How is a Fairchild jfet identified from fakesters?

[lowvolt]

I've been buying (or at least attempting to) some jfets online, how does one discern the real deal from the flakey fakeys?

And, in the end, does it even matter?  How are they tested to verify QC?  I should say, how do I test them to verify QC?  Is it just an "ear test" by actually installing them in a test circuit of choice?  Or is there some standard by which to measure them against?  Kindof like HFE gain testing of BJTs.

There are so many (for instance) 2N5457 jfets out there with a friggin "F" on them, some of the "Fs" are solid, some are just outlined (as far as the typeface used on the component itself to ID it by brand).  ARG!!

So, I guess I'm asking two things .... 1.) how does a person ID proper Fairchilds, and 2.) does it even matter as long as the jfet meets some srtof standard that is measurable by the end-user?

Thanks.    Every day is a learning experience.

[brett]

Hi
tricky question...
the characteristic that we are mostly concerned with is the Vgs(on/off). Bias and gain (in a common source circuit), are both affected by this Vgs. Unfortunately, this characteristic is VERY difficult to control in manuacturing. I can imagine Fairchild might try to make a bunch of 2N5457s and accidentally half of them will be 2N5456s. Someone here will know how they allocate them (by batch?? I'm assuming they don't test each one).
So...brand doesn't matter much. I've never heard comments like "Brand X is noisey" or "Brand Y's 2N54xx" has really a really tight range of Vgs(off)".
If you want to build a test rig, I suggest testing the Vgs(off). It will be high (-1V) for the JFETs known for high gain (J201 etc), and low (-6V) for the JFETS with other uses (2N5458, MPF102, etc). I'm fairly sure that the JFET matcher at geofex.com will give the voltage for a certain amount of conduction (ie just 'on'), but I can't quite remember the details, so I won't guarantee it.
good luck

[smallbearelec]

By buying from a source that offers the guarantee that I do:

All of my JFET stock is major brands: Fairhild, National, Vishay, etc., most purchased from franchised distributors. The part that came from the broker market is almost all in the original manufacturer packaging. If there are any duds--not likely--you will know whom to blame. No doubt I charge more than some other on-line sources, but how much is peace-of-mind worth?

[PRR]

> Fairchild might try to make a bunch of 2N5457s and accidentally half of them will be 2N5456s.

I think they make them all together, then sort. Like making cookies. Some come out crunchy, some come out soft. People will buy either type, but they like to know which they will get.

> I'm assuming they don't test each one

They *may* get lucky, a whole wafer comes out on the '56 end of the spec.

However it is a _long_ way from wafer to tape-reel. Things go wrong.

At least long ago, EVERY part got a basic test.

Why?

Consider that you done good, your final product works 99 out of 100 parts.

Nobody wants parts, they want whole systems. Say you solder a hundred 99%-good parts into a large PCB, an industrial controller or a PC motherboard. A hundred parts at 99% good each is a 36% good final board. You must reject 64% of these PCBs that you worked on so hard! Two-thirds of your money wasted.

So you test at every stage, *before* you put more time/effort into the production.

(Not so severe in the DIY world, where projects are typically smaller; but parts are produced for today's pretty-complicated commercial products not we DIY folks.)

This also means there is a good supply of rejects. Of course if a part is a little outside the spec for one type-number, it can probably be sold under another type-number. Historically the makers don't even put a type-number on the part until they see the orders, then mark the "best" part-number that a part will qualify as. OTOH you get parts which won't suit any type-number. Historically these got sold by the pound, and backroom shops sorted them for very low-performance very cheap products like pocket radios and toys. And at the end, there's piles of no-good parts just begging to be sold anonymously to naive buyers.

How do you know it is Fairchild?

You may not care. JFETs are not rocket science. Fairchild does not personally make JFETs in Portland, they contract-out. The same contractors supply other brands. Their prime-parts are good.

You can test. Vgs and Idss will tell you if a part is in the ballpark you are looking for. But buying from strangers, you better test for gate leakage, breakdown voltage, and noise. This is a fair amount of work (if you don't have automated tester/sorter systems). Manual testing is MUCH more expensive than buying known-good parts in the first place.

Look at the logo? What does that prove? Semiconductors are made in the same areas where printing was invented. They can print as good as they wish to. Maybe colored threads or holograms? When MicroSoft tried that on retail Windows boxes, the counterfeiters had excellent fakes on the market right away.

Even Intel chips have been faked. A real wake-up for the distribution industry.

Steve "Bear" is right. You trace the source hand-to-hand. Or since the big boys won't talk to little buyers, you find a supplier you _trust_. Mouser DigiKey Allied are major players, and their business would collapse if they were dealing fakes. Being major, they don't focus on old-school stuff. But there are also the stomp-box specialists, like SmallBear. All their customers hang together. Again a taint of fakery would collapse the business. Also nobody gets rich/quick selling a good line of pedal parts, better to sell used cars or credit-repair. That Steve has chosen to stick it out in a small market says he loves the pedals, and their builders, and takes care to do-it-right.

