Incorrect labeled 1N4148 diode

[mawyatt]

While making some diode capacitance measurements we've uncovered an improperly labeled 1N4148 diode. This diode is showing over 9pF zero bias capacitance which is well over the max value of 4pF per data sheet. Here's a couple images showing the improperly label diode and a good 1N4148.

These diodes were purchased from LCSC, and the improper labeled diode has a LCSC identification of  32-B06-C, and they have been notified.

Best,

[strawberry]

SMD M4 1N4004

[TimFox]

In my presbyopic old age, I have great difficulty reading the part number on my legitimate 1N4148 diodes in glass DO-35 packages.
Can you read any numbers on your apparently misrepresented diodes?

[mawyatt]

It reads only "T4" and is in a SOD-123 package. These were purchased from LCSC and not eBay or AliExpress.

Edit: Eyes are also bad and we use component camera/display or headband magnifiers.

Best,

[strawberry]

reverse recovery could tell what type diode it is

[mawyatt]

Well capacitance has already told that it's not a standard 1N4148. These are heading for the junk pile along with other unnamed components.

Thank goodness these weren't used in our latest builds which do use a few 1N4148

Best,

[tooki]

SMD M4 1N4004

How is this a useful reply?

[MK14]

These diodes were purchased from LCSC

I hope I don't come across as nit picking, but I would like to point out and ask, a number of things.

LCSC are the seller/distributor, just like if you had said they were from Digi-key.  I'd be more interested in knowing (as well), who the manufacturer of the diodes, is.  Because like Digi-key, LCSC supply from a huge number of different manufacture's, perhaps tens of thousands or hundreds of thousands, I'm not sure, exactly.
I.e. If I said I bought a transistor from Digi-key, and it was different to the datasheet.  It wouldn't be too helpful, as they sell so many different types and manufacture's of them.

Hopefully rarely, but it is possible, that the wrong parts were put in the plastic pack.  If I recall correctly, there have been a number of threads on here, from a number of different suppliers, when stuff like that (or similar), has occurred.

Has your test instrument and configuration (which I'm not familiar with, as such), heated up the junction temperature, significantly above room temperature (around 25 deg C).  Which might (only might), affect the readings, relative to the datasheet.  Some parameters, are hugely varied by temperature, such as leakage currents, in some cases.

Was this bad capacitance reading, just on one of the components, or on many of them ?

If it was a Chinese make.  Was it important to check if they were a recommended Chinese make or not ?

I'd prefer to see/know the actual LCSC part number, possibly beginning with the letter C, like C####### something, where #### are often numbers.  I looked at your pictures, but someone seems to have scribbled large black marker pen over it, so I can't make it out.

[thm_w]

If it says T4, then they didn't pick the wrong part.

Datasheet states "5pF typical", no maximum. So OP you got exactly what you paid for.
https://lcsc.com/product-detail/Switching-Diode_Shandong-Jingdao-Microelectronics-1N4148W_C115103.html

If you want a part with a maximum rating, then you can find other manufacturers that are 2pF max (on LCSC or elsewhere):
https://www.smc-diodes.com/propdf/1N4148W%20N0571%20REV.E.pdf
https://datasheet.lcsc.com/lcsc/1809291522_Guangdong-Hottech-1N4148WS_C181133.pdf

[Someone]

If it says T4, then they didn't pick the wrong part.

Datasheet states "5pF typical", no maximum. So OP you got exactly what you paid for.
https://lcsc.com/product-detail/Switching-Diode_Shandong-Jingdao-Microelectronics-1N4148W_C115103.html

That even shows a typical capacitance vs voltage which heads off past 10pf as the bias drops below 0.1V, which seems consistent with the questionable measurement:

This diode is showing over 9pF zero bias capacitance

Brand substitution always risky!

[moffy]

Perhaps you could use them as tuning diodes?

[mawyatt]

If it says T4, then they didn't pick the wrong part.

Datasheet states "5pF typical", no maximum. So OP you got exactly what you paid for.
https://lcsc.com/product-detail/Switching-Diode_Shandong-Jingdao-Microelectronics-1N4148W_C115103.html

If you want a part with a maximum rating, then you can find other manufacturers that are 2pF max (on LCSC or elsewhere):
https://www.smc-diodes.com/propdf/1N4148W%20N0571%20REV.E.pdf
https://datasheet.lcsc.com/lcsc/1809291522_Guangdong-Hottech-1N4148WS_C181133.pdf

Well when we searched for an 1N4148 diode per LCSC, and it lists various 1N4148 diodes and we specify a SOD-123 package then we expect the die in the package to be a 1N4148 die. We have at least 4 different 1N4148s types in various packages included leaded. This is the only device that shows a 9pF zero bias capacitance, all the others read less than 1pF. All this indicates that this is NOT a 1N4148 die but some substitute die.

The 1N4148 is a registered JEDEC part number and Vishay shows a maximum C of 4pf, so does Fairchild, so does Microsemi, so does Diodes Inc, so does Taiwan Semi, so does Multicomp and so on, all less than 4pF MAX.

The device MPN on the bag is 1N4148W, Wiki lists the 1N4148W as a 1N4148 die in a SOD-123 package, thus the "W" designation. This part is not a 1N4148 die in a SOD-123 package, but some other die!! Check out every single reference in Wiki, not a single one shows a maximum capacitance above 4pf, some have a 2pF maximum.

https://en.wikipedia.org/wiki/1N4148_signal_diode

Might expect this from eBay or AliExpress, but not LCSC, nor Digikey, Mouser, or Newark. Sure they are just distributers and just "middlemen", and have no way of knowing the quality of the components they offer, but they do need to know about this kind of mislabeling and why we alerted them to this issue.

If we ordered a 2N3904 in a SOT-23 package instead of the normal TO92, then why would anyone expect anything other than a 2N3904 die, putting another NPN die in a SOT-23 package and listing it as a 2N3904 isn't acceptable, neither is a JEDEC registered 1N4148 in a "W" package using another die.

Hopefully this will prevent someone from using these diodes in a mass produced product and having to deal with potential resulting issues caused by the excessive capacitance.

Best,

[MK14]

Your supplied source, seems to present a partially different view point.
https://en.wikipedia.org/wiki/1N4148_signal_diode

however, today most manufacturers list common specifications.

Because most, doesn't mean all.  So, maybe checking the ACTUAL datasheets specification, would be a good/better idea, rather than assuming everything is completely standardized, for 1N4148's.

[TimFox]

I believe that the JEDEC-registered 1N4148 is in a DO-35 leaded package.  Other packages are 1N4148 “equivalent”.

[mawyatt]

Your supplied source, seems to present a partially different view point.
https://en.wikipedia.org/wiki/1N4148_signal_diode

however, today most manufacturers list common specifications.

Because most, doesn't mean all.  So, maybe checking the ACTUAL datasheets specification, would be a good/better idea, rather than assuming everything is completely standardized, for 1N4148's.

Different view, don't think so!!

Here's a direct quote from Wiki.

Packages[edit]
The JEDEC registered part numbers 1N914 and 1N4148 were originally only available in an axial package, but over time similar parts became available in surface-mount packages too.

Through-hole package
1N4148 in DO-35 glass axial package.[10][11][9][12][13]
Surface-mount packages
LL4148 in MiniMELF package.[12][14][15]
1N4148W in SOD-123 package.[16][17][18]
1N4148WS in SOD-323 package.[19][20][21]
1N4148WT in SOD-523 package.[22][23]
Note: Some 1N4148 family surface-mount packages are marked with "T4" text.[16][17]


Directly from the above Wiki references for 1N4148W:

Ref [16]: Diodes
Low Capacitance: Maximum of 2pF
Ref [17]: MCC
Junction Capacitance Max
CJ VR = 0V, f = 1MHz 2pF
Ref [18]: Vishay
Diode capacitance Max
VF =VR =0V CD 4pF


Best,

[mawyatt]

I hope I don't come across as nit picking, but I would like to point out and ask, a number of things.

LCSC are the seller/distributor, just like if you had said they were from Digi-key.  I'd be more interested in knowing (as well), who the manufacturer of the diodes, is.  Because like Digi-key, LCSC supply from a huge number of different manufacture's, perhaps tens of thousands or hundreds of thousands, I'm not sure, exactly.
I.e. If I said I bought a transistor from Digi-key, and it was different to the datasheet.  It wouldn't be too helpful, as they sell so many different types and manufacture's of them.

Doubt we'll ever find out who the actual die manufacture is, semiconductors are often rebranded, and likely the wafers were processed by one and packaged by another, and even rebranded after packaging.

Hopefully rarely, but it is possible, that the wrong parts were put in the plastic pack.  If I recall correctly, there have been a number of threads on here, from a number of different suppliers, when stuff like that (or similar), has occurred.

Has your test instrument and configuration (which I'm not familiar with, as such), heated up the junction temperature, significantly above room temperature (around 25 deg C).  Which might (only might), affect the readings, relative to the datasheet.  Some parameters, are hugely varied by temperature, such as leakage currents, in some cases.

The test waveform voltage was very low (0.1vrms @ 1MHz) and should not increase the junction temperature.

Was this bad capacitance reading, just on one of the components, or on many of them ?

We checked a couple diodes, they were the same high capacitance.

If it was a Chinese make.  Was it important to check if they were a recommended Chinese make or not ?

