EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: devttys0 on March 04, 2015, 04:18:00 pm
-
While troubleshooting a simple single-transistor amplifier circuit recently, I found that I have what appears to be a bad batch of 2N2222As.
The DC biasing of my original circuit is attached (bias1.png). The base voltage should be at 3.6v, putting the emitter voltage at 3v, and emitter/collector current at .5mA. The base bias resistors are much stiffer than they need to be, so the impedance looking into Q1’s base (R3 * (hFE + 1)) should have negligible effect on the base voltage, even with abnormally low values of hFE (the lowest minimum hFE listed in the data sheet is 35).
With Q1 omitted from the circuit, I do in fact read 3.66v at the junction of R1 and R2. However, with Q1 inserted into the circuit, the base voltage drops to 3.39v. This means that R2 || (R3 * (hFE + 1)) must be ~2.4K ohms; since R2 is 2.7K, then (R3 * (hFE + 1)) must be ~24K ohms, making hFE only 3.3 (5600 * (3.3 + 1) = 24000)!
Granted, the hFE values listed in the data sheet are given for a particular VCE and IC, so just to check, I replicated the data sheet’s specified configuration (bias2.png). The base voltage should be ~11v, putting the emitter voltage at ~10.4v, making the emitter/collector current 1mA. VCE is then ~10v, matching almost identically to the conditions specified in the data sheet for a minimum hFE of 50.
Again, the measured base voltage with Q1 removed was as expected (11.00v); however, with Q1 inserted, the base voltage rose to 12.30v. This means that there must have been some base-collector current, and the base-collector resistance was low enough to drop the total resistance of R1 || Q1 significantly.
All voltages are well within spec, so there shouldn’t be any breakdown of the collector-base junction, but clearly there is. Changing the power supply voltage for the circuit in bias2.png, it seems that the base-collector break down voltage is about 9v (a noticeable increase in the base voltage occurs when Vcc is at about 18v).
Now, if this was happening to one transistor, I’d just figure that it was one that maybe I had blown at some point in the past and had somehow managed to find its way back into my parts bin. But in fact I have five transistors that exhibit this exact behavior, all marked “P2N2222A E02” (picture attached). Their markings seem to be consistent with those from ON semiconductor (http://www.onsemi.com/pub_link/Collateral/PN2222-D.PDF (http://www.onsemi.com/pub_link/Collateral/PN2222-D.PDF)), though I couldn’t tell you where I got the transistors from; most stuff I get from digikey, but I do have parts in the junk box from various “grab bags” and the like.
I’ve replaced the offending P2N2222As with other BJTs (Fairchild 2N2222As, 2N4401, MPSA18, 2N5210), and they all work as expected in the described circuits, having practically no effect on the base voltage.
I've never seen this before, so I was wondering if anyone here has any insight/experience with something like this? Is it not unheard of to get a bad batch of transistors? Is there some plausible way they could have all been damaged in the same way? Or is this just the price you pay for those “grab bag” specials? (I’m guessing the latter… :P)
-
I would guess fake or counterfeit parts.
-
Are you sure your transistor is not installed in reverse ie. collector connected where emitter should go and vice versa?
-
You bought them on eBay?
I know I've never bought transistors off eBay, but I have gotten parts from surplus places, which is probably just as bad. Like I said, I can't be sure of the source; they've probably been sitting in my parts tray for a while.
Are you sure your transistor is not installed in reverse ie. collector connected where emitter should go and vice versa?
Positive. They *do* work to some extent; the original circuit I was building was a 16MHz Pierce oscillator. With the transistors in question it didn't work at 16MHz, but does function at frequencies below 10MHz or so (with proper transistors in place, the circuit works at 16MHz as intended).
-
This is what I would expect from a bjt operating in reverse active region (c-e reversed) .You get a beta of ~ 1-7 ,and reverse e-b (which is now c-b) breakdown about ~7 to 9V.
Regards
-
This is what I would expect from a bjt operating in reverse active region (c-e reversed)
Well, so much for me being "positive". :P Sure enough, flipped the collector and emitter and they work as expected now. :palm:
Now I really wish I knew where I got these from, as their collector and emitter are reversed from every 2N2222 datasheet I've ever seen.
-
Ah, found it. They're Motorola: http://www.solarbotics.net/library/datasheets/2N2222.pdf (http://www.solarbotics.net/library/datasheets/2N2222.pdf)
-
I recall a similar problem many years ago when I worked in an electronics shop. I think it related to a BC212 Vs BC212A but I may be wrong and can't be bothered to check ;D We had customers returning the 'A' version as faulty. They were in fact all fine but customers were using the BC212 pin-out for a BC212A which was different. This can be a royal PITA so it is always best to find the exact pin-out for the transistor used and not just the 'generic' pin-out. I mote the transistor from Motorola is prefixed with a 'P'. I have not seen that version notation before and totally understand your assumption that it was a standard 2N2222A.
