CA2074950A1 - Method for the automatic selection of containers and a measuring arrangement for this, as well as a plant with such a measuring arrangement - Google Patents

Abstract

ABSTRACT

In order to increase the reliability of detection of gaseous contamination in containers, gas (G) from the container is subjected to several different methods of analysis (54a to 54d), each of which emits differing output signal courses (I1 to I4) depending on the contaminants and their concentration. A vector of the state vector type (P Gas) is formed with the output signals (I1 to I4) as a state variable and checked to see whether it defines a permissible or an unacceptable state of contamination (56). On the basis of this examination the decision is taken whether a container is acceptably or unacceptably contamined.

Description

- ` - la- ~a7~9~0 Method for the automatic selectio~ of contai~er~ and a measurinq arran~ement for this, as ~ell as a plant with such a measur1nq arra~ement.

Th~ pre~ent i~vention relates to a ~ethod in accorda~ce with the introduceory part of claim 1, a mea~urin~ arrangement in accordance ~ith that of claim 17, ag ~ell a3 to a plant with such a measuring arrangemen~ according to claim 31.

From the EP-A-0 306 307, ~hich herewith is declared an integral part of the present descripeio~, it is k~own, within the framework o~ the recycli~g of co~tairler~, to detect on empty contai~ers, in particular pla~tic container~, e.~. plastic bottles, whether any contaminations are pre~e~t inside the container.

To this e~d it is proposed to detect ~uch co~tami~ations with the aid of an ionizatio~ tech~ique, e.g. flame ionization or photo ioniza~ion irl the UV-rarlge, and i~ need be to eliminate the contaminated co~tain-ers before re-filling them.

From the W088~0086 a selectio~ method of containers is known, with which either dis~illed ~ater i5 sprayed into the co~tainers~ water that may ~ow be contami~ated i removed from the contai~ers and a~a-lysed, or water i5 ~prayed i~to the containers, shaken in same and then analy~ed.

2~7~

'~, i~crease the analysi~ redund~ncy, it iq ~ell k~o~n to use two di~ferent a~alysi~ ~echnique~. The outpu~ signal of the detector~ of the e~o ænalysi~ teehni~u~ first of all compared ~ith reference data, the compari~o~ re~ult i5 ~hen co~p~red with a correlation value range ~hich depends on ~he ~peciic analysis ~echnique and a specific product ~hich originally ~a~ present in the con~ainer.

Orl ~hich b~si~ the deciding correlatio~ factnr~ are determined accor-ding to this publication, i~ not indicated. Described is the use sf detector~ ~hich specifically detect a given substance, e.g. a sugar analy3er, and the providing of two such detector~ that respond with a narrow band to a substance ~ould make it po~sible to detect compou~d contamination~.

In principle, with regard to the technique described in the W088/
0086, it must be regarded as di~advantageous that liquid samples must be eaken from the con~airlers in questio~, ~o that the test method ba3ed thereon i~ extremely ~low. On the other hand, because of the slow~esq resulting from the use of a liquid, there is ~ufficient time to test the liquid sample for co~taminations u~der conditio~s ehat are practically similar to those in the laboratory.

The present inverltion proceeds from a method as described in the abovementioned EP-A-0 306 307, according to which gas from the containers is analysed. Already because of the flow properties of - 2 ~

ga~, compared to tho~e of a liquid, thc la~t~r m~thod re~ult~ i~ R
co~ iderabl~ shor~e~ing of the proce~s cycles compar~d to the ~ethod of W088/00862, ~hich i particularly important or the ~eles~i~g of containers that are fed to the li~e in quick succession.

Proceedi~g from ~uch a yas analysi3 proces3, the prese~t inventio~
proceed~ rom the recognition of the problem that known analysis ~ethods for gas analyRes 3upply output signals that depend both on ehe contaminatiag ~ubsta~ce ~hat ig bei~g detected as well as on its co~centration, i.e. o~ two variables. Thi3 cau~es ambiguity in the evalua~ion of such individually viewed signals. Often it is ~ot pO5-~iblP to differe~iate whether a detected output si~nal i~dicates a co~centra~ion a of the subseance A or a conce~tration B of the sub-stance ~. With a measuring operation that is being considered, the co~centration a of the substa~ce A may provide the same result as the concentratio~ 8 of the subs~ance B.

Under the a~pect of reliabiliey, it i3 the object of the pre~e~t in-ve~tion to ~olve this problem.

This is achieved with the method of the above~entio~ed type when one proceeds i~ accorda~ce wi~h the wording of claim 1, or with a uitable measuring arra~g0ment i~ accordance with the wordin~ of the characterizins part of claim 17.

It was recognized that differe~t ~as analysis technique~ give output Yignal~ that depend o~ the co~ce~tr~tion and o~ the contaminati~g substance, re pectively. Thes0 differ0rt output ignals, becau~e of the diff~rent analy~is e~chuiques, are line~rly indepe~dent of o~e anoeher in ~he sen~e tha~, for example, there doe~ not exist a simple proporeiorlality be~een the different sis~als. The transfer charac-teri~tics of ehe different gas analy~is techniques are characteristi-cally differe~t i~ their depende~ce on the variables ~su~sta~ce~ and "concentrationa. The ter~ noutput 6ignal" in this connection denotes all signal parameter~ that can characteriz2 a signal, e.g. amplieude, pha~e, step re~pon~e, pulse response.

Accordirlgly, uieh the proposed procedure ~o redundancy i~ created, ~hich always con~ t of increasing, in the statistical sense, the reliability of the overall measuring operatio by a plurality of the ~ame type of measurements, but a tese result is created only by uSi~s differene analysi~ techniqueq, e.g. in a X-, Y-, Z-coordinates system three ~easurements each for X-, Y- and Z-position coordinates define the result, i.e. the positisr~ vector.

With the procedure accordi~g ~o the invention the selectio~ method a~d the ~easuring arrangement used for ehi4 become extremely reliable i~ that, in particular, no containers are selected as admissibly con~
tami~ated ~he~ th~y ar~ inadmissibly contaminated. The greater the number of independently used analysis techniques, the greater the aforemeneisned reliability becomes.

- 2~g~3 Becau~e it i9 po~Yiblo to carry ou~ the different analysis technique3 i~ par~llel, i.e. ~i~ul~a~ou~ly or qussi-~imultaneou31Y! there i~ ~o sigrLificarlt ~lowing do~ of the selection method on the gas i~ide or outside the respective cor~tainer. A~ already mentio~ed, this is o~
decisive importarlce for container3 that quickly succeed one ano~her durin~ line examir~ations.

Accordir~g to the uordi~g of clai~ 2, gas analy~is tech~ique ~hat ca~
be u~ed are: infra-red ~bsorptio~ measuremerLt, ~easurement6 by mean of semi-conductor sas se~ors, ~easurements by meaus of electro-chemical cells, ionizatio~, especially photo io~izatio~ arldJor spark ionization, a~d measurement of the resultan~ gas ionization, or pos-sibly measureme~t by means of mass spectroscopy. Preferred, ~ecause of the simplicity and rapidity, i9 the mea3uring com~ination by means of semi-conductor gas sensors, photo iorlization and spark ionization.
Ie is al50 readily pos~ible to use, withi~ the framework of the pre-sent irlventiorl, di~ferently respondirlg semi-co~ductor gas sersors i~
the se~e of two different a~alysis technique~, and/or to use, in the indicated sense, the photo ionization or spark iorization or another of the mentioned analysis tech~iques ~wice or ~everal times with dif-ferent outpu~ ~ig~als irl the se~se of the present inven~io~.

The group of preferred techniques also includes the infra-red ~b-~orptior mea~ure~ent, e.g. wi~h infrared ~emi-conductor se~sors, as ~arkeeed~ for example, ~y the firm ~ohl Se~sors Incorporation, 70 W Barham Avenue, US-Santa Rosa, which, fitted wi~h ~arrow-baud, opeical fil~er~ arld ~he~ providing an i~fra-red trarlsmitti~g ~ource ~hat giv~s off light in the IR-ra~ge irl questiorl, determ1~e ~hether o~ prodeter~ined absorptio~ bands the transmitted r~diatio~ i~ a~-sorbed or ~ot by the ga~, based on ~hich a spQcific conclu~ion i8 reached regardi~g the pre3erce or abse~ce of specific substarlce co~-eaminaeions ard their corcerltratio~s.

