EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: eti on November 05, 2021, 03:33:13 am
-
I've got a cheap Vichy VC97 dmm, and another one by "AstroAI", and when I shove a P2N2222A in the hfe tester, I get wildly different gain readings. On one I get 253, and the other is 150 - any idea what gives? And yes, the pins are correct, I've tested it 5x
(and I'm also aware that the *P*2N2222A* variant has a different pinout.
Thanks.
-
Seems like hFE should be in the range of 50 to 300, so both indicated values are within the expected range.
-
Seems like hFE should be in the range of 50 to 300, so both indicated values are within the expected range.
Ah okay, but why the vast difference - the device isn't going to change its gain between meters, is it. Also the sockets are horrendous on these cheap DMMs 😂
-
It is going to change its HFE at different collector currents or different VCEs and you have no guarantee that those conditions are the same for both meters (and knowing the calibre of the VC97 I don't expect you'll find those conditions listed in the meter's data sheet/manual, or the data sheet/manual for the even more anonymous AstroAI which I've never even heard of before).
This is a question about pretty basic transistor theory, or rather a lack of sufficient grasp of it, and as such would have been more at home under "beginners" where people are generally more minded to be forthcoming with material of a tutorial nature - which is what it looks like is needed here.
-
Hfe is not well controlled between units, and it also varies significantly with collector current, with different transistor types optimized for different collector currents. A lot of small signal transistors are optimized to operate with a collector current of about 10 milliamps, but low noise and high gain types will be optimized for 1 milliamp or even 100 microamps. You can get a good idea of the design current from the datasheet by looking at the hfe specifications to see which ones are fully specified at a specific collector current.
-
Take a look at the data sheet. https://www.onsemi.com/pdf/datasheet/p2n2222a-d.pdf (https://www.onsemi.com/pdf/datasheet/p2n2222a-d.pdf)
You could guess from the graph of hfe vs collector current that these testers are putting a few milliamps through the device in test. You could even naively assume that one uses roughly 2 mA and the other about 15 mA. But that would require assuming that the junction temperature was the same in both tests. And that Vce was the same for both meters.
Some things you might do to investigate. Measure the current (you have two meters so this is just a wiring issue.) Measure the voltages. Check repeatability on both meters. Look for variation over time from self heating.
-
Thanks everyone :))
-
Reverse engineer the hFE circuits in those meters :-/O
-
I've got a cheap Vichy VC97 dmm, and another one by "AstroAI", and when I shove a P2N2222A in the hfe tester, I get wildly different gain readings. On one I get 253, and the other is 150 - any idea what gives? And yes, the pins are correct, I've tested it 5x
(and I'm also aware that the *P*2N2222A* variant has a different pinout.
Thanks.
In a nutshell, the hFE measurement of a multimeter is useful for binning transistors (sorting by similar gain), if you need specimens with similar values.
-
The hFE is unimportant for most applications. When a circuit calls for matched transistors, it's normally more critical to choose parts with the same VBE, at the same collector current. This can be determined using a diode test function. Connect the base and collector, with the positive probe on the collector & base and the negative probe on the emitter, for NPN transistors, or the opposite polarity for PNP transistors.
-
Hfe varies with tempreature. Put your finger on the transistor to see the warming effect
-
Those cheap multimeter hFE measurements are a joke.
An almost correct way of doing it is testing the transistor in an op-amp current sink, where the emitter current is precisely controlled. This is not 100% correct (it should be the collector current), but close enough. The emitter current is set to an appropriate value, and the base current is measured to compute hFE.
More difficult: it should be a pulse measurement to avoid heating of the transistor.
A fun project for spending time :)
-
Usually, hFE is a stronger function of IC than of VCE, and typically it shows a maximum value for some reasonable value of IC.
-
The hfe measurement on a DMM is a joke only if you try to make it more than it is. It is a quick way to find if a transistor is dead, and gives an indication if the gain is enough to be useful on an unknown part and if it is generally what it should be for a part where you can get the datasheet. Quite useful if you are away from the lab and have only basic tools at hand.