Glad you got it sorted out
Aurora
Correction. I just checked.... it was the BC212 versus the BC212L. As in your case, different pinout !
BC212 and A version = E B C
BC212L = E C B
-
This is what I would expect from a bjt operating in reverse active region (c-e reversed)
Well, so much for me being "positive". :P Sure enough, flipped the collector and emitter and they work as expected now. :palm:
Now I really wish I knew where I got these from, as their collector and emitter are reversed from every 2N2222 datasheet I've ever seen.
The handy $20 LCR/transistor tester that is a popular post going even today here would have validated the pin out of any npn/pnp device.
-
I have not seen that version notation before and totally understand your assumption that it was a standard 2N2222A
I hadn't either, but it turns out that even Wikipedia specifically notes that parts with the "P2N" prefix have a different pinout. http://en.wikipedia.org/wiki/2N2222 (http://en.wikipedia.org/wiki/2N2222)
The Wikipedia image showing the different pinouts and part prefixes was also the second hit in Google images when searching for "2n2222 pinout". |O
-
The handy $20 LCR/transistor tester that is a popular post going even today here would have validated the pin out of any npn/pnp device.
I second that, +1
Funny, as soon as a "non working" part is the subject, the first thought of some people is "ebay, fake!"
What would be the point/profit in "faking" a $0.005 transistor?
-
The handy $20 LCR/transistor tester that is a popular post going even today here would have validated the pin out of any npn/pnp device.
Heh. that is quite literally the main reason I got that transistor tester. it goes right next to my parts bin to verify transistors as i pull them out. Ever since my bulk order of 2n3904s was once about 10% BC548s.
now if i can modify the firmware to recognize lm317, lm78*, lm334, and lm431s I'll finally be able to sort my TO-92 bucket.
-
Still more reason to leave the stupid 2222 alone! Like the 3055, it is a historical fossil only, and utterly irrelevant today.
Tim
-
What's better in a to-92 package? Gain is nothing great but Ic isn't bad, and something like .14 each for a hundred real ones.
(I do think it's funny when you find a can version (2n2222) used rather than the TO-92 version though!)
-
Yeah, the first mistake is using 2N2222(A), which is a TO-18 metal can part -- way too expensive for the inferior specs!
2N4401/3 is my preferred "medium size" jellybean. It's much better specified; what PN2222 would've been if it were developed some years later than it was introduced.
Even better, 2N3904/6, at about 1/3 the size, is correspondingly faster for small signal and logic applications.
Not always well specified (I don't know that the JEDEC spec is any tighter, but some manufacturers' are?), but still better.
For more nuanced applications, there are still other excellent choices; 2N5088 for higher hFE (and specified -- not really 'low', but known -- noise performance), MPSH10 for higher speed (MMBTH10 = SMT), etc. Or 2N7000 (in TO-92; 2N7002 = SOT-23 -- yes, a true JEDEC part in SMT) or BSS84 (P channel) for MOSFETs of similar spec.
All are prodigiously produced, and pleasant on the purse. ;)
Tim
-
Thanks for the opinion. I'll make a note of those parts for myself. :)
-
In regards to the original post:
Can't you compare the diode drops for the 2 junctions using a DMM and get a sanity check on which is the BC and BE junctions?
When I do this the BC junction (being larger I suppose) has about 10mV less DC drop.
-
however, with Q1 inserted, the base voltage rose to 12.30v.
The issue is likely with pin-out or bad measurements. That's not supposed to happen with a npn.
-
In regards to the original post:
Can't you compare the diode drops for the 2 junctions using a DMM and get a sanity check on which is the BC and BE junctions?
When I do this the BC junction (being larger I suppose) has about 10mV less DC drop.
This is about a typical difference, yes. I think it's more due to the difference in doping, which gives 10-20mV less "built-in potential", which also gives rise to the not-quite-zero saturation voltage offset (i.e., Vce(sat) goes down to ~10mV but not much below, for any combination of Ic and Ib). So, the one that's the higher breakdown voltage (lighter doping) gets identified.
Testing it as a zener would be more reliable, since that gets you breakdown directly. Beware of symmetrical junctions, though... then again... nevermind, because if it's symmetrical, who cares, right? ;D (Rarely found among silicon parts, but some exist; mainly, old Ge transistors were made with symmetrical or near-symmetrical junctions.)
Tim
-
Testing it as a zener would be more reliable, since that gets you breakdown directly. Beware of symmetrical junctions, though... then again... nevermind, because if it's symmetrical, who cares, right? ;D (Rarely found among silicon parts, but some exist; mainly, old Ge transistors were made with symmetrical or near-symmetrical junctions.)