Although ~ith cer~ai~ con~ai~er4 ~ith co~ainer ~alls that transmit in ~he IR ~aveband ir~ question, an IR ab40rptior. measurement could be carried out ~y irradiati~g the container, also then the transmis~ion conditiorl~ of the container wall are subject to such great specimen dispersions that a reliable ~etectins of the said sub3tar~ce co~tami-nation i~ the ga~ is no~ readily possible For this reason, with this procedure, i.e. ~hen chooRing the IR absorption measurement as one of the arlalysis eech~iques, the gas i4 subjected to the test as a gas sample, or, using light conductors o~ a lance IR-light i~ beamed into the container, recorded on this la~ce after traversirlg a gas path, a~d a correspondins ~ig~al i5 tapped off for the e~aluation.

Furthermore, a~ indicated i~ claim 3, i~ is proposed to provide, with particular prefere~ce, as one of the ~aly5i9 techniques at least one ~emi-co~ductor ~ensor, in particular a semi-cor~ductor gas se~sor, which en_ures the uqe of a particularly simple ~rLalysi~ tech~ique.

As a ~atter of fact, semi-co~ductor ga~ 3er~sors are k~own, as made arld marketed, for example, by the comparly Figaro Engiueering, Osaka/

. - 7 ~ ~ f~

Japa~. Such semi-co~ductor ga~ ~e~30r8 c~n be i~troduc0d ~xtremely easily and, bec~use of heir small size, al~o directly i~to the co~tainer or ~lo~gside a flow path for test ga~ from the co~tai~0r~, at a~y point, for th~ analysis of the gas samples. Under the aspect of reliability, ~y providing one or several such ~emi-co~ductor gas 3e~sor~ it is po~ible to al80 realize ~i~h such 3ensors differe~t arlalysis techniques in ehe 3e~se of the prese~t invention, or to create higher redundancy.

However, semi-conductor sensors, a~d i~ particular semi-co~ductor gas sensor-~ have relatively lo~g step reqpo~se times, i.e. ~hen a sudde~
cha~ge i~ the gas occurs o~ the input side, its outpue sigual changes similarly to that of a low-pass filter, a~d relaeively 510wly moves a~ymptotically towards the correspo~di~g end value.

This problem, which from the point of view of the rapidity of the process for~ed an ob~tacle fsr the u~e of semi-co~ductor gas ~e~sors, is eliminated if o~e proceeds a.~ indicated i~ claim 4, i~ that the output sig~al of the at least o~e semi-conductor gas se~sor is diff~-rentiated ~ith re pect to time, and the result of this differe~tia-tion, i.e. ~he initial sig~al climb i~ evaluaeed for the selectio~.

Si~ce as output sig~al with such semi-co~ductor gas se~sors usually the output resi3ta~ce varies, the cha2ge with respect to time of it ou~put resista~ce is, therefore, evaluated.

~7~f~
.
A~ ehe ti~e differ~ti~tio~ of the ~mi-conductor ga~ 90~80r output ~ignal~ correla~es ~ith the ~axi~u~ valu~ of the output 8i~nal ~hich it trie~ to reach, already ~hortly after there occurs on the i~pu~
~ide a cha~ge in a gas conce~tration ard/or i~ a ~ubstance, the selectio~-e~fect1ve sig~al car~ be ascertained from the ~aid diffe-rentiation.

From the above it can now be noted, among oth0rs, that ~hen a semi-condu~tor gas sensor has detected a ga~ contamirla~ion which moves its outpu~ 3ignal in the direction of a new end value, this gas sensor, because of it~ ~e~ory", ~ill now analyse a further gas sample fal-~i~ied ~ith the re3ult of previously de~ected mea~ureme~ts. This ~ould mean that a provided gas ~e~sor would again drastically 310w down the process cycle, as i~ i3 necessary to ~ait until the effect of a previou~ ga~ a~alysis has died away.

When proceedi~g i~ accordance with the wordi~g of claim 5, this is prevented i~ that at least two sets comprisirlg a~ least o~e semi-conductor ga~ sensor each are provided, a~d the tes~ gas from successive corltainers i5 fed to different sensor sets, so that ~he irldividual sets are given ~i~e to re-se~ their output signals to a basic value without increasirlg the time of the process ~ycle from co~tai~er to corltainer.

So that, ~hen proceeding in this marmer, the supply li~es and the semi-co~ductor gas sensor itself can be cleaned, i~ is proposed, as indicated i~ claim 6, that after a meaYUreme~t the semi-conductor ~ ~ 7 ~
- g -gas e~sor, a~d therefor~ al~o the supply lirL~, ar~ rin3ed ~i~h a rin~i~g ga~. With ~uch a ga~ ri~ing, because of ~he type of the rinsirLg ga-~ arLd/or its ~lo~ along the se~sor, there occur on the ~e~80r a behaviour similar to that ~hich occtlr.~ duri~g the afore-~entio~ed detecting of a contamination. As a result thereof such a gas ~en or, because of the rinsing operatio~, ayairl ca~not be used for ~ome ti~e for eonta~inatio~ mea~urements.

It rlo~ i~ proposed, irl accordance with the further wordi~g o~ claim 6, to adapt the type of rinsing gas and/or the ri~si~g gas flow to the flow of u~contamirlated test gas from the contai~er in such a way that, whe~ changing over from rinsing to mea~uring or vice-versa, this change produces an only mini~al, if any, change i~ the signal at the output of the ~emi-conductor se~sor. Accordi~gly, the serlsor does not ~experience" a cha~ge from testing cycle to ri~sing cycle or vice-versa ~hen the ga~ f~d in duriny the testirLg cycle is no~
coneaminaeed.

In accordance with the wordi~g of claim 7, to remove the ~as from the container preferably a carrier gas is uqed arld, in the se~se of what has been said ~ith reference to cla1m 6, the carrier gas i3 used as rlnslng gas.

Preferably, in accordarlc with the wording of claim 9, the mentioned adjustment is carried ou~ such that duri~g the cha~ge-over rinsi~g/
mea~uring a~d vice-versa, the output signal o~ the ~emi-conductor 2 ~) 7 ~
~ensor re~ains ~ubsta~tially co~ant.

From the EP-A~O 306 307 mentioned ~t the ou~set it i~ k~owrL ~o exa-~ine the ga~ sample from ~ container by flame ionizatio~ follo~ed by a~ analysi~. Thi~ i3 a relatively slo~ proce~s and in addition al~o di~advantageous from the simplicity point of view. On th~ one hand, as a ~atter of fac~, during the flame ionization with a hydrogen flame, the ~as flo~ing pa~t the flame ~ay not sigr~ifica~tly di~turb the flame, which s~ts limits on the flo~ velocity a~d ther~fore the rapidity of the ~ea~urir~g operatio~, and furthermor~ the supplying of flame gas is expensive.

For this reason, in accordance with the ~ording of claim 10, as o~e analy~is technique the ~as is preferably exposed to an electrical discharge gap and ie~ di3charging behaviour and/or the di~charge-related gas ionization is evalua~ed a3 an output sig~al for the 3election. The providi~g of an electrical discharge gap, similar to the spark plug of a combustion engine, is extremely simple as this can be miniaturized, i~ not ~usceptible to co~tami~ation and, being flexible, electricity can be supplied to it practically anywhere.
The mea3uring operation ~i~h thi3 i3 very quick as it i~ at l~ast ~ithin li~it~ independent of the flow velocity of the test gas, and for certain applicatio~s the ~park ionization, urllike the flame ioni-zation, can be u~ed in the to be tested container itself.

If, following the wordinq of claim 11, the gas i~ ionized ~ithin the framework of one of the a~aly5i5 techniques, a simple evaluatio~ i5 2 ~

realized b~ ~e~uring the mobility o~ the io~s.

A preferred embodimene of the method ~ccordirlg ~o the inv~ntion is specified further i~ clai~ 12. According to thi~, ~ith each of ~he a~aly~i~ tech~ique3, u.~ing calibra~ed gas 3amples, in ~ n-dimen3io~al seate Bpace ~ith ~ > = 2, ~ith coordinates that each corre~pond to the output ~igr~als of the analysiq techrlique3, at least one area is defined, ~i~h pointq correspondi~g to coordinate values which define ad~i~sible contami~ation~, and point6 outside the at lea3t one area ~hich defi~e a gas sta~e corresponding to inadmi~sible contamina-tion~. The output ignals ascertained on the gas by me~ns of the technique~ are auto~atically evaluated, a~ coordinate values, to determine ~hether they jointly de~ine a gas state within the admis-sible area or no~.

