Your analyzers are not too old.They were never given a way to connect.
Nobody double-gowns for fun. If you have ever suited up for a clean room, you know the routine: gown, gloves, gown again, and fifteen minutes or more of your day gone before you have touched a single thing.
Now think about why somebody makes that trip. In a lot of facilities, it is to read numbers off an analyzer, because the instrument that measured them has no way to hand them to your modern systems on its own. So the numbers ride out on a clipboard, and at the end of the shift somebody types them into the manufacturing execution system, your MES, by hand.
I have walked that round with customers, and the fix is smaller than you would expect.
Your older analyzers and measurement tools still measure beautifully; they just have no way to connect to today’s systems on their own.
The bridge is a person: report files read by someone, typed into the MES, and then checked by a second someone, because typing fails at a rate that never reaches zero.
When the instrument sits in a clean room, one reading can cost a double-gown and fifteen minutes at the door before you ever see the screen.
01 · The connection
The instruments are fine. The connection is missing.
Your instruments are qualified, calibrated and checked on a schedule, and each one has a paper trail behind it. None of that is the problem. The problem is that most of them were bought before the systems they now need to talk to.
The chiller was in the building before the MES was. The analyzer’s vendor supported a file export and left the rest to you. The serial port on an older instrument has been putting out the same line of text for fifteen years, and the Modbus registers behind the chiller’s panel hold the same values the panel shows to whoever walks past. The equipment is doing its job. It writes a CSV or a fixed-width report file into a folder that nothing is watching, and that is where the automation ends.
Nobody designed it this way. It accumulated one purchase at a time, and every one of those purchases was a reasonable decision on the day it was made. But your instruments are not too old to share their data. They were never given anywhere to send it.
02 · The bridge
The bridge is a person, and you pay twice
So how does a number get from the report file into your MES today? A person reads it and types it. And a typed number costs you twice, because copying fails in ways you already know: two digits swap, a reading lands one row off, a value goes in against the wrong batch. It fails at a rate that never reaches zero, so a second, qualified person has to check the copy against the source, and most of that checking ends in confirming a number was copied correctly.
You have a vocabulary for what this does to the record, and your auditors use it fluently: ALCOA+. A number typed at the end of a shift is attributable to the typist, not the instrument. It may not be legible by the form’s fourth photocopy. It is not contemporaneous, because the record was made when the typing happened, hours after the measurement was true. The original stays in a folder or on a clipboard while your MES holds a copy, and the record is only as accurate as the transcription. The plus asks the rest: whether the repeated run is complete in the folder, whether two instruments keeping two clocks stay consistent, whether a workstation disk is enduring enough to be an archive, and whether the record is available to anyone but the person who filed it. 21 CFR Part 11 frames the electronic side of all this, and a record made by retyping paper inherits everything the paper step did, however capable the software downstream happens to be.
None of that is carelessness. It is what copying is. Every number a person retypes is a number a second person has to check.
03 · The trip
The reading is free. The trip is not.
Now add the gowning. When the analyzer sits in a clean room, collecting its numbers is not a walk down a corridor. It is a double-gown at the door, and just getting into the room can take fifteen minutes or more before you have seen a single reading.
That cost shapes behaviour in ways nobody writes down. Readings get collected on the schedule the gowning allows rather than the schedule the process deserves. And when something goes out of trend, the first question anybody asks is what the instrument read and when it read it. If that answer lives on a form in a binder, your investigation starts by reconstructing the measurement instead of examining it. If you have to double-gown to collect a number, the trip costs more than the number.
04 · How we help
How can we help?
The move we make with customers is smaller than a capital project: read what your equipment already produces, where it already produces it. The CSV or text file can be watched and parsed the moment it lands. Then this data needs to be delivered securely with a full audit trail with acknowledgements into your MES system.
Flowgate is where we built this. It reads the report files your equipment already produces, and carries the readings out to where your MES or LIMS can consume them. What a reviewer sees came from the instrument rather than from a keyboard, stamped with the moment it was captured instead of the moment somebody got to a workstation.
The delivery path holds up its end of that requirement. A node inside the facility opens the connection outward, so the path across the boundary carries no inbound rule into the facility. Readings buffer where they were captured while a link is down, which turns a dropped connection into late data instead of missing data. Sign-in binds to the Active Directory you already run, so an action carries a name instead of a shared login. And the audit trail is hash-chained, so an entry altered after the fact breaks the chain so that it is easily identifiable and then can be investigated.
And where an instrument speaks a protocol from another era entirely, RetroBridge reads it and serves it onward, and Flowgate carries it from there.
And here is the boundary, because I would rather set it out myself than leave you to find it later. Software is not compliance, and we do not offer it as though it were; what a platform can give you is capabilities and records built to support a process your organisation owns. Validation is yours. What gets validated is a system in its operating context, your instruments, your configuration, your procedures and your people, and no supplier can bring you a validated state. Whether reading an existing output counts as a change to your system, and what your change control requires when it does, is your quality unit’s question to answer, and we would not try to answer it for you.
05 · Sources
Sources
06 · Take with you
The three things to take with you
Your instruments are not too old to share their data; they were never given anywhere to send it. Every number a person retypes is a number a second person has to check. And if you have to double-gown to collect a number, the trip costs more than the number.
If you want one next step, walk one scheduled round this week and count two things: how many numbers somebody wrote down, and how many minutes it took to reach them. Those two numbers are the size of this problem at your site.
If you would rather describe the capture step you actually have and hear what we would do with it, write to us at info@h2innovations.ca.