I'd prefer to see/know the actual LCSC part number, possibly beginning with the letter C, like C####### something, where #### are often numbers.  I looked at your pictures, but someone seems to have scribbled large black marker pen over it, so I can't make it out.

These were purchased a long time ago, and just used as general stock, not for any specific project or product, so no research into the brand or manufacture was conducted at the time. The marking are from a long time ago when the item was checked against the order and used to identify the diode type and package. We wouldn't have discovered this without doing some capacitance measurements on various zener diodes and decided to measure a couple regular diodes. Actually quite surprised at the zener diode measured capacitance, had thought this was much less than what we are measuring.

Best,

[Someone]

Your supplied source, seems to present a partially different view point.
https://en.wikipedia.org/wiki/1N4148_signal_diode

however, today most manufacturers list common specifications.

Because most, doesn't mean all.  So, maybe checking the ACTUAL datasheets specification, would be a good/better idea, rather than assuming everything is completely standardized, for 1N4148's.

Directly from the above Wiki references for 1N4148W:

Ref [16]: Diodes, Low Capacitance: Maximum of 2pF
Ref [17]: MCC, Junction Capacitance Max, CJ VR = 0V, f = 1MHz 2pF
Ref [18]: Vishay, Diode capacitance Max, VF =VR =0V CD 4pF

That some/many/most 1n4148 parts have a maximum capacitance specification of 4pf (or whatever) does not make that the standard. You can easily find the opposite, manufacturers happily advertising 1N4148 with no maximum capacitance specification:
ROHM: 1N4148 4pf (non-specific)
STAR SEA: 1N4148 4pf typical
Jingdao Microelectronics: 1N4148W 5pf typical

So what does EIA 1N4148-1N4154 (the JEDEC registration) actually contain? I'm not going to pay just to see it.

[MK14]

So what does EIA 1N4148-1N4154 (the JEDEC registration) actually contain? I'm not going to pay just to see it.

I thought the idea is that the JEDEC standards are open and free.

https://www.jedec.org/standards-documents   with 1n4148 or 1n4148w, seems to find no results.

[MK14]

Different view, don't think so!!

Thanks for all the detailed explanations and answers.   
I now, much better understand, where you seem to be coming from, with your point of view.
I largely agree with what you have been saying, now.

Which seems to be, the 1N4148W, is a standardized part, so should adhere to the specifications of JEDEC.

[gamalot]

I don't have a high frequency LCR and can only test at 100KHz. I tested the 1N4148WS I bought on Taobao (sister site of Aliexpress) which is in the SOD323 package and has 'T4' mark on it and result was measured to be about 0.63pF.

For reference I tested another switching diode, the BAV21W, which is in SOD123 package with 'A82' on it, the typical junction capacitance given in the datasheet is 5pF, and I measured it to be about 1.8pF.

---

I did another test, even the 1N4148 in DO-35 glass axial package has a junction capacitance of less than 1pF.

[thinkfat]

I guess the take home for this is, if your design relies on a specific property of a part, check the actual datasheet.

[NETeagle]

Hi,

If you read the datasheet, the capacitance of this 1N4148W :
https://lcsc.com/product-detail/Switching-Diode_Shandong-Jingdao-Microelectronics-1N4148W_C115103.html

The capacitance may varie from 3 to more than 10pF for a Vr of 0.1V to 80V.

[magic]

That some/many/most 1n4148 parts have a maximum capacitance specification of 4pf (or whatever) does not make that the standard. You can easily find the opposite, manufacturers happily advertising 1N4148 with no maximum capacitance specification:
ROHM: 1N4148 4pf (non-specific)
STAR SEA: 1N4148 4pf typical
Jingdao Microelectronics: 1N4148W 5pf typical

So what does EIA 1N4148-1N4154 (the JEDEC registration) actually contain? I'm not going to pay just to see it.

The standard probably does contain 4pF max.
ROHM's spec is unclear, but characteristic curves show 1~1.5pF typical, so they likely meant it as a maximum spec.

As for WanHungLo Electronics, obviously you expect them to deliver the cheapest junk which will still work for 90% of customers.
1N4148 is mostly used as a generic low power diode so it's likely that less than 10% customers care if it's 1pF or 10pF.

[free_electron]

Here's a direct quote from Wiki.

seriously ? wikipedia ? for electronic components data ?

We should introduce new labels to put on electronics "designed by internet" , "Arduino inside" and "wikipedia tested"

Use the manufacturers datasheet. And buy diodes from a known source with a known datasheet.
You can't simply go manufacturer xyz does not meet datasheet abc ...

As for the JEDEC thing. what standard are you referring to ? JESD282 ?
A lot of manufacturers only refer to JD95 which is a packaging standard. it says nothing about the electrical properties.

JESD282 only classifies
- forward current (not even surge current)
- reverse voltage
- storage and operating temperature
- outline

if a part number has a letter suffix ( like 1N4148A instead of 1N4148) it means that also some additional parameters are specced. Forward voltage, Reverse current and recovery time. . some other too. Can't remember. look it up.
that's it. capacitance is not one of them if i remember right.

And even with things like peak voltage you have to be careful. that is all given at 60hz half sinewave for a fixed number of cycles and a cooldown period.

In reality the only data you can expect to be fixed for a "type" is reverse voltage and forward current. the rest ... pinch of salt.

[mawyatt]

I don't have a high frequency LCR and can only test at 100KHz. I tested the 1N4148WS I bought on Taobao (sister site of Aliexpress) which is in the SOD323 package and has 'T4' mark on it and result was measured to be about 0.63pF.
I did another test, even the 1N4148 in DO-35 glass axial package has a junction capacitance of less than 1pF.

Similar result to what we've measured, thanks for taking the time to verify this.

BTW interesting LCR meter, does it support DC bias?

Best,

[gamalot]

I don't have a high frequency LCR and can only test at 100KHz. I tested the 1N4148WS I bought on Taobao (sister site of Aliexpress) which is in the SOD323 package and has 'T4' mark on it and result was measured to be about 0.63pF.
I did another test, even the 1N4148 in DO-35 glass axial package has a junction capacitance of less than 1pF.

Similar result to what we've measured, thanks for taking the time to verify this.

BTW interesting LCR meter, does it support DC bias?

Best,

Its built-in bias voltage is limited to 2.5V, luckily the bid I won included an external ±42V ±40V bias adapter (ZM2328).

[MK14]

Well capacitance has already told that it's not a standard 1N4148.

What/which standard exactly, is it disagreeing with ?

Ideally, be very specific, and include links (free) to it.  A Wiki, although very useful and nice.  Is NOT really a reliable source (as already mentioned, by another poster), when there are disagreements like this.

I will counter, with a counter example, against what you have been saying.  I hope you respect Vishay Semiconductors, as a counter source. Here is a copy, of one of their datasheets, for a 1N4148 diode:

https://www.farnell.com/datasheets/2613993.pdf

I can't find any mention of it being conformant to a JEDEC 1N4148 standard.  But what it does say, is as follows:

Electrically equivalent diodes:
1N4148 - 1N914

There are all sorts of standards, that have been created over the years, decades and even centuries.  But, just because such a standard exists, doesn't mean that it is necessarily used/recognized in modern times.  Some standards, seem to have just disappeared from living memory.

E.g. There was an original RS232 standard, complete with standardized 25 pin connector (which may NOT have been aprt of the standard).  Over the years, some of the pins got dropped, through lack of use (as modems eventually mostly disappeared from common use), its voltage requirements became more relaxed (lower), and 9 pin varieties, became the 'standard'.

The so called "2N3055" 'standard' Power Transistor, differs for partly complicated reasons, from the now ancient (possibly 1968), original version/specifications.  Due to manufacturing practicalities.

[mawyatt]

That very Vishay reference shown below in a screen capture clearly states diode capacitance 4pf MAX! This is what just about every reference to a 1N4148 shows, regardless of package, some of the surface mount types even have a maximum capacitance of 2pF.

If you have another source for a reference other than Wiki, please provide.

Remember this 1N4148 is referred to as a "Small Signal Fast Switching Diode", not a "General Purpose Rectifier" like the 1N4001, and thus junction capacitance IS important in it's intended usage, although it has become the general purpose small signal diode of use by many, myself included.

Anyway the 1N4148 is like a diode analogy of a M3 bolt, if I go to any hardware store and buy (or order from anywhere) a M3 bolt I expect the threads to be the correct diameter and pitch, the heads and length may be different for different types, but the threads better be correct. That's what I expect and same goes for the universal 1N4148, and 9pF capacitance isn't a 1N4148 in my book, of course YMMV.

Best,

[free_electron]

Op is most likely referring to JESD282.
That defines some key parameters under a part number. capacitance is not part of the key parameters.
Reverse voltage , forward current and thermal range are standard. sub classification can add some other parameters. The standard specifies all testing conditions for each of the parameters.

The standard is publicly available and free. Register at jedec.org ( you can even use a gmail address ) and you can grab anything.

[tunk]

Anyway the 1N4148 is like a diode analogy of a M3 bolt ...

Maybe not the best analogy, it doesn't specify the bolt material.

[MK14]

That very Vishay reference shown below in a screen capture clearly states diode capacitance 4pf MAX! This is what just about every reference to a 1N4148 shows, regardless of package, some of the surface mount types even have a maximum capacitance of 2pF.

If you have another source for a reference other than Wiki, please provide.

