ST only makes the TO-220 cased 3055 now, the metal can now obsolete. ON is the only TO-3 maker with any real volume these days. Even then they don’t stock many of any given type anymore in DK or Mouser.
Everybody and their dog seems to make a BD139/40 of sorts. Nothing wrong with the ST or ON versions, but the fT’s are not as high as the old NXP/legacy Philips. Since the parameter is NOT specified in the data sheet, it’s not technically “wrong“. They will work where BD139/40 are called for - you‘re not going to end up with some 3 MHz TIP31C equivalent from ON. It will be better than that, and I suspect they are even from China and India. They are just not that hard to make properly. If you don’t like what you end up with from CDIL, you know where to get “better” ones.
Luguang may indeed make some good copies. But treat them on a case by case basis until you know what you get. If they started copying old Sanken MT200’s I’d take them with a very big grain (more like a whole ocean) of salt till they pass the test of time. So far no one else has.
Everybody and their dog seems to make a BD139/40 of sorts. Nothing wrong with the ST or ON versions, but the fT’s are not as high as the old NXP/legacy Philips. Since the parameter is NOT specified in the data sheet, it’s not technically “wrong“. They will work where BD139/40 are called for - you‘re not going to end up with some 3 MHz TIP31C equivalent from ON. It will be better than that, and I suspect they are even from China and India. They are just not that hard to make properly. If you don’t like what you end up with from CDIL, you know where to get “better” ones.
Luguang may indeed make some good copies. But treat them on a case by case basis until you know what you get. If they started copying old Sanken MT200’s I’d take them with a very big grain (more like a whole ocean) of salt till they pass the test of time. So far no one else has.
Have you encountered transistors that went bad because of age ?. ( properly used, not overpowered, overvoltage etc. )
I have not tried these, but have bought other stuff from TME, some CDIL transistors for example. All works fine for me.
Not on anybody else’s.
Nobody else’s makes it to 190 MHz, so they don’t state what it is. There’s no min/max on the Philips data sheet, and everybody else’s just omits it. Technically, in a court of law, it’s not lying. Old SGS types came in around 50. Fairchild usually about the same. Has it gotten better? I don’t know. Good enough for most uses when you need a generic driver transistor. Or a Vbe multiplier that doesn’t need a |-*(%ing insulator. Old ones actually did, but not the current crop of them. These parts are just generics, so I tend to treat them that way. Don’t expect the world of it and don’t sweat it when you don’t get it. Most places where I’m worried about fT being north of 100 MHz require more than an 80 volt part so I’d be looking further up the food chain.
Nobody else’s makes it to 190 MHz, so they don’t state what it is. There’s no min/max on the Philips data sheet, and everybody else’s just omits it. Technically, in a court of law, it’s not lying. Old SGS types came in around 50. Fairchild usually about the same. Has it gotten better? I don’t know. Good enough for most uses when you need a generic driver transistor. Or a Vbe multiplier that doesn’t need a |-*(%ing insulator. Old ones actually did, but not the current crop of them. These parts are just generics, so I tend to treat them that way. Don’t expect the world of it and don’t sweat it when you don’t get it. Most places where I’m worried about fT being north of 100 MHz require more than an 80 volt part so I’d be looking further up the food chain.
You know what? I have never said any manufacturer is junk. What I said was to watch out and actually look at what you are buying and TEST THEM, TEST MANY AND OFTEN, until you know them.
I have received many transistors over the years marked as originals, and they were not. Many marked with type numbers that clearly were not even close the the original part. Eventually I gave up trying to find honesty. As I hinted at earlier, if the data sheet is incomplete, you can bet these are not the exact type as marked. They may not even be a reasonable substitute. I've run into that often enough, and I have thrown out hundreds of transistors at my expense. Same for ICs (some non-functional). So if the clowns of the world are willing to sell that garbage, what else are they willing to flush into the world market? Really barren datasheets? Run.
I don't care who you are in this world, if you make a good product you do stand behind it and want the world to know. No excuses. If the product is less expensive, there will be a very good reason for it.