While you can get more information with enough effort and understanding (just as you can measure your latitude with time, a couple of sticks, and a tape measure) you are better off getting a better instrument if you need more than an "it is alive" indication.
-
On simple ICL7106 DMMs it tends to be reasonably accurate, maybe up to 10% off.
The measurement is performed at Ib=10µA (220k pullup to 2.9V IIRC) and the number displayed is simply 100x the collector/emitter current in mA. For a typical reading of 200 it's 2mA.
A 50% difference is somewhat surprising, it has to be significantly different circuits in these meters.
-
The measurement is performed at Ib=10µA (220k pullup to 2.9V IIRC) and the number displayed is simply 100x the collector/emitter current in mA.
Which is totally useless. Collector current is the important controlling parameter. Just supplying some current to the base and measuring the IC is laughable.
But the feature looks nice in the sales brochure.
-
I'm telling you that the collector current is literally displayed on your meter during the measurement so you can correlate it with plots in the datasheet to estimate performance at other bias conditions ;)
And that the current used is fairly reasonable for typical small signal transistors with typical β of 100~300 (ends up being 1~3mA).
-
I understand completely what you're telling me.
I'm just saying that that's not an hFE measurement. It's just "something" that you can "correlate" with the data sheet.
The data sheet specifically lists the collector currents to be used for hFE measurement.
-
I understand completely what you're telling me.
I'm just saying that that's not an hFE measurement. It's just "something" that you can "correlate" with the data sheet.
The data sheet specifically lists the collector currents to be used for hFE measurement.
The data sheet tells the current at which the manufacturer specifies hfe. But hfe is a parameter that exists at all currents and temperatures. And what the DMM reports is a very good approximation of hfe at one operating point. Whether that number is useful depends on your application. I agree that it is nearly useless for evaluating specification compliance. At best it can say maybe yes or almost surely not.
-
regarding different subscripts hFE and hfe - forward current gain CE config. V2 short circuited - I see from beginners video attached that a large capacitor in parallel shorts the ac output, presumably for the dc gain measurement you would literally just short with a wire? (a beginner's question - tagged on here)
https://www.youtube.com/watch?v=SPmaLlHwAZg&list=LL (https://www.youtube.com/watch?v=SPmaLlHwAZg&list=LL)
-
By definition, the “FE” subscript denotes the (DC) ratio of IC/IB, while the “fe” subscript denotes the (AC) derivative dIC/dIB, both evaluated at a given value of VC and IC.
-
open circuit output? or something in between?
-
no reply -probably not a great question - but it looks like hFE matches up well with a simple test:
https://www.youtube.com/watch?v=nD7fYhdISGI (https://www.youtube.com/watch?v=nD7fYhdISGI)
-
The data sheet tells the current at which the manufacturer specifies hfe. But hfe is a parameter that exists at all currents and temperatures. And what the DMM reports is a very good approximation of hfe at one operating point.
The DMM does NOT report hfe. That's an AC value.
It leads you to think it's reporting hFE (which is a DC value), but in the wrong way.
Uppercase and lowercase play a very large role in science and engineering.
The almost correct way of measuring hFE is this:
-
The data sheet tells the current at which the manufacturer specifies hfe. But hfe is a parameter that exists at all currents and temperatures. And what the DMM reports is a very good approximation of hfe at one operating point.
The DMM does NOT report hfe. That's an AC value.
It leads you to think it's reporting hFE (which is a DC value), but in the wrong way.
Uppercase and lowercase play a very large role in science and engineering.
The almost correct way of measuring hFE is this:
You are absolutely and pedantically correct. Though I don't know when this distinction was made. When I was in university (during the transition from vacuum tubes to transistors) I don't recall both terms being identified. And there is much useful (but approximate) analysis that can be accomplished without making this distinction.