The gas ta~e i5, therefore, ascertained by means of a ~state vec~or"
and i~ is evaluated whether this state vector lies in an admissible or inadmissible range.

To further~ore preve~t with to be tes~ed containers that dominant gas portions or gas portions ~hat are present in too high a concentration will make ~he ~ubsequent analyses dificult, and to therefore reduce the concentrations in question to 3uch a~ extent that the provided analysis techniques can operate in the provided ranges of measurin~
characteriseics, it is proposed ~o proceed a~ indica~ed in claim 1~.

lf one procecd~ dicat~d in clai~ 14, i.e. if the co~tainer i~
heat~d, it become~ po~sibl~ ~o expel i~to-the ga~ ~o be tested co~-ta~inaeion compon~nt~ that have bee~ ab~orbed by the cont~iner ~all.

Furthermore, ~ith the said selection there frequerl~ly exi~ts a pro-blem irl that the original filling, al~o after emptying the co~tai~r, produces or may produce such a high degree of contaminatio~ that this co~ceal~ other contami~ation~, in the ~ense of a signal-to-noise re-flection. It ~ould, therefore, be extremely advantageous if a ~imple, reliable procedure could ~e found for ascertai~ing what was ~he ori-ginal conten~ of a corltairler.

Thi~ i~ achieved by the procedure indicated in claim 15.

As a result thereof the Analysis work is reduced in that, by reading the indica~ed marki~g, the origi~al co~tent becomes kr~own arld corr~s-po~ding co~tami~atiorls can be selectively suppre4~ed or take~ ints co~sideration in fsvour of other contaminations or the corresponding sigrlal poreions.

With the aforementiorled semi-conductor se~or , in particular semi-conductor gas 3erl30rs, the "memory" behaviour of which has beerl explained, there al30 occur~ a ~pecial problem when o~e or several of ehe semi-conduceor ~ensors detece a high coutamination value that pu~hes up its output ~ig~al, 30 that such a ser~sor theu al~o requires a correspondi~gly lo~g time to swin~ back eo its origi~al value. As a ~`~ 207~ o r~ult thereof 3uch a ~mi-conductor 8en80r ~ould the~ aqain ~ot be ready for ~ubsequ~nt examination~, and the exa~ination eycl~ ~ould be extended accordirgly u~til tho said semi-conductor ga~ 3en~0r ha~
again reached itR readirle~s to ~e~ure.

To prevene this, in accordQ~ce ~ith the ~ording of claim 16, th output ~ignal~ of the provided ~emi-conductor sen~or ~ets ~re teYted to 3ee whether they exceed a predeter~ined value. If ~o, the set i~
question i~ di~abled at lea3t or the i~mediately ~ollowi~g analysis.
Orle of ehe other semi-conductor sensor sets which iq ready ~o ~easure is ~hen used.

With ~he preferred time differentiatio~ of the output signals of the semi-conductor 4e~30rs ~ na~urally the time di~erentiatio~ of this signal is tested to see whether it exceeds the predetermi~ed value, so that also here one need not ~ait until the output sig~al of the 4emi-conductor se~sor level~ out o~ the output sig~al level that correspond~ to the contamination.

Since any~ay, preferably a~d according to the invention, successive gas sa~ples are fed sequentially to different semi-corLductor sensor sets, so that, for example, ehose ~hat have just beer~ used can in the meantl~e be rinsed, in the ca es mentioued here preferably more tha~
o~e measuring cycle i~ left out u~til ehe set which i5 over-saturated i~ ehe indicated se~se is again ready to measure, which carl easily be deter~i~ed by monitoring its output signal, whilst the subsequent ~ ~J ~ 3 ~easuring cycle~ are carried out unaffocted on oth~r 3~ts.

A~ ~as ~entioned more to~ard~ the begin~ing, a preferred ~aly~is technique u3ed in conn~ction with the present i~vention consi~t~ of the u e of electro-chemical measuring cells, a~ marketed for example by AMS A~alysen-Meqs-Systemtechnik, D-Dielheim, by means of ~hich, the presence or abserlce of ~pecific gas components can be detected ~ithin a ~arrow band.

A measuring arrangement according to ~he invention i~ ~pecified in the clai~s 17 to 29.

A tast pla~t according to the inve~tion ~i~h a measuring arra~gement according to the irventio~ is specified ir~ claim 31, with which a co~veyor arrangement is provided for plastic bottle3 that are con-veyed as containers i~ streamlir,e ~ashion to and from the measuring arra~geme~t, and ~ith which every bottle can be tested with great reliability and i~ a fast rhythm, unlike spot checks which, in par-ticnlar in con~ection with the re-u e of food containers, ca~not be used for reason4 of safety.

In the following the i~vention will be explained, by way of example, with reference to figures.

Theqe show:

~ ~) 7 ~
~ 15 -Fi~. 1 diagra~matically a di~charge gap u~ed preferably ~ one of the gas ~naly~ echnique~ ~ith the method accordi~g to the invention or on the mea3uring arrangeme~t according to the inve~tio~ for ehe ionizatio~ and simulta~eous deter~i~ation of a mea~urable variable tha~ is at least co-~ignificant for the co~tai~er selection accordi~g to the i~ventio~.

Fig. 2 proceedi~g ~rom the illuqtratio~ of Fig. 1, a further embo-diment wherei~ the discharge current is regulated and the said mea~urable variable i~ determined from the behaviour of the regulating circuit, Fig. 3 diagrammatically, the u e of the dischar~e gap for the ionization of the contai~er gas in tha co~tai~er itqelf, ~ig. 4 diagrammatically, an embodi~erlt for ehe di~charge iorization of the gas and subsequent, electro-static ion separation, for detcrmi~i~g a measurable variable as a preferred ana-lysis tech~ique or detector device, Fig. 5 a~alogously to the illu~tratior~ of Fig. 4, a further embo-di~ent, wherein ion separations are detected in depe~dence o~ respective ion ~obilitie~ as measurable variables, Fi~. 6 diagrammatically, ~or use i~side a ~o be tested coritainer, a di~charg~ ionization device, followed by an elec~ro-~tatic, 2~7~o ~obility-selec~ively operati~g ion sep~ra~ion d~vics, Fig. 7 diAyram~atio~lly, the provisio~ of a pre-~election to prevent explosions i~ the ca~e of certain contamina~ion ~ubs~a~ce3 ~nd ~ith an electric di3charge gap in~ide, (a), or out~ide, ~b), the contai~er, Fig. 8 u~der the main aspect of the present inYention, a signal flow/fu~ction block diagr~m of a ~electio~ deYice according to ehe invention, operating by the ~ethod according to the invention, Fig. ~ a sig~al flow/functior~ block diagram of an analysi~ unit uith 4emi-conductor sen~ors, in particular semi-conductor ga~ sensors, Fig. lOa the qualitative response behaviour of a -~emi-co~ductor gas sensor to ri~sing ga5/te5t gas eycles, Fi~. lOb the adjusted behaviour of the semi-conductor gas sen or, Fig. 11 diagrammatically, the blocX diagram of a ga~ sampling u~it.

A~ me~tio~ed at the outRet, the prosent invention relate~ to the problem of irlve~tigatiny the state of contamination, in particular o~ empty contai~ers. For exa~ple, with plastic bottle~ uhich are 2~7~9~0 received for re u3e, ~her~ ~xi3t~ greQt uncertair~ty ~8 to how they ~ere used a~t~r their origin~l contont, ~,g. mi~er~l ~ator, fruit juice~, etc., had been o~p~i~d. It i8 known ~hat such bottle~ are ofte~ u~ed for oth~r purpo~e~, for example i~ the household, e.g.
for qtoring soap ~ater, herbicides, ~gi~e oil, acids, petrol, benzene etc. If ~uch substances were stored in containers that are made available for re-use ~ith a new ori~i~al filli~g, ~ith certai~
cate~ories of co~tami~ation substances arl adverse e~fect on the taste of the ~ewly filled origi~al conten~ ca~ be expected, or such a co~tairler can no longer be used for re-filling because of the in-compaeibility of the contamination or because they may be harmful to people's health.