J-FETs are often symmetric, there are variations of the schematic symbol to differentiate when you need a symmetric or non-symmetric one. whether the gate is centered or next to the source on the symbol.
John
-
BJT pinouts are a total disaster for a hobbyist. If you have parts that you are not 100% sure about, the only way is to test the pinout, e.g. using a multimeter with a transistor tester capability (very handy for finding the pinout). It shows about the "expected" Hfe when it's correctly connected - otherwise it shows way too little or way too much.
When you order the parts from a proper source and check the exact datasheet with all the possible numbers and letters in the product code (they might not be printed on the transistor), then you can trust the pinout in datasheet.
I guess there are reasons why they produce same type of transistor in many different pinouts, some being technical but most historical or political - different manufacturers, different era, counterfeit parts, etc.
With mosfets, it's the same pinout 99.9% of the time and it's easier to measure the body diode; the reverse mosfet is just almost a short and does not "kind of" work like a reverse bjt.
-
Both 2n2222 and 2n2222a work in the same way, unless both come with different absolute maximum ratings. I'd suggest you have a look at
https://www.theengineeringprojects.com/2017/06/introduction-to-2n2222.html (https://www.theengineeringprojects.com/2017/06/introduction-to-2n2222.html)
they have discussed everything you need to know about this transistor with proteus simulation.
-
In regards to the original post:
Can't you compare the diode drops for the 2 junctions using a DMM and get a sanity check on which is the BC and BE junctions?
When I do this the BC junction (being larger I suppose) has about 10mV less DC drop.
You are the first person in thirty years I've come across that knew about that. That's my "trick" question I like to ask people. Most people only know how to find the base and polarity.
I had the same problem with a bunch of Radio Shack 2N2222s that had an incorrect pinout on the package. I was sitting in on an intro electronics course as a break from looking through a petrographic microscope all day. One day the instructor who was a semiconductor physicist by training said that the BC junction had a lower voltage drop because the greater area of the collector trumped the higher doping of the emitter. I had asked everyone I knew who had any electronics knowledge how to verify the pinout, but no one knew. Needless to say I was quite excited and immediately went home to try it. Now it's habit to check the pinout before soldering.
A lot of DMMs now won't turn on the diodes on the resistance range so you have to measure the voltage drop using the diode test. But at the time all I had was a VOM so I measured the forward resistance. The collector was lower. My confusion was dispelled.
-
now if i can modify the firmware to recognize lm317, lm78*, lm334, and lm431s I'll finally be able to sort my TO-92 bucket.
I've been thinking about porting the device to a Pro Micro clone (or any other ATmega32u4 microcontroller with native USB support, or a much more powerful but 3.3V only Teensy LC), with one of those nice 128x32 or 128x64 OLED displays, so that when standalone, it'd work exactly like the existing one (but with a nicer display), but when connected to a PC, a simple Python3+Qt5 (or Python3+Gtk3) application would allow one to list the devices to be recognized, so that if one of them is recognized and within spec, it'd be shown on both the OLED display as well as in the application, and one could use it to sort their parts bin...
The software part is easy, but the electronics side is one of my "forever projects", which will never see the actual light of day ... :(
-
In regards to the original post:
Can't you compare the diode drops for the 2 junctions using a DMM and get a sanity check on which is the BC and BE junctions?
When I do this the BC junction (being larger I suppose) has about 10mV less DC drop.
You are the first person in thirty years I've come across that knew about that. That's my "trick" question I like to ask people. Most people only know how to find the base and polarity.
I had the same problem with a bunch of Radio Shack 2N2222s that had an incorrect pinout on the package. I was sitting in on an intro electronics course as a break from looking through a petrographic microscope all day. One day the instructor who was a semiconductor physicist by training said that the BC junction had a lower voltage drop because the greater area of the collector trumped the higher doping of the emitter. I had asked everyone I knew who had any electronics knowledge how to verify the pinout, but no one knew. Needless to say I was quite excited and immediately went home to try it. Now it's habit to check the pinout before soldering.
A lot of DMMs now won't turn on the diodes on the resistance range so you have to measure the voltage drop using the diode test. But at the time all I had was a VOM so I measured the forward resistance. The collector was lower. My confusion was dispelled.
Thanks for that suggestion. It's much better than the method I used to use for identifying the collector and emitter, until you posted this. I used to mess around measuring the Hfe, by connecting the transistor across a 3V power supply and connecting the base to positive supply, for an NPN transistor, via a 100k resistor, then exchanging the other pins, until I found which way would draw the most current, therefore have the higher Hfe, which would be when the collector is positive and emitter negative.
