ives s uch a hig h ra tioapart from a w orm drive. It is mathematically impossible to achieve theratio w ith tw o or three gears w ithoutusing an unreasonably large numberof teeth. At lea s t three s ta g es ofg ea ring d ow n a re req uired w ithma ny ind ivid ua l a rb ors tha t mus teach have good quality bearings.This difficulty, and the unreas ona b ly la rg e a mount of p la y in s uch amecha nis m (a nd hence p rob lemsw ith repeatability), w ould cancel outa ll the a d va nta g es of our id ea . Butyou w ould not ha ve this a rticle inyour hands if the problem could notb e overcome even d es p ite the firmb elief of the exp erts ; a nd w ithoutemp loying s uch a rca ne d evices a sflexib le comp onents fitted b etw eengears w ith intricately-cut teeth, ballbearings moving in contorted orbitson even more contorted g uid es , orbelts w ith different patterns of teethon either s id e. If you w a nt to knowmore, there is plenty to read on thissubject on the Internet: you w ill findid ea s there cons id era b ly more curious tha n our d rive s ys tem. Wea chieve the d es ired g ea ring ra tious ing four g ea rs . The p rincip le ha sb een know n for a very long time,a nd there is even a VDI (Germa nAssociation of Engineers) documenton the s ub ject (s ee b ox); how ever,p ra ctica lly no-one ha s yet rea lis edthe potential of the idea.Our d rive us es the p rincip le ofs ub tra ction. Ima g ine a tra vela torsuch as those found in airports. Ourtra vela tor is circula r a nd runs w ithcons ta nt s p eed . You a re running onthe tra vela tor in the op p os ite d irection to its rota tion; the nea rer yours p eed to tha t of the tra vela tor, thelow er your s p eed in rela tion to afixed point. If you match the speed ofthe travelator exactly, your net speedis zero: you ha ve a ma ximum ‘g ea ring ratio’, the tw o speeds cancellingone another out.The s itua tion is s imila r in ourdrive system. A spur turns w ithin anannular gear w ith internal teeth: thea ctua l va lue of the g ea ring ra tio isnot important. Fixed to it is a seconds p ur g ea r, w hich turns in a s econdannular gear. Note that the tw o spurgears are fixed to the same axle. Thetw o comb ina tions of tw o g ea rs —i.e., the ratios of spur gear to annualgear — are slightly different from onea nother, b ut the more s imila r theya re the g rea ter the overa ll g ea ringratio obtained. Since the second spurgear turns at the rate determined bythe first stage rather than that determined b y the s econd , the s econda nnula r g ea r mus t ma ke a comp ens a ting motion. In fa ct, w e s ub tra ctthe tw o speeds, just as in the travelator example. This explanation mays eem ra ther comp lica ted , b ut theillustration in Figure 2 should makematters clear.Normally more teeth are requiredto a chieve a hig her g ea ring ra tio.Here, how ever, tha t is not the ca s e,b eca us e it is merely the d ifferencebetw een the tw o ratios that matters.The a uthor ca lls this d rive s ys tem‘a xle d rive’ a nd a p a tent ha s b eena p p lied for. The cha nces of s ucces slook slim, how ever, since the designhas been know n for a long time.A g la nce a t the interna ls of thed rive s ys tem in Fig ure 3 immed ia tely s ug g es ts ma ny p os s ib ilities .For exa mp le, there is enoug h s p a cein the housing to include the parts ofa motor; then one w ould ha ve a nincredibly small unit w hich w ould benot jus t a gearbox but a motor w itha slow ly-turning output shaft w ith aElektor Electronics3/2001

