Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
Is it possible that the 151-0367-00 transistor story is horribly wrong?
|
After solving the children's diseases on my tektronix, struggling with the proper electrolytes in the low voltage line,
and of course with a lot of help from the people on this forum - the thing worked. Of course, like everyone who comes here, I've read a lot of texts about the problematic transistor that has a 100% failure rate. According to the advice available I replaced these transistors, and I noticed that the trigger is no longer as good and stable as before. Then I started reading texts about the SKA6516 transistor, wanting to find its characteristics. In the book Semiconductors-catalog-1982 I found that it is a transistor A5T3571 which has the following characteristics: Ft min - 1,2 GHz Vce / Ice - 6/5 Hfe min - 100 Vceo - 30V Vcbo - 40 V Vebo - 3 V Icbo - 10nA Ccb - 1,5 pF Vce(sat) - 300/20 mV/mA Ic max - 50 mA I've also read on the forums that people have swapped them with 2N2369 even 2N2222. Most often, the advice was to replace it with KSP10BU, which is relatively easily available today. I was also advised to try replacing them with BF199 which was used in CRT TVs and is also still relatively easy to obtain. I grabbed my box from the CRT TV days, and found a bunch of BF199, BF198, BF311. Then I got hold of that famous Chinese tester and I was amazed - absolutely everything and one on the tester was declared defective, and it shows c-e leakage as an additional diode. Also, the identical behavior as with Tektronix transistors can be observed, depending on the direction in which you put it in the tester, one way it will show that it is a PNP, if you turn it the other way around it will show that it is an NPN transistor. And this is where my doubt begins – 1. how is it possible that absolutely the entire stock high frequency transistors I have is defective, and that not a single one is good? 2. why did tektronix decide to hire semiconductor manufacturers to make such a transistor with custom characteristics, when at that time all of transistors what is listed as replacement today, was available at the time of manufacture of the oscilloscope? 3. Is it possible that these Chinese testers cannot handle the characteristics of these transistors, so they declare them defective? 4. why didn't I know or feel the need to open my tektronix until one of the big electrolytes failed and the device stopped working because of it, and most likely these transistors were in the same state the whole time? 5. if those transistors are so bad, why does my oscilloscope work at all when they are in there? 6. I put the original tektronix transistors back on the trigger board (luckily I didn't throw them away), and why does the trigger work better now? 7. Why does it say - selected from stock, in some places next to these transistors in the manual. Does that mean that in addition to being custom-made, they also paired them? What are your thoughts on these issues? |
|
I assume you know the pinout of the KSP10BU is different. Putting it in like the original will cause the device to not work. The leakage of the original 0367 transistors is a diode from C to E. It will check and work that way. I have done this to some in the past. You will likely notice the oxidation on the leads of the older ones with round leads. That oxidation will creep into the case and cause leakage. The leads with square leads do not have the oxidation problem. It was the manufacturing quality at the time and quality of clean rooms. If you work on older stereos from the 70's, you will see many signal transistors that are bad by listening to it, testing or both. Many times the leads will be black with oxidation. Newer numbers had a better process of manufacture and better clean rooms.
I suspect the hfe of the 0367s was well out of normal. That C to E leakage is the cause of that. I have seen triggering and other problems from the original 0367s. Once replaced, the problems went away. You may have to touch up controls a bit to adjust for the new transistor. Resistors out of tolerance or bad in another way, can cause triggering problems. Solder joints and contact problems are also to blame. Selected from stock can mean a certain hfe range. Ones made in recent times will have better hfe range than older ones. The selected could be ones that are close in hfe. If you have a 7D15, the 0402 is the same as the 0367. That is the only plugin that uses a different number for the same transistor. It will also be leaky in the same way as the 0367. Mark |
|
I also agree and have confirmed Marks findings about the 151-0367-00 transistors. I have confirmed the issue on a Tektronix 577 curve tracer as well as a PC ATLAS PRO component tester. I video on the topic can be found here:
toggle quoted message
Show quoted text
Zen -----Original Message-----
From: [email protected] <[email protected]> On Behalf Of radiobero.bb@... Sent: Wednesday, September 13, 2023 11:32 AM To: [email protected] Subject: Re: [TekScopes] Is it possible that the 151-0367-00 transistor story is horribly wrong? Thanks for the reply Mark I would like to know what device you used when you determined that the transistors you tested had c-e leakage? |
|
I used a VTVM to test the transistors. Notice the older style case is larger than a typical TO-92. These are the ones with the round tarnished leads. Zen went further to prove these are bad. The LCD meter shows the diode from C to E on the screen. It looks like the horizontal outputs TVs would use with the internal damper diode. This is a part that is best changed on sight. Depending upon how much Tek. gear you have, get a qty. of KSP10BU. A bag of 100 is not expensive. The 7B92A has a number of these 0367s in them.
Mark |
|
All excellent questions. You are right to be skeptical about a 100% failure rate, and your following that up with tests on a BF199 indict the test method more than the transistor.
toggle quoted message
Show quoted text
It's relevant to mention that many high-ft transistors have vey low breakdown voltages in inverted mode (i.e., collector and emitter exchanged). Depending on how those Chinese component testers do their analysis, it's possible that the transistor is actually driven into reverse avalanche breakdown, which could very well be (mis)interpreted by the tester's algorithm as a diode being present and forward-biased when the emitter voltage is above that of the collector. For grins, I tried some 5GHz microwave transistors (Toshiba 2SC3302) that are known good (my students use these by the bushel in the microwave circuits class I teach). They (the transistors, not the students) all behave the same on the Chinese component tester as the "bad" 151-0367-00 transistors. The tester insisted that these transistors had the infamous C-E diode. A standard DMM diode test does not show this diode, because the DMM's applied voltage is too low to provoke reverse breakdown. So, the tester is the problem, and not the transistor. Based on this set of results, an automatic "replace on sight" policy for these transistors based solely on a component tester's say-so seems unjustified, especially since the tester is most likely to make an error when evaluating possibly expensive high-ft devices. -- Cheers Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/13/2023 05:33, radiobero.bb@... wrote:
And this is where my doubt begins – |
|
I should also mention that one suggested replacement for the 151-0367-00, the KSP10BU, has something like twice the C-B capacitance, so any circuits that are sensitive to that parameter will suffer degraded bandwidth.? It could be that trigger stability might be impaired at high frequencies, for example, if the transistor is used in a critical part of a sweep circuit. You want a sub that has not only a similar ft and voltage rating, but also a similar max Ic (since that is a strong correlate to Ccb).
toggle quoted message
Show quoted text
-- Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/13/2023 23:49, Tom Lee wrote:
So, the tester is the problem, and not the transistor. Based on this set of results, an automatic "replace on sight" policy for these transistors based solely on a component tester's say-so seems unjustified, especially since the tester is most likely to make an error when evaluating possibly expensive high-ft devices. |
|
Thanks for the reply Zen.
I am a regular viewer of your channel, I really appreciate your videos, I have watched almost all of them related to the field that interests me. The quality of your video works is really fascinating and I learned a lot from them, and I believe that many other electronics enthusiasts are of the same opinion. Thanks to you too, Mr. Lee, for your reply. As for these transistors, I have for now sent back all the ones that looked good (no corrosion), and tested good the old fashioned way with a multimeter. There was only one that showed some strange symptoms when tested with a multimeter, and I temporarily replaced it with a 2N3069 because I don't have another at the moment. Maybe I'll replace it with the BF199, I'll see how the device behaves. Maybe I didn't search hard enough, but I don't know if you've noticed that if you're looking for that type of transistor that goes up to 1.1GHz frequency, you couldn't find that there is a transistor available today with these characteristics, and that at the same time its HFE min.=100 It seems that they still had a reason to order a custom-made transistor, according to the characteristics listed below. However, I am not an expert enough to say why it was necessary or what it was intended to achieve. Beros |
|
Hi Beros,
toggle quoted message
Show quoted text
It is more probable that the high-beta devices were found from sorting through a population of off-the-shelf devices, rather than having a special one custom made. And that's how you would find a suitable replacement here. Buy a bunch that are otherwise close in parameters, and cherry-pick the ones with adequate beta. A beta of 100 is not a crazy-high value, so a fair fraction would probably meet that requirement. (Of course, Murphy guarantees that you'll just happen to get a batch that all happen to have the same, low value of beta...) -- Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/14/2023 02:19, radiobero.bb@... wrote:
Maybe I didn't search hard enough, but I don't know if you've noticed that if you're looking for that type of transistor that goes up to 1.1GHz frequency, |
|
I worked at Cincinnati Electronics in the late '70s. We were building
toggle quoted message
Show quoted text
military radios. One obsolete transistor was from Motorola. Since this was a NATO design, they agree to produce a run of the obsolete part, but would only ship them untested.For an order of 10K transistors, they would ship 12.5K. Then we had to have a tech sit there testing them, before shipping them to our plant in Mexico. This was easier than the process to get an ECO approved by multiple nations. In my view, an extra 25% shipped showed little confidence in that run. On Thu, Sep 14, 2023 at 5:33?AM Tom Lee <tomlee@...> wrote:
Hi Beros, |
|
I'm dying to know: What yield did the tech ultimately find? That would tell us whether or not a 25% sandbag was overkill. So if you happen to? remember, please share the punchline.
toggle quoted message
Show quoted text
Thanks! -- Cheers, Tom Prof. Thomas H Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/14/2023 3:32 AM, Michael A. Terrell wrote:
they agree to produce a run of the obsolete part, but would |
|
When I served in the US Army fixing radios(early 70s) we had a transistor problem like that. We would order replacement lots by the box of a hundred and then the using an X-Y home brew tester would inspect the new box of parts when we had had a slow period. 30-50 percent failure out of the lot. Whatever the design, they also failed regularly. It was somewhat of an effort to keep enough working spares available.
Bob |
|
Two thoughts:
- High-ft transistors in (Chinese?) testers: good chance they oscillate at some v high frequency? Might be worth an experiement with a ferrite bead or two on the base. - In my experience, when Tek selected devices from a hi-frequency part, the selection parameter was generally related to an HF/switching characteristic (ft, Ccb etc) John |
|
I did a little more digging in that lab today. I have a 475A late serial number on the bench and the 151-0367-00 transistors were not tarnished on the legs. I dug up a data sheet and found that the transistor is a MPSH05 This is speced to have the following:
toggle quoted message
Show quoted text
C->E breakdown voltage 80 Vdc C->B breakdown voltage 80 Vdc E->B breakdown voltage of 4 Vdc E->C Breakdown voltage Not specified 80Mhz switching frequancy Transistor measured with a DCA pro shows the reverse diode currently assumed bad. Moved to a Keithley DMM6500 C->E open circuit as expected E->C 0.754 Volts or 750mV with a test current of 1mA out of the DMM 6500 Just some more thoughts and testing on this matter but this proves that the DCA pro might be getting confused but it IS showing relevant data as confirmed by other instruments of a more refined pedigree. I am interested in doing additional testing if anyone has ideas on what kind of testing could solve this questions definitively once and for all. I can't do SMU mapping of 3 pole devices as I am short the Keithly 2470 but it is on the list. So I am limited to only a Tek 576 and 577 for curve tracing devices over 12.5 as the DCA Pro can only do a collector voltage up to about 12.5 VDC Zen -----Original Message-----
From: [email protected] <[email protected]> On Behalf Of John Sykes via groups.io Sent: Saturday, September 16, 2023 3:42 AM To: [email protected] Subject: Re: [TekScopes] Is it possible that the 151-0367-00 transistor story is horribly wrong? Two thoughts: - High-ft transistors in (Chinese?) testers: good chance they oscillate at some v high frequency? Might be worth an experiement with a ferrite bead or two on the base. - In my experience, when Tek selected devices from a hi-frequency part, the selection parameter was generally related to an HF/switching characteristic (ft, Ccb etc) John |
|
Hi Zen,
toggle quoted message
Show quoted text
I've seen the MPSH05 mentioned online elsewhere as the part number, but I am dubious. Do you have a Tek reference showing that part number? The 1 Oct. 1986 RPR lists it as something quite different -- SPS8811. The specs of the MPSH05 seem inconsistent with the requirements (e.g., speed) of many circuits using the transistor, so I am guessing that a typo somewhere is responsible for some confusion. That said, component testers are too crude to clear up the situation here. If you have a curve tracer, that's the best way to see what the transistor is doing in all regions of operation. Of course, the reverse region behavior is irrelevant circuit-wise unless there are circuits imposing reverse voltages on the transistor (which there should not be). But as far as understanding why the component testers are getting confused, a curve tracer is the right instrument. Better to look at the raw I-V data, rather than some algorithmic inference from the data. What I predict is that you'll see what I see when I curve-trace microwave transistors: Very low reverse-avalanche voltages. If the component testers ever apply, say, a half-dozen volts or so as they do their search, it is quite probable that they'll break down a microwave transistor in the reverse direction. The ultra-heavily-doped B-E junctions will often show significant conduction with just 3-4 volts in reverse, so an "upside-down" NPN will start to conduct at just a VBE above that. And if the transistor avalanches, then the voltage between collector and emitter could snap to a fraction of the initiating voltage and fool the component tester into thinking that a C-E diode just turned on. As for the DMM6500 result, I've never used that instrument, but I note that the current source in the diode test mode has a compliance spanning at least a 12V range, which can be enough to trigger reverse avalanche breakdown in microwave transistors. If you have a cheapo DMM, try using its diode test function and see if you still see a C-E diode. Most cheap DMMs impose a 2-3V maximum in diode test mode. The MPSH05 shouldn't show this behavior, given the datasheet numbers. -- Cheers Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/18/2023 11:47 AM, Zentronics42@... wrote:
I did a little more digging in that lab today. I have a 475A late serial number on the bench and the 151-0367-00 transistors were not tarnished on the legs. I dug up a data sheet and found that the transistor is a MPSH05 This is speced to have the following: |
|
Tom,
toggle quoted message
Show quoted text
The reference document showing the MPSH05 is Tektronix's own transistor cross reference document. Located right side of the page. However a search of the document also cross references to a A5T3571 AND a SKA6516. So the plot thickens some more. In looking at the A5T3571 The tek xref has this part number specified at 1.2Ghz I was not able to pull a data sheet for a SKA6516 In testing a newer one with out tarnished legs it behaved fine on the curve tracer and I think we have another clue as to the "real part number" given that the curve tracer is well in spec and dialed in I pushed the transistor to mild breakdown. Not to damage but just to when breakdown was starting. The 151-0367-00 that came out of the 457a on the bench currently breaks down right at 25Vdc so that is in line with the A5T3571 data sheet. Which would point to this being a 1.2Ghz part. Zen -----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Tom Lee Sent: Monday, September 18, 2023 10:39 PM To: [email protected] Subject: Re: [TekScopes] Is it possible that the 151-0367-00 transistor story is horribly wrong? Hi Zen, I've seen the MPSH05 mentioned online elsewhere as the part number, but I am dubious. Do you have a Tek reference showing that part number? The 1 Oct. 1986 RPR lists it as something quite different -- SPS8811. The specs of the MPSH05 seem inconsistent with the requirements (e.g., speed) of many circuits using the transistor, so I am guessing that a typo somewhere is responsible for some confusion. That said, component testers are too crude to clear up the situation here. If you have a curve tracer, that's the best way to see what the transistor is doing in all regions of operation. Of course, the reverse region behavior is irrelevant circuit-wise unless there are circuits imposing reverse voltages on the transistor (which there should not be). But as far as understanding why the component testers are getting confused, a curve tracer is the right instrument. Better to look at the raw I-V data, rather than some algorithmic inference from the data. What I predict is that you'll see what I see when I curve-trace microwave transistors: Very low reverse-avalanche voltages. If the component testers ever apply, say, a half-dozen volts or so as they do their search, it is quite probable that they'll break down a microwave transistor in the reverse direction. The ultra-heavily-doped B-E junctions will often show significant conduction with just 3-4 volts in reverse, so an "upside-down" NPN will start to conduct at just a VBE above that. And if the transistor avalanches, then the voltage between collector and emitter could snap to a fraction of the initiating voltage and fool the component tester into thinking that a C-E diode just turned on. As for the DMM6500 result, I've never used that instrument, but I note that the current source in the diode test mode has a compliance spanning at least a 12V range, which can be enough to trigger reverse avalanche breakdown in microwave transistors. If you have a cheapo DMM, try using its diode test function and see if you still see a C-E diode. Most cheap DMMs impose a 2-3V maximum in diode test mode. The MPSH05 shouldn't show this behavior, given the datasheet numbers. -- Cheers Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/18/2023 11:47 AM, Zentronics42@... wrote: I did a little more digging in that lab today. I have a 475A late serial number on the bench and the 151-0367-00 transistors were not tarnished on the legs. I dug up a data sheet and found that the transistor is a MPSH05 This is speced to have the following: |
|
Hi Zen,
toggle quoted message
Show quoted text
Thanks for the data. Looks like there was indeed a typo at Tek back in the day and it didn't get corrected until later (the RPR document I referenced earlier). The cross-reference doc you linked to doesn't have a date that I could find, but I'm guessing it is a lot older than the RPR doc because it still lists a few Shockly (sic) 4-layer diodes without stating that these should not be used for new designs. By the time the RPR doc came out, Shockley diodes hadn't been manufactured for quite a few years. If you get a chance, would you throw one of the (assumed) dead transistors on your curve tracer and see how it looks in both forward and reverse mode? I'm curious to see if it is basically ok in the normal mode, but has a very low breakdown when collector and emitter are swapped. --Cheers, Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/18/2023 9:34 PM, Zentronics42@... wrote:
Tom, |
|
It's a muddle. For 151-0367-00, the 1982 edition of the Common Design Parts Catalog (Semiconductors volume) says Similar to A5T3571. 151-0367-01 is Similar to SKA6516. The 1988 edition says 2N3571 and SKA6516. I assume the latter is a private part number. A couple of manuals that use 151-0367-00 describe the part as "selected from 3571TP" then give the mfr part number as SKA6516. The scan of the Transistors RPR is dated 1986 and says Motorola SPS8811, which is no doubt another private number. A manual using 151-0367-01 says "SKC0536". In the RPR, 151-0367-01 is listed as TK0040 made by Trio-Tech Reliability in Benicia, California. It's used in comparatively few instruments but many of the -00 references say Use 151-0367-01. Make of all this what you will.
This was discussed here before, in 2020; see "151-0367-00 and 151-0402-00 transistors". Dave Wise ________________________________ From: [email protected] <[email protected]> on behalf of Tom Lee <tomlee@...> Sent: Monday, September 18, 2023 9:58 PM To: [email protected] <[email protected]> Subject: Re: [TekScopes] Is it possible that the 151-0367-00 transistor story is horribly wrong? Hi Zen, Thanks for the data. Looks like there was indeed a typo at Tek back in the day and it didn't get corrected until later (the RPR document I referenced earlier). The cross-reference doc you linked to doesn't have a date that I could find, but I'm guessing it is a lot older than the RPR doc because it still lists a few Shockly (sic) 4-layer diodes without stating that these should not be used for new designs. By the time the RPR doc came out, Shockley diodes hadn't been manufactured for quite a few years. If you get a chance, would you throw one of the (assumed) dead transistors on your curve tracer and see how it looks in both forward and reverse mode? I'm curious to see if it is basically ok in the normal mode, but has a very low breakdown when collector and emitter are swapped. --Cheers, Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/18/2023 9:34 PM, Zentronics42@... wrote: Tom, |
|
Tom,
toggle quoted message
Show quoted text
Your thoughts on the behavior are correct. The transistor does function "normally" when driven correctly C->E however the breakdown is quite low and agrees with the Keithley 6500 0.75Vdc in the E->C The transistors will also break down with no base drive at about the 0.7 Vdc. I am mildly familiar with avalanche circuits having not needed to work with then out side of 1 or 2 lab circuits, if I am out in left field let me know. But my understanding is that you would avalanche a transistor from C->E to keep it healthy and not shorted. I have not worked with many Ghz + circuits so I am a bit out of my depth here. But isent any E->C current leakage considered a fault. I don’t know of any situations where I would want to avalanche a transistor E->C, but I might have not seen any yet. This is where the plot thickens in terms of 4xx series scopes even more. In a 475 there are 15 of these of which 13 are on the trigger board in the main triggering path. The logic side of the triggering board gets fed from the +15,-15,-8 and +5 power rails. So depending on exactly where the supplies are fed and how I could see some strange things happening. The other 2 locations where this transistor is used there is one in the Z axis amp as well as one in the Horizonal drive section if I remember correctly. Do not have many concerns about these to as they don’t need to deal with + and - wave forms but the triggering sections is a bit weird. Zen -----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Tom Lee Sent: Tuesday, September 19, 2023 12:59 AM To: [email protected] Subject: Re: [TekScopes] Is it possible that the 151-0367-00 transistor story is horribly wrong? Hi Zen, Thanks for the data. Looks like there was indeed a typo at Tek back in the day and it didn't get corrected until later (the RPR document I referenced earlier). The cross-reference doc you linked to doesn't have a date that I could find, but I'm guessing it is a lot older than the RPR doc because it still lists a few Shockly (sic) 4-layer diodes without stating that these should not be used for new designs. By the time the RPR doc came out, Shockley diodes hadn't been manufactured for quite a few years. If you get a chance, would you throw one of the (assumed) dead transistors on your curve tracer and see how it looks in both forward and reverse mode? I'm curious to see if it is basically ok in the normal mode, but has a very low breakdown when collector and emitter are swapped. --Cheers, Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 420 Via Palou Mall Stanford University Stanford, CA 94305-4070 On 9/18/2023 9:34 PM, Zentronics42@... wrote: Tom, |
|
Hello Zen,
last night i got a little more time and i tried this tip that someone wrote below, to put a ferrite ring on the base of the transistor when testing. I use a Chinese component tester, the same as in your video. There were no consecutive results on the basis of which anything could be concluded. But when I use the external crocodile clips that I mounted on that tester, and when I connect E-C with my fingers, the phantom leakage diode disappears. This has worked for me enough times that it could be said to be a general behavior. I don't have enough data, but what I notice is that all the transistors I have, and they are specified for high frequencies, behave like this. I opened another 475 on which I started work, this time I was particularly interested in these transistors. I noticed that they have different colored dots on the back, in some places in the oscilloscope where they are next to each other. I checked them and interestingly, in three places I noticed that the HFE is identical to the color code, that is, it matches exactly the same hfe number. I don't believe it's a coincidence, this really had to be chosen that way during production. Most often, the yellow/purple combination of markings prevails, there are some that also have three dots, whatever they mean. greeting |
Michael A. Terrell
John Sykes
to navigate to use esc to dismiss