For this reason it must be ascertained ~hether and which re-qidual contamina~ions are prese~t in the containers, so that a selection car~ be carried out betwee~ container~ that can no longer be used for a new origi~al filling, those that, for example, firs~ have to undergo a special cleani~g process, and those that can quite safely be re-filled.

Ir~ this co~nection it must be borne in ~ind ~hat, depending o~ the material of the container, in particular with plastic bottles, certain of ~he me~tioned co~tamirlation substa~ces are absorbed by the wall material, and the contamination is 810wly desorbed i~to a freshly filled co~tent.

lt~ -2~)7l~9~D

A~ i~ certai~ C~8e5 al~o ~he content o~ a contair~r m~y be conta-~i~a~d, ~d the ~a~ lyi~g above this i~ then cont~i~ated, the invention can, ~ieh regard to all it~ a~p~cts, also be u~¢d o~
co~tainers that have already bee~ filled. The procedure accordi~g to the invention i~ explai~ed ~ith reference to Fig. 8. ~ere, a co~bi~ation of different analysi~ tech~iques is used, a~d their output ~3ignal9 8re evaluated combined.

Fir~t of all, with reference to Figure 1 to 7, 9, preferred ana~
lysis tech~iques are described ~hich, performed accordin~ to the invention, are especially suitable for the intended u~e within the system according to the invention.

Fig. 1 sho~s diagra~atically an em~odiment of a device for a~cer~
tairLing a ~ea~urable varia~le ~hich at least is co-significant as to whether ehe gas presen~ in a container does ur does not contai~
contaminations in a correspondi~g conce~tration o~ a specific sub-stance group.

By way, for example, of a sampling line 1, a gas sample G i5 draw~
off from a not illustrated, empty or partially filled co~tainer, possibly also one from outside the cQ~tainer which is in direct contact therewith or with lts ~illing, a~d is moved past a dis-charge gap 3 uith arl electrode pair 5. The gap 3 is operated by mea~s of a current source 7~ The discharge i3 produced as a corona discharge or a ~park discharge.

2~7~5~

If, by the noe illu~trated ~uctio~ de~ic~ - in thi3 connectiorL ~ee ~ig. 11 - th~ ~a~ ~a~pl~ G i~ ~uckod up fro~ the container and moved p~3t the di~charge gap 3, it~ discharge voltagc uill cha~ge. This vol~aye UF is ~ea~ured ~ith a Yoltage mea~uri~g device 11.

The output sign~l of the voltage ~easurirg device 11 is evaluated as a ~ea~urablç variable and to ~his end i3 fed, for example, to a compara~or u~it 13, to ~hich other reference ~ignals can be ~ed by 8 referer~ce si~ral unit 15 Selected accordi~g to the discharge voltage UF, output ~igral~ A1, A2 . are giver. off as measurable variable~, which are relevant ~or specific sub-groups of co~tamina-tion ~ubstances or even for 3pecific co~taminatio~ substa~ces, or for specific contamination concentrationq The reference sig~als are determired by calibration measuremertq and are adjusted based 02 starldard co~eaminated gas 3amples In the diagram o~ the left in Fig l the voltage UF is illustrated qualitatively ~ith points P1 - P3 corresponding to UF-values at which ~he spark gap ignition takes place ir~dependently o~ the various contaminations acting there-o~ .

Proceeding from t~e illustratiorL of Fig 1, Fig. 2 shows a furthermeasurable variable determinatio~ on as discharge gap 3 Here, by mean~ of a controllable high-voltage source 7a, a di~charge is main-tained bet~een the electrodes 5 of the spark gap 3 With a current mea~uring device lla, the di~charge curre~t iF i3 measured and compared on a comparator unit 17 with a current reference value iFSOLL tha~ can be set on a re~ere~ce ~ignal u~it 19 ~7~ O

The di~f~re~ce ~ig~al L~ ~certaiued on the comparator unit 17 i~
passed o~ a~ regu~ati~g difere~ce, pos~ibly by ~ay of a regulator 21, as djust~o~t variable, to the controllabl~ voltage source 7a ~hich no~ acts ~ arl adju~ti~g elcme~t in the current re~ulati~g circuit, in ~uch ~ ~ay that ehe discharge curre~ iF follo~8 the reference value ~hae ca~ be ~e 0~ th0 reference signal source 19 aq ~ominal value adjust~r, and preferably corresponds ~o the cons~a~t adjusted reference value ;SOLL

The regulating dif~ererlce sig~al ~ or the adju~tment ~ig~al RU for the voltage source or the output voltage of the voltage source 7a is evaluated as ~ea~ured variable. Thi~ ~ea~red variable is, as ex-plai~ed with reference to Fig. 1, fed in eurn to a comparator unit 13 with ~uperpo~ed refererlc~ ~ignal uni~ 15, a~d deperding on the sig~al range in ~hich the mea~ured variable ascertained on the regulating circuit lieq, a conclusio~ i3 reached regardi~g the prese~ce or abse~ce of contaminations of various substance groups or regarding ehe presence of contamirlatio~ of various co~ce~trationR i~ the gas sample G.

As can be rloted frQm Fig. 1 and 2, here the diRcharge behaviour of the discharge gap 3 and its electric actuatio~ is used directly as a tese value for the measurable variable.

With the e~bodiments according ~o Fig. 1 a~d 2, a corona-AC or DC

di3charge is produced.

~1 -~7~

As illuatrated i~ the Fig. 1 arld 2 and 11, ~he g~ ~ample G ca~ be tapped off through a sampli~g li~e 1 fro~ the to be testod co~tainer, However, according to Fig. 3 it is also pos~ible, s~eirlg that the discharge ~ap 3 can ~asily be ~iniaturized, to i~troduce the di~-charge gap 3a i~to the to be te3ted corltairler 25, e.g. Nith the aid of a te~t larlce 23 illustrate~ diagrammaeically i~ Fig. 3, a~d to then proceed in accordance ~ith the informatiorl furni3hed in respect of Fig. 1 arld 2.

The taps 27 on the lance 23 according to Fig. 3 correspond to the taps which irL Fig. 1 and 2 are 3ho~n ~i~h the same refererLce numeral 27 on the discharge gaps 3 illustrated there.

Fig. 4 sho~s a further embodiment of an arrangement used according to the irlven~ion for performirLg ~he method according to the invention, wherein by ~ea~s o~ the discharge gap the ga~ is ionized and, in con-tra~t to the em~odimerLts of Fig. 1 and 2, the ionized ga~ i5 exami~ed away fro~ the discharge gap.

By way of the ~amplirlg li~e 1, the gas sa~ple G is ta~en from the to be tested contairler or it~ direct vicinity a~d fed to ~he discharge gap 3, operated ~i~h the current source 7. A condenser arrangement, e.g. a cyli~drical conde~ser 29, i~ provided after the discharge g2p 3, in the direction of flow of the ga~.

-- ~,G --~)7~ja It compri~e~ the cyli~drical outer conde~3er shell 29a ~d the co-~xi81, in~ide ~andrel 29i.

The corlde~er 29 i8 ch~r~ed to a predeter~ined voltage value by mea~s of a~ adjustable voltage ~ource 31, 80 that an electric field E i~
formed on ehe co~denser. Because of the gas ionization on the dis charge ~ap 3, depe~ding on the polariey and strength of ~he electric field E, ions of the one polarity are driven to one of the conden~er plates 29a, 29i, a~d ions of the o~her polarity to the other plate.
The balance of the ion~ driven to the capacitarlce plate~ 29a, 29i produces, in ~he ex~ernal circuit connected to the cyli~drical con-de~er 29, a curre~t i. Thi3 is measured as current i~tegral by a charge amplifier 32 or, as irLdicated by broken lines, by a curre~t amplifier 32a.

Whe~ a charge ampli~ier 32 is provided, the i~tegratio~ time T, duri~c which the current ~lowi~g through the condenser 29 i~ inteyrated, is pre-set, and ~his i~terval T is ~et off by any ~ignal ST defi~ing the start of the measuri~g cycle, e.g. at the start of the sucking off of gas or whe~ a specific ~urge front of the current i occurs.

