Abstract
Abstract-Thc dingnosis of eleclrical motors lhrough lhe deleclion
of faull frcc¡ucncy signal ures in lhc currcnl's speclrum
has bccomc an establishcd standard in the field of industrial
mainlenancc syslems. Ne\'crlheles, ils implemenhtlion on devicrs
wilh low com¡mting po11·cr remains a praclical challenge. Industrial
conlrollers, such as progrnmmable logic conlrollers, or
modern, low cost controller harchntrc, such as lhc Arcluino or
lhc Raspbcrry Pi open sourcc hardware proposals, lack bolh lhc
on-boarcl memory and thc high speed data ncc¡uisition hardware
lo perform nn nccurntc spec tral rmalysis of lhc mnchinc's curren!,
in order lo idenlify the speclral componenls producrcl by rach
lypc of faull. In lhis paper, a signa! conditioning unit, basecl on a
nonl clownsampling melhod of lhe curren!, is prese nted. This unit
reduces lhc amounl of curren! samples lhnt must be processed
by lhe dingnoslic unit lo a mere srnnple JlCI' cmTrnl cycle,
mainlains lhe sub·hcrlz accuracy ncedecl lo rcsoh·e faull, ami
com·erls lhe mains componen! inlo a constan! vrt lue thal can be
casily climinatcd without using :my additional filt ct·. Bcsidcs, il is
implemenled using low cost dcvices, just resistors ami opcraliomtl
mnplilicrs. Thc proposed mcthod is th eo rclirrt lly dcvcloped in lhis
pnpe1·, and il hrts been valiclaled using incluction moto1·s with
broken bars fcd dirrc tly by lhe mains or lhrough varirtble speed
dri\'es.