|Elliott Sound Products||Counterfeit Semiconductors - 2|
Last Update - 04 Feb 2006
Sanken 2SA1216, 2SC2922
25 May 2003 - From Singapore, Michael Chua provided this information ...
|The Sanken on the right is the original part in each photo, with the fake on the left, as indicated on the photo.
Obvious Features of Original (Front) ...
1) The lettering is thicker and smaller.
Less Obvious Features (Front) ...
3) Overall size - the original is slightly bigger, about 0.5mm on either edge.
Obvious Features (Rear) ...
1) In the original, the back metal plate is slightly frosted.
Common Features: Both weigh approx 10 grams.
|... And when cracked open, the difference is very obvious
1) The die on the original is attached to a TO3P type heatsink which is in turn bonded to a larger heatsink contact face.It is difficult to tell exactly, but the die in the fake looks to be around 3mm² which seems to be a common feature of most of the fake devices seen. That it is a great deal smaller than the original, and that it has less ability to spread the heat to the metal heatsink face is quite apparent.
I can only assume that the metal backing plate is slightly thicker in the fake, so that the weight of each unit is reasonably similar. This will not necessarily assist heat removal though, and these fake devices would be lucky to withstand 100W dissipation (based on tests I have done on other fakes with a similar die size).
When I asked if it was ok for me to 'borrow' the info from his site, Mike sent me the following e-mail ...
I am delighted with your offer. For the sake of the audio community, it is important that these people are exposed. I was told these fakes originated from China. What surprises me is how close they are to the real thing. I will keep you posted if I come across more fakes.
See the original page at AmpsLab, and my thanks to Michael for allowing me to use his photos and info. If these are available in Singapore, you can guarantee that they will be in wide circulation very quickly, since Singapore is a major distribution centre for Asian semiconductors.
12 April 2004 - From the UK, Mark W has provided the following ...
I regret to report to you that you may now add 2SA1943 to the list of possibly fake transistors. My story will probably sound familiar. I was powering up an amp of my own design (more or less) for the first time when the fuses blew. I removed the output devices and discovered that one had failed short. I checked the schematic and the pcb layout for design/construction errors. I did this twice. I first decided that the fault was mine since earlier I had a small problem because of forgetting a sil-pad (though not on the failed 2SA1943). But also felt that the circuit and layout were essentially OK - it wasn't THAT original. I try to power up again and when the rails reached ±35V the prior story repeated. I was still inclined to suspect my own ineptness but on a lark, I smashed open one of the 2SA1943's if only to alleviate my frustration.
I won't send the samples. You could mention however that one visual characteristic of the fakes is that the Toshiba making is on a smooth shiny rectangular area of the package. On the fake it is easier to read the markings than the real ones.
What I found looked exactly, I MEAN EXACTLY, like the photo on your site of the inside of a fake 2SA1302. I conclude it's the same bastards just printing a different label on the package, At first I just thought I couldn't find the die or didn't know what I was looking for. Then I saw this 3mm x 3mm square thing pasted on with white glue. There is a clue on the outside. On a real one (I sacrificed one just be sure) the Toshiba label is typically hard to read on the dull surface. On the fake the surface has a glossy area where the label is quite clear.
The 2SC5200s from the particular supplier seem to be real. They don't die when abused and the exterior is dull flat black, somewhat difficult to read unless the light is right.
This was a foregone conclusion - I knew that the criminal bastards would go for the latest Toshiba devices sooner or later, but I must admit I'm a little surprised that it took them so long. This means that it is inviting disaster to use any Toshiba power transistors unless you are 100% certain of the source - unlikely in the extreme.
This is a pity, because they have a very good performance (well, the genuine ones do), but the criminals have once again ruined the reputation of a perfectly good transistor, and created a situation where it would be folly to use them in a design.
23 Jan 2003 - Alleged NTE devices have been found in the US (NTE36/37) that failed instantly, and just to add insult to injury, took the bridge rectifier and transformer with them. An edited quote from the reader who found these latest gems ...
I just got done reading your article on the counterfeit semiconductors and it all clicked for me. A little while back i was repairing an old amp of mine and purchased 2 pretty expensive matched pairs of BJTs from <name withheld> here in the States. NTE37 and NTE37 are the same as C2581 and A1106 devices. Well these NTE brand semis looked weird to start off with - one transistor of a matched pair was in a green case but the rest were in black ... under the NTE36 and NTE37 I could see a shadow of a device number on each semi that had been taken off. I had no clue about this counterfeiting thing at the time and put them in my amp - they immediately burned out but this time it took out the transformer and rectifiers with them. I thought i was just loosing my touch at fixing stuff at the time but now i see what the real problem is 'cause I cracked open one of the cases a few minutes ago and it definitely didn't look like the original semiconductor die. Tomorrow I'm going to contact the supplier and let them know, because I paid them US$40 for 4 transistors and ended up with about US$150 in damage.
A reader in Sweden sent me some pictures of fake Toshiba devices, purchased from a local dealer - this shows just how bad these counterfeits can be ...
Nice and Flat? Not Likely!
Notice in particular the rightmost picture - the transistor base (the heatsink surface, not the internal connection ;-) is so convex that it won't sit even close to flat. In this case, there is about 0.5 mm convex curvature, which is so completely unacceptable that words fail me!
I have heard other reports of bases that are concave, so the device will never make proper contact with the heatsink, but you can't see it. Either is unacceptable (in the extreme). The printing on the 2SA1302 looks like it was done with a felt-tipped pen (well, maybe a little better than that) - not quite what one expects from a reputable manufacturer, is it? These devices passed the acetone test, so the markings are quite permanent (not all do though - the printing can be removed quite easily from some fakes).
This is an appalling state of affairs. I also recently had an e-mail from a reader in the UK. He built the P68 subwoofer amp, and was only running it into 8 ohms and it failed. After a quick exchange of e-mails, it transpired that he purchased some 'Toshiba' devices from a local supplier for less than the normal price - say no more! His first set came from a reputable dealer (but not the distributor), so almost no-one can be trusted on this score.
From New Zealand ...
A reader sent me a sample 2SA1302 and 2SC3281 to test for him, after his P68 sub-woofer amp blew up during quite gentle testing. The 2SA1302 died well before I could reach my target of 5A at 30V (the limit of the SOA curve for these devices for steady state current at that voltage). The tests were done with the transistor firmly clamped to a heatsink, and were of short duration to prevent the die from overheating - but it still failed!
The die is 2.5mm square! This is tiny, and I am actually surprised that the transistor managed even to get to 2A at 30V before it blew. This is well short of the specification, and obviously the reason the amp failed. One can hardly expect a 60W (at best) transistor to provide close to 120W output (the approximate power expected from each pair of devices). Naturally, the devices that I was sent were clearly branded as Toshiba, and are just as clearly fakes.
The 2SC3281 actually managed to survive my SOA (Safe Operating Area) test, but given that the printing was almost identical to the other device, I would be highly unwilling to trust it - it may well be genuine, or simply a "better class" of fake. There were also subtle differences in the case construction of each device, with the 2SA1302 being almost identical in all respects to the Chinese devices that I have (not branded as Toshiba, but also incapable of the rated power during an SOA test). The 2SC3281 was different from any of the other samples I have.
Just in case anyone was wondering, I know it's not my P68 amp design, since I would have had a great many complaints by now if the design were flawed, so that only leaves the transistors as being highly suspect. My original amp works fine as well, and has been "punished" many times without failure.
Please take great care when buying semiconductors - especially 'premium' devices. Since these have the highest markup (i.e. they are fairly expensive), they are the ones most likely to be fakes.
From another reader ...
"Add 2SA1302 and 2SC3281 (Toshiba) to your counterfeit list. I found them (counterfeit ones, that is) here, in dinky Malaysia! The ink printout is totally different from the real thing, being WHITE in colour."
Toshiba plastic transistors are usually marked in white, so this could be misleading. However I do know for a fact that Chinese (unbranded) 2SA and 2SC devices are available, but these make no pretence at being Toshiba. Perhaps (although I think we can be definite on this score) someone has bought the Chinese ones and re-branded them as Toshiba - a worthwhile effort for the criminal element, since the Chinese devices are quite cheap.
Since the Chinese devices are not branded, they cannot be deemed counterfeits, but what sort of quality you could expect is anyone's guess, so one should be wary. It is possible that these transistors are OK, but equally they may be completely useless.
Further Update (06 Jan 2002)
I have checked the Chinese versions, both a load test and visual inspection. The device I checked blew up at well below the normal peak power, and a look at the innards revealed a silicon die about 3mm square - too small for the documented power rating.
On the positive side (if there is one), at least the base plate was copper, and passably (!) flat, unlike the new fakes shown above. These devices would probably be OK in a low power amp, but cannot be used at anywhere near the full capabilities of the real Toshiba transistors.
Are these being purchased by the unscrupulous and re-labelled as Toshiba - you can count on it !
One thing that is known, is that Toshiba has not made 2SA1302 and 2SC3281 transistors since 2000 (or thereabouts), so the chance of obtaining genuine devices is very low. I would like to be able to suggest that MJL1302/3281 or the current Toshiba 2SA1943 and 2SC5200 be used, but fakes of the Toshiba devices are already spreading, and even the latest ON-Semi devices can't be far behind).
Where (or what) next? I received an e-mail from a reader in India, who purchased some premium opamps (at a premium price, naturally). Having paid for OP-07 opamps, one would be disheartened to put it mildly to discover that they were really 741s. I don't know if this has happened anywhere else, but it is fair warning that you could be next.
Now we have Toshiba branded 2N2773 power transistors. This in itself is interesting, as a search on the Toshiba site reveals that they don't even seem to make this transistor! It would be unusual for a Japanese manufacturer to make a '2N' device at all, but doubly so since this is a very old device now, and seems to be discontinued by just about every other maker.
Again, these have all the earmarks of counterfeits - and naturally enough someone was caught out, and his amp failed with these transistors installed. If you happen across any of these components, be afraid - be very afraid!