EDIT:
I'm surprised LTSpice models this properly. I was originally only doing to use it to draw the schematic, for illustrative purposes, but I tried simulating it for curiosity's sake and it worked! I haven't actually tested a real 2N2222 to see if this is correct, but the figures do look reasonable: IB + IC of 96μA when connected backwards and 4.67mA when connected correctly. IB was about 24μA, so that gives an Hfe of 3, connected backwards and an Hfe of 195, correctly.
(https://www.eevblog.com/forum/projects/bad-batch-of-2n2222-transistors/?action=dlattach;attach=457927;image)
-
I can still feel my elation at acquiring that tiny bit of knowledge. I find it very sad that it is not well known.
-
It’s actually taught in the art of electronics supplementary lab manual right back from the first edition in about 1980. Very very useful bit of knowledge.
-
I didn't get AoE until ed 2 at which time I already knew about it. And I never did anything with the lab manual even though I have it.
-
SPICE uses Gummel-Poon model, standard. It's a more general model, and yes, includes inverted operation! You're effectively measuring the BF and BR parameters of the model. :)
However, as is the case with any model -- the parameters fitted may not be most representative for your situation. Example: low-Vce(sat) transistor models typically have BR ~ 5, but real devices are closer to 100-150.
Tim
-
Testing it as a zener would be more reliable, since that gets you breakdown directly. Beware of symmetrical junctions, though... then again... nevermind, because if it's symmetrical, who cares, right? ;D (Rarely found among silicon parts, but some exist; mainly, old Ge transistors were made with symmetrical or near-symmetrical junctions.)
Back before JFETs became available, they made "chopper" bipolar transistors with symmetrical or at least more symmetrical junctions. Like JFETs, they are useful for wide differential input voltage differential pairs, audio muting and switching, and synchronous modulation and demodulation applications like chopper amplifiers.
They are still manufactured and can be found if you know what to look for:
https://www.mouser.com/ProductDetail/Toshiba/2SC3326-ALF?qs=%2fha2pyFaduiyTvpx89IGbfH1eNAndtaXwZN3qxNTDg4%3d (https://www.mouser.com/ProductDetail/Toshiba/2SC3326-ALF?qs=%2fha2pyFaduiyTvpx89IGbfH1eNAndtaXwZN3qxNTDg4%3d)
https://www.mouser.com/ProductDetail/Central-Semiconductor/CMPT404A-TR?qs=sGAEpiMZZMshyDBzk1%2fWiw99kSkYzPxmA%252bTeVhEzuE0%3d (https://www.mouser.com/ProductDetail/Central-Semiconductor/CMPT404A-TR?qs=sGAEpiMZZMshyDBzk1%2fWiw99kSkYzPxmA%252bTeVhEzuE0%3d)
-
I have found various pinouts for these transistors. Always check a datasheet from the actual manufacturer.
-
BJT pinouts are a total disaster for a hobbyist. If you have parts that you are not 100% sure about, the only way is to test the pinout, e.g. using a multimeter with a transistor tester capability (very handy for finding the pinout). It shows about the "expected" Hfe when it's correctly connected - otherwise it shows way too little or [bold]way too much[/bold].
Interesting, does this mean there are some transistors with higher current gains when reverse connected or is it an artifact of the measuring technique?
EDIT: Fixed quoting error.
-
Meters measure collector current for some collector voltage and base current. If there's an erroneous sink of "collector" current at that voltage, it'll look like hFE is huge. Which, isn't wrong...
I don't know of any that have higher inverted hFE, but some do have very useful hFE: purpose-made "muting" transistors (higher Vebo, inverted hFE over 100 usually, and specified in terms of Rce(on) -- used much like JFETs for analog switching, but with base current rather than gate voltage), and low-Vce(sat) [power] switching transistors, which are not specified for inverted operation (and have ordinary Vebo), but are often rated in terms of Rce(on), and typically have quite high inverted hFE (presumably, a consequence of their low Vce(sat), because saturation isn't a one-sided thing, it's symmetrical around zero and to do well in one direction, it should do ~equally well the other way too).
Tim
-
Meters measure collector current for some collector voltage and base current. If there's an erroneous sink of "collector" current at that voltage, it'll look like hFE is huge. Which, isn't wrong...
Tim
But how could you get an erroneous collector current (unless the junction had broken down due to excessive voltage - very unlikely)? Surely hFE measurement is about as simple as it gets in most meters - a fixed base and collector resistor to VCR and measure the collector voltage? A more sophisticated meter might use a true constant current source for Ib but I still can't see how it could cause an erroneous reading.
-
Certainly:
- Shorted or leaky transistor
- Reversed, breakdown (at DMM test voltages, would probably be a superbeta or RF type with Veb <= 3V say)
- Actually a FET (probably JFET or depletion MOS)
- Actually an SCR
etc. :)
Tim