Florin ordered some PCBs from Iteadstudio, a pcb prototyping service. As part of their service, they claim that all PCBs are tested before they are delivered. However, many have been bringing this claim into question. [Florin] found a complete lack of any markings indicating actual probes had been used on his boards. Though they claim that 100% of the boards are e-tested, they replied to his inquiry somewhat cryptically.
All of the pcb’s have a stripe on the edge,it does not mean they have all been electrically tested ,it is the same as outline .
People on the Dangerous Prototypes Forum have been finding similar results. The company has commented, on this page, that that they will be responding to this soon (back in February)
45 thoughts on “Are Your PCBs Being Tested As Claimed?”
He’s not heard of pogo pins then ;)
The “cryptic” response sounds like a standard sort of line from someone for whom English is not their first language – translation: “The black mark is not related to testing”.
For $1 per PCB I’d relax.
If you are a company with a significant customer base speaking a language, it behooves you to have staff available to translate with a modicum of correctness.
Also many places will test BEFORE tin flooding and soldermask/silk screening. so those marks will not show up on the tinning.
Looks like an iTead board!
Why are they buying from that places? they are more expensive than sparkfuns service.
You sir are poorly mistaken. I ordered from BatchPCB once. 2 PCBs for $30. I ordred from iTead for my next order. 10 Boards for $24. That’s $2.40/PCB as opposed to $15/PCB from BatchPCB. How exactly is that more expensive?
Wrong, I just ordered 10 boards for $30 from ITead. That is much better than the 1 board for ~30 I got from batchpcb. ITead is actually the best priced boards for hobbyists, IMO. And they turned out great.
I’m not sure about leaving marks or not, but I have had the experience where a board house stamped in ink (ET) indicating that the board had been tested and we found it had some serious connectivity issues (shorts if I recall). We called them on it and they admitted to lying. This was a big name board house and they never saw our business again.
So for a hobbyist, you either have to trust the board house or inspect it yourself. For any commercial boards, make sure the testing house is NOT the same as the manufacturing house. Which is a lot harder if they are also populating the boards…
Apparently they also just sit on completed boards before sending them out. Got an e-mail asking if I wanted to buy the extra 14 boards that they made by mistake, I said no and they replied by saying they’d thrown them out. So, obviously the boards were done but it still took two weeks before they were sent.
There were a couple of other issues as well but I wont bore you with the details.
I will be looking elsewhere for the next order.
I sent a design to them that had an error with the file. It took them a whole month to tell me I had to fix the problem and resend the file to them.
I have the opposite experience with the prototype company we use. They charge extra for e-testing but I have found the boards come back with the tell-tale signs of the little pin prick on every pad created by the flying probe machine regardless of whether I do or don’t pay this extra fee. I now don’t pay for the e-testing under the assumption they seem to test it anyway. Not had a single reject to date on the many boards I have ordered supposedly without the e-testing!
Are you at liberty to share with us the name/address/cost of the prototype company you use?
Yeah, they’re called PCB Pool and are based in Ireland I believe. Not the cheapest but the quality is top notch from my experience.
Pay nothing, get nothing. Simple as that. If somebody makes an offer which at this price couldn’t be real, chances are it isn’t.
I bought 10 boards from ITead literally last month, and they arrived 100% working. And this is a very complicated high speed digital board. I have had no problems with them so far and they are super cheap.
Here’s a bad “electrical test” experience I had many years ago.
You can see 3 traces are shorted together (the traces are 15 mils wide with 10 mil separation), and the “ET” stamp nearby. The vendor’s identifying mark in the upper right corner. That vendor is Advanced Circuits: http://www.4pcb.com I paid quite a bit extra for the electrical test service… and I’ve saved this image all these years, because I’m still unhappy about it.
Hey Paul, couldn’t help but reply when I saw the pjrc domain. I’ve still got an 8051 board that I prototype on from roughly 10 years ago (my first credit card purchase). Glad to see you’re still in the game. I will need to get my hands on one of those bad-ass Teensy systems soon instead of messing around with those lame Arduino systems.
Glad to see your boards are coming out better these days with proper vendor electrical testing.
I think everyone should chill the f* out on Itead and Seeed quality. 10 boards for 15$… if only 1 works out of those 10 you still have a decent deal… and it for sure is usually more than half that work.
Also file errors, slowness in notifying etc, yup. Price is cheap, service is slow and cheap, deal with it or pay more, you decide, but quit all the moaning please. These guys are providing a fantastic possibility with PCBs at these prices and I doubt they can quite make it work at those prices if they were to provide top service or guaranteed quality.