Remember this 1N4148 is referred to as a "Small Signal Fast Switching Diode", not a "General Purpose Rectifier" like the 1N4001, and thus junction capacitance IS important in it's intended usage, although it has become the general purpose small signal diode of use by many, myself included.

Anyway the 1N4148 is like a diode analogy of a M3 bolt, if I go to any hardware store and buy (or order from anywhere) a M3 bolt I expect the threads to be the correct diameter and pitch, the heads and length may be different for different types, but the threads better be correct. That's what I expect and same goes for the universal 1N4148, and 9pF capacitance isn't a 1N4148 in my book, of course YMMV.

Best,

I agree with you.  I've looked further into the (what I suspect the device datasheet of what you actually bought, was), and it breaks too many of the 'standard' 1N4148 diodes characteristics.  Even if, those characteristics, are NOT part of the JEDEC speficiation (mentioned in other posts, in this thread, i.e. here:  https://www.eevblog.com/forum/projects/incorrect-labeled-1n4148-diode/msg4168855/#msg4168855  ).

You have convinced me, because of what you said about it being a "Small Signal Fast Switching Diode".

The M3 bolt analogy, has issues.  Because, some aspects of an M3 Bolt, may or may not be, part of its 'official' specification.
E.g. Precise weight, specific materials used in its construction, bolt/nut full tighten, release, life expectancy, precise colour, etc.  May NOT be part of the specification.

The situation is still complicated, as the devices that you bought, (presumably) meet its available datasheet specification.  But I agree, they are not standard 1N4148 diodes, even if no direct link to such a document, has yet been posted (but there is a link to where you can get the documents, via free email registration, so I can register later, and look for myself).

The inferior 1N4148's manufacturer, seem to have decided to make the 4pF maximum, a 5pF typical value, measured at 1.5V rather than 0V.  What is also surprising, is that the same manufacturer, seems to sell plenty of very fast, low capacitance diodes, although they seem to be more expensive than the 1N4148's.  So, I don't really understand why they didn't stick to the 1N4148's (somewhat or fully) standard definition.

Maybe there is some technical explanation, such as they are really some other part numbered diodes.  Which is, they fail the test(s) for what they are suppose to be e.g. higher voltage, low leakage etc.  Are then reclassified/sold as 1N4148's, if possible, to ensure they can sell the highest possible number of salvageable (working) dies.

[MK14]

Op is most likely referring to JESD282.
That defines some key parameters under a part number. capacitance is not part of the key parameters.
Reverse voltage , forward current and thermal range are standard. sub classification can add some other parameters. The standard specifies all testing conditions for each of the parameters.

The standard is publicly available and free. Register at jedec.org ( you can even use a gmail address ) and you can grab anything.

Thanks, it is not your fault.  But the completely crazy website, keeps on refusing me permission to register, because I can't fathom out, their ridiculous captcha nonsense thing.  So, I've given up, after perhaps 10 failed attempts.

The answer you entered for the CAPTCHA was not correct.

I don't get on well with CAPTCHA systems.  Basically, see below:

[mawyatt]

You have convinced me, because of what you said about it being a "Small Signal Fast Switching Diode".

Actual I didn't create that, this is directly from Wiki's description, Vishay's description, Diodes description, Microsemi and others.

The M3 bolt analogy, has issues.  Because, some aspects of an M3 Bolt, may or may not be, part of its 'official' specification.
E.g. Precise weight, specific materials used in its construction, bolt/nut full tighten, release, life expectancy, precise colour, etc.  May NOT be part of the specification.

Probably not the best analogy, but I would expect the threads to match.

Best,

[free_electron]

Probably not the best analogy, but I would expect the threads to match.

They do match, but you are nitpicking about one bolt having unified thread, the other sharp-V. They are both the same pitch , depth and slope , just that one is a bit rounded at the top. They will both fit the same nut. So the chinese versions have a bit of rounding because they use a more relaxed quality control..
If the junction capacitance is so important : buy from a known good source. For 99% of the people this is not a critical factor.
You expect the best quality manufactured under the tightest of QA , then go off and buy chinesium parts from the cheapest vendor out there .

[TimFox]

One of my pet peeves is selling "1N" or "2N" devices that have different packages from the original JEDEC registration, without changing the prefix.
For example, plastic or SMD re-package of a 2N2222, which is a TO-18 metal package through-hole device.  Of course, there are legitimate "PN2222" and similar part numbers for plastic-package equivalent devices.

On the other "thread", note that the modern American "Unified National" thread form is defined identically to the modern ISO metric thread form in terms of geometry.  https://www.boltscience.com/pages/screw6.htm

[MK14]

My understanding, is that there are two different ways, a 1N4148 could be designed into a circuit.  I'm showing two extremes.

One is, that it is basically what some/Dave(EEVblog) likes to call Jelly Bean parts, such as the 1N4148.  E.g. A 1N4148, being pulled to the 5V rail, via a 1K resistor, with almost 5mA flowing through it.  I.e. The circuit is nowhere near any datasheet limits, and a simple jelly bean part, fits the bill just fine.  That is where you could use your cheapest, worst manufacturer, riskiest supplier(s) for.  Depending on how you value the quality of your final product(s).

Alternatively, it is a critical part of a circuit, really accurately, dependably and reliably needs to respond within 2.000 nanoseconds, for every pulse.  In a circuit which really needs the reverse recovery time, diode capacitance and other circuit parameters, to be within certain specified limits.  In which case, I'd suggest considering a specific, very high speed diode, characterized for such uses, with suitable design margin, for best quality/dependability and reliability.
Which, if it is decided to stick with a 1N4148, should then be a specific manufacturers's part (number), with a suitably good quality pedigree, behind that manufacturer.

It is exciting, using powerful test equipment, to measure tricky things.  Either because the test equipment originally cost $20,000, 15 years ago, and it allows amazingly accurate and detailed tests to be performed.  Or to design your own test equipment, that measures complicated parameters, with your own, home brew modifications.

But, to use either of the above equipment, to measure, what probably arguably is a generic jelly bean part, in some fairly rare, partly obscure way.  Then complain, because (what seems to be), the cheapest possible supplier, from the cheapest possible component of that type they sell, then disregarding their claimed datasheet values.  Then making a panic, panic, panic, a jelly bean part differs from a specification, because they measured it at 1MHz/1.5V, and gave a typical value (on the datasheet), of 4 or 5pF, rather than at 1MHz/0V, and had a maximum value of 4pF, is perhaps a little bit of an over-reaction.  Sorry, if this causes offense.

As I see it, the critical part,  is that it DID agree with their published datasheet, if I've followed this thread, correctly.  Therefore is arguably, at least a little bit unfair, to create a complaint thread, without clearly stating that fact, right from the start.

My basic understanding, is that jelly bean parts, are NOT 100% bolted down, in overall specifications, across ALL manufacturers and available parts.  They are what was easily and very cheaply being made available, at that time.  E.g. 741 op-amp or 555 timer.  Yes, they do have somewhat basic specifications, bolted down.  But if you get into the fine detail and/or need certain parts of its specification, to be EXACTLY what you desire.  You need to choose an exact/specific 741 (partly obsolete these days) or 555 timer, from a specific manufacturer.  Then carefully go over their datasheet and perhaps discuss the situation with the supplying company, if your purchase volumes and/or contacts in the industry, permit.

TL;DR
No datasheet violation = Arguably no big problem, as the (specific/actual) datasheet should really be consulted, if certain parts of the component specification, really matter, in the actual use case.
There are standards, but my understanding is, that those should NOT be relied upon, too much.  I.e. Datasheets should be used and/or consulting with the actual supplier(s) and their application engineers.

[thm_w]

That very Vishay reference shown below in a screen capture clearly states diode capacitance 4pf MAX! This is what just about every reference to a 1N4148 shows, regardless of package, some of the surface mount types even have a maximum capacitance of 2pF.

So its not really standardized among the major manufacturers. Some are 1.5, 2, 4pF max.
I get the frustration, but this is not so different from buying a 2pF 1N4148 and then getting pissed when you buy another and a 4pF diode shows up.

edit, I did manage to find one on digikey, diotec semi 4pF typical https://diotec.com/request/datasheet/1n4148.pdf

[mawyatt]

Alternatively, it is a critical part of a circuit, really accurately, dependably and reliably needs to respond within 2.000 nanoseconds, for every pulse.  In a circuit which really needs the reverse recovery time, diode capacitance and other circuit parameters, to be within certain specified limits.  In which case, I'd suggest considering a specific, very high speed diode, characterized for such uses, with suitable design margin, for best quality/dependability and reliability.
Which, if it is decided to stick with a 1N4148, should then be a specific manufacturers's part (number), with a suitably good quality pedigree, behind that manufacturer.

If a standard 1N4148 diode (read less than 4pF and 4ns) does the job then no need for a specific very high speed diode, here's exactly where the 9~10pF and 8ns "jelly bean diode" might cause a problem tho!!

It is exciting, using powerful test equipment, to measure tricky things.  Either because the test equipment originally cost $20,000, 15 years ago, and it allows amazingly accurate and detailed tests to be performed.  Or to design your own test equipment, that measures complicated parameters, with your own, home brew modifications.

Nothing tricky about measuring diode capacitance, nor requiring $20K equipment, it's relatively simple and straight forward if you understand the solid state characteristics of PN junctions and the synchronous techniques to make such measurements

But, to use either of the above equipment, to measure, what probably arguably is a generic jelly bean part, in some fairly rare, partly obscure way.

As stated and referenced earlier the 1N4148 is a "Small Signal Fast Switching Diode", note the "Fast Switching" terminology, which directly implies things such as recovery time and capacitance are important, certainly not "fairly rare" nor "obscure" as you stated, since after all it is a Fast Switching Diode not a general purpose rectifier

Then complain, because (what seems to be), the cheapest possible supplier, from the cheapest possible component of that type they sell, then disregarding their claimed datasheet values.  Then making a panic, panic, panic, a jelly bean part differs from a specification, because they measured it at 1MHz/1.5V, and gave a typical value (on the datasheet), of 4 or 5pF, rather than at 1MHz/0V, and had a maximum value of 4pF, is perhaps a little bit of an over-reaction.  Sorry, if this causes offense.

Almost all 1N4148 sources show the maximum reverse recovery time a 4ns (Vishay, Semitech, Diodes, Rohm, On Semi, MCC, and others), Wiki lists this directly in the 1N4148 specifications, yet the diode in question, the 1N4148W from Jingdao Micro shows a 8ns maximum recovery time and also shows a graph of capacitance of over 10pF at 0.1V bias, while others sources mentioned show an absolute maximum C of 4pF at zero volts and 4ns maximum recovery!!

Anyway, there is no panic, nor overreaction from knowledgeable folks, they know what a 1N4148 diode should be, and a 9~10pF with a 8ns recovery time diode is NOT a 1N4148 in my book, and frankly that's all that matters here at the Labs!!

Best,

[Someone]

If a standard 1N4148 diode (read less than 4pF and 4ns)

Yet you cannot point us to this "standard".

Please do share with us the source (beyond your personal belief) of what 1N4148 actually specifies.

[mawyatt]

If a standard 1N4148 diode (read less than 4pF and 4ns)

Yet you cannot point us to this "standard".

Please do share with us the source (beyond your personal belief) of what 1N4148 actually specifies.

What we consider a loose "standard" and you may differ, but if the vast majority of all sources show a 1N4148 with less than 4pF and under 4ns reverse recovery and only one (maybe a few others) shows 9~10pF and 8ns, I'll side the majority

Of course this isn't a "proper" standard, but does have some backing as a tried and true 1N4148 dating back 50 years ago and we've used them since then

Even a "Jelly Bean Diode" using JEDEC nomenclature 1N4148 shouldn't justify typical capacitance spread of over 14 to 1. I don't believe the usual "Jelly Bean Parts", such as 1N4004, 1N5817, 2N3904, 2N2222, 2N3906, and so on show this kind of important parameter variation, if you know of some please enlighten us!!

BTW the 1N4148 cap ratio of over 14 to 1 points to a completely different 1N4148 core die design. This is not a process variation, but a completely different die design, probably with a completely different diffusion profile, and hints that maybe a substitute die is being used!!

Best,

[mawyatt]

Think we've wasted enough of our and others time and BW on this subject. If you chose to view the 1N4148 as a generic jelly bean diode without much regard to capacitance and reverse recovery parameters, go ahead and do so

We chose to call this particular Jingdao Micro 1N4148W a rogue 1N4148 and will sent them to the junk pile or trash can!!

Anyway, our original intent was to alert folks to this excessive capacitance in the Jingdao labeled 1N4148W diode, not create a pi&&!ng contest 

BTW if anyone wants these Jingdao 1N4148W diodes, just send a SASE and we will happily provide them!!

Best

[Someone]

BTW the 1N4148 cap ratio of over 14 to 1 points to a completely different 1N4148 core die design. This is not a process variation, but a completely different die design, probably with a completely different diffusion profile, and hints that maybe a substitute die is being used!!

Different manufacturer puts different thing inside "equivalent part", only you seem shocked/outraged by this. There is not a designated 1N4148 design/die/factory/process, its just a designation for some common set of specifications, which are in a standard, but you want to "own" instead.

Stop using "standard" for your position and this false authority, there is already an authority which publishes the standard you are ignoring and refusing to acknowledge.

Feel free to sign up and extract the standard if you are so wedded to designing by the designation 1N4148, and do share with us what it actually contains. Another poster says it does not specify capacitance. Likely it doesnt but you continue to promote your ideal that is does/should (without evidence).

Part dies change often enough that designing to off spec parameters (as you are doing) leads to problems. Instead of acknowledging this and accepting you didnt check (ever?) the standards and specifications, you make out like its the manufacturers fault for not achieving your ideal imaginary "standard".

We chose to call this particular Jingdao Micro 1N4148W a rogue 1N4148 and will sent them to the junk pile or trash can!!

And I'll keep pointing out that its your unsubstantiated opinion on that matter, sure seems like a 1N4148.

[free_electron]

for the nth time : Capacitance is NOT part of a jedec classification unless you drill into the suffixes.
a 1N number , without any trailing digits specifies 4 parameters.

For rectifier diodes, the Type Administrator will insist that a new type number, not a suffix letter, be used, if a
new device differs from an older registered device in any of the following respects:
1) Any change in forward current rating, except surge current.
2) Any change in reverse voltage rating.
3) Any change in maximum storage or operating temperature rating.
4) A new rating or characteristic not required of the older device and for which industry agreement for
reregistration of the older device could not be reached.
5) A major change exceeding the greatest change permitted for suffix designations in one or more of the
characteristics listed.
6) New outline dimensions.

Capacitance is not in that list.

There is this clause too:

Commercial data describing that device must identify by asterisks all parameters which appear on the JEDEC registration.

For example this datasheet : https://www.onsemi.com/pdf/datasheet/1n4001-d.pdf explicitly calls out the JEDEC parameters.  Vrrm , Vrsm , Vr , Io , IFSM.

To get a new JEDEC number your part needs to be substantially different from any existing part BEFORE they will create a new classification. so, if a part is close enough to an existing part you don't get a new number.

For parts that have letter suffixes (1N4148A 1N4148R etc)
The letter R means it is a reverse polarity , the letter M is a case deviation. other letters are new types.
The suffix denotes a tightening of an original specification. you can replace a 1N4148 with a 1N4148W , but not the other way around.

To get a letter suffix there are also rules.

Forward Voltage (VFM) 0.8 - 0.9 x VFM (reg)*
Reverse Current (IRM) 0.05 - 0.20 x IRM (reg)*
Surge (Nonrepetitive) Peak Forward Current (IFSM) 1.50 - 3.50 x IFSM (reg)*
Reverse Recovery Time (trr) 0.10 - 0.50 x trr (reg)*

So a 1N4148x can have 1/10 of the original recovery time.

There are also so called "Slash sheets"  1n4148/696 would be equivalent to an automotive qualified part AEC-Q101
The slash sheet defines a number of parameters that are brought under control. The original 1N4148 does not conform to any of those. These are additional "tightening" parameters ( And accoupling testing methods. JEDEC specifies the testing method. you can't just do whatever, that would be meaningless for comparison)

Many manufacturers violate JEDEC by not clearly marking the controlled parameters or by creating part numbers that violate JEDEC nomenclature. 1N4148-DO35 would be such a violation. It needs to be a single letter, a dash and number sequence or slash and number sequence. Anything else is junk


In short, if you would have bought a 1N4148-xyz and it does not meet capacitance specs you could complain. but 1N4148 is not "tight" enough of a specification to include the capacitance. You'd have to check RDF22 to see what nomenclature includes that. unfortunately i can;t help, i don't have access to those anymore. I left the whole semiconductor stuff 8 years ago

[mawyatt]

Different manufacturer puts different thing inside "equivalent part", only you seem shocked/outraged by this. There is not a designated 1N4148 design/die/factory/process, its just a designation for some common set of specifications, which are in a standard, but you want to "own" instead.

No joke, of course it's a different die design, how else could it have 14X capacitance difference!! I'll wager that under noopy's microscope the Jingdao 1N4148 die looks completely different than the other 1N4148 die from Vishay, On Semi, Diodes, and so on. Likely there aren't many semi fab houses producing 1N4148 wafers for obvious $ reasons, and most sources are using these few fabs for wafers and packaging in different formats and also likely exchanging packaged parts. Would not be surprised to find that the Jingdao die is from another part type, not even intended for the 1N4148 diode.
 
Where do you get off thinking I'm shocked for outraged, or what do I "own"?? And how is a 14 to 1 capacitance "equivalent"?? As for specifications, did you happen to notice a parameter called reverse recovery time, it's a specification and listed even in the early 1N4148 predecessor called the 1N914 (see original TI data sheet from 1960). The Jingdao is specified as a max of 8ns whereas everyone else is specified as a max of 4ns. Maybe this is "equivalent" to you, as is the capacitance??

Stop using "standard" for your position and this false authority, there is already an authority which publishes the standard you are ignoring and refusing to acknowledge.

I'll use standard all I want, it's an acceptable level as I stated for what I expect a 1N41418 to be, what authority are you??
Care to show your standard as you mentioned? If it's JEDEC, this requires a fee.

Part dies change often enough that designing to off spec parameters (as you are doing) leads to problems. Instead of acknowledging this and accepting you didnt check (ever?) the standards and specifications, you make out like its the manufacturers fault for not achieving your ideal imaginary "standard".