When the iQtegratio~ time T has expired, the re-~ettirLg switch orl the charge amplifier - illus~rated diagrammatically in Fig. 4 - is closed.

The output sig~al, whether i~ corresponds to the current integral, if the charge amplifier 32 i~ provided, or to that o~ the provided -' .CJ --~ ~3 ~ f3 curre~e ~plifier 32~ f~d, irl the manner alr~ady de9crib~d ~ith r~r~nce to Fig. 1, to ~ compar~or unit 13, o~ the output 3ide of ~hich, ~elec~ed according ~o the magnitude of the occurring i~put ~ignal E, output 8ig~al3 Al, A2 stc, occur a~ ~ea3urabl~ variable.

Here the spar~ gap 3, arra~ged either in a to be ~ted container its~lf in accordanc~ ~ith Fig. 3, or, a~ illustrated in Fig. 4, i~
the sampli~g line 1, i~ used only for the io~ization of ~he ~o be teste~ gas.

Thi~ procedure ~akes it po~sible, becau3e the dischar~e gap can be miniaturized, to provide the ~a~ io~ization in a con~tructionallY
flexible ~anner at any poist of a 3elec~io~ plant. The separation take3 place at the ~ame place, ei~her along the sampling line, or i~ the to be tested container i~elf, or the positio~ thereof is located away from the ionization.

Wherea~ ~he procedure described with reference to Fig. 4 only per-mits a lump sum determination of, a~ measurable variable, the charge balance of the gas that occur~ as a result of the spark ioni~ation, to ~hich end, if the conden~er arrangement 29 is arranged alongside a samplir~g line 1, the ga3 must be fed irl at a predetermined flow velocity, Fig. 5 shows in pri~ciple a procedure by means of which, after ionizatio~ of the ga~ from the container, either by the use according to the invention of a discharge gap, or al~o i~ the known manner, e.g. by flame ioniza~io~ or, preferably, photo ionizatiorL

2 ~

by ~ear ~ of W, aIl ~valuatio~ ~ake~ plsces of the ion~ orm~d irL the ga~ sccordin~a to their ~obility . A~ ~ re~ult thereof co~tamir ation~
of differe~t ~ub~a~ce3 or ~ub~tance ~roupg ca~ be detected more e-lec tively .

To this end the ioniz~d gas G* i~ fed to an electro ~atic 3eparator sta~e 35, con3tructed subRtantially as sho~n i~ Fig. 4, which, for exa~ple, again con-~igt~ o~ a cyli~drical co~de~er arra~gsme~t. This compri~e~, for example, a larg~ inter~al mandrel 30i a well as a plurality of cyli~dric~l gurfaces 30a arrarlged in~ulated behind or~e another. All conde~ers, for~ed by the common inter~al mandrel 30i and o~e cylirldrical surface 30a each, are preferably plac~d under the 3ame electro-static voltage by ~earL~3 of the voltage ~ource 31, so that ehe ~ame field ~tre~gths E lie above the respective co~densers 30i, 30a.

If the gas enter~ the conde~ser space 30z with ions of a differe~t mobility, as illusera~ced diagrammatically, alld these experience in 3ame, becau~e of the homoge~eous f ield 3tre~gth E, provided that the iorls have the same charge~, al~o ide~ical deflec~ion forces, then the more mobile io~s are def lected more per axially traversed path than the le95 mobile ones. Accordingly, the curre~ts il, i2 ...
led o~f from the re~pective conde~sers are, as mea~urable variables, a~ i~dicatio~ for the ion~ def lected seque~tially in the directio~
of the ga3 f lo~, ~herei~ io~s of a decreasirlg mobility contribute an increasing amount to the curreD.t of ~he conde~er arrangeme~ts posi-tioned dossn~rea~a with respect tc the directio~ of f low of the gas .

o The ~apped-of~ curre~ explai~ed ~ith referç~c~ to Fig. 4, are de~ect~d by ~ charqe ~mplifi~r or current amplifier, a~d proce~s~d further a~ ~ea~ur~bl~ variable~ for the co~tai~er sel~ction.

Fig. 6 ~ho~s a~ embodime~t for di~charge ionization of the gas and electro-s~aeic separation measureme~, directly in a to be tested container. In a ~urther developmerlt of the arrangemerlt described with refere~ce to Fig. 3, on the lance 23 ~ith at it~ end a discharge gap 3, on the upper part, a plurality of metallic surfaces 3~i, insulated from one a~other, i8 provided, and coaxially to same, a metallic cy-lindrical ~urface 33a.

A~ illustrated diagrammatically, the lance ~hich has bee~ developed further in thi~ manrer i~ let irlto a ~o be te~ted container, a~d near the bottom thereof ~he gaR iB ionized by means of the discharge sap 3. Already becau~e of the resultant heati~g of the gas in3ide the container, there occurs a ga3 flo~ in the directior~ of the co~tainer ope~i~y, i~ which sectio~ lies the separator stage formed by the corL-denser~ 33i, 33a.

Preferably, i~ addition to thi~, a forced flow of the iorLizing gas G~ is brought about by feeding in ~ further ga8, a carrier gas, e.g.
through dia~ra~matically illu trate~ opening~ 37.

The electriciey supply to ~he spark gap la and the co~den~er arrange-ment a~ ~ell as the curre~t tap~ for tappi~g off the curr~n~
i2 e~c. are passed ~hrou~h the lance 2~, and the ~ame applies to a line to the ga~ outlet~ 37.

A~ ~e~eiorled, ~ith ~he e~bodima~ accordi~g to Fig. 1 ~o 3 pr~-ferably a corona di~charge is produced. With those accordi~g to Fig.

4 to 6 both a corona di~charge as ~ell a~ a spark discharge c~ be produced, i.e. ~hen ~he ionization of the ga4 is ~easured. When ope-rating ~i~h 0park di~charge, for a mea ureme~t, preferably a 3eries of a predetermi~ed nu~ber of sparks i6 produced, and i~ the flowi~g ~a3 G* io~ized by this the io~ d0n~ity is mea~ured and averaged over a predeter~ined period, so a~ to obtain, i~ particular, more reliable re3ults.

With certain contaminatio~ Aubsta~ces the discharge ionizaeio~ ac-cordir~g to the inventio~, or al50 a k~o~n flame ionizatio~, ~ay cau e a~ explo3io~. Becau~e of thi~, for rea~ons of 3a~ety, when u~ing these ionizaeiorl techniques on the occurring empty co~taiuers, a pre-~election mu~t be carried out. This is, for a measuring inside ~he container, illu~tra~d diagrammatically in Fig. 7a. According ~o this the to be tested co~tainers, eg. plastic bottles, are moved o~ a conveying i~stallatio~, either a conveyor belt or a carrouAel system, past a first ~easurirg s~ation 40, where, either by the taking of gas sample~, as illustrated, or by immersing a probe into the container in question, the presence of specific, explosive co~taminations is detec~ed.

To this e~d, accordi~ to the i~ventio~ preferably semi-conductor gas sen~or~ or electro-chemical cell~ are u~ed, adapted to the detectin~

- ~7 -~7~ 0 of k~o~n e~plo~iv~ corltamlna~ious. I~ a cont~iner ~ith ~xplosive co~ta~inaCion~ ia d~et~d, th~, as illustr~ted diagr~atically, .g. by mearLs of a co~veyor ~hunt, the co~tai~er in que~tion i7 re~OV2d 80 that it ~ ot be te~ted further. Co~tainers that are r~cogrlized a~ safe i~ thi3 re~pect are passed on to the ionization ~ea~urinq ~tatio~ 42 ~ith the la~ce 23.

Based on the a~certair~i~g of further corlt~inatiorl~ a~d a corre pon-ding evaluation of the relevant measuri~g sig~al6 o~ an evaluation unit 44 r a further conveyor shunt i5 ac~uated, and inadmis3ibly con-taminated container~ are removed or passed on to a special cleaning proce~s, ~hilst o~ly containers with corltaminatiorls of an admissible type are pa3sed o~ for re-filli~.

As was merltiorled at the out et, c@rtain coneamirlatiorl substances are absorbed by certairl ~all materials of the coneainers, and in particular by plastic, and are released agai~ into the irlside of the cor~tairler only 510wly and in depe~dence on ehe temperature. Without special ~easure~ the contamirlation co~cer~tratio~ inside the co~tain-er, viewed at a given time, may be difficult to measure. ~oweYer, if the contai~er ha3 bee~ filled and ha~ been stored for quite a lo~g time, there nevertheless occurs, for example, arL adverse effect on the taste of the contant of the co~tainer.