If you check data sheets it is somewhat uncommon to find both specified. And when they are it is common to find a wide range of spec values for both, with a very large overlap in the ranges. So it would be unusual to make a production design that depended on a difference in the values.
Of course, when it matters you need to be aware of the difference. And you can then buy large numbers of transistors and select the ones that meet the needs of your circuit. Or spend the time to measure each transistor and adjust the circuit to match the transistor. Not the usual way to build things, but for certain applications might make sense.
-
You are absolutely and pedantically correct. Though I don't know when this distinction was made. When I was in university (during the transition from vacuum tubes to transistors) I don't recall both terms being identified. And there is much useful (but approximate) analysis that can be accomplished without making this distinction.
If you check data sheets it is somewhat uncommon to find both specified. And when they are it is common to find a wide range of spec values for both, with a very large overlap in the ranges. So it would be unusual to make a production design that depended on a difference in the values.
Of course, when it matters you need to be aware of the difference. And you can then buy large numbers of transistors and select the ones that meet the needs of your circuit. Or spend the time to measure each transistor and adjust the circuit to match the transistor. Not the usual way to build things, but for certain applications might make sense.
Surely a comparison can be made here with diodes, where "resistance" could be measured as either V/I or as dV/dI, and clearly they have different values for any given value of I or V.
Similarly, if Ic/Ib is more or less constant for varying values of Ic, then dIc/dIb would have the same value as Ic/Ib. However, if this is not the case, and if dIc/dIb could have significant variation compared to Ic/Ib, then the distinction matters.
I imagine if you were doing a "small signal" design around a particular bias point on the transistor, then dIc/dIb would be the parameter of most interest?
-
To first approximation, given the wide range of hFE and hfe from unit to unit for a given part number, the numerical values for both will be similar (within the spec range). However, the "fe" values are for AC small signal analysis and the "FE" values are for switching large-signal analysis.
-
You are absolutely and pedantically correct. Though I don't know when this distinction was made. When I was in university (during the transition from vacuum tubes to transistors) I don't recall both terms being identified. And there is much useful (but approximate) analysis that can be accomplished without making this distinction.
If you check data sheets it is somewhat uncommon to find both specified. And when they are it is common to find a wide range of spec values for both, with a very large overlap in the ranges. So it would be unusual to make a production design that depended on a difference in the values.
Of course, when it matters you need to be aware of the difference. And you can then buy large numbers of transistors and select the ones that meet the needs of your circuit. Or spend the time to measure each transistor and adjust the circuit to match the transistor. Not the usual way to build things, but for certain applications might make sense.
Surely a comparison can be made here with diodes, where "resistance" could be measured as either V/I or as dV/dI, and clearly they have different values for any given value of I or V.
Similarly, if Ic/Ib is more or less constant for varying values of Ic, then dIc/dIb would have the same value as Ic/Ib. However, if this is not the case, and if dIc/dIb could have significant variation compared to Ic/Ib, then the distinction matters.
I imagine if you were doing a "small signal" design around a particular bias point on the transistor, then dIc/dIb would be the parameter of most interest?
You are right. There can be factors of difference between the two. But I think the real point many are trying to get to is that designs which depend on hfe or hFE being within a factor of three to five of a specific value are unlikely to be successful. The variation from transistor to transistor is likely to be as large or larger than the difference between hFE and hfe.
My design courses in school emphasized techniques to make the circuit insensitive to the specific value of these parameters. The work I have seen and done since then continues this emphasis. And in designs which are insensitive to large differences in the value there is usually little sensitivity to the difference between the two parameters.
To me the right analogy with respect to the subject of this thread are those printed battery testers seen on one brand of battery. The one that has a heating element and a liquid crystal coating over a printed red-yellow-green strip. Shorting it across the battery terminals heats the display changing the opacity of the liquid crystal and revealing a good/bad indication for the battery. It can be argued that this is a crude current meter or a crude voltmeter, though it properly does neither. But it does provide a useful indication of the battery condition. The hfe test in these DMMs does not technically measure either hFE or hfe correctly, but it does give a number in the neighborhood of both, which is good enough for some purposes.