Never stated we designed anything around this diode. As stated earlier, if you bothered to read, this was purchased long ago as a general stock item with quite a bit of other items from LCSC, likely a couple years back. Only reason we discovered this was while measuring zener diode capacitance and did a few other diodes, we found this excessive capacitance. The Jingdao is a rogue 1N4148 in our book, and that's all that matters here!!

And I'll keep pointing out that its your unsubstantiated opinion on that matter, sure seems like a 1N4148.

Can you justify the 8ns reverse recovery (goes along with the higher capacitance), vs the 1N4148 specification of 4ns (see image)??

The Jingdao might seem like a 1N4148 to you, but not here

BTW note the TI 1N916 (1N4148 predecessor) data sheet from 1960, they use milli-microseconds for nanoseconds and uuF for pF!!


Best,

[mawyatt]

for the nth time : Capacitance is NOT part of a jedec classification unless you drill into the suffixes.
a 1N number , without any trailing digits specifies 4 parameters.
...
Forward Voltage (VFM) 0.8 - 0.9 x VFM (reg)*
Reverse Current (IRM) 0.05 - 0.20 x IRM (reg)*
Surge (Nonrepetitive) Peak Forward Current (IFSM) 1.50 - 3.50 x IFSM (reg)*
Reverse Recovery Time (trr) 0.10 - 0.50 x trr (reg)*

...

So a 1N4148x can have 1/10 of the original recovery time.

However, the reverse recovery time is a specification as you've shown, and the Jingdao 1N4148W has a specification of 8ns maximum, and the max specification from Wiki for a 1N4148 is shown as 4ns which jives with all the other 1N4148 sources mentioned except Jingdao! Since a larger junction capacitance implies a longer recovery time this makes sense.

So again we consider the Jingdao 1N4148 a rogue 1N4148 and the higher reverse recovery time supports this  

Edit: You noted: "The suffix denotes a tightening of an original specification. you can replace a 1N4148 with a 1N4148W , but not the other way around."  This won't apply with the Jingdao 1N4148W because the Jingdao is well outside the 1N4148 reverse recovery specification as shown and thus would fail in many applications.

So the Jingdao 1N4148W should not be considered a 1N4148W as it fails the 1N4148 replacement.

Best,

[MK14]

The Jingdao is a rogue 1N4148 in our book, and that's all that matters here!!

Actually, it's NOT all that matters here!!

Thanks, you've just given me evidence, that it is indeed 1N4148 compatible**.
**Because I'm reasonably happy with the original 1N914's performance, as a high speed diode, instead of a 1N4148.  The datasheet(s) you seem to supply, would seem to show, that the 1N914 was originally specified as 3.5ns TYPICAL reverse recovery time.  Which in my book, can mean around 7 or 8ns, is a suitable estimate as a maximum value, in very rough, back of envelope, non-committed, terms.

TL;DR
I have rather liked the 1N914's (long story).  So, illustrating that this 'rogue' 1N4148 has similar reverse recovery times to it, has made my day!

[mawyatt]

From just above your post, within quotes from free_electron that we posted just before your post.

Edit: You noted: "The suffix denotes a tightening of an original specification. you can replace a 1N4148 with a 1N4148W , but not the other way around."  This won't apply with the Jingdao 1N4148W because the Jingdao is well outside the 1N4148 reverse recovery specification as shown and thus would fail in many applications.

So the Jingdao 1N4148W should not be considered a 1N4148W as it fails the 1N4148 replacement.

The original 1960 data sheet for the 1N914 and 1N916 from TI preceded the introduction of the 1N4148 which was introduced in 1968 according to Wiki. The present 1N4148 specs are listed in Wiki and reverse recovery is listed as 4ns maximum which jives with various sources. Since reverse recovery and junction capacitance are tightly related, with the measured non-Jingdao 1N4148W capacitance that we and others measured, one would expect lower reverse recovery times than 4ns.

We measured a SMD 1N4148W that had a 0.657pF zero bias capacitance, this low capacitance was confirmed by another. With this low a zero bias junction capacitance the reverse recovery should be very quick indeed, maybe under 2ns

BTW, "here" refers to our Labs here, not on this site!! Obviously everyone's opinion matters on this site, although some seem to be questionable

Best,

[free_electron]

Capacitance is NOT part of the RDF for a base JEDEC number. And please stop using wikipedia. nobody designs using Wikipedia "specs"
You also cannot refer to 1N914 datasheets. The reason there is a 1N914 or 1N916 is that they are deemed sufficiently different to merit their own classification. you cannot shove a 1n4148 under a 1n914. They have different numbers again because they are sufficiently different.

The 1N4148W can be used as a 1N4148. neither capacitance nor recovery time are part of the original 1N4148 base number.

The real question would be if the jingdao 1N4148W meet other 1N4148W specs. You need to compare fuji apples to fuji appels. not mix boscop or honeycrisp with fuji. they are all apples. but they have distinctive subclassification.
As for measuring the capacitance : that needs to be done with a test frequency of 1MHz ... i don't think your LCR bridge goes that high. You need a semiconductor parameter analyzer for that.

[mawyatt]

Capacitance is NOT part of the RDF for a base JEDEC number. And please stop using wikipedia. nobody designs using Wikipedia "specs"
You also cannot refer to 1N914 datasheets. The reason there is a 1N914 or 1N916 is that they are deemed sufficiently different to merit their own classification. you cannot shove a 1n4148 under a 1n914. They have different numbers again because they are sufficiently different.

Sorry, but don't have a JEDEC or another spec source, care to provide one? Otherwise I'll keep using Wiki and various known manufactures such as On Semi, Vishay, Diodes, and so on. The 1960 TI reference was to show how far back this type of High Speed Switching Diode dates, not to use it as a reference even tho it does show capacitance curves and indicates Reverse Recovery Times. So this was referred to as the 1N4148 predecessor which showed up in 1968, but not to be compared with.

The 1N4148W can be used as a 1N4148. neither capacitance nor recovery time are part of the original 1N4148 base number.

Interesting, since this is directly from one of your above posts, quoted directly below?  Seems to indicate Reverse Recovery Time as a parameter, along with the usual other diode parameters. Please clarify what part of your posts should we accept and what part should we reject?

The suffix denotes a tightening of an original specification. you can replace a 1N4148 with a 1N4148W , but not the other way around.

To get a letter suffix there are also rules.

Quote
Forward Voltage (VFM) 0.8 - 0.9 x VFM (reg)*
Reverse Current (IRM) 0.05 - 0.20 x IRM (reg)*
Surge (Nonrepetitive) Peak Forward Current (IFSM) 1.50 - 3.50 x IFSM (reg)*
Reverse Recovery Time (trr) 0.10 - 0.50 x trr (reg)*

So a 1N4148x can have 1/10 of the original recovery time.

The real question would be if the jingdao 1N4148W meet other 1N4148W specs. You need to compare fuji apples to fuji appels. not mix boscop or honeycrisp with fuji. they are all apples. but they have distinctive subclassification.

Well we've compared apples to apples and according to other 1N4148W specs (Vishay, Diodes, Semtech, MCC, On Semi, and so on) the answer is the Jingdao 1N4148W does not match or even approach the capacitance and reverse recovery (maybe more parameters, but we didn't check) specs!! Exactly why IMO this should not be considered a 1N4148W.

As for measuring the capacitance : that needs to be done with a test frequency of 1MHz ... i don't think your LCR bridge goes that high. You need a semiconductor parameter analyzer for that.

Yep, that's what the diode data sheets show and all our diode capacitance measurements were at 1MHz, and the LCR meter used was the Hioki IM3536 which supports up to 8MHz. We designed and built a special test fixture structure to support DC Bias measurements for this purpose (details posted on another thread), since the IM3536 only supports internal +-2.5VDC bias. We use either a KS34465A or Keithley DMM6500 to monitor the DC bias across the diode DUT with proper isolation so as not to affect the capacitance measurements.

We've been doing semiconductor chip design and testing for a long long time, from advanced ADCs, to the 1st MW Silicon Single Chip receiver, 1st RF SoC, 1st Ad-Hoc RF Networking Transceiver (MESH), to 2nd most complex chip DARPA had done (CLASS), and quite a few others I can't mention. Used various leading CMOS fabs, IBM SiGe BiCMOS (7,8,9HP), Bell Labs CBICV2, even InP, so we know what we are doing, even in retirement 

Best,

[T3sl4co1l]

What letter suffixes does JEDEC define?

Usually just -A, -B, etc. for I assume incremental or process improvements, possibly die shrink sort of stuff.

1N4148W is an SMT, not at all possibly compatible with a 1N4148 -- I would comfortably assume the -W is a wholly separate part created by manufacturers, in a nod to the original, but in no way dependent upon its spec.

For sure, I mean, there's 2N7000, but not 2N7000W (or whatever suffix one might use for SOT-23), nor MMBT7000; but 2N7002, an entirely different part number, is in the JEDEC system, for an identical(?) electrical spec with different package.

Interestingly, there's a MMBT7002 from a handful of small manufacturers, so it seems.

MMBT I don't know the history of, but for sure it's not JEDEC's normal numbering system.

As for JEDEC electrical parameters -- even when given, they're usually notoriously loose.  Consider 2N2222 fT.  Only min is given.  No constraint is provided, it could be anything.  It probably wouldn't be economical to supply RF dice under this number -- but it wouldn't violate the spec, assuming everything else is met.  (Which, for this particular example, I suppose some kind of transistor with same doping level and abruptness, thus same voltage limits, but finer interdigitation would be needed to give a concerningly high (600MHz+?) fT?  Or, any interdigitation at all, really, given most 2222s just use a blob or star sort of pattern.)