It is, therefore, furthermore proposed, as illustrated diagrammati-cally in Fig. 7 at 46, that prior ~o carrying out the co~tamination 2~ 14~.jO

dse~c~ion, co~ tion sub~ta~c~ that have bee~ absorb~d by the ~all~ of the containers ~hould be oxpell~d. According ~o the in-vention, ~his i~ done by heating the co~tainer~, ag illustraeed by th~ h~a~ flo~ Q, ~hich ca~ be done by infrared radi~tion, a~d ~ith pla~tic co~tai~ers i~ particular also by micro~ave heatin~, by Yapo-rization or ga~ification of the i~side o the container and/or from ~he outside, e.g. by letti~g in normal hot air.

I~ certain ca~es it is anyway indicated to rinse the containers with a gas, preferably ~ith air, in particular purified air, and to rin_e out cer~ain amount_ of residual gas qte~ming from specific origi~al co~tent~, ~hich other~ise could conceal other contamina-~io~s during the contamina~io~ detection.

Contami~ations ~temmin~ from original coneents, e.g. from fruit juiceq, ca~, a~ was found, cause considerable interference durlng the detection of other contamination substances. An ex~remely Rimple possibility for overcoming this problem consists in pro~
viding ~he contain.ers ~i~h a marking, e.g. a moulded-in code, correqpor~di~g to the original content. If this is provided, such a marking can easily he read during the selection of the empty containers, so thae informaeio~ is then at hand regarding the type of the origi~al content.

I~ accordarlce therewith, contamination signals can be fileered out in a narrow ba~d, adju~ted to this one product, so as to reduce th -- 2g --- 2V7~V

~e~uring i~terf2re~ce cauqed by the original cont~nt.

I~ this way the probl0~ of the original co~tene-rel~ted measuring interference can be ~olved, for th~ maki~g availahl~ of ~ea~urable variable3 for ~he co~tai~er ~election di~cu~ed here.

A~ has already been men~ioned, the main problem in achievi~g the objectives ~et her~ i~ that most analy3iR techniques, unle~s one goe~ to great expe~se, e.g. by selective infrared ~pectrography or by u-qing expeAsive ma~ spectrometers, supply measuri~g signals that are dependent o~ the ~ype of ~he contamination substance as ~ell aq on its concentratiorl. This means, in other word-, that often there exi~s ambi~ui~y as to whether a substance A with the co~centratio~ a iq pre~e~t or a ~ub~ta~ce B with the concentra-tion B, as orl the ~ame Kdetector~ both co~ditions may lead to the sa~e output sign~l.

If o~e now looks at analyqis tech~iques, which are photo ionization or flame ioniza~io~ followed by an io~ density determination, - qpark gap io~izatio~ followed by a not further ca~egorized ion density determinatio~ or ion density de~erminatio~ taking into account the mobility of the ion~, . ~ 30 -2~7~

- detectio~ of cont~minatio~ ~ub~tanc~s ~ith semi-co~ductor ~a~
8~80r~ or by means of i~frared.ab~orptio~ ~easurement ~ith ~emi-co~ductor infrared ~ensor~, in principl~ possibly al80 ~a38 qpectroscopy, det@ction of certain contamination qub~tances ~ith electro-ch~ical cel 13, i.e. in the light of the coneai~er selection problem of the type di3cussed here, it ~ill be noted that al30 if th~ above case occurs with the o~e tech~ique, ~ith at lea~t one of the other technique~ the si~nal for the contamina~ion -~ub-~ta~ce B with the concentration B
will not be the ~ame as that for the ~ubstance A with the concen-tration a.

By the combined use according to the inventio4 of at least two of the mentioned techniques, as will be explained in the following, ~he ~electivity reliability i5, therefore, considerably increased, or rather is only now obtai~ed.

This takes place, according to Fig. 8, under the main aspect of the present inventiorl, in that, a~ illu~trated diagramma~ically, gas fro~
the to be te~ted contai~er iY fed to a ~umber n of different analysis station.~, ~ith n ~ - 2, e.g., aq illustrated, to a station 54a based on ga~ ionization, a ~ation 54b based on the use of semi-co~ductor ga~ ~ensors, a station 54c ba~ed o~ the use of electro-chemical cells, 2 ~J 7 ~
a ~t~eiOn 54d ba~ed on the u~ of inrared ab~orption me~urement~
etc., or also arlslysis ~ations of the same type, but ~ith differen~
~eaAurin~ charac~eri~tic~.

For a ~as sample, the n s~ation3 produce meaqured value~ I~ .... In that are i~for~a~ive ~or the ~electio~. These signals I defi~e in a n-dimensiorlal 3tate ~pace the stat~ P of the gas in question.

On an evaluation computer a ~-dimen3ional "space" i5 stored in a no-minal range store 56, and in ~ame state ranges that are admissible and that are inadmissible. This is illustrated diaqrammatically in the block 56 in Fig. 8 in a three-dimensiorLal ~space" ~ith the co-ordinaees corresponding to I~ , I3 and the admissible range ZUL.
The value~ Il to In~ ~hich defir~e admis~ible and inadmissible conta-minatios compo~itions a~d co~centratio~s of the gas, respectively, are ascertained beforehand by calibration measuremenes with stardard-ized gas and 3tored i~ the store 56.

When measuring the state P of a ga~ occurring at a specific moment, the ga~ sample state defi~ed by the measured value~ I1 to In is compared in a comparaeor unie S8 with the admis~ible state vectors PZUL for the ga~ sample, stored in the store 56. If the state vector P of ehe gas sample tested at that moment lies within the space range stored in the ~eore 56, then on the oueput side of the comparator urlit 58 the selection i~ decided i~ the affirmative se~se, i.e. the contai~
er that has ~u~t been tested i5 released as acceptable for re filli~g.

7 ~ ?~ 5 r) Oth~r~i3e the contai~er i~ questio~ i3 ~limi~ated.

Th~ inpue ZUL (Il, I2 ....)ln on the ~toraga block 56 repr~
8ent5 the inpu~ for the a~certai~ed coordin3te values I, ~hich define the admi~ ible vector ~p~ce range ZUL. Further~ore, a~ter the sel~ction on the selection bloc~ 56, on the path for i~admi~-sibly con~aminated contai~er~, a further selection ~tage 68 may be provided, ~here, for example, without ti~e pressure, it is checked once again under laboratory like condi~ions whether a container coming in on this path 70 i9, in fact, inadmi~ibly contaminated or not. If it is really inadmi~ibly conta~inated, it i~ elimi~ated.
If not, its vector coordinate values I ~ill be s~ored in an i~er-mediate store 72, fed back to the storage block 56, to thus defi~e in an automatic learning process th~ admissible space range ZUL in a more refi~ed ~ay.

Extremely ~uitable for such a procedure is a neuronal computer network, wherein a~ ini~ial rough ~odel, corresponding here to the ad~i~sible space range ZUL, is refined by an automatic learnin~
process.

As was me~tioned at the out~et, one of the preerred a~alysis techniques i5 based on semi-conduc~or se~sor~. When using such semi-conductor elements there exists, as explained at the outset, a problem in that it~ step respo~se is relatively slow. If, duri~g the flowin~ pa~t of contaminated ga~, a contamination pulse i~

~7~9~J~

produced orl the input ~ide of such a ~en~or, ~he ~emi-co~ductor 3e~-~or ouepue ~ignal ~ill ~ove up relatively slowly to a corr~po~di~g ~aximu~ value, to th~n drop agai~ ju~t a~ 810~1y.

The~e problem~ may al~o occur ~i~h other ~easuring techniques, e.g.
~ith the infrared ab~orption measurement uith 3emi-conductor infrared ~enSOrQ, 80 that the follo~ing explanation3 al o apply to the~e.

A3 can be noted from Fig. 9, the ou~put ~igrLal~ of the 3emi-conductor sensors 60a, 60b and 60c illus~rated there are ~uch ~hat, depending on the occurri~g contamination, they move towardq the ~aximum value A~x, ~hich however takes relatively lorlg.