[Kesh]

why fake penny transistors?

[lowvolt]

Hey man, I'm not trying to duck under Small Bear or anything like that at all.  I'm just attempting to educate myself on more of the aspects of this stuff that I seem to be so far behind everyone else here on (wow, that was HORRIBLE grammar).

I'm not insistant on buying any name brands, I even detailed that in the opening post.  I'm just trying to learn how to tell which blankets have smallpox so I don't get sick.

Even Small Bear can unintentionally ship inop components  So I am simply attempting to brain up so I'm not at the total mercy of say-so.  But from what I've gathered so far, you pays yer munnies - you takes yer chances.  You buy them from someone trustworthy (such as Small Bear) and you hope for the best.  You're ultimately trusting Small Bear's trust in his supplier, who may be trusting his own supplier.  There's a lot of trust going on.

Thanks for the help.  I now see it's a matter of using best judgement and selecting a vendor that has a dog in the fight and stands to lose something valuable should they sell poo.

Got it. 

[lowvolt]

why fake penny transistors?

There's money in rebranding floor sweepings.  When a company can buy D-stock for pennies per pound and sell them for $1 each in 20 pc minimums there's a few bucks to be had.  Especially when that's their whole gig.  Look at the ebay sellers from overseas that sell any number of different types of semiconductors (such as CA3080s, and other difficult to locate items) in 20pc lots for $10 per package.  It's what they do so they're set up to do it profitably.  Especially with sweat-shop ethics involved.

[Ronan]

There's a thread from a few months ago about J201's. Maybe do a google search "fake j201 site:www.diystompboxes.com"

[darron]

i even got some from one of our reputable sources that were fake! found them in my good bin so can't say who for sure. no matter...


the fake j201's don't have the similar character stamp on the back.

and they are different. made some boogies with them and had to dial the gain down.


http://www.diystompboxes.com/smfforum/index.php?topic=97600.0

[PRR]

I don't mean to diss you or bless the Bear.

Fake chips has been a REAL problem for 5 or 8 years (that we know of).

Reported 3 years ago:

http://eetimes.com/electronics-news/4087628/U-S--Fake-parts-threaten-electronic-market

That report focuses on Defense Dept purchasing. A fake FET in a pedal is one thing, a boatload of FETs needed for missiles or aircraft could ground the DoD's whole business.

3 weeks after that report, NewEgg sold 300 counterfeit Intel CPUs. Now, sure, NewEgg are low-ballers but they know not to ship bricks. (One hard-drive company literally did that, a long time ago.) These were not even electronic, just plastic castings. NewEgg trusted their wholesaler, but somebody dropped the balls.

> You're ultimately trusting Small Bear's trust in his supplier, who may be trusting his own supplier.  There's a lot of trust going on.

This is true.

[Jdansti]

I wonder how mobile these Asian counterfeit operations are and whether the governments are cracking down in them.

I once bought a box set of a TV series for my niece as a Christmas gift-the least expensive set online. When it arrived, the disks had blurry labels and were in plain paper sleeves that were stuffed in a brown mailing envelope post marked Malaysia. No fancy box or plastic case.   

There was no way that I could give it as a gift, so I decided to check the disks to see if they had anything on them. Only half of the 10 disks were playable.   

I wanted a refund from the company that I had bought them from, but they had no phone number or email address on their website. 

So I called American Express (the credit card I had used), and they refunded my money immediately.   They said that these counterfeiting companies move so quickly, the governments can't keep track of them.  I went to Fry's and bought a legitimate set for the gift.

The real kicker of the whole fiasco happened about two months later. I received a phone call from Amex and they asked me if I was trying to purchase four airline tickets from Malaysia to China (I wasn't).   Fortunately they had flagged the attempted purchase as suspicious and refused it until they could confirm it. They overnighted a replacement card to me.

Anyway, the point of this long story is find out if anyone knows how these parts counterfeiters operate and whether there are any governmental efforts to curb them.

[PRR]

> trying to purchase four airline tickets from Malaysia to China

I think you answered your own question. Less than a day to fly anywhere in the world.

[Jdansti]

^For DVD pirates that would be true because all you need to make fake DVDs is a few banks of DVD replicators-it could be done in someone's house.  I was wondering, though, if the equipment that the parts counterfeiters use is very mobile. In other words, how much equipment and floor space does it take to fake transistors and other components?

[lowvolt]

Ok, so just so I MAKE SURE I have this like ~rock solid~ in my bucket-of-a-head .. there's no way to test/verify a FET, right?  Other than actually installing it in a circuit and checking actual performance?  There's no little gadget I can buy, no little tester I can make that will provide a set of measurments that I can compare to a known ~good~ FET used as a baseline or a control.  Do I have this correct?

(PRR ... no sweat ... these danged computers and text-screens can convey the wrong emotions and attitude so easily.  I'm good if you are! )

[PRR]

> there's no way to test/verify a FET, right?

You can test anything.

Get some batteries, current-sources, meters.

