development process

It's sort of okay, but biased towards very high volume manufacturing of fairly complicated parts. It is possible to get a printed circuit layout right at the first attempt, and you have to be building more than a thousand parts a year to justify going through multiple printed circuit layouts to get to an optimum solution. Engineer time is expensive.

Marked separation between product development and manufacturing.

I guess if you don't actually 'make' anything any more, that this is a common philosophy.

. . . . no actual customer either.

RL

IME, most engineers are decoupled from manufacturing. Bigger shops have "manufacturing engineers" who delight in denigrating the "naive" design choices they made. Smaller shops just slap together "modules" that someone else has sorted out how to manufacture, economically (to some level of reliability).

I recall working on a design in the 1980 time frame where I

So, I brought it up in a project meeting and the guy in charge of manufacturing chimed in "Sure, I can do that! But, servicing will be difficult..." "No problem as its an intentionally disposable unit!"

How often do you see a manufacturing line that does anything more than assembly and final inspection, nowadays? Sheet metal shop? Paint shop? Shake-n-Bake? Especially for any significant *quantity*.

ARM merited a chuckle when they pioneered (?) fabless design centers. But, hasn't most product development/manufacturing taken the same approach? Even to the point of having product drop-shipped from YOUR vendor's shop?

We have a full pick-and-place line, and a secondary selective solder machine for thru-hole parts. And VOA and xray machines for inspection.

My manufacturing people don't mind putting parts on both sides. It doesn't make testing difficult, especially if the bottom parts are mostly passives, like bypass caps.

We never make prototypes. We design a thing and release all the revision A drawings to manufacturing and they make a bunch of first-article products, which we test. We expect some of those FAs to be sellable.

That's why I find the attitude of iterating for years to be so weird. If you assume that the first few designs will be defective, they will be.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

They shouldn't be defective. They may well be sub-optimal, but you have to manufacture in volume for this to be worth correcting. The bigger the volume, the smaller the sub-optimality that is worth correcting.

They left out certification.

Cheers

Good point. If a product needs certification, all that will need to be redone every time the design is iterated. Yet more fun.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

At least for Part 15 unintentional radiators, since it's all SDoC anyway I think the question is more like what changes are you comfortable making post-compliance testing, under the assumption that such changes won't materially change the validity of the test results backing up your self-declared conformity.

For the relatively simple designs I'm doing now (2-sided, 10 square inches, sub 20 MHz clocks) compliance testing is not an insubstantial part of the budget, $2500 USA/Canada and another $3000 CE, so avoiding re-tests is nice, but the good news is there's only so many ways to revise a relatively simple design. It's very unlikely the i2c bus layout or the clock frequency is going to substantially change soon, and all the parts should be available another 5-10 years at least.

Yeah changing some DC current sense resistor values or swapping the uP pins the switches go to or something could change the EMI profile. So could looking at it funny, phase of the moon, etc..

But, they change the product. If your customer/client -- or a litigious actor -- opts to challenge those changes, you expose yourself to adverse outcomes. You should be asking yourself "why do I think I *need* to make this change". Esp as the folks making that determination may not be competent to understand the inconsequential nature of that change.

"Do you admit to making a change to the product after certification?" "Er, ... yes (as seen in the published/archived photos of the article) "Why do you think a 'snapshot' of the design is archived in that process? Obviously SOMETHING has changed else you wouldn't have altered the product."

I did QC at a factory for a summer job. Policy was that if *I* touched an article, it was scrap -- even if all I did was verify its color! This to ensure product that was sold all went through the same process and no deviations that may have affected merchantability or liability.

Your customers absorb those costs. If your goal is to make a better product FOR THEM, then it's a non-issue. The mistake was the PREMATURE certification as it didn't DO anything for them.

Or expose a hazardous voltage to an accessible location, etc.

It's easier just to put testing where it belongs in the development process -- at the end. How confident would you be in a product that made a few "insignificant" changes to its software AFTER testing/validation was complete?

What legal theory/tort is the hypothetical litigious actor proceeding under to "challenge the changes" in court? The FCC handles FCC infractions, you can't sue somebody for "not being FCC compliant." Breach of contract? Unfair competition? False advertising? That I'm being a big jerk?

It's the hypothetical litigious actor's responsibility to obtain sufficient evidence of non-compliance and find an attorney willing to pursue a case, for whatever that they can actually sue over, to whatever end. Sounds like it could be expensive and time-consuming to even begin the process. What are they hoping for, precisely? There's no guaranteed payday here.

I suppose a litigious actor with deep pockets could sue me right out of business if they were determined enough to clobber all competition but I figure a litigious actor with deep pockets could likely always find an avenue to go about that, hunting for EMI violations from minor PCB revisions seems like a perverse way to go about it.

Literally every product that has upgradable firmware does this, all the time. How confident I would be depends entirely on what changes were made. Send it out for re-testing every time the splash screen says "Good Morning, Dear User" instead of just "Hello"??