To ~ow ge~erally ~horten the mea3uring cycle ~ime, use i5 ~ade o~ the fact that the climb of the output ~ig~al increases when the reached maximum output si~nal value becomes hi~her. Because of thi~, ~ith such sensors the se~sor output signal i5 not evaluated directly, but its time differentiation 61 i3 evaluated as the mea~urable variable A~, as illustrated in Fig. ~.

As with ~emi-corLductor sensor~ the variable is it~ reRistance, A cor-responds to the re~istance pattern.

As ca~ furthermore be ~oted, the ~i~e which the output signal of such sen~or~ requires to again a~sume it~ initial value is the lo~ger, the higher the reached maxi~um value A~x. To now nevertheles~ be able ~,J~J.i 20 74~ O
.
to dra~tically shorte~ the ~a~uri~g cycle time, indepe~derltly o~
~hi~, according tn Fig. 9 t~o or ~u~h such 3e~sors or set~ o~ 3uch sen~or~ are u~ed, e.g. cyclically, for successive gas sample analy4es This is co~trolled by a control u~it with a cyclic regi3ter 62, Pre-ferably it i~ monieored, e.g. ~ith the comparator u~it~ S4, uhether the outpue sig~al of orle of the se~sors or ~et of sensors assumes an i~a~miqsibly high value, and this one gen30r or ~et of ensors i5 ~he ~aken out of the cycle for a prede~ermi~ed ti~e 1.

Accordingly, sets 60a, ~ ... of at least one semi-corlductor sensor each are provided, which are used sequentlally for successive gas samples G. If the output ~ignal of a semi co~ductor sensor or its time differentiation moves beyo~d a threshold value pre-set on com~
parator unit 64, the~ the sen~or or 3et of ~e~sors i~ question will be swieched of~ for a predetermined number of subsequent sample gas measuring cycles.

As illustrated by broken li~es, in this co~rLectio~ it is readily poqsible to morlitor the output signal values A, e.g. with a further comparator 65 indicated by broken li~es, and, as illustrated for set 60c, by way of example, eo determi~e, in accordance with the mome~-tary output sig~al value, ehe time duri~g which a semi-conductor gas sensor ~et ~us~ remain switched of~. In other words, such a sensor.
~et will only agai~ start to ~easure whe~ its output sig~al value again drops ~elow the ehreshold v~lue set on ~he threshold value unit ~5 .

~ ~) 7 ~

A further proble~ ~ith semi-conductor ga3 3ensors or possibly al~o radiation ~emi-conductor ~en~ors, as usod for the in~r~r2d ~bsorption ~ea~uri~g, i thae on ~he one ha~d Rupply line for the oampl~ ga~
a~d housi~g arrarlgements in which the ~e~ors are arra~ged must ~e rinsed to ~i~imize the i~fluencP of a preceding meaqurement on a ~ub-sequent ~easure~ent, but that on the other ha~d such semi-co~ductor 9e~50r~ react to a ri~si~g gas flow with a 510~ output signal, of the ~ype as illu~rated at A i~ Fig. 9. This would mean, therefore, that ~hen such semi-conductor sen~ors are rinsed, in particular rinsed with gas, preferably with purified air, after such a rinsing cycle they must remain out of operation for just as long as after a mea-suring cycle, i.e. the ~umber of provided semi-cor~ductor sensor ~ets 60 accordirlg to Fig. 9 would have to be doubled ~o obeain the same throughputs.

Fig. lOa illustrates qualitaeively, over the time axis t, a rinsing gas flo~ S~ hatched, and by dot-dash lines the resultant pattern of the output sigr~al A of a semi-conductor gas sensor. From this ie ca~ be noted tha~ only after expiry of a fall ~ime, a new measuri~g cycle with the test gas supply G can be started on the semi-conduc~or gas sensor in question. However, for time-economy reasons, one should aim at Letting ~easuring cycles immedia~ely follow rinsing cycles and vice-versa.

According to Fig. lOb in co~junction ~ith Fig. 9, this now becomes po~sible according to the invention i~ that the test--gas ~low G and 2 ~ 7 ~

the rinsi~g gas ~lo~ S ~re ad~pted to o~e a~other by mea~3 o~ flo~
adju~t~ent el~ment~, as illu~rated diayramm~tical~y i~ Fig~ 9 at VG ~nd Vs, in ~uch a ~ay that the ~emi-co~ductor ga3 sen~or expe-riences a ~ub~ta~ially continuou~, co~tant flow. With this, ~he test ga3 flo~ is preferably produced by the flow of a carrier gas, to which is added gas from ~he co~tairler that i3 being te~ted. Pre-ferably, as rinsing gas the .~ame gas i3 ehen used as the carrier gas, for example and preferably dry, purified air i~ u~ed for both. If differen~ ~ases are used for the rin~ing and as carrier gas, it has bee~ found that hy changing the flow ratio of the test ga3 G and the rinsi~g gas S, the influence of the different gaq types can to a large extent be compensated.

Fig. lOb illustrates dia~rammatically, for identical carrier and rinsing gaqes, ri~sing cycles S, a measuring cycle G with uncon-taminated ~as, i.e. carrier gas, then a measurirlg cycle G with contaminated gas. Taki~g into accou~t the semi-conductor output signals, the adjustmene i5 carried out 3uch tha~ duri~g the uc-cessive cycles ri~sir~g gas/carrier gas or u~contaminated test gas, e3~entially no output ~ignal or po~sibly a substantially time-constant output signal appears on the semi-co~ductor gas se~sors, which makes it po3sible to test and rinse successively in the se~se indicated above.

The use of a carrier gas ~akes place, for example, a~ illustrated in Fig. ll, by co~necti~g, e.g. by means ok a sealing connecti.on 74, a carrier gas tank 70 to the contai~er 71, which is shown posieion~d on ~ ~ 7 ~

a co~veying devi~e 72. By ~ans of a pump 76, c~rrier g~ to~ther ~ith ga~ co~tained i~ ~he contai~r i3 ~d to the ~asu~in~ a~range-~ene according to the invention, as illu~trated at 78. Naturally, it i~ also po~ible to utilize the water jet pump principle ~ith the carrier ga~ d~ pump ga4.

T~e use of the carrier ga~ a~ ri~si~g ga can take place, for example, in sn ~xtremely ~imple manner by providing a controllable change-over valve Vcs, by means of ~hich the contai~er i~ ~ridged during rin~i~g phase~.

Furthermore, the ~ulti-parameter evaluation explained with reference to Fig. 8 and the corresponding procedure can be modified as follo~s:

By mean-~ of sample ga~ measureme~ts, divided into admissibly conta-minated and inadmissibly con~aminated, admissible and inadmissible combination3 of ~he output ~ignals Il to In are ascertained. With the~e I-values ascertained on the calibrated gas ~amples, a suitable mathematical funGtion is now determined, in such a way that the func-tion value, in dependence on the me~tioned variables, ca~ be divided unequivocally into at least one value range for admissibility and value ranges for inadmissibility.

Instead of seOring a~ admissible multi-dime~sional range, as was ex-plained ~i~h reference to Fig. 8, possibly to save storage space, the found ~athematical fu~ction i~ stored, and the measured ga~ values J ,~ _ 20rl4 .
are entered i~to thi~ fu~ction a~ variablc~. A~ter doi~ ~u, it i~
exa~i~ed ~ho~her ths re3ulta~ u~ctiorl value lies i~ the ad~issiblc or i~ the inadmi~sible ~ur~ctio~ value ra~ge.

If ~urther~ore, ~ e Fig. 7a, a ga~ sa~ple G* is take~ from the container according to Fi~. 7b, prefersbiy the testing for explo~ive cont~ir~ation~ takeg place on ehe gas sa~ple in ~uestion before it is pas3ed o~ eo the unit 41 for ~he di3charge or flams ionization. The staeion therl controls, for example, a valve 45 provided ahead of the unit 41.

By utilizing various ~rans~er characteristics, in particular al~o of the different, described arlalysi~ techniques, both with regard to contami~atio~ subse~ces a well as their corlce~tration~, it is made po~3ible, by a co~bined consideration and evaluation of the measured variable~ of at lea~t t~o of these ~tations with differer~t characte-ristics, to e~qure a considerably greater certai~ty as to whether a qpecific corltai~er can be used again or ~ot. The sig~al controlling the ~election i8 a u~iform signal co~posed of various components.

~f Y

Claims (32)

Patent claims:

1. Method for the automatic selection of containers, depending on weather these containers or possibly their contents are contaminated with an admissible or an inadmissible group of contaminated, with which method gas from the containers is analysed, the result of the analysis is compared with a predetermined reference, and then, based on the comparison result, the selection takes place, characterized in that:

- the analysis is carried out by means of n analysis tech-niques (54a, b, ...) with n > = 2, each of which produces different output signal patterns (I1 ...) as a function of the contamination substance as well as its concentration, - at least one n-dimensional range (ZUL) with state variables (I1, I2 ...) corresponding to the output signals (I) for states of admissible contaminations is determined, - the analysis results of the analysis techniques (I1, I2 ...) are evaluated by checking whether they define a state in the admissible state range (ZUL).

2. Method according to claim 1, characterized in that as analysis technique on gas from the inside of the container, or on gas containing gas from the inside of the container, at least one of the following techniques is used:

- infra-red absorption measurement, - measurement by means of semi-conductor gas sensors, - measurement by means of electro-chemical cells, - ionization, especially photo ionization and/or spark ioni-zation, and measurement of the resultant gas ionization.

3. Method according to claim 1 or 2, characterized in that as one of the analysis techniques at least one semi-conductor sensor (HL, IR) is provided, to which is fed gas from the inside of the container or gas that contains gas from the inside of the container.

4. Method according to claim 3, characterized in that the output signal of the at least one semi-conductor senor (HL) is dif-ferentiated with respect to time (61), and the result of the differentiation is evaluated for the selection.

5. Method according to claim 3 or 4, characterized in that at least two sets with at least one semi-conductor sensor (60a, b ...) each are provided, and the gas out of or from succes-sive containers is fed to different sensor sets.

6. Method according to any one of the claims 3 to 5, characterized in that after measuring the gas, the semi-conductor senor is rinsed with rinsing gas (S) and that by adapting the type of rinsing gas to the gas of the container and/or adjusting the flow of rinsing gas along the semi-conductor sensor to the flow of the gas from the container, the influence of the rinsing gas on a sub-sequent measurement on the gas from the container is minimized.

7. Method according to claim 6, characterized in that for gas out of the container a carrier gas is used, and as rinsing gas the carrier gas.

8. Method according to claim 6 or 7, characterized in that as rinsing gas purified air is used.

9. Method according to any one of the claims 6 to 8, characterized in that by adjusting (VG/VS) the flow of rinsing gas and the flow for gas from the container, when changing from rinsing to measuring or vice-versa, the output signal (A) of the at least one semi-conductor senor is kept substantially constant.

10. Method according to any one of the claims 1 to 9, characterized in that as one analysis technique the gas is exposed to an electrical discharge gap (3) and the discharge behaviour (IF) and/or the discharge-related gas ionization (i) is evaluated as an output signal for the selection.

11. Method according to any one of the claims 1 to 10, characterized in that as one analysis technique the gas is ionized and an output signal (i1, i2 ...) dependent on the mobility of the ions is evaluated for the selection.

12. Method according to according to any one of the claims 1 to 11, characterized in that with each of the analysis techniques, based on calibratior gas samples, in the n-dimensional state space with n > = 2, with state variables that each correspond to the output signals of the analysis techniques, at least one area is defined with states that correspond to state variable values which define admissible conta-minations, and states outside the at least one area which define a gas state corresponding to inadmissible contaminations, and in that the output signals determined on the gas with the techniques are automatically evaluated as state variable values as to whether they define a gas state within the admissible area or not.

13. Method according to any one of the claims 1 to 12, characterized in that at least prior to the use of at least one of the analysis techniques the containers are rinsed (S), preferably with water, steal, a gas, preferably with air.

14. Method according to any one of the claims 1 to 13, characterized in that prior to the use of at least one of the analysis techni-ques the containers are heated (?), preferably by means of infra-red, team, water, a gas, air or by means of microwave energy.

15. Method according to any one of the claims 1 to 14, characterized in that before using up the original content the containers are marked with a marking corresponding to their original content, and during the selection the marking is read and also used for the selection.

16. Method according to claim 5, characterized in that on the semi-conductor sensor sets (HL) the output signals are checked to see whether they exceed (64) a predetermined value, and if so the set in question is not used at least for the immediately following analysis.

17. Measuring arrangement for the automatic selection of containers, depending on whether these containers or possibly their content are contaminated with an admissible or an inadmissible group of contaminations, comprising a detector arrangement for the analysis of gas from the containers, a comparator unit to which are connect-ted the output of the detector arrangement and a reference signal storage unit, characterized in that the detector arrangement comprises n detectors (54a, b ...) with n > = 2, which generate different output signals (I1, I2, ...) as a function of the contamination substance and its concentration, that the comparator unit (58) compares the substance and concentration dependent out-put signals (I1, I2, ...) of the detectors (54a, b ...) with output signals of the storage unit (56) associated with the detec-tors, which latter output signals depend on the admissibility of contamination substances and their concentration.

18. Measuring arrangement according to claim 17, characterized in tha at least one of the detectors is formed by one of the following arrangements:

- an infra-red absorption measuring device, - a measuring device with semi-conductor gas sensor, - a measuring device with electro-chemical cells, - a mass spectrometer, - an ionization device, especially a photo ionization and/or spark ionization device, with an ionization measuring device.

19. Measuring arrangement according to any one of the claims 17 or 18, characterized in that at least one of the detectors comprises at least one semi-conductor sensor (H, IR).

20. Measuring arrangement according to claim 19, characterized in that a time differentiation unit (61) is interposed between the semi-conductor sensor and the comparator unit (61).

21. Measuring arrangement according to claim 19 or 20, characterized in that the detector comprises at least two sets (60a, b ...) with at least one semi-conductor sensor each, and that selectively controlled feed elements are provided for feeding the gas to the semi-controlled sensor sets.

22. Measuring arrangement according to claim 21, characterized in that a control unit (62) is provided, which controls the controllable feed elements in such a way that the gas from successive contain-ers is fed sequentially to different sets (60a, b ...).

23. Measuring arrangement according to claim 22, characterized in that the outputs of the sets are led to a threshold value sensitive unit (64), the output of which acts on the control unit (62), in such a way that when an output signal of a set exceeds a threshold value preset on the unit, this set is disabled at least for a sub-sequent measurement (?).

24. Measuring arrangement according to any one of the claims 19 to 23, characterized in that a rinsing gas line (S) communicating with a rinsing gas source opens out in the area of the semi-conductor sensor.

25. Measuring arrangement according to any one of the claims 19 to 24, characterized in that a rinsing gas line (S) from a rinsing gas.
source and a line for the gas open out, possibly jointly, in the area of the semi-conductor sensor, and that at least one flow ad-justing element (VG, VS) for adjusting the ratio between the flows from the rinsing gas line and the line for the gas is provided in the area of the semi-conductor sensor.

26. Measuring arrangement according to any one of the claims 17 to 25, characterized in that a carrier gas feed line from a carrier gas source (70) leads into the container area, and s suction line for carrier gas and gas from the container leads from this area to the detectors (78).

27. Measuring arrangement according to any one of the claims 17 to 26, characterized in that an ionization device (3), preferably a spark cap, is provided for the gas.

28. Measuring arrangement according to claim 27, characterized in that along a flow path for the ionized gas at least two electro-static ion separation stages (30) are provided behind one another, as detectors.

29. Measuring arrangement according to any one of the claims 17 to 28, characterized in that a code reading device is provided for a code provided on the container.

30. Method according to any one of the claims 1 to 16, characterized in that at least a part of the container, the state of which does not lie in the admissible range, is subjected to a further selec-tion (68), and with the results of the further selection and the variable values of the respective containers (71) the admissible state space range (ZUL) for admissible contaminations is checked in the sense of an automatic learning process, and is possibly changed.

31. Plant comprising a measuring arrangement according to any one of the claims 17 to 29 and a conveying arrangement (72) to and from the measuring arrangement for plastic bottles which as containers occur in a streamline fashion.

32. Use of method according to any one of the claims 1 to 16, 30 or of the measuring arrangement according to any one of the claims 17 to 29 or of the plant according to claim 31 for the testing of plastic bottles for foodstuff.