Hi xXBrunoXx,
When I bought Japanese transistors, I knew the guy personally that went to Japan and bought them. For 15 years I had maybe three bad transistors. He asked why I didn't complain about his higher prices ... well, it was a lot less expensive to buy known good products. Especially when your reputation rides on it. I mark the parts up anyway, so within reason it doesn't matter to me except in my inventory carrying expenses. Turns out to be a very good business model for me. Customers could absolutely rely on me to deliver well executed, reliable service, and I had a number of techs working for me. Train them well, feed them good parts and supply good instruments for them to use. There is no other way to get good results. Long term it paid off.
I have received many transistors over the years marked as originals, and they were not. Many marked with type numbers that clearly were not even close the the original part. Eventually I gave up trying to find honesty. As I hinted at earlier, if the data sheet is incomplete, you can bet these are not the exact type as marked. They may not even be a reasonable substitute. I've run into that often enough, and I have thrown out hundreds of transistors at my expense. Same for ICs (some non-functional). So if the clowns of the world are willing to sell that garbage, what else are they willing to flush into the world market? Really barren datasheets? Run.
I don't care who you are in this world, if you make a good product you do stand behind it and want the world to know. No excuses. If the product is less expensive, there will be a very good reason for it.
Hi xXBrunoXx,
Yes I have, many times. Even under warranty with infant transistors. Less so today using On-Semi or Toshiba, and yes. I'll pay extra for having reliable parts. I even test each part before I install it using good parts. Testing is much less expensive than cleaning up after a failure, and especially if you have to troubleshoot it.
When I bought Japanese transistors, I knew the guy personally that went to Japan and bought them. For 15 years I had maybe three bad transistors. He asked why I didn't complain about his higher prices ... well, it was a lot less expensive to buy known good products. Especially when your reputation rides on it. I mark the parts up anyway, so within reason it doesn't matter to me except in my inventory carrying expenses. Turns out to be a very good business model for me. Customers could absolutely rely on me to deliver well executed, reliable service, and I had a number of techs working for me. Train them well, feed them good parts and supply good instruments for them to use. There is no other way to get good results. Long term it paid off.
Curious here, Ana. If someone came in with an old blown amp with a pair of 40636’s driven by the old TO-39 drivers (and all, both channels, were fried) what would you actually use to replace them? Assuming of course, the customer was attached enough to the piece to pay $200-300 for the job? (most would simply pitch it).
Related question - an old console stereo with PNP Ge’s in a classic totem pole. Driver was some unknown house number TO-92 darlington. One dead, other still working but put your finger on it and it’s obviously being run within an inch of its life. Replaced both with D40K2’s from a surplus house - tested good, used in various projects over the last 10 or 15 years. Still had about 30 left. Considered a “hack” or not?
Related question - an old console stereo with PNP Ge’s in a classic totem pole. Driver was some unknown house number TO-92 darlington. One dead, other still working but put your finger on it and it’s obviously being run within an inch of its life. Replaced both with D40K2’s from a surplus house - tested good, used in various projects over the last 10 or 15 years. Still had about 30 left. Considered a “hack” or not?
Anatech, would you recommend a cheap automatic transistor tester, good features, but priced low enough for hobby use?
And what do you think of the transistor test function found in digital multi-meters, are they adequate for initial testing?
Also, if there is not much difference in price between 3055 and 3773, would it be a good option to substitute with a 3773 or equivalent European / MJ series transistor with higher voltage / current / Ft ratings, same case?
And what do you think of the transistor test function found in digital multi-meters, are they adequate for initial testing?
Also, if there is not much difference in price between 3055 and 3773, would it be a good option to substitute with a 3773 or equivalent European / MJ series transistor with higher voltage / current / Ft ratings, same case?
Hi wg_ski,
I'd look up the 40636 to find the basic specs (looks like a high voltage 2N3055). Then I would attempt to find a transistor I had on hand, or something I could get in real inventory (not Ebay or AliE). I might go with an MJ15022 for example. It's faster, so the amp would need recompensating. If I couldn't get the correct parts or good substitutes, that is the end of the job. Search time is as valuable as bench time. No sense beating your head against the wall.