You took an action. Doing so means you assume liability for that action. All a lawyer has to do is convince some average joes that your action

You don't know what they will be claiming as injury. "We bought 100 of these devices. Paid our staff to install them. Then discovered a flaw in them that someone-who-will-testify-for-us claims could be the cause of the problem. We want to be reimbursed for the items, the labor to install -- and uninstall them, plus our inconvenience or possible lost business"

Regardless, being *named* in a lawsuit will likely cost you far more than the $2500 (you're trying to avoid) to respond.

Early in my career, I designed a medical device. DM+DL was in the $400 range. Firmware updates required swapping ROMs. A trip "into town" to service a unit would cost the client $600. I'm sure he could PROVE these costs. Do I want to be on the hook for $600+ for each firmware upgrade he undertakes to fix something I've done wrong? So, I *lose* $200 for every unit sold??

No, the assumption is that the product WAS tested prior to the upgrade being rolled out. How *effective* that testing is will vary with the complexity of the product.

But you don't KNOW what has been changed! You may not observe any changes and yet know it's not the same product as "yesterday".

"Little changes" often aren't. I've met many developers who have a tale of some "simple change" that bit them in the ass, disproportionately.

Again, "WHY make the change" is the operative issue. IMO, everyone should have some experience working in industries that eschew changes and/or make them very expensive to implement. When its an effort to make a change, you think really hard about WHY you are making it.

Datron made a very nice DVM called the 1061A. They also made a 1061. When I asked their rep why they sold both versions he explained that the A version had a bug fixed and a few small improvements. However, some big military customers had qualified the 1061 for their applications and did not want to have the substantial expense of requalifying the meter. They preferred to live with the known bug rather than have any uncertainty associated with the upgrade. Everyone else preferred to have the better model. So they made both. John

Yes. There are many industries with explicit or implicit regulatory requirements where change is discouraged and any known change has to be carefully qualified.

Aero/astro, defense, medical/pharma, gaming and, lately, automotive are good examples. "New and improved" arent things they look for!

I keep entire development environments (OS, toolchain, source code, etc.) for every project I've designed. I *need* to be able to reproduce an exact copy of the

I was tasked with making some changes to a grey-area gaming device many years ago. In the process, I discovered that the original design had made some errors in the probability computations.

In theory, certain types of wagers would produce BETTER outcomes (for the bettor) than strictly expected. Of course, the only way for a user to know this would be to make lots of wagers and compare expected outcomes with actual before the discrepancy would be noticeable.

In a *regulated* market, this would have been a serious problem as it indicates the prior implementation was not a "fair" game. (laws mandate this) But, as it was a grey market, the change could be introduced without any practical consequences.

You can't _freely_ change a product, because then you're acting negligently. I think the value of certification is it's a part of how you would demonstrate to relevant parties that you're making a best-faith effort at not designing products negligently which s*it up the spectrum with EMI.I don't believe it was intended to be a formal statement of God's truth on the matter.

Yep, anyone can try to sue you for anything at any time and claim any injury they like. The burden of of proof for damages tends to be negligence, not "the manufacturer had in place generally reasonable engineering practices but yet made an error regardless." You have to determine what a reasonable standard is for your industry and design.

But you act like re-testing after every change, is some iron-clad defense against disgruntled customers, is it? Why should they even believe what _my_ test reports say about anything, anyway? I'm the one paying the lab after all. IIRC FCC unintentional radiator requirements don't even require a lab be accredited..

Making a revision utilizing sound engineering judgement and documented rationales which then results in some bizarre chain of causality which results in a substantial lawsuit because the massive EMI from a formerly compliant product caused substantial damages I think is why we have product insurance and LLCs, I'm not convinced these worst-case scenarios can ever be entirely mitigated at the design level. You pays your money and you takes your chances in life.

It's unclear who "they" is here, I'm talking about a competitor. Why would my botch cost them anything to "handle."

Someone assumes that my e.g. TV manufacturer sends the mainboard out to an anechoic chamber for comprehensive EMi profiling every time they roll out a firmware upgrade, which is like every week? Who assumes this??? I definitely don't assume that.

Well, certainly be more vague.

Yeah I'm definitely not reporting any "errors" I find in the probability calculations of some Mafia-designed gaming machines to anyone, if that's not part of my job description, lol..

Gaming devices aren't "mafia designed". And, in regulated industry are highly controlled, tested and ethical. Remember, a flaw in a dweisgn can just as easily be used to screw over the party *operating* the device.

This is what typical gamblers fail to understand in thier thinking that they have a "system" enabling them to win: if the game wasn't "fair" (i.e., had some exploit that could be abused), then no one would want to operate *or* play it.

Games are profitable for their operators because gamblers can't "walk away" -- even after a net gain.

