My Transistors, original or copy?

sanken fakes

Dear fellows.

I analized some sanken transistors that I bought in different places , and besides that the "suspects" ones have a low beta, I discovered something that i would like to confirm .

I have attached two photos, I marked with a red circle the zone of interest. one has a number engraved inside this small circular recess, and the other is blank.

the one with date code 78P is definitely a fake.

Could this be considered a "rule" for detecting malicious sanken's ?

any comentarie will be appreciated

regards

Alexander rubli
alex@rubli.net
 

Attachments

  • SANKEN1.jpg
    SANKEN1.jpg
    15.1 KB · Views: 629
  • SANKEN2.jpg
    SANKEN2.jpg
    14.8 KB · Views: 629
if it had been the TO-3P packages, i would have better data for you on the dimples in the plastic. however, the second picture shows a dimple with rounded edges, as if the case had been stripped with a wire wheel or sandpaper, so that device is suspect. the dimple in the first picture has sharp edges.
 
Ex-Moderator R.I.P.
Joined 2005
If theres a "6" or "9" there could be a difference

On the fakes I have seen shown here especially the number 6 and 9 are more rounded, which they are not on originals

Also it seems the small dot holes you have marked should be sharp edged and smooth almost shiny at their bottom

And all letters/numbers are nicely rounded at all ends

Mind you, Im no expert, I just happen to have some old burned ones :eek:
 

Attachments

  • sanken.jpg
    sanken.jpg
    23.2 KB · Views: 588
Last edited:
info about the mold marks (dimples) can only be compiled if you have access to a lot of known good parts over a span of about 5 or 10 years. manufacturers do NOT publish any info on mold marks. one very good reason that mold mark data isn't published would be to keep counterfeiters guessing, but i suspect the real reason it's not published is that it's too much information to keep track of. i was lucky to find a correctly laid out drawing and description of the lot/date code/ beta range ID markings for sanken devices. toshiba keeps more accurate data on their marking layouts, and publishes the data in their "Reliability Program Guide", because their company has an active program to monitor the reliability of the devices they manufacture (and discover counterfeits as a side benefit).
 
the company i work for has decided to hire a testing facility that specializes in counterfeit detection. there are a lot of methods the facility uses to detect counterfeits.
1. X-ray machines to look inside the device without having to break it apart, and even look at the emitter pattern of the die.
2. transistor curve tracers which can even identify what transistor used to make a fake might have been originally
3. accurate dimensional measurements including the flatness of the heat sink tab.
4 capacitance measurements (this is currently one of the methods i use)
5. mold marks and printing (also one of the methods i currently use.
6 materials testing. are the fakes made with different materials than the original?

the biggest part of solving the problem has been to identify the sources of the fakes, bypass the processes that allow these sources to provide certain components and order the devices directly from the manufacturer of the equipment. so far, so good, but there's a steep learning curve for management, because nobody in the company has ever identified this problem before, and nobody wants to admit that an 800lb gorilla is regularly getting poked in the eye.
 
the company i work for has decided to hire a testing facility that specializes in counterfeit detection. there are a lot of methods the facility uses to detect counterfeits.
1. X-ray machines to look inside the device without having to break it apart, and even look at the emitter pattern of the die.
2. transistor curve tracers which can even identify what transistor used to make a fake might have been originally
3. accurate dimensional measurements including the flatness of the heat sink tab.
4 capacitance measurements (this is currently one of the methods i use)
5. mold marks and printing (also one of the methods i currently use.
6 materials testing. are the fakes made with different materials than the original?

the biggest part of solving the problem has been to identify the sources of the fakes, bypass the processes that allow these sources to provide certain components and order the devices directly from the manufacturer of the equipment. so far, so good, but there's a steep learning curve for management, because nobody in the company has ever identified this problem before, and nobody wants to admit that an 800lb gorilla is regularly getting poked in the eye.