Itead advertises 100% e-testing on their product pages. Because of that, the price isn’t important. If they don’t test 100% of the boards they’re misleading consumers. Which is exactly what youritronics complained about.
This exchange appears to be a practicality versus ethics issue. Both are right, both use common sense.
The only reasonable solution as I can see is to understand lowest pay equals lowest quality in one way or another. Stay aware and make others aware of the lie and choose your provider according to your values.
Some of you don’t understand the reason behind my blog post. I understand perfectly that this is a cheap service and you get what you pay for but the real problem is false advertising. Based on that you could assemble a project with some expensive parts and have to trash them all because of a bad pcb which you thought will never reach you because of the advertised 100% E-TEST.
Seeed Studios (who clearly use the same or similar board houses for fabrication) state that only 50% of boards are tested by default. Mine have always come through with half of the boards marked on the edge. I’ve never had any faulty boards but I’ve never bothered to check for signs of testing. Others have had one or two faulty boards that I know of, but they have always been ones not marked as tested.
One run of boards I got from Gold Phoenix was clearly tested properly, as 2 of the 150 individual boards on the 3 panels were marked with a big white X mark across them, and sure enough, two pins of the 44 pin QFP footprint had a small bridge. Maybe not as cheap as itead/seeed (certainly not for protos) but they are top notch quality on a budge :)
Just checked the boards from my last Seeed Studios order. The boards marked on the edge have clear probe marks on all the pads, and the ones without the edge stripe have no probe marks :)
+1 to Seeed Studios
Sorry for the semi-blurry pics. Macro extension tubes + no tripod != real macro :D
have any one had received a faulty PCB from them?
i looked at ones i have ordered from seeed and all had the stripe and all had clear tool marks … but i did get a PCB of an arduino uno SMD without parts on it that both had toolmarks AND had several shorts in it … tho it was not made by arduino
So for those of us looking to dip our toes into pcb fab on the cheap…… Where should we go?
This is shaping up like one of those “any publicity is good publicity” situations.
Some of the prices being tossed around here are very appealing.
That picture that were attached in the post have been clearly tested. There are at least 12 probe points that can been seen there. They do point to point testing, not pad by pad. That means that they verify connectivity by testing the entire length of the trace using pads at either end. There is no need to verify intermediate points between of course.
I think Florin owes iTeadstudio an apology at this point.
Florin appears to be having trouble seeing the probe marks so I’ve circled a few of them which are visible in his photo : http://imgur.com/bOfzQ
what you’ve circled are not test marks. Everyone who’s seen a tested pcb before knows that test marks are obvious and easy to spot.
What you’ve circled are only irregularities commonly seen with a HASL finish.
oddly new obligatory xkcd: http://xkcd.com/1053/
What does “testing” of a PCB actually MEAN, anyway? A board house gets gerbers, which don’t include any signal or connectivity information. I’m not sure how they could “test” at all, but it wouldn’t surprise me if there is something called “test” in the PCB industry that tests SOMETHING, without really giving you too much of an indication of whether your board has errors. (inspect one board manually for errors. If none are found, designate that the “gold” board. Now throw it on a x*x matrix of test pins, and compare electrical connectivity to all the other boards. Those that match are tagged “good.” That doesn’t seem particularly unreasonable. But it also has lots of room for errors…
It’s not hard to use the copper layer and soldermask layer gerber files to identify which exposed pads should be connected together. Combined with the drill file to identify layer to layer connecting vias it’s easy for them to generate a “schematic” of which pads are electrically connected. I don’t know if they even test both sides of the board at once. Faulty vias are probably visually inspected, and the flying probe test is just to find faults in the copper layers.
It’s just a continuity test between all the pads that should be connected together, against all the pads that shouldn’t be connected together. Surprisingly, this can be an issue, even with professional boards.
It’s entirely understandable. The complexity of moving from an AND test to a NAND test of the whole board would require a custom fixture, testing all points at once, which would run very expensive very fast.
not testing? that may be good thing if you are going to build some illegal electronics (locally or federally) say like radar jammer or cell phone jammer cable descrambler phreak boxes (though obsolete) or hardware based drm circumventer like scms ( toslink) or hdcp (hdmi).
it is like using your local film developer to develop pictures that are illegal as long as they never look at them.
now if the test is just to check for connection along the path of the trace simple pad to pad continuity test both along the trace path and between the layers of the board then you are ok.