I agree, some of us actually have to turn a profit.

Particularly for a small business you can't just write off development costs and take a net loss year after year, the IRS has thought of that and reclassifies your "business" as a "hobby" pretty soon. They're not wrong that's kind of what a lot of hobbies are, a structured way of having fun losing money.

So, your argument is that it was good enough to pass certification... but you FREELY CHOSE to change it?

You're focusing on the certification as the issue that would be challenged in a lawsuit. It can STILL pass certification -- possibly with even an cleaner spectrum (in the case of emissions).

But, you made *a* change and the plaintiff claims harm as a result.

"Why didn't you leave it the way it was?"

I worked for a hand-tool manufacturer. At the time, we were sued by someone who was up on a ladder using a WOOD chisel to cut ALUMINUM gutters. A piece of metal flew into his eye. He sued because the manufacturer should have designed their tool (or, included a warning against such use) so that his "loss" could be avoided.

It's relatively easy to convince a jury that it is "reasonable" to expect someone to MISUSE a tool in this manner. After all, how many have used a slotted screwdriver to try to remove a philips fastener? Or, as a prybar? Or, a chisel? Or, the *handle* of a tool as a hammer?

Was your product involved in a loss? Is it possible -- or even likely? -- that the loss can be partly or wholly to blame on your product's design? Does evidence exist that you took an active role in CHANGING your design?

Once the case is on the court calendar, you're already out of pocket more than the cost of recertification -- even if that had no bearing on the liability issue.

And, as a result, doctors, pharmaceutical companies, police, etc. should all have immunity based on THEIR self-assessment of their products? Welcome to the real world!

So, you want to spend those monies in a manner that gives you the most bang for your buck, right?

I'm sure you can argue that a design hasn't changed "substantially" some months and number of iterations after a particular point in its development. So, why not get the certification out of the way ASAP? After all, *you* are convinced that all of the "improvements" thereafter won't affect THAT aspect of its performance.

Do you think an attorney wouldn't get you to concede that such an impact is POSSIBLE? And, the other party has a DEMONSTRABLE loss?

I'm talking about a litigant. You can't control what their claim(s) will be.

Do you recall:

Forget the merits of the case. A product was effectively canceled by litigation.

They likely have those facilities in house. And.or experience with tens of thousands of units on which to base their assumptions.

If you've not heard such tales, then you must have a very small set of colleagues willing to share details of their experiences with you.

I was asked to build a "computer lab" for disadvantaged students. A place where they could "drop in" and do homework, etc. The organization had received a dozen or so "identical" computers and monitors from a (business) donor. Had set aside a space for them and the students. But, needed someone to install the software, print services, etc. so the setup would be self-maintaining (despite the desire of students to visit web sites that might not be safe, install software that could potentially corrupt the system, etc.

I built a dozen identical disk images. Legally acquired licenses for them (MS has a "store" for non-profits to license software at low/no cost). Set up a networked printer. Arranged for the system to "restore" itself after each reboot (so each new user saw exactly the same environment as the previous user, regardless of any "changes" the previous user may have made).

And, I added a nice *accessible* USB3 hub so students could plug in multiple thumb drives (e.g., to copy files between) as well as their phones.

I left the grunt work of physically positioning the machines, monitors keyboards, network cabling, switch, printer, etc. appropriately.

And, had them purchase additional copies of the USB3 hub (I only had one to donate).

A while later, I got a call that none of the hubs were working! So, a 40 minute drive to the site (plus another 40 to return home) and, sure as s*it, the hubs aren't being recognized!

Except the first one!

Yet, they are all identical in appearance, part number, etc.

TL;DR -- the manufacturer had made a change to the hubs' implementation and VID/PID had changed. But, nowhere was this evident on the devices! You only discovered it when you plugged the hub in and noticed that it wasn't recognized.

So, I can find the appropriate driver and reimage all of the machines to recognize the "new" hub (retrieving my donated item). Or, find some OTHER hub that would work with the systems as built.

Regardless, I "lost" a day of my time to a problem that I had already "solved" -- because a manufacturer thought it acceptable to change THEIR product without changing a part number, product name, etc. in a way that would allow buyers to determine if they were purchasing the same product.

It takes you a lot of words to get to the crux of your argument. If I were to redesign you I would make it so you used less words to get to the point. There it is! Finally..

I was glad to hear your perspective, though, it differs from a number of the other tech people I've talked with who said startups should consider just ignoring getting small runs of new products certified entirely, given that the average FCC fine costs about as much as the certs do...and that was back when we had an FCC who vaguely gave a s*it.My bad for making a best-faith effort to comply with the law, I guess.

No I've never been to China.

As far as I can tell from your posts that's precisely what someone might consider _you_ at this time as well, Mr. Pot/Kettle! That you feel a need to keep a server farm in your residence doesn't change this. Rather like domestic TV manufacturing your factory QA days are long gone.