So, if they don't spec on trr or Cjo, yeh, it's technically correct.  Shite spec, but technically correct.


You've done quite a lot of semi work, I'm guessing just not with such bog standard parts, or with standards the likes of JEDEC?

That or most of this shit is so old (1960s, over half a century ago!) it was long settled by the time you, even joined the workforce maybe?  Hah.  (Well maybe not that old comparatively, you're retired now wasn't it?  So most of this was probably concurrent eh, nevermind.)

Tim

[MK14]

Sorry, but don't have a JEDEC or another spec source, care to provide one? Otherwise I'll keep using Wiki and various known manufactures such as On Semi, Vishay, Diodes, and so on.

No, you can't do that, you need to provide links/information/indications of a specific SOURCE, so that people can nod their heads in agreement.  Otherwise, things could/would/might get out of hand.

If you want to chuck your own components/stuff, for ANY reason you decide, into the junk pile, that is entirely up to you.  But, if you want to create a thread on a public forum, suggesting/hinting that others do the same.  You need to justify it, ideally with actual visible/touchable/existing sources of information.  Of suitable quality/reliability/trustworthiness.

As far as I understand things.  It is entirely normal, for the top three component manufacture's, of a specific component, to have datasheets, which in at least one area of concern, are noticeably better than the worst three component manufacture's, of that component type.

Hypothetical (made up) examples, as I don't want to spend ages checking through datasheets.

The best TTL/CMOS/similar device manufactures, may have a specific parameter, buried deep in the datasheet, which says the worst case leakage current is no more than 0.5 microamps.  The cheapest competitors datasheet, may give that parameter, as a looser, 2 microamps.  True, that is worse, but in the vast bulk of applications, that will work out just fine.  If not (i.e. it does matter), you can either use one of the better parts, in that circuit, or redesign it slightly, to have a lower value pull up resistor or similar.

TL;DR
It is NOT up to US, to provide you with the 'claimed' specification sheet(s), which support your claims.  It is up to you.  We are perfectly entitled to reject some or all of your sources, such as Wikipedia and other similar, datasheets.  As they are NOT part of any official standards, as such.

[free_electron]

Reverse Recovery Time

recovery time yes. capacitance ? no. and before you start saying "they are related" , they are not the same.

Well we've compared apples to apples and according to other 1N4148W specs (Vishay, Diodes, Semtech, MCC, On Semi, and so on) the answer is the Jingdao 1N4148W does not match or even approach the capacitance and reverse recovery (maybe more parameters, but we didn't check) specs!! Exactly why IMO this should not be considered a 1N4148W.

Good. i dug a bit deeper because this intrigues me ( i deal with this kind of crap day in and day out , supply chain , pcb libraries )
So it turns out that the governing standard for letter suffixes is RS-370-B which specifically controls diodes.

The letter structure of a JEDEC number

- 1 digit specifying the number of "useful electrical connections" minus 1. a diode has 2 terminals so it becomes 1 (2-1) . a transistor has 3 terminals so becomes 2.
- the letter N or C. N meaning a registered part , the letter C designating the bare die form. so a 1C4148 would be a unpackaged diode die. stick it in a package and you get a 1N4148....
- two or more digits that are a sequence number in the catalog.

if no other suffixes : this is the "root part" for which only 4 parameters are defined. (already discussed earlier) if a registered jedec package is used this is part of the RDF.
so certain types can have a hard package designator called out. others may skip that.

Suffixes - and / are package / regulation classifications.(see earlier posts.)

Letters:

Letters A to K (except I, that is not allowed because of potential confusion)  are revisions of the diode with additional controlled parameters not part of the original RDF (Registered Data Format in JEDEC parlance)
Letter L means leads longer than root package spec
Letter S means leads shorter than root package spec
Letter R means this diode sits reverse in the package. for example a sot23 typically has the cathode at pin 3 , anode at pin 1. the R would reverse the orientation. you will find this more with transistors since it makes single sided layout easier (stereo amplifiers for example: you can mirror layout.)
Letter M meaning microwave diode. they deviate from the above lettering ( this is historical before they started drawing a formal standard , so it gets messy...)
Letter P followed by one or more digits : package variations. Same electrical characteristics just different package. This is also historical. no new parts are allowed with this letter scheme.

the sequence is also specificed : [terminal count][N/C][identifier numbers]<optional variant A..K><optional R reverse ><optional M microwave)
in that order.
so 1N909BRM would be a  1n909 , b variant , Reverse mount , microwave
1N909RBM would be invalid format. (reverse before variant. )

Anything else is not JEDEC but just arbitrary stuff the manufacturers slap on to . so that W means exactly nothing ... where JEDEC is concerned. You'd expect a 1N4148W from abc to be the same as a 1N4148W from xyz , but that is not necessarily the case. And since capacitance is not part of the RDF ...
There are many numbering schemes in circulation that are not JEDEC. everybody seems to slap on whatever and nothing is really controlled.
One clause specifies that JEDEC registered data needs to be marked with an asterisk in the datasheet, and a footnote stating the asterisked parameters are JEDEC registered. I can't find any datasheet where this is marked for the 1n4148w (see the example i gave for the other diode. ). pretty interesting. So all these parts out there are not truly JEDEC 1N4148 . They are close, but no cigar...

JD282 and RS-370-B are the controlling documents. read at your own leisure

[thm_w]

Think we've wasted enough of our and others time and BW on this subject. If you chose to view the 1N4148 as a generic jelly bean diode without much regard to capacitance and reverse recovery parameters, go ahead and do so

I don't see it as time wasted.
I think your capacitance measurements were interesting, and resulting discussion of value.

[mawyatt]

Anything else is not JEDEC but just arbitrary stuff the manufacturers slap on to . so that W means exactly nothing ... where JEDEC is concerned. You'd expect a 1N4148W from abc to be the same as a 1N4148W from xyz , but that is not necessarily the case. And since capacitance is not part of the RDF ...

So here's your exact quote from earlier:

The suffix denotes a tightening of an original specification. you can replace a 1N4148 with a 1N4148W , but not the other way around.

So obviously you are completely contradicting yourself here!!!

In one case you state you can replace a 1N4148 with a 1N4148W but not the reverse, then soon after state that the W means nothing, so how can you claim you can replace a 1N41418 with a 1N4148W when as you state the "W" means nothing 

This is all we need to know about the value your responses, and what your true knowledge base is!!

Best,

[MK14]

This is all we need to know about the value your responses, and what your true knowledge base is!!

You're being ridiculous!

You are turning other people/engineers, into simple on/off, black and white 'boxes'.  That are either 100% correct/right/accurate all the time, or potentially/fully wrong.  People/engineers etc, are NOT as simple as that.

The person who sweeps the floor of someones labs, might say a particular component is WRONG, because it looks a funny colour.  They may have absolutely no knowledge of electronics.  But that DOESN'T mean that what they say is wrong.  Maybe (probably not), that component is WRONG, it could be the case.

Analogy/example, let me poke fun at you:

1N41418

You can't even 'spell' 1N4148, so I'm going to ignore all future responses you make.  As you can't even get 1N4148 correct!.  (Joke, for dramatic effect).

[CatalinaWOW]

This thread identifies yet another level of marketing BS.

Most people are aware that 'typical' values in a data sheet rarely are, and only specified limits to performance are meaningful.

What everyone also needs to be aware of is that with Wikipedia or other source which has no contractual traceability the most you can say is that the quoted values are 'typical'.   

As you have thoroughly demonstrated there is one manufacturer who slaps a 1N4148 label on a part that does not meet the 'typical' specifications.  In this context I am using the OPs statement that most of the specifications on the Wiki are similar. Would the OP be happy if the vender for this part had posted their data sheet on the wiki?  An action that anyone can do.

The fundamental message is what free_electron has stated over and over again.  Do not count on performance that is not traceable through contractually enforceable documents.  Many engineers have learned this lesson the hard way.  Most critically when they have implemented some clever design that depends on some components out of specification performance.  And then the manufacturer "upgrades" their design, or the original vendor goes out of business, or the original vendor learns of the special value of their part and bins them into another higher priced part number, or some market elephant corners the supply of that version of the part and the alternatives just don't work.

Railing that some vendor didn't follow the herd won't help.  They sold their parts, and their market volume is unlikely to be affected by your complaint.

[mawyatt]

This is all we need to know about the value your responses, and what your true knowledge base is!!

You're being ridiculous!

You are turning other people/engineers, into simple on/off, black and white 'boxes'.  That are either 100% correct/right/accurate all the time, or potentially/fully wrong.  People/engineers etc, are NOT as simple as that.

The person who sweeps the floor of someones labs, might say a particular component is WRONG, because it looks a funny colour.  They may have absolutely no knowledge of electronics.  But that DOESN'T mean that what they say is wrong.  Maybe (probably not), that component is WRONG, it could be the case.

Analogy/example, let me poke fun at you:

1N41418

You can't even 'spell' 1N4148, so I'm going to ignore all future responses you make.  As you can't even get 1N4148 correct!.  (Joke, for dramatic effect).