I didn't spend a great deal of time researching your example. I don't have time to do that. Each case is different as well, everything from the damage, the circuit, how it is put together and the customer. Some people will kill the most robust equipment, you can't win. I walk on those before repairing it.
In this case it isn't a hack, or maybe an understandable hack. The big thing is that you are using known, good quality parts for a long lasting and reliable repair. I'll assume you checked the characteristics and they were suitable. I think I would select a larger package than the original, or attach a heat sink so it ran at a lower temperature. Same for the working one as we know it has been run hard (abused by design). Again, a judgement call on the tech who is working on it. I'll never throw darts at a person who makes responsible decisions.
One fact we need to consider is that not everything is reasonably repairable, and we do not have to fix everything.
I'd look up the 40636 to find the basic specs (looks like a high voltage 2N3055). Then I would attempt to find a transistor I had on hand, or something I could get in real inventory (not Ebay or AliE). I might go with an MJ15022 for example. It's faster, so the amp would need recompensating. If I couldn't get the correct parts or good substitutes, that is the end of the job. Search time is as valuable as bench time. No sense beating your head against the wall.
I didn't spend a great deal of time researching your example. I don't have time to do that. Each case is different as well, everything from the damage, the circuit, how it is put together and the customer. Some people will kill the most robust equipment, you can't win. I walk on those before repairing it.
In this case it isn't a hack, or maybe an understandable hack. The big thing is that you are using known, good quality parts for a long lasting and reliable repair. I'll assume you checked the characteristics and they were suitable. I think I would select a larger package than the original, or attach a heat sink so it ran at a lower temperature. Same for the working one as we know it has been run hard (abused by design). Again, a judgement call on the tech who is working on it. I'll never throw darts at a person who makes responsible decisions.
One fact we need to consider is that not everything is reasonably repairable, and we do not have to fix everything.
Hi NareshBrd,
There isn't one tester that covers everything. I have various testers, one I use a lot is the Heathkit IT-18 (1970s). It measures leakage in two modes, plus rough beta. Being analogue you can heat or cool the semiconductor to see trends and sudden changes. It is a basic instrument, but you can do a great deal with it. I wore out my first one, got another and have one as backup. I'd say they are very useful if you understand how they work and how transistors work along with failure modes.
Modern terters can be useful, and I have a Tek 577 along with the old Heathkit tracer. I also designed a jig to match signal transistors. That gets used a lot as well.
Beta test in cheap meters can be a rough sort, but transistors are extremely sensitive to temperature. You cannot match transistors using those testers. The Heathkit IT-18 is one of the better tests to see if a transistor is funky or not.
Replacing outputs can be a fun exercise. Even parts with the same number may sometimes not be stable. Equipment built in the 60s through 80s could be very sensitive to drivers and outputs. Old techs are very familiar with this (one reason an oscilloscope is a critical piece of equipment). Sometimes you just need some equipment, hobbyist or not. I started with a 500 KHz tube oscilloscope, single trace and no horizontal sync. Army surplus, I did a lot with it. You find a way, and sacrifice.
The best instrument is between your ears. The more you know, the easier everything is and you can figure out how to test parts and get real answers.
There isn't one tester that covers everything. I have various testers, one I use a lot is the Heathkit IT-18 (1970s). It measures leakage in two modes, plus rough beta. Being analogue you can heat or cool the semiconductor to see trends and sudden changes. It is a basic instrument, but you can do a great deal with it. I wore out my first one, got another and have one as backup. I'd say they are very useful if you understand how they work and how transistors work along with failure modes.
Modern terters can be useful, and I have a Tek 577 along with the old Heathkit tracer. I also designed a jig to match signal transistors. That gets used a lot as well.
Beta test in cheap meters can be a rough sort, but transistors are extremely sensitive to temperature. You cannot match transistors using those testers. The Heathkit IT-18 is one of the better tests to see if a transistor is funky or not.
Replacing outputs can be a fun exercise. Even parts with the same number may sometimes not be stable. Equipment built in the 60s through 80s could be very sensitive to drivers and outputs. Old techs are very familiar with this (one reason an oscilloscope is a critical piece of equipment). Sometimes you just need some equipment, hobbyist or not. I started with a 500 KHz tube oscilloscope, single trace and no horizontal sync. Army surplus, I did a lot with it. You find a way, and sacrifice.
The best instrument is between your ears. The more you know, the easier everything is and you can figure out how to test parts and get real answers.
Was more or less wondering what you do about parts that are unobtainable through the standard channels. Like the TO5/39’s…. Central Semi, Solid State Inc… or God forbid NTE128/9 (the direct replacement)? You can get them from Digikey, but do you trust them for customer repairs? Or punt all of them and use whatever modern TO-126 driver you happen to stock. I figured as much on the MJ15022/4.
That poor little TO-92 darlington was being run at 30V and 28 mA. EEEEEYOWCH! No wonder it failed and not the 400 year old Germaniums (Which could easily develop tin whiskers which may or may not safely clear). A brand new MPSA13 would be just as much at risk, and everything bigger that’s still in production has the built in base-emitter resistors so they won’t bias correctly. Ahhh.. what about these D40k’s? I’ve even seen them in that same position before. It was either that or “make” one out of a 2SC3902 and a KSC1845. The D40K’s did come from Skycraft eons ago, and if they were NFG that would have shown up by then as I‘d been using them for personal projects and they behave perfectly normal.
That poor little TO-92 darlington was being run at 30V and 28 mA. EEEEEYOWCH! No wonder it failed and not the 400 year old Germaniums (Which could easily develop tin whiskers which may or may not safely clear). A brand new MPSA13 would be just as much at risk, and everything bigger that’s still in production has the built in base-emitter resistors so they won’t bias correctly. Ahhh.. what about these D40k’s? I’ve even seen them in that same position before. It was either that or “make” one out of a 2SC3902 and a KSC1845. The D40K’s did come from Skycraft eons ago, and if they were NFG that would have shown up by then as I‘d been using them for personal projects and they behave perfectly normal.
Hi wg_ski,
I have a lot of old parts, including Ge devices. I would substitute a TO-5/39 with a TO-126 part and commonly do. I have some real 2N3440 and 2N5415 on hand, and some other TO-5/39 parts I bought on "last time buy". Not many as they were expensive by that time.
However if I couldn't find a real part from a known source, I would either put together a composite part as you suggested or walk on the deal. Definately I would use a larger device under the conditions you mentioned. There is no point in fixing something in a way where it would simply fail again. Your reputation takes a hit and you end up with a repair that keeps coming back. I don't want to be married to a bad design.
I know NTE and ECG parts well. If I see one anywhere in a repair that has a proper replacement, those are on the estimate for replacement. I refuse to use them, and I will not risk a repair with one installed. Only TV guys, or poorly trained techs use those things. Some industrial guys do as well, but they are not good parts. Who even knows what they are really supposed to be.
D40K, seeing the package brings back memories, duowatt devices. I would use a TO-220 and turn the package, bending leads to make them line up. A small heat sink if they run hot. A TIP102G might work well. I have used some Central parts and they have been okay. They make good rectifiers. The Central D40K would be a neater repair.
The ones you have are familiar to you. You have history and experience with them, so it's all good. I wouldn't pull them, just ECG, NTE or devices clearly intended to be fakes. A laser etched old signal transistor number for example, typically hFE measures about 200. It seems they use the same part for all kinds of remarked transistors.
I have a lot of old parts, including Ge devices. I would substitute a TO-5/39 with a TO-126 part and commonly do. I have some real 2N3440 and 2N5415 on hand, and some other TO-5/39 parts I bought on "last time buy". Not many as they were expensive by that time.
However if I couldn't find a real part from a known source, I would either put together a composite part as you suggested or walk on the deal. Definately I would use a larger device under the conditions you mentioned. There is no point in fixing something in a way where it would simply fail again. Your reputation takes a hit and you end up with a repair that keeps coming back. I don't want to be married to a bad design.
I know NTE and ECG parts well. If I see one anywhere in a repair that has a proper replacement, those are on the estimate for replacement. I refuse to use them, and I will not risk a repair with one installed. Only TV guys, or poorly trained techs use those things. Some industrial guys do as well, but they are not good parts. Who even knows what they are really supposed to be.
D40K, seeing the package brings back memories, duowatt devices. I would use a TO-220 and turn the package, bending leads to make them line up. A small heat sink if they run hot. A TIP102G might work well. I have used some Central parts and they have been okay. They make good rectifiers. The Central D40K would be a neater repair.
The ones you have are familiar to you. You have history and experience with them, so it's all good. I wouldn't pull them, just ECG, NTE or devices clearly intended to be fakes. A laser etched old signal transistor number for example, typically hFE measures about 200. It seems they use the same part for all kinds of remarked transistors.
In diyaudio everyone wants to use genuine, top quality components but source, availablity & cost force them to use what is available to them.
I was looking for latfets but i left disappointed because it is not available here. Good news is very recently my niece went to UK for study, so now it's become easier for me to bringing those laterals. Because who wants to pay 10x more than the price of parts for custom duty and shipping? Certainly i'm not in that list.
I was looking for latfets but i left disappointed because it is not available here. Good news is very recently my niece went to UK for study, so now it's become easier for me to bringing those laterals. Because who wants to pay 10x more than the price of parts for custom duty and shipping? Certainly i'm not in that list.
Last edited:
Fyi
Of late I see Mouser selling Diotech brand semis, which is a German semi manufacturer. From reviewing the datasheets they look to be a reputable mfr, anyone have experience with them?
One issue with postal shipping is putting a value on the item which is required for insurance purposes, if you pay for a tracked item, so if they get lost or stolen or damaged you are covered.
Problem is if the item is identified as a sale item which then has the customs add duty and taxes. I identify the items as commercial samples
Of late I see Mouser selling Diotech brand semis, which is a German semi manufacturer. From reviewing the datasheets they look to be a reputable mfr, anyone have experience with them?
One issue with postal shipping is putting a value on the item which is required for insurance purposes, if you pay for a tracked item, so if they get lost or stolen or damaged you are covered.
Problem is if the item is identified as a sale item which then has the customs add duty and taxes. I identify the items as commercial samples
Last edited:
I used their bc327/337 transistors, no complaints. The company was founded in the seventies by people from ITT, and I have no reason to think they don't make the usual german quality stuff.
About the BD139/140 and it's mysterious Ft value. I have heard it said that they are originally to-92 BC639/640 devices put in a to-126 package. If you compare the datasheets, this seems indeed plausible, the parameters are basically the same. Interestingly, for the BC639/640, the Ft value is always in the datasheet. It is not always given in the same way though. Sometimes as minimum 100Mhz, other times 200 Mhz typical, or different values for NPN and PNP.
Now, it is claimed by some that modern (fake?) BD139/140's are not made in the same way, but with a bigger die, which makes them slower. It is my theory that this is not probable when a manufacturer makes both the BC639/640 and the BD139/140, like CDIL and Luguang. Why make different dies if you can use the same small one for both types?
ST, formerly SGS thomson, however never made the BC639/640 I think. Suspect...?
How could one do a quick and dirty measurement without an oscilloscope?
Now, it is claimed by some that modern (fake?) BD139/140's are not made in the same way, but with a bigger die, which makes them slower. It is my theory that this is not probable when a manufacturer makes both the BC639/640 and the BD139/140, like CDIL and Luguang. Why make different dies if you can use the same small one for both types?
ST, formerly SGS thomson, however never made the BC639/640 I think. Suspect...?
How could one do a quick and dirty measurement without an oscilloscope?
Last edited: