Well... what if I am testing a fake.... thereby "both" are MJ15032... and have the same SOA (in theory to be tested)
My contention is that lets say I hit it at the peak of a point in the SOA, (hypothetically assuming 100v and 0.5amps) and pulse it... I am saying that what if the fake is resistant to pulses at 100v and 0.5amps, my test is in error.....right? Becaus ethe fake would survive it, leading me to a false conclusion.
Thereby a better way would be to let the sucker rip for an hour on a heatsink, if it survives, I want it even if it be a fake.
My contention is that lets say I hit it at the peak of a point in the SOA, (hypothetically assuming 100v and 0.5amps) and pulse it... I am saying that what if the fake is resistant to pulses at 100v and 0.5amps, my test is in error.....right? Becaus ethe fake would survive it, leading me to a false conclusion.
Thereby a better way would be to let the sucker rip for an hour on a heatsink, if it survives, I want it even if it be a fake.
For me, it's a (long) pulse test. A 2 second pulse and that's it.
I don't really use a heat sink because in 2 seconds. the thermal dynamics of the case don't let the case temerature get too high. I actually hold the transistor in my hand so that if it heats too quickly, i'll know about it.
This test has worked for me:
3A Ic, 50V Vce,
for 2 seconds to test for an MJ15003/4. This is just inside the DC SOAR of the genuine transistor.
A 2N3055 will fail this test in less than 1 sec.
I use the DC SOA curve, since I'm not sure how reliable a test might be using the other lines. Besides, i'd have to supply higher currents, higher voltages, and supply them from high power semiconductor switches, if I used one of the pulse SOA lines, which would radically increase the complexity of my test jig.
You don't need more than 2 sec and don't need a heatsink, because then you'd be including thermal dynamics as part of the test.
We know the fake could probably pass thermal dynamics testing becuse it's made by a factory with capabilities to accurately clone a TO3 case (which is not your average grandpa's garage). The only way the fake would fail a thermal is if the die was stuck with spit to the substrate (which is not entirely impossible).
The importance of SOA qualification is that non-destructive testing still allows you to return the part, should it fail the test.
Andy
I don't really use a heat sink because in 2 seconds. the thermal dynamics of the case don't let the case temerature get too high. I actually hold the transistor in my hand so that if it heats too quickly, i'll know about it.
This test has worked for me:
3A Ic, 50V Vce,
for 2 seconds to test for an MJ15003/4. This is just inside the DC SOAR of the genuine transistor.
A 2N3055 will fail this test in less than 1 sec.
I use the DC SOA curve, since I'm not sure how reliable a test might be using the other lines. Besides, i'd have to supply higher currents, higher voltages, and supply them from high power semiconductor switches, if I used one of the pulse SOA lines, which would radically increase the complexity of my test jig.
You don't need more than 2 sec and don't need a heatsink, because then you'd be including thermal dynamics as part of the test.
We know the fake could probably pass thermal dynamics testing becuse it's made by a factory with capabilities to accurately clone a TO3 case (which is not your average grandpa's garage). The only way the fake would fail a thermal is if the die was stuck with spit to the substrate (which is not entirely impossible).
The importance of SOA qualification is that non-destructive testing still allows you to return the part, should it fail the test.
Andy
Hi,
A pulsed SOA will catch fakes as easily as a 2 sec test. To be reliable you would want to exercise the parts to the listed breakdown voltage and maximum current as well. Adding a pot check for fT and DC beta would then catch all fakes. A capacitance test is good for catching anything that deviates from the boogie die used in the part.
Normally the fake has a smaller die than the original part.
-Chris
A pulsed SOA will catch fakes as easily as a 2 sec test. To be reliable you would want to exercise the parts to the listed breakdown voltage and maximum current as well. Adding a pot check for fT and DC beta would then catch all fakes. A capacitance test is good for catching anything that deviates from the boogie die used in the part.
Normally the fake has a smaller die than the original part.
-Chris
I forgot to mention...
Absolutely true, several points on the SOA need to be tested.
In my experience, all the fakes I found were really bad transistors, they failed on the first test point. But this may not always be true. So if it passes the first test point, you go on to the next.
And of course, testing capacitance, hfe at lo and high currents (especially at hi currents), and other parameters doesn't hurt.
It's called getting the signature of the component. I suppose a PIC microcontroller project could be the perfect way to implement an automated transistor signature tester. Any takers?
(certainly more useful than using PICs to make egg timers and rain water gauges.)
I'm not sure what fakes you've seen, but the ones I've seen were MJ15003's ($2.00). And all other audio power transistors that I know (which are all good counterfeit candidates) are priced higher than that part. There was a time not long ago when the MJ15003 was very expensive. Some fakes may date back to that time.
Andy
Clem:
If you "keep getting fakes" there may be something to question with your sourcing method.
All the testing in the world won't help you if you source from underhanded vendors, who are not likely to accept returns.
The power transistor users community has been getting increasingly more savvy in the last 5 years or so. Internet communication and dissemination of counterfeit information are making them a harder prey for felons. More buyers are choosing to pay premiums and buy from safer channels. More are returnig bad stock. More are asking tough questions before bying, some threatening legal action when burned by vendors in their own country.
As a result. both new counterfeit stock, and an increasing backlog of older, unsold, or returned counterfeit stock is accumulating in illicit channels. The pressure to sell this stock at fire-sale prices (especially on anonymous marketplaces like Ebay) is mounting.
Which means you'll find the parts you're looking for at the prices you want to pay from the most willing sellers. Which is something you should be concerned about.
I wouldn't be surprised if on this very forum there was someone spinning a story about the 8 pairs of "Toshibas" he has for sale from an old amp project he never got around to building...
Andy
If you "keep getting fakes" there may be something to question with your sourcing method.
All the testing in the world won't help you if you source from underhanded vendors, who are not likely to accept returns.
The power transistor users community has been getting increasingly more savvy in the last 5 years or so. Internet communication and dissemination of counterfeit information are making them a harder prey for felons. More buyers are choosing to pay premiums and buy from safer channels. More are returnig bad stock. More are asking tough questions before bying, some threatening legal action when burned by vendors in their own country.
As a result. both new counterfeit stock, and an increasing backlog of older, unsold, or returned counterfeit stock is accumulating in illicit channels. The pressure to sell this stock at fire-sale prices (especially on anonymous marketplaces like Ebay) is mounting.
Which means you'll find the parts you're looking for at the prices you want to pay from the most willing sellers. Which is something you should be concerned about.
I wouldn't be surprised if on this very forum there was someone spinning a story about the 8 pairs of "Toshibas" he has for sale from an old amp project he never got around to building...
Andy
Hi Chris! Yes, I'm here - thanks for the link!
Andy - we're willing to pay, but the parts simply aren't available, or so it seems. Either the store sells it to you real cheap, or charges and arm and a leg, and you still have a pretty good chance of winding up getting a fake. There aren't too many sourcing options where I live for those who just need a couple of transistors.
Someone mentioned a "PIC project" - I'm for it, I can handle the programming - someone volunteer to do the interface circuits?
Cheers!
Clem
Andy - we're willing to pay, but the parts simply aren't available, or so it seems. Either the store sells it to you real cheap, or charges and arm and a leg, and you still have a pretty good chance of winding up getting a fake. There aren't too many sourcing options where I live for those who just need a couple of transistors.
Someone mentioned a "PIC project" - I'm for it, I can handle the programming - someone volunteer to do the interface circuits?
Cheers!
Clem
Hi Andy,
Just about every popular japanese line - 2SC/2SA stuff, far as I can tell. We haven't confirmed if the newer OnSemi 211xx is fake as well; hence the interest in the ability to test.
I was discussing this with Chris a couple of e-mails ago that it would be nice if the tests could be non-destructive. The market here is a bit different- you have rows and rows of retail stores that sell the parts over-the-counter. You go into one, ask if they have a particular part number; if they do, they bring it out. If not, go to the store next door. They are happy to sell you one 1/4W resistor if you so required. The salesperson generally doesn't know very much about what he or she is selling, and probably wouldn't believe you (or care) if you made a claim that a part is a fake. Bottom line - if you blow up a transistor (through SOA test), you pay for it. So, if you have a non-destructive test, all the better - you have the chance of returning the part and never forking out any money for it...
Cheers!
Clem
Just about every popular japanese line - 2SC/2SA stuff, far as I can tell. We haven't confirmed if the newer OnSemi 211xx is fake as well; hence the interest in the ability to test.
I was discussing this with Chris a couple of e-mails ago that it would be nice if the tests could be non-destructive. The market here is a bit different- you have rows and rows of retail stores that sell the parts over-the-counter. You go into one, ask if they have a particular part number; if they do, they bring it out. If not, go to the store next door. They are happy to sell you one 1/4W resistor if you so required. The salesperson generally doesn't know very much about what he or she is selling, and probably wouldn't believe you (or care) if you made a claim that a part is a fake. Bottom line - if you blow up a transistor (through SOA test), you pay for it. So, if you have a non-destructive test, all the better - you have the chance of returning the part and never forking out any money for it...
Cheers!
Clem
Sounds cool, this retail electronic marketplace of yours.
Sort of like the semiconductor Mecca in Tokyo.
We don't get that here in the US. All parts are bought sight unseen by mail. (at least those for power amps)
I would expect counterfeiters to find good opportunity for unloading their cr@p in such a place.
I don't suppose any of those merchants are on the OnSemi Authorized Distributor list?
If you cannot reliably source audio transistors anywhere, I'd be willing to send you some, no charge. But I only have the MJ21193/4 (hi SOA - low hfe), and the MJ0281/0302 (low SOA - hi hfe), also the Fairchild FJA4213/4313. Because of counterfeit issues, I've long avoided all Japanese types.
I don't think there's anything extraordinary in Japanese transistors that can't be found anywhere else.
I find it amazing that those street retailers would accept returned parts because of SOA failure. And you want to bring a PIC tester with you when shopping for parts? I'm just trying to imagine what this would look like...
You trying to explain to the merchant what you intend to do to his transistors with the PIC contraption you brought along... ("You want to do WHAT!!??"...)
And then explaining that green light means pass, red light means SOA failure. And you just got 3 red lights... ("You come here, destroy transistor, and then you want refund??!!" followed by an invitation that you perform an anatomically impossible act on yourself...)
Sorry, I just couldn't help it...
Andy
Sort of like the semiconductor Mecca in Tokyo.
We don't get that here in the US. All parts are bought sight unseen by mail. (at least those for power amps)
I would expect counterfeiters to find good opportunity for unloading their cr@p in such a place.
I don't suppose any of those merchants are on the OnSemi Authorized Distributor list?
If you cannot reliably source audio transistors anywhere, I'd be willing to send you some, no charge. But I only have the MJ21193/4 (hi SOA - low hfe), and the MJ0281/0302 (low SOA - hi hfe), also the Fairchild FJA4213/4313. Because of counterfeit issues, I've long avoided all Japanese types.
I don't think there's anything extraordinary in Japanese transistors that can't be found anywhere else.
I find it amazing that those street retailers would accept returned parts because of SOA failure. And you want to bring a PIC tester with you when shopping for parts? I'm just trying to imagine what this would look like...
You trying to explain to the merchant what you intend to do to his transistors with the PIC contraption you brought along... ("You want to do WHAT!!??"...)
And then explaining that green light means pass, red light means SOA failure. And you just got 3 red lights... ("You come here, destroy transistor, and then you want refund??!!" followed by an invitation that you perform an anatomically impossible act on yourself...)
Sorry, I just couldn't help it...
Andy
vectorplane said:
No, no one seems to be claiming they are "authorized" dealers - I think the closest ones would be in HK or Singapore...
The preference for Japanese transistors is simply born out of the fact that there was very little 'audio-grade' US-made transistors back then (here), and a much larger selection of japanese stuff (i.e. 2SD555/2SB600 comes to mind, back in the early 80's - versus 2N3771s or 2N3055/2955??!!!). I know of at least one fairly reliable supplier of OnSemi, but the man doesn't stock much audio-oriented stuff, sadly. I guess I might have to turn to the last option - buying direct from the manufacturer and paying the extra cost of S/H and blind taxes...
Regarding the PIC thing - as I said, I'd really prefer the test to be non-destructive, you can see why!!
... So an alternate but reliable test would be good, and we might be able to get away with saying "uh, the transistors you have don't have enough ft... or gain... " - salesperson wouldn't have the foggiest idea what that means... Cheers!
Clem
)
A photo of my "Toshiba" 2N3055 ....
the chip is directly welded on the metal of case ...
My transistor Led
An externally hosted image should be here but it was not working when we last tested it.
the chip is directly welded on the metal of case ...
My transistor Led
An externally hosted image should be here but it was not working when we last tested it.
Gold_xyz:
This is a real pearl of basic research and discovery. You could turn this into a real work of electronic art.
K-amps:
I just noticed the Krell Amp Wiki you mention. What an incredible collective effort! I suddenly want to build one...
Clem:
My offer still holds, in case your part search is unsuccessful. I can ship at no charge (from my employer's mailroom )
I'll consider it a go for that project using a PIC to test transistors. I'll put together a schematic and go from there. Are you ok with programming some of the new dsPIC devices? Or should we rather stick with 16F series parts, for low cost?
I like the multiple independent PWM outputs in the dsPICs, but ultimately, any PIC can implement PWM in software.
PS.
If someone can explain to me what (facts) make testing on the pulsed SOA curves superior to testing on the DC SOA curve, I'll consider that in the design; otherwise DC it is. (2-second DC pulse)
I have 3 quabbles with pulse testing:
1. Compared with the variable voltage supply you need for DC testing, pulse SOA requires more jig circuit complexity to generate high power pulses. The amplitude of both voltage and current also need to be higher than for DC, since pulsed SOA boundaries lie farther out on the graph.
2. You have to worry about pulse rise and fall times. (these are very high power square waves--think of the driving slew rate you'll need) If they are not fast enough, you'll take the DUT through V/I excursions that may fall outside the SOA, even for a genuine device. Remember that your testing is already occuring awfully close to the SOA line. Doesn't take much to cross it. How can you be sure you're not generating a false positive test (destroying a good one) because of test waveform slew-rate limitations?
3. Because pulsed testing gives the die some "breathing" room (time to dissipate heat between pulses), it factors some thermal conductivity into what you're actually measuring. Since you'd expect the fake to have similar if not identical thermal dynamics to the original (the cases on these things are decidedly NOT what gives them away), inclusion of any thermal figure of merit into the measurement, dilutes the potentially significant V.I difference you're trying to capture between a good and a fake.
This is why I think the DC SOA is the toughest and most revealing test for a transistor.
Andy
This is a real pearl of basic research and discovery. You could turn this into a real work of electronic art.
K-amps:
I just noticed the Krell Amp Wiki you mention. What an incredible collective effort! I suddenly want to build one...
Clem:
My offer still holds, in case your part search is unsuccessful. I can ship at no charge (from my employer's mailroom
)I'll consider it a go for that project using a PIC to test transistors. I'll put together a schematic and go from there. Are you ok with programming some of the new dsPIC devices? Or should we rather stick with 16F series parts, for low cost?
I like the multiple independent PWM outputs in the dsPICs, but ultimately, any PIC can implement PWM in software.
PS.
If someone can explain to me what (facts) make testing on the pulsed SOA curves superior to testing on the DC SOA curve, I'll consider that in the design; otherwise DC it is. (2-second DC pulse)
I have 3 quabbles with pulse testing:
1. Compared with the variable voltage supply you need for DC testing, pulse SOA requires more jig circuit complexity to generate high power pulses. The amplitude of both voltage and current also need to be higher than for DC, since pulsed SOA boundaries lie farther out on the graph.
2. You have to worry about pulse rise and fall times. (these are very high power square waves--think of the driving slew rate you'll need) If they are not fast enough, you'll take the DUT through V/I excursions that may fall outside the SOA, even for a genuine device. Remember that your testing is already occuring awfully close to the SOA line. Doesn't take much to cross it. How can you be sure you're not generating a false positive test (destroying a good one) because of test waveform slew-rate limitations?
3. Because pulsed testing gives the die some "breathing" room (time to dissipate heat between pulses), it factors some thermal conductivity into what you're actually measuring. Since you'd expect the fake to have similar if not identical thermal dynamics to the original (the cases on these things are decidedly NOT what gives them away), inclusion of any thermal figure of merit into the measurement, dilutes the potentially significant V.I difference you're trying to capture between a good and a fake.
This is why I think the DC SOA is the toughest and most revealing test for a transistor.
Andy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.