Ridiculous? Don't think so!! Not turning anyone into anything, we'll let them do that by themselves, yourself included

Yes we misspell often, certainly not the first nor the last time. However, we don't spew out a bunch of misinformation as has been shown, and frankly don't have the time nor patience for this kind of BS.

What we got out of this measurement was; not all 1N4148s are equal, and one was way different with measured capacitance 10~14 time greater than the rest, and some SMD 1N4148 have very low capacitance, under 0.7pF, and so on. Thought this might prove useful to some folks, but found a few seem to get really ticked off about this.

Before we started this post we expected a 1N4148 of any flavor to have <4pF capacitance and <4ns reverse recovery based upon many decades of use, despite all the prior discussion and arguing this hasn't changed, as we still consider a 1N4148 to fit within 4pF and 4ns, and frankly that's all that matters here at the Labs!!

BTW keep tabs on our misspelling, I'm sure you'll collect plenty of examples

Best,

[MK14]

Before we started this post we expected a 1N4148 of any flavor to have <4pF capacitance and <4ns reverse recovery based upon many decades of use, despite all the prior discussion and arguing this hasn't changed, as we still consider a 1N4148 to fit within 4pF and 4ns, and frankly that's all that matters here at the Labs!!

I can appreciate that way of working.  You're expecting 1N4148 diodes, to behave with reasonably 'tight' specifications, as mentioned above.  Which means that when it is designed into production units.  They will behave quickly enough to NOT allow inputs to be potentially (ignore pun) damaged, because the reverse recovery time was too slow, or not allow the full bandwidth capability, again, because of performance/capacitance issues.

So, from that point of view, they are NOT 1N4148 spec devices, and so need to be kept away from 'real' 1N4148's.  I.e. put in the junk component pile, or similar.

Others (including me), who say use the manufacture's datahseet, are still right (in some senses).  But its characteristics, are more like a 'substitute' slightly/partly equivalent 1N4148 part, rather than a close equivalent one.

Ridiculous? Don't think so!! Not turning anyone into anything, we'll let them do that by themselves, yourself included

Yes we misspell often, certainly not the first nor the last time. However, we don't spew out a bunch of misinformation as has been shown, and frankly don't have the time nor patience for this kind of BS.

Well, in that case, let me throw criticism(s) back at you then.

Your measurements, are showing perhaps two datapoints (test diodes), that you tested.  But the real issues, if you go into mass production with these devices.  Is that, that ONLY represented a particular batch (in all likelihood).  The real $64,000,000 question, is how far can the worst possible such diode, be.  Which is still able to get through all binning, testing by the manufacturer, followed by any binning/testing at your (production) end.
That is why/where the actual datasheets come into play.

I.e. Although a couple of the test diodes (from a better manufacturer of 1N4148's), may meet your 4pF and 4ns recovery measurements.  That doesn't necessarily mean, every single one, supplied for all your production runs, will.

So your tests, are only really part of the solution to keeping your circuit design and products, of reasonable or better quality.

Or to put it into different words.  Your measurements, only really gave you some limited typical values, rather than the real/true maximums, that apply to those components.  Which could be a lot worse than what you measured, especially across different batches and/or much larger quantities of those diodes.

In some cases, real problems with the components, don't actually show up until significantly higher (or lower) temperatures, are involved.  E.g. Leakage currents, can significantly be connected to the junction temperatures.

So your testing was rather limited really.

EDIT: But in all fairness to yourself.  I can't really complain.  You happened to test a couple of samples, of what were supposed to be 1N4148's (in your perception).  On finding out they weren't (so called standard 1N4148's), you decided to report your findings, here in this thread.  My criticism is therefore fairly unfair, as a couple of outliers, are enough to spot problems with those diodes.

[free_electron]

So obviously you are completely contradicting yourself here!!!

Letter suffixes in general mean a tightening of the spec. I gave you the list as per RS-370-B. It stops at K for revisions. there are some other letters that have special meaning.

This is all we need to know about the value your responses, and what your true knowledge base is!!

way to go Mr "wikipedia 1n4148". i'm out.

[MK14]

So obviously you are completely contradicting yourself here!!!

Letter suffixes in general mean a tightening of the spec. I gave you the list as per RS-370-B. It stops at K for revisions. there are some other letters that have special meaning.

This is all we need to know about the value your responses, and what your true knowledge base is!!

way to go Mr "wikipedia 1n4148". i'm out.

The OP should of been digesting/listening/reading your nice/sensible/good responses, and sensibly replied to them, himself.  Instead he seems to use certain criterion, that puts people on his, 'special' list of people to not pay much attention to.

The OP seems to remind me of engineers, who never, ever make mistakes (I mean that in the sarcastic sense), and anyone who disagrees with them. Even fractionally, gets added to their 'special' list.  I suspect the internet, is making it easier for them to do that.  I'm not sure they are taking enough account of the fact that basing such choices, on some limited text, that people are making, in posts.  Can lead to all sorts of errors of judgement.  It is probably a poor quality way of behaving and dealing with fellow people/engineers, as well.  But these observations I'm making, could easily be wrong, as the internet/forum medium, hides some of the details, and anyway, who knows what a person is really thinking about.

[T3sl4co1l]

You are turning other people/engineers, into simple on/off, black and white 'boxes'.  That are either 100% correct/right/accurate all the time, or potentially/fully wrong.  People/engineers etc, are NOT as simple as that.

Um?

I am manifestly aware of, and quite fully support the fact that, I am 100% potentially wrong, all of the time.  Like, there was any other option?  Get real.

But herein lies a microcosm of this thread... there's always going to be those who are wronger than most.  Learn to spot them, and ignore them.

Tim

[MK14]

Um?

I am manifestly aware of, and quite fully support the fact that, I am 100% potentially wrong, all of the time.  Like, there was any other option?  Get real.

But herein lies a microcosm of this thread... there's always going to be those who are wronger than most.  Learn to spot them, and ignore them.

Tim

Reluctantly, I have to admit, you are raising some good point(s), there.  But, there are many different types/sorts of engineers (and non-engineers), from educational background, and their ratio of theoretical stuff to practical work.  Then age, experience, country and many other factors, can also get thrown into the mix.
But anyway, thanks for the response.  I suspect I'm automatically, and subconsciously making similar decisions, to some level/extent.  Even if I'm not 100% aware of doing it.

EDIT:  Also, to OP.  SORRY, if I've caused any offense.  When you can't see the person/people involved face to face, it can be all too easy, to offend people.

[mawyatt]

@MK14

We did measure a few 1N4148s in different packages (DO-35, SOD123, SOD323), the Jingdao was outliner. All the data sheets we checked show a maximum Tr of 4ns and a maximum C of 4pF, a couple show 2pF, only the Jingdao shows no max C or Tr only a typical C of 5pF and Tr of 8ns. What's interesting about this data sheet is that it shows a graph of C vs V where C is ~10pF at 0.1V, yet states a typical C of 5pF with no reference to bias voltage nor measurement frequency. Please note that almost all sources indicate a max C and Tr, only Jingdao uses typical for both.

gamalot measured an unknown 1N4148WS acquired from Taobao and confirmed a low C of 0.63pF and also a DO-35 1N4148 of less than 1pF also confirming our measurements, see post #19.

We've made mistakes, and will likely do so in the future, but certainly won't try and cover up such and will man-up to the facts!! Most of my later career involved SOTA IC design, a mistake here could easily cost well over $5M, since the mask sets alone cost $5~20M and that was a decade ago, today likely 10X as much. You quickly learn who to trust with critical design tasks, and since many times new folks to the group get "assigned" by management you quickly need to access their individual skill and knowledge level, so a skill learned by necessity!!

Edit: Should mention the good that came out of all this "discussion" was the exceptionally low C of certain 1N4148Ws, I mean a Cj of ~0.66pF @ 0 volt bias

Anyway, no offense taken and none intended

Best,

[MK14]

almost all sources indicate a max C and Tr, only Jingdao uses typical for both.

Well, during the course of this thread, I did a very quick (online) investigation of that manufacturer, and it seems they are very big, as regards diodes.  A huge number of their range (quick look by me), easily perform at very high speed, and with very low (zero volt) capacitances.
So, I'd suspect that there is some reason, why those so called '1N4148W's, have such (relatively) poor specification.  (Speculation) It could be they are 'rejects' from other diode lines, or other ways, poorly performing diode dies, became available for their 1N4148W's.

Probably we will never know.  Or someone gets overly curious, and investigates the diodes further, in an attempt to identify what they really are.

Most of my later career involved SOTA IC design, a mistake here could easily cost well over $5M, since the mask sets alone cost $5~20M and that was a decade ago, today likely 10X as much. You quickly learn who to trust with critical design tasks, and since many times new folks to the group get "assigned" by management you quickly need to access their individual skill and knowledge level, so a skill learned by necessity!!

Thanks, that is a great explanation!   

It is a sad reality of life, that some engineers hold great skills, and are really brilliant, in many ways.  But, they just have this habit of sometimes being too lazy, and don't bother to check certain things, with the datasheets and/or ask for help when needed (over-confident).  Which can lead to overly expensive mistakes.  Which can then make them unsuitable for some tasks, such as the $5,000,000+ masks, you mentioned.
Also on reflection.  I DON'T feel anything like, as bad about you using Wikipedia, as a reference source, as I originally did.   Although there are still some lingering doubts, and there are better, more definitive sources.  At least you had a set of sources that the Wikipedia article gave.  Which you were able to leaf through and nominally check out.