What kind of meter are you using to do the capacitance measurements?
 
the company i work for has decided to hire a testing facility that specializes in counterfeit detection. there are a lot of methods the facility uses to detect counterfeits.
1. X-ray machines to look inside the device without having to break it apart, and even look at the emitter pattern of the die.
2. transistor curve tracers which can even identify what transistor used to make a fake might have been originally
3. accurate dimensional measurements including the flatness of the heat sink tab.
4 capacitance measurements (this is currently one of the methods i use)
5. mold marks and printing (also one of the methods i currently use.
6 materials testing. are the fakes made with different materials than the original?

the biggest part of solving the problem has been to identify the sources of the fakes, bypass the processes that allow these sources to provide certain components and order the devices directly from the manufacturer of the equipment. so far, so good, but there's a steep learning curve for management, because nobody in the company has ever identified this problem before, and nobody wants to admit that an 800lb gorilla is regularly getting poked in the eye.

By this thread is to see a curve tracer concept:
http://www.diyaudio.com/forums/solid-state/151253-diy-curve-tracer-pc.html
If it is possible to implement a appropriate software to get the p-Spice parameters about this curves, then there are a good precondition to find out the quality standart and further more realistic simulation results. Additional I have a good tool to check the values of the manufacturer if he publishes it.
 
some of the transistors in question have Vce ratings of up to 300V, and Ic ratings of up to 20A (Id for some FETs go up to 60 or 80A or more). and don't forget that for PNP and P-channel devices, these are negative values. so a curve tracer that can do that isn't a simple piece of equipment.
 
i just discovered something today about the font Sanken uses on their parts. it has to do with the curvature of the hook and the loop of the 6 and 9, the curvature on the leading edge of the 2, and the hook on the 1. study them carefully, and you will see the font used by Sanken is quite unique. the loop of the 5 is unique as well, but not as obvious. i haven't seen a 4 or a 7 yet, and the 3, 8, and 0 are not unique.



another thought i had while mulling over the counterfeit problem, is that (and i want to run this idea by everybody here, so i may also start a thread about it) is that one simple method that might be able to non-destructively detect fakes, might to be to find it's Early Effect voltage. so my question about this would be "is it reasonably safe to assume that even with variations within the limits of a particular part number's range of parameters, that the Early voltage remains fairly constant?". or will the Early voltage vary radically between samples of a particular transistor that has an 8:1 acceptable range of beta according to the data sheet? if the Early voltage is reasonably constant within normal device-to-device variations within a part number's normal range, then a counterfeit would most likely be very far from the target range of an original, because of differences in die structure and size. that's my idea at least, if it's not workable please tell me so. i was thinking of a test setup with a couple of preset constant current sources to establish base current and a variable voltage power supply. the setup would check a short series of data points of Ic vs Vce, and use a simple calculation to derive the Early voltage.(or graphic extrapolation of Va?)
 
i just discovered something today about the font Sanken uses on their parts. it has to do with the curvature of the hook and the loop of the 6 and 9, the curvature on the leading edge of the 2, and the hook on the 1. study them carefully, and you will see the font used by Sanken is quite unique. the loop of the 5 is unique as well, but not as obvious. i haven't seen a 4 or a 7 yet, and the 3, 8, and 0 are not unique.

I don't understand this - especially the follow:
"font Sanken uses on their parts"
 
if you look at genuine sanken parts, their print font has some unique characteristics that set it apart from the printing used on counterfeits, namely the shapes of some of the digits. i thought what i said was clear enough, maybe it doesn't translate well...
now I understand - you mean the kind of labeling resp. lettering like the two different versions as show in the attachement from
Counterfeit Transistors
This is one of the features. But please note, there are more than one company, that make copies. Surely even those where no differences to an original Sanken are visible in the front label.
I have heard, Sanken is a franchise partner from Semelab/Magnatec
Magnatec. Power System Semiconductors
Therefore I would ask there for a trusted distributor; I think that's the only way for maximum security.
Another way would be to create p-spice parameters (only early effect isn't enough - so I think) with help of curve tracer and pc based analyse system. But in the moment nobody can do this.
 

Attachments

  • Sanken Fake and Genuine.pdf
    79.7 KB · Views: 198
Last edited: