Design Transfer & use of "prototype" components in production devices.

[ThatSinc]

Hi All,

I have a query regarding Design Transfer, particularly with regards to 7.3.2.c. during the planning of design and development and transfer activities at "each stage" of development , and 7.3.8 for the overall process of final-design transfer to production.

As a part of the development process, prototyping, and ensuring production capability, during the design process "Company A" will manufacture components to the drawn specifications at any given time. These will be issued a batch number, and the technical drawings will have a release issue.
Rather than manufacture 5 or 6 of the prototype parts, a small production size run (~250) will be machined.

My experience of the development process has largely been at huge companies where prototype runs and production validation runs are often considered scrap of the design process and discarded.

would it be acceptable / in conformance with the design and development requirements of 13485 to use these production run sized batches, manufactured prior to final device verification, validation, and design transfer processes taking place - so long as the specification of the part manufactured at that time is the same as the specification at the time of final transfer, assuming all V&V work has been successful?

Where 7.3.2.c talks about transfer activities at each stage of development - would it be considered a design transfer activity to manufacture these prototypes if they were intended to be used as production if they meet the requirements, which would then subject them to the requirements of 7.3.9 on control of design and development changes?

Assuming all purchasing requirements are met for any raw materials, and all traceability is maintained - what additional concerns need to be contended with?

I really hope I'm over thinking this and there's a far more simple solution right in front of my face that I can't see.

Any and all guidance would be sincerely appreciated.

TS.

 

[indubioush]

If the lot of components meets all requirements of the currently approved specification, it is okay to use that lot for production. However, this isn't so simple, and your individual quality system processes will come into play. Is the supplier manufacturing process the same now as it was when the components were machined? Is there a first article inspection process that was not done before that now needs to be completed? What is the status label on the components? If it says R&D use only, you will need to go through a relabeling exercise. Is the supplier approved? Is the supplier process required to be validated? Are the raw materials supplied from the same place?

Sometimes this status change activity it performed by re-receiving the components to the current specification. Hope that helps.

 

[ThatSinc]

Fortunately, everything manufactured for R&D/Development use is treated as if it's already a production part - so the machining processes would be the same, the inspections would be the same, suppliers would all have been qualified in accordance with local procedures required for production etc.

Items are labelled with a part number and batch number, no other status markings as they're physically segregated away from the manufacturing areas.

Would it be within the realms of acceptable practise to simply include as a part of the final design transfer process a check of all "prototype batches" manufactured against current specs and review whether they are suitable for production use and transfer them through as a part of the design transfer record?

 

[Hi_Its_Matt]

@indubioush hit on several important factors to consider. It's not just whether the final part meets spec, but was the overall production and handling process the same (or justifiably similar enough)?

...they're physically segregated away from the manufacturing areas.

I'm not sure if the end device is sterile, but sterilization validation is based on a particular bioburden level, which is a function of the manufacturing process and controls in place. If the area you're segregating these parts off to is significantly different from a cleanliness/bioburden perspective than the regular manufacturing area, then that would call into question whether your sterilization process could adequately sterilize the previously-segregated-to-a-dirtier-area parts.

I have to assume cost is a factor driving this question, yea? It would obviously be cheaper from a materials standpoint to use the stuff you already have, but are you going to eat away at that savings simply by having people work on an investigation and justification to use the parts?

 

[ThatSinc]

I'm not sure if the end device is sterile,

Fortunately not sterile, and the environment in which they are segregated is the same (or cleaner, actually) as the manufacturing environment so the requirement on maintaining the level of cleanliness necessary for the device (I can't remember what ER/GSPR that is off the top of my head) is met.

Cost and "we've been doing it like this for 20 years and we're not changing, so figure out the process so that we can still do it"

It's a really small outfit and the owner/designer/tea maker says "once a component is drawn, it's considered done and if I want to make a full production run, I will".

For all intents and purposes EVERY prototype part is manufactured identically to how it is manufactured as a production part, so for near enough every part the investigation and justification will be a very quick exercise.

 

[indubioush]

Would it be within the realms of acceptable practise to simply include as a part of the final design transfer process a check of all "prototype batches" manufactured against current specs and review whether they are suitable for production use and transfer them through as a part of the design transfer record?

Yes, this is okay.

 

[Hi_Its_Matt]

It sounds to me like you have a pretty clear reason to use the parts.