Sometimes, a Wikipedia article, is the ONLY available source of information.  In which case you either use it, or accept not having any information at all, at least for the time being.  Some new things in Electronics and Computing, is initially only mentioned in Wikipedia articles, and some (dubious) websites.  Until it is officially announced, by the proper/official sources.

Edit: Should mention the good that came out of all this "discussion" was the exceptionally low C of certain 1N4148Ws, I mean a Cj of ~0.66pF @ 0 volt bias

That is amazing!  Particularly, as it is still a super cheap and available (at least BEFORE the chip shortages), component (1N4148W).
Discussions often create all sorts of new knowledge, and stuff, that was not realized before.

[mawyatt]

Well, during the course of this thread, I did a very quick (online) investigation of that manufacturer, and it seems they are very big, as regards diodes.  A huge number of their range (quick look by me), easily perform at very high speed, and with very low (zero volt) capacitances.
So, I'd suspect that there is some reason, why those so called '1N4148W's, have such (relatively) poor specification.  (Speculation) It could be they are 'rejects' from other diode lines, or other ways, poorly performing diode dies, became available for their 1N4148W's.

Likely some "other" die that was not designed to be a 1N4148 type?

Probably we will never know.  Or someone gets overly curious, and investigates the diodes further, in an attempt to identify what they really are.

Maybe noopy can preform his magic de-encapsulating and imagery on the Jingdao and another 1N4148W diode, so can have a look? He's quite good at this, and has mastered the chip imaging skills

Also on reflection.  I DON'T feel anything like, as bad about you using Wikipedia, as a reference source, as I originally did.   Although there are still some lingering doubts, and there are better, more definitive sources.  At least you had a set of sources that the Wikipedia article gave.  Which you were able to leaf through and nominally check out.

Sometimes, a Wikipedia article, is the ONLY available source of information.  In which case you either use it, or accept not having any information at all, at least for the time being.  Some new things in Electronics and Computing, is initially only mentioned in Wikipedia articles, and some (dubious) websites.  Until it is officially announced, by the proper/official sources.

We know what Wiki is and certainly not a reference to be utilized in an IEEE journal technical paper. We didn't have access to JEDEC, and Wiki did give a nice history from the original TI 1N914-6 way back in 1960, and had references to some manufactures data sheets. So it seemed an acceptable reference in this discussion, not knowing the very mention of Wiki would seem to enrage some folks!! 

Edit: Should mention the good that came out of all this "discussion" was the exceptionally low C of certain 1N4148Ws, I mean a Cj of ~0.66pF @ 0 volt bias

That is amazing!  Particularly, as it is still a super cheap and available (at least BEFORE the chip shortages), component (1N4148W).
Discussions often create all sorts of new knowledge, and stuff, that was not realized before.

Agree, that lovely 1N4148W is one low capacitance diode indeed. As mentioned much earlier, wouldn't be surprised to find that some special low-cap diodes are actually 1N4148 die and sold at a premium 

Best,

[edavid]

Interestingly, there's a MMBT7002 from a handful of small manufacturers, so it seems.

MMBT I don't know the history of, but for sure it's not JEDEC's normal numbering system.

Motorola Miniature Bipolar Transistor - from the people who brought you MPS, MPSA, MJ, MJE and many other famous non-JEDEC parts.

That makes MMBT7002 a real 

[TimFox]

Yes, Motorola issued quite a few plastic and SMD packaged versions of chips originally used in TO-92 and metal TO-18, TO-5, etc. packages.
Their data sheets were believable for their production parts.
Years ago (before they were acquired), a field salesperson for International Rectifier cautioned us about "IRFxxx"-numbered parts from other manufacturers, claiming they did not meet the specs in IR datasheets.

[mawyatt]

Yes, those old Motorola discrete parts were good, as were the data sheets in the Data Books!! Also the old TI Orange Data Book and the famous National Linear and other Data books were on every engineers desk

Best,

[CatalinaWOW]

I certainly was not enraged by the use of Wiki as a source of information.  I use it also, and often and believe in the concept sufficiently to donate annually.

But, I remain concerned about belief in Wiki as definitive.  Wikipedia both benefits and suffers from its crowd sourced methodology.  In the entertainment of "who's who" areas as well as recent history it is dangerous to place any trust in the articles.  Technical issues are much more reliable, but still have to be used with a grain of salt.

[MK14]

Now the dust has settled.  In NO WAY, am I trying to disagree, with anything or anyone, which has already been mentioned in this thread.  I'm just providing something I just found, while looking for more information, about 1N4148 specifications.

The following video (just for DISCUSSION, I'm in no way saying it is definitive), can't be taken as definitive, by any stretch of the imagination.  But it seems to suggest, what I can believe is vaguely close to the reality, and where this thread seems to have headed towards (very approximately).  It seems to say that the JEDEC standard, is only defining the naming scheme for the diode, rather than its specifications.  It also seems to say there are a minimum set of basic characteristics, for it to be a low cost 1N4148.  Such as, no more than 4ns switching speed, at least 150mA current capability, and at least a 100 volts, working voltage capability.

[RoGeorge]

This thread went surrealistic.   

[MK14]

This thread has made me VERY curious, about the OP's suspect 1N4148W's dies, and where they have come from.  So I've been doing some digging.  I decided to go through the (problematic 1N4148W's) manufacturer's (Jingdao Microelectronics), diode range.  In an attempt to find a suspect, for what the diode die/type, really might be.  So, VERY wild speculation on my part.  I stopped, as soon as I found a possible candidate, there could be others.
The datasheets reverse recovery time, is quoted as a lot higher (50ns), but that could be because it is measured at different ratings, for the more powerful, at least voltage wise device.  The OP's part, doesn't even seem to define the test conditions for its reverse recover time, making it rather difficult, to easily/directly compare these diodes, as regards that part of the specification.

It seems, they sell a '1N4148' like diode, in relatively high voltage terms.  Presumably, because of the way the internal structure of the high (relatively) voltage die is.  That has caused or is related to the increase in junction capacitance.  It is rated for 300(continuous)/350(peak) volts.
The alternative diode I found, has the same package and same junction capacitance, datasheet specification, as the OP original problematic part(s).

Called a BAV3004W, also in a SOD-123 case.  I presume (wild guess), it failed the high voltage and/or leakage tests.

Here is the datasheet, of the possible, real die diode:
https://datasheet.lcsc.com/lcsc/2009141737_Shandong-Jingdao-Microelectronics-BAV3004W_C779377.pdf

The possible original part  (  https://datasheet.lcsc.com/lcsc/1809291712_Shandong-Jingdao-Microelectronics-1N4148W_C115103.pdf  ), at a quick glance, doesn't even seem to define the test characteristics, for the reverse recovery time.

[T3sl4co1l]

Curiously, they're not even any good at that one [BAV3004W] either; compare Diodes Inc's version, under 1pF!

FYI, t_rr changes surprisingly little with If, depending more on ratio of forward to turn-off (recovery) current, or ramp rate (dI/dt).  So, although you'd prefer recovery tested at level (i.e., 10A for a 10A diode), doing it at e.g. 0.5A, or say 50 ohms in a test jig (for signal diodes), is really not as unrepresentative as you'd think for such a nonlinear device.

Tim

[gamalot]

As an electronic engineer from China, based on my experience, I suggest that when you buy discrete semiconductor devices in LCSC, choose Changjiang Electronics products, they have a longer history and should have more reliable product quality.

https://lcsc.com/product-detail/Switching-Diode_Changjiang-Electronics-Tech-CJ-1N4148W_C2099.html

[MK14]

As an electronic engineer from China, based on my experience, I suggest that when you buy discrete semiconductor devices in LCSC, choose Changjiang Electronics products, they have a longer history and should have more reliable product quality.

https://lcsc.com/product-detail/Switching-Diode_Changjiang-Electronics-Tech-CJ-1N4148W_C2099.html

Thanks for that recommendation/tip, I'll bear it in mind, in the future.

That is EXACTLY, why I said in an earlier post:

If it was a Chinese make.  Was it important to check if they were a recommended Chinese make or not ?

Because I had a feeling it has been discussed on here before.  You really need to know what you are doing, when buying from Chinese suppliers.  Probably more so than western makes.  There use to be a guy on here, who seemed to give very good advice about stuff like that, but unfortunately they are not on here any more.  (Blue*****, partly hidden, as they want to remain private, now I believe).

[gamalot]

As an electronic engineer from China, based on my experience, I suggest that when you buy discrete semiconductor devices in LCSC, choose Changjiang Electronics products, they have a longer history and should have more reliable product quality.

https://lcsc.com/product-detail/Switching-Diode_Changjiang-Electronics-Tech-CJ-1N4148W_C2099.html

I have immigrated to Australia for many years, and I just noticed that they spun off the discrete semiconductor division of Changjiang Electronics in 2018, and the new name is Changjing Electronics.

https://www.jscj-elec.com/en

[thinkfat]

As an electronic engineer from China, based on my experience, I suggest that when you buy discrete semiconductor devices in LCSC, choose Changjiang Electronics products, they have a longer history and should have more reliable product quality.

https://lcsc.com/product-detail/Switching-Diode_Changjiang-Electronics-Tech-CJ-1N4148W_C2099.html

Oh, that's good to know I happen to use their 1N4148WS in a design (LCSC part number C2128). Not that the switching speed would matter in that particular application, though.