I don’t think this argument makes much sense. How does the PCB manufacturer know that U1 is a microcontroller, for instance, rather than a quad flip-flop? Unless you have every chip marked, or the PCB is labeled with a description of its purpose on the silkscreen, the fab house generally doesn’t know exactly what you’re making. They could possibly make an educated guess, but there’s not really that much difference between the PCB of a legal radio and an illegal jammer.
Indeed, even if they could reverse engineer your PCB layout and somehow correctly guess all the component values to create a schematic (perhaps like dowsing with a rod/fork for underground water), if you’ve used a microcontroller, programmable logic, or any non-volatile memory, they’d have to be psychic to guess the code or data you’ll program.
Most average, run-of-the-mill psychic readings run about $1 to $5 per minute. I’m sure any service which could divine technical details accurately (and perhaps future stock valuations?) would cost considerably more.
Ordinary non-psychic reverse engineering is also quite expensive. Generally it takes a lot of time and manual effort by an experienced engineer to turn a PCB layout into a reasonably accurate schematic with reasonable guesses at component values, not to mention IC part numbers.
It’s pretty amazing this service can be so incredibly cheap. Certainly the process must be very automated, as even the low labor rates in China must be a factor. How they can even make this profitable is pretty mysterious. Whatever they’re doing, it certainly doesn’t include a budget for anyone to spend much time or effort to review the details of each PCB !
This isn’t isolated to custom pcb makers, I was replacing the control board to a Frigidaire front load washer when i discovered the door sensor connection to the pcb had missed the tin bath. I had two options, take the hit on an hour one way service call and wait for a new control panel or get out the soldering iron. wheres the quality control on this stuff anyway?
They have pogo pins with rounded tops that will not leave marks on a PCB pad. Takes very little pressure to make contact.
Having said that, I doubt seriously, that they’re gonna pay a guy/girl, to sit there, and test all the connections, for shorts, or breaks or anything else, while they charge $25 or what ever.
Nor are they going to have a full array of test pins where they can just pop any board on it and program it to test it.
Also, so what?? You show me a board, that is clearly faulted, that SHOULD HAVE BEEN FOUND, then you come back and complain about it.
The idea that they’re complaining, when they found nothing wrong? Other than there’s no marks on my board?
Get a life dude. Go find something to cry about.
But crying because you can’t find anything to cry about??? Good grief!!! The board doesn’t have a problem does it?? Then MOVE ON!!!
There are automated “flying probe” machines which do this sort of testing very quickly and cheaply. I believe you just give it some sort of DRC data along with the gerbers.
So next time your mechanic doesn’t do an oil change when he should yet your car keeps running just fine anyway despite the very real potential to cause serious damage to the engine are you going to get yourself a life too?
It’s quite simple. If you don’t do a continuity test on the boards, don’t advertise that you do.
If I order a board that hasn’t been tested I look over it very carefully with a loupe to see if it’s fine. Yet if I see a board with test marks and a test marking on it I don’t bother.
99% of the time this means nothing to me. That last 1% of the time it may just blow up in my face.
But hey if you like paying for a service you don’t get I’d be more than happy to give you a long distance massage every day for the next year if you transfer $100 into my paypal account.
Hyperbole. You are comparing a ~$1 piece of electronics to an automobile and saying its the same thing.
“It’s quite simple. If you don’t do a continuity test on the boards, don’t advertise that you do.”
I can’t be certain but I’m pretty sure [Microguy] isn’t a manufacturing company.
“You show me a board, that is clearly faulted, that SHOULD HAVE BEEN FOUND, then you come back and complain about it.”
Go to the Dangerous Prototypes forum link provided in Caleb’s writeup, first post on page #1. There’s your clear fault that should have been caught by 100% E-test. Which is why people are trying to determine if the E-testing is consistently being done as advertised.
Just got some boards in from seeed studio today. All of the pads on all 10 boards show marks from the flying probe machine.. and I didn’t pay for the 100% test.
All boards were marked with red marker on one side.
I got 20 boards from Seeed a few weeks ago, and also didn’t pay for the 100% e-testing (50% is free). All of the boards have 2 black marks on the side, but none of them have marks on any of the pads.
I got your flying probe test. What are you making? single sided? double sided? kids toys. Wait until you have a board that is 28 layers, and fails…that originally passed a test and after a few years shows a dead short….amateurs.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)