Design validation shall be conducted on representative product. Representative product includes initial production units, batches or their equivalents. The rationale for the choice of product used for validation shall be recorded

In fact, it sounds to me like you could think about this completely the other way around...
Basically you (for some bizarre reason) defined and put in place first all the production controls, and you even built a production batch of parts. Then you used a small number of those parts to verify and validate the design, validate your manufacturing process and test methods, etc.

You're not trying to reclassify prototype parts as OK for final use... you just built a batch of "production parts," quarantined 95% of them, and used the other 5% to do all the design and manufacturing verification/validation activities necessary to make it OK to use the 95%.

 

[Ronen E]

You're not trying to reclassify prototype parts as OK for final use... you just built a batch of "production parts," quarantined 95% of them, and used the other 5% to do all the design and manufacturing verification/validation activities necessary to make it OK to use the 95%.

That's all nice and well for a post-rationalization, but one thing to remember (in general; not necessarily applicable in this specific instance) is that in validated processes there are sometimes in-process controls, or pure monitoring activities, that become evidently necessary, and are put in place for production, only after the actual process validation. If "the production process if fully put in place first, then retroactively validated using a sample from the initial production run", it's possible that such necessary controls/monitoring were not in place during that run. Just sayin'.

 

[ThatSinc]

Basically you (for some bizarre reason) defined and put in place first all the production controls, and you even built a production batch of parts. Then you used a small number of those parts to verify and validate the design, validate your manufacturing process and test methods, etc.

That's exactly what's happening. The guy that runs the company has been working in machining for >40 years and somehow if he draws it, knows it'll do what he wants and the machineability is built into the drawing in the first instance. More often than not, he's right.
There are some occasions when it'll be scrapped, but his approach is that he'd rather take a few hits like that than set up and prototype each part to then have to go back and repeat.

With this in mind would you consider that to get each component off of the drawing and into the machine shop as a production batch, you would need to design transfer that component during whatever phase you are in at that point in the design process?

s that in validated processes there are sometimes in-process controls, or pure monitoring activities, that become evidently necessary, and are put in place for production, only after the actual process validation

Near enough every part manufactured, be it "prototype" or production, has every nth part measured for critical dimensions, based on the complexity and number of machining processes involved. If it was reviewed whether any additional process checks/controls were put in place following the manufacture of that batch, further checks could be performed on it prior to release.

 

[Ronen E]

I'm starting to get a weird sense of "no matter what, I'm not going to change anything; maybe chuck up some paperwork retroactively if I have to, but not really change anything in my process", so maybe what I'm about to say is a waste. I understand that your boss is one of those veterans who "knows better than everybody" (including the entire QA community) because they've been around for decades and they've done well for themselves; and maybe there's nothing you personally can do about it. But there's a reason these approaches are in place, at least in medical devices. Sometimes you only discover that something is wrong after somebody dies as a result, and sometimes it's not just the boss having to scrap some workpieces.

There are some occasions when it'll be scrapped, but his approach is that he'd rather take a few hits like that than set up and prototype each part to then have to go back and repeat.

In a nutshell, this is the old QC mindset (as opposed to QA). Let's make something and then see if it turned out okay. It probably will. Worst case, we'll scrap it, but it probably won't 'coz we know what we're doing. And if it looks right, it's probably right. Right?
Yes, things didn't work too bad in the 70s so I guess there's no harm in keeping that state of mind. Why bother with fancy terms and mountains of paperwork?

With this in mind would you consider that to get each component off of the drawing and into the machine shop as a production batch, you would need to design transfer that component during whatever phase you are in at that point in the design process?

Design Transfer is not a step, it's a set of activities or a process, that is not always linear or solid (i.e. it's sometimes a little fragmented). It's about taking a set of barely-mature design outputs and converting/growing them into viable serial manufacturing specifications. I'd say that you'd definitely need a subset of Design Transfer done before any production that might end up as a commercial (or at least clinical) one.

Near enough every part manufactured, be it "prototype" or production, has every nth part measured for critical dimensions, based on the complexity and number of machining processes involved. If it was reviewed whether any additional process checks/controls were put in place following the manufacture of that batch, further checks could be performed on it prior to release.

Not every process aspect can be post-inspected in the resulting part. I'm not saying these examples are relevant in your case, but in some machining processes things like wear of the machining tool or the workpiece temperature need to be monitored during machining. These are just examples, so no need to tell me that your specific process doesn't call for these; the point is to say that in-process monitoring is a real thing that sometimes has no good substitute. Even if you end up with only traditional controls (e.g. good old workpiece dimensions measurements) the point is you've considered all foreseeable risks upfront and decided that that's all you need. Upfront, because if your analysis ends up differently you can't go back in time to measure the process.