... And Still They Come ... From a reader in India, and reproduced (almost) verbatim:
I was reading the article about duplicate/fake transistors. Well, they started faking ICs too!. Don't get me wrong though. I live in India. In my best knowledge, there are no IC/Transistor making factory anywhere in India. So I don't think the fakes are MADE here. But there is a possibility that India is a kind of dropoff-point.
The real purpose of this mail is to add one more IC to the known frauds. (Hundreds more maybe there). I bought this IC, LM3915, supposedly made by National Semiconductor for your LED VU Meter project. It costs about 50Rs (our currency, that's about $US1.00 ). But it burned out the instant I connected it to the 15-0-15 supply. I bought another one from another store. It looked a bit different. Anyway after reading your article, I got suspicious. I used a simple knife to remove the top cap like thing of the IC (Normal ICs cannot be stripped like that).
I found another IC. When I scratched the silicon, I saw what I was expecting, LB1405. A vastly inferior (in my experience) and cheap IC. I got ripped by 5 times the cost. I might not have found out this if I didn't power it with 15 volts. I don't know how 'THEY' managed to do this. But it wouldn't have worked anyway. The pin configs are very different. I showed it to the store owner. He discarded it as my mischief. But I couldn't help my poor friend who was making a 10 channel EQ. Poor fellow. He burned out all of the ICs he bought from this store. He doesn't have the budget to replace all the chips. So he's using the EQ without the VUs. Poor chap.
The above was actually received some time ago, and I forgot about it. I have now remembered :-) As you can see, this is widespread, and many store owners are unlikely to admit that they have fraudulent stock.
I think I can say with reasonable certainty that this is the tip of the iceberg. How much reject stock (factory seconds, out of tolerance, incorrectly marked, etc) is gathered up by unscrupulous dealers and sold off as first quality? My guess is - a lot.
Always remember ... Any deal that seems too good to be true almost certainly is too good to be true!
From Canada ...
About three weeks ago we received a batch of transistors from Digikey... To be more specific: 2N3773's... (about... 100 of them at $1.25 ea.)
I have worked with the original MOT's and I know the way they are built and labeled. These "new" parts, didn't look like anything I have ever seen, and I have been in this field for almost 20 years working in audio related goodies.
The Manufacturer: MEV (you tell me if you know them)
Case: Steel or something like that
The finishing: lead immersed. the whole case looked as if it had been immersed on molten solder to "give" it a "silver coat" look alike. (the pins even looked as if they were used devices and had been cleaned off to strip excess solder material.
The label (markings): looked like cheap paint barely stamped onto the top of the case. some acetone and it rubbed off. (YIKES!!!) And it looks like this:
And, if this wasn't scary enough yet, here's the best part of the movie...
I installed one new pair on a switching amp used on a GE servo. Each board makes 1 half of an H Bridge. So a total of two boards are necessary to form a dual direction servo unit (Each amp uses a total of 12 2N3773's for a total of 90 Amps at 90Vdc, at full load when the trannies are completely either on or off depending on the direction of rotation) all the original devices on the amp were ok, except two that were shorted. After double and triple checking of the board, I installed it on a test bench we have built. (to simulate the working conditions required by GE's service manuals).
The "new" trannies lasted 15 seconds... they started off fine and gradually deteriorated until they went off with a bang!!! The rest is history... replaced them again with two more from the same batch and they worked for an hour...
So I decided to crack the first pair open... (considering i had read your stories on your website...) The Dies are smaller than those of a 2N3055. 25% smaller than the original Motorola devices.
Silicon is Silicon any way you slice it and (normally; did I hear... counterfeit???) current densities are the same from one device or manufacturer to another ... regardless of it's use or purpose. Once you go beyond this set parameter, you're in trouble. Even worse if the TO-3 case (like this aforementioned device) has a coin no bigger than 10mm wide by 2.5 mm high. (Yes I love Metric system too.)
Footnote: I just remembered another device I ran into that same day... A 2N3055 (supposedly MOTOROLA, as it was labeled. Yet the ink used for the label was the cheap kind.) that looked almost identical to a genuine MOT device, BUT it was made in MEXICO. So far that sounds believable... Right???
Wrong!!! I opened the casing after i had blown one up at only 6 amps, and came to see that the die was slightly bigger than that of a TIP 41C. Like I might have said before, I know very well the dies in these devices. I cracked many of them open to see their guts after they blow. Weird, eh? No coin, or any internal heat spreader at all. The chip looked like it was glued to the case. No traces of the usual solder material that's normally used.
My dear friend...
As much as I love my hobby and my profession (which happens to be the same. I HATE THIS CRAP!!!! (Excuse my french!). I mean... seriously. What's next?????
Let's hope that someone sees this and takes some action!!!. I am, and WILL do my part! I hope this matter (someday) might be resolved, or at least tamed. (Yeah, right.)
I apologize for my rather dry sense of humour. Yet this is no laughing matter. Please feel free to modify this email at your will and post it on your site (if you feel it's worth the effort.)
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2000-2004. Reproduction or re-publication is allowed due to the importance of ensuring that everyone should be aware of these fraudulent practices, on condition that the name and URL of the original page and author (Rod Elliott) of the information herein is not removed or replaced.|