GENERALINTERESTvery hig h torq ue. One need onlythink of the countless applications invehicles , s uch a s for electric mirrora nd s ea t a d jus ters , or for electricw indow s, w here a flat drive systemis req uired to fit b etw een the d oorpanels. In robots; in component handling systems; this drive is ideal any-w here w here a s low b ut p ow erfulmotion is required.The Drive in ActionOur d es ig n not only ha s cons tructional advantages over normal linears ys tems , b ut a ls o w orks b etter inmany w ays. What happens w hen the drillinghea d of a linea r ma chine is a t one end of itstra vel a nd mus t move to the op p os ite end ?The head must be moved the entire distance.In our design, things are circular: define 360 as one ‘end’ of the travel and 0 as the other,a nd the tw o a re the s a me! If w e a re a t 360 a nd need to move to 10 , w e d o not need toThe drive system: how it worksWhen we started looking for a simpledesign for the drive system to give preciselythe high gearing ratio required for ourmachine, we were met by shaking heads:the experts told us that high ratios couldonly be obtained by using multiple separatestages, each with a small ratio, multiplyingtogether to give the total desired.Using an ordinary planetary drive a practically-realisable gearing stage with a ratio ofup to about 7:1 can be built. This is due tothe fact that multiple spur wheels must turnwithin the annular gear and must clear oneanother. The diameter of the spur wheelsmust be less than half that of the annulargear to allow clearance, but must of coursenot be too small of they will have too fewteeth to run smoothly. If higher ratios arerequired, multiple stages can be cascaded:with two stages we can obtain a ratio of7x7:1, or 49:1. Add another stage and wehave at most 343:1 (although for constructional reasons the maximum obtainable isonly just over 300:1). Each stage takes upspace, and a three-stage drive with a diameter of 40 mm with attached motor mighteasily have a total length of over 100 mm.Further, each stage brings with it play in the mechanism andangular error, which all add together. Economical three-stage planetary drives have a play of about 3 degrees in the output shaft.Although that may sound small, in our design it would translateinto an unusable 12 mm of error at the drilling head.Precision drives in the middle price-bracket — say around onehundred pounds — reduce this play down to just below onedegree. For our application, this is unfortunately still completelyunusable.Take a look at the figure and imagine the following sequence ofmotions.We will use different tooth counts from those shown in the figure. Suppose the lower annular gear has 45 teeth, its spur wheelhas 10 teeth, the upper annular gear has 50 teeth and its spurwheel has 11. These values will provide an excellent example ofhow the drive works.The lower annular gear is fixed, and the long main shaft, shownpassing through, is free to turn. This is the motor’s output shaft,which turns the plate on which the small spur wheels aremounted.The spur wheel is free to turn on bearings in the plate, and ifthe main shaft is turned then as the spur wheel runs around theinside of the annular gear, it turns on its own shaft. If the mainshaft is turned through 360 , then the spur wheel will have turnedon its own axis through 4.5 revolutions since its gearing ratio is3/2001Elektor Electronics45:10, or 4.5:1. Although the two spur wheels are free to turn relative to the plate they are fixed to one another. So, when thelower spur wheel turns through 4.5 revolutions, the upper onemust do so too.Now if the upper spur wheel, with 11 teeth, turns through 4.5revolutions, that equates to 49.5 teeth. But the upper spur wheelturning through 49.5 teeth in the upper annular gear causes aproblem: the annular gear must make a compensating motion, andthe size of this motion must be 50 minus 49.5 teeth, or exactlyhalf a tooth.Since the upper annular gear, with 50 teeth, moves by a halftooth for each revolution of the main shaft, this means that 100revolutions of the main shaft are required to obtain an entire revolution. With this combination of 45:10 and 11:50 ratios we havetherefore obtained, in one stage, a gearing ratio of 100:1. In ourapplication we use ratios of 60:22 and 23:63, and with no othermodifications to the construction obtain a gearing ratio of 231:1.The diagrams show the construction to scale and with exact teethcounts.In the 19-page VDI (German Association of Engineers) guidelinedocument number 2157 of September 1978, which deals withplanetary drives, basic constructional rules and calculation methods are given. The exact arrangement dealt with in that documentis called a ‘simple planetary drive’ which for technical reasons isnot suitable for higher gearing ratios.17

GENERALINTERESTFigure 3. The ‘axle drive’ as used in our machine.move through 350 , but only through 10 . Forthis rea s on our s ys tem is four times fa s tertha n a conventiona l CNC ma chine w orkingover the same area. Also, the mass moved inour s ys tem is much les s , w hich a llow s themachine to w ork faster still.The construction of the machine, as can beseen in the photographs (Figure 4), aimed ama ximum s tiffnes s a nd minimumw eig ht. The a rm a nd the ta b le ca nbe moved at 70 mm per second, andthe s tep p er motors only req uire as imp le d rive circuit b eca us e theydraw less than 1 A.Na tura lly, our ma ch

Elektor Electronics is proud to present an unprecedented home construction project: a CNC PCB drilling machine which is economical, accurate, quick, and bristling with new ideas. . the tool magazine to drop off the old drill and pick up the new one; even here,