2N5457's commercial spec-sheet has 12 parameters which could safely be measured.
http://www.onsemi.com/pub_link/Collateral/2N5457-D.PDF

Rating Limits:
Vds
Vgs
Ig
Characteristics:
V(br)gss
Igss
Vgs(off)
Vgs
Idss
|Yfs|
|Yos|
Ciss
Crss

Doing just a few parts, from a bare bench, I'd figure 20 minutes per parameter for the first part, 5 min each for a few more, say 10 min per parameter. Times 12 parameters, 2 _hours_ per part. I've worked cheap, but even in my youth I got $4.30 for 2 hours, which is 5X or 10X more than known-good JFETs cost.

Having measured these specs, you'd know if the part in hand fit the published specs for 2N5457. You still don't know "who" made it. You don't know for-sure if there's something wrong (or wrong for your use) about it.

Noise is not specified on the 2N5457 sheet, although there is a "typical" which you can check; noise-tests are pretty fussy measurements. The curvature of Gm/|Yfs| is also given only as "typical". You can measure these, but if your part fails, you can't complain "this isn't a 2N5457!!" because Typical is not Guaranteed.

You may also wish to check the Dimensions. Are the leads actually 0.407mm-0.533mm across one way, 0.39mm-0.50mm the other way, and 1.15mm-1.39mm apart? (Gotta fit your holes!) Height of blob 4.32mm-5.33mm just might be critical in a tight design. Realistically you might accept a 0.55mm lead and a 4mm or 6mm blob, and instead of testing to-spec you test against your product (does it fit?). Too-thin drill bits break, so your holes are likely to be 2X-3X bigger than the specified lead, so an out-of-spec lead is acceptable in this application.

Some more of these parameters may be don't-care. In a JFET phase-shifter the Vds applied is very-very small, nowhere near 25V; likewise we don't put more than about 5V in the Vgsr direction (we still want to know if a part has absurd-low breakdown values). OnSemi gives two Vgs specs, in many applications one or the other is all we need. When a spec shows both Gm and Ron, *usually* they come to the same thing one way or the other.

Edu Lab workbook on measuring basic JFET parameters:
http://socrates.berkeley.edu/~phylabs/bsc/PDFFiles/bsc4.pdf

Manual tester for the three basic parameters:
http://www.forsselltech.com/media/attachments/JFET_Jig1.pdf

Arduino-based auto-tester, collects two numbers on each JFET:
http://www.ssguitar.com/index.php?topic=2060.0

Automatic tester, BJT or FET, only measures three parameters of a JFET.
http://www.neazoi.com/semiconductortester/download_190.pdf

[lowvolt]

So ... with all that having been said ... it seems most time-efficient to simply use sockets in each construct and test empirically.  Stick the danged things in there, and play the guitar.  If it sounds ~good~, we're done.  If not, change them until the right combination is found (phase shifters or other circuits requiring ~matching~ exempt, of course).

Yea?

[Jdansti]

If you take the "socket and try it" approach, you might want to consider that each JFET you test is properly biased. You might want to place a trim pot in place of the drain resistor in your test circuit and adjust it for the best sound for each JFET you test. If you want to get anal about it, you could measure the trimmer resistance and label each JFET with it.

Having said all of that, I've purchased JFETs from the DYISB store, Mouser, and Tayda, and never had a problem.  I suspect that Small Bear's would be fine too. I still socket most of my JFETs anyway just in case.

I think that this is one of those cases where you have to weigh the time and effort of pre-testing every JFET you purchase VS the time and effort of discovering a bad one when you install it in a circuit.

[lowvolt]

Well, my thing was attempting to maintain control over consistancy from build to build, really.  Not so much finding a "bad" one.  The drain resister trimmer sounds appealing.

Being somewhat naive about some of this stuff, I figure since I was testing/using caps within certain go/no-go ranges, it might be a good idea to adopt the same approach with fets (jfets specifically).  But from what I am picking up from the posts here I may be worrying over nothing.  It seems to be more or less a "it either works or it doesn't" attitude.

Ok ... cool .... sounds fine. 

[PRR]

> using caps within certain go/no-go ranges, it might be a good idea to adopt the same approach with fets

You can do that. If your JFETs are basically good, sort for both Vgs(off) and Idss.

However. Apparently (I am not sure why), JFETs vary SO much that even the makers (with millions of parts in hand and hi-speed testers) generally will NOT supply tight-spec parts, the way cap-makers do.

An alternative to excessive selection is to use circuits which tolerate wide variations. Instrumentation amps force-bias their FETs to do what the designer wants. RF amps generally can idle wide-open and shut-down as needed when signals happen. However in pedals we do have several legacy designs which are quite fussy.

[kwijibo]

> using caps within certain go/no-go ranges, it might be a good idea to adopt the same approach with fets

However. Apparently (I am not sure why), JFETs vary SO much that even the makers (with millions of parts in hand and hi-speed testers) generally will NOT supply tight-spec parts, the way cap-makers do.

Discrete JFETs are somewhat niche nowadays, a lot of designers seem to think of them as obsolete parts. Caps are everywhere though.