-
The variation from transistor to transistor is likely to be as large or larger than the difference between hFE and hfe.
That depends on the frequency and GBWP of the transistor. At higher frequencies, there will be a significant difference between the hFE and hfe.
To me the right analogy with respect to the subject of this thread are those printed battery testers seen on one brand of battery. The one that has a heating element and a liquid crystal coating over a printed red-yellow-green strip. Shorting it across the battery terminals heats the display changing the opacity of the liquid crystal and revealing a good/bad indication for the battery. It can be argued that this is a crude current meter or a crude voltmeter, though it properly does neither. But it does provide a useful indication of the battery condition. The hfe test in these DMMs does not technically measure either hFE or hfe correctly, but it does give a number in the neighborhood of both, which is good enough for some purposes.
Yes, the hFE test on a DMM is good enough to check whether a transistor works, or not. Unfortunately it tends to be the cheaper meters which have it, probably because it compromises the CAT rating, which isn't important if you're only using it on extra low voltage circuits.
Those battery test strips are often better, than a proper multimeter, because they actually provide a load for the battery. Measuring the voltage with a meter is often a poor indication of whether the battery is good, or not. It's not unusual for an AA cell to measure 1.5V, but have a high impedance, causing the voltage to drop under load.
-
That depends on the frequency and GBWP of the transistor. At higher frequencies, there will be a significant difference between the hFE and hfe.
If we're all going to get pedantic about it then the significant difference is between the hFE at DC and the hfe at higher frequencies. :)
-
That depends on the frequency and GBWP of the transistor. At higher frequencies, there will be a significant difference between the hFE and hfe.
If we're all going to get pedantic about it then the significant difference is between the hFE at DC and the hfe at higher frequencies. :)
Fair enough my wording could have been better, but I would have thought it would be obvious that hFE is nearly equal to hfe at low frequencies and there's only a big difference at higher frequencies. For the purpose of this exercise, what constitutes high and low frequencies, depends on the transistor. A big and slow 2N3055's hfe might drop significantly at the upper end of the audio band, whereas a PN2222A's hFE and hfe can be treated the same, up to a few MHz.
-
A big and slow 2N3055's hfe might drop significantly at the upper end of the audio band, whereas a PN2222A's hFE and hfe can be treated the same, up to a few MHz.
"Plummet" is the word you're looking for:
(https://www.eevblog.com/forum/chat/measuring-hfe-of-p2n2222a-varies-wildly/?action=dlattach;attach=1318799;image)
Yes, a guaranteed minimum hfe of 10 at 1kHz. There's a reason nobody uses 2N3055s in audio any more (if they have any sense).
-
Only problem with that is that a part branded as 2N3055 might actually be a very fast part, not the original slower than paint drying planar device, but the modern part, while meeting or exceeding all the headline specs for a 2N3055, might actually have quite a bit of gain at 30MHz, and can bite you. After all, all you need to brand a device as 2N3055, is have a peak collector current of 15A, gain greater than 3 at that current, and low leakage at 60V, which is very easy today with most modern process plant. But the data sheet does not reflect this at all, as you do not get typical values for a lot of those parameters, just the minimum JEDEC spec for the part number.
-
That snippet is actually from the datasheet of a modern 2N3055 from OnSemi. That second column is a "Max" column, it was just too much faff to screen grab the table headers and integrate them into the upload.
Nevertheless, your point is well made.
-
Just out of curiosity I went and looked at the current OnSemi 2N3055AG data sheet. No mention of hfe at all. Just a gain bandwdith product spec (.8 MHz min to 6 mHz max for base part, 2.2 MHz min and 18 MHz max for a select part)
As always parts and their data sheets are a moving target, with traps for the unwary in any on going production situation.
https://www.onsemi.com/pdf/datasheet/2n3055a-d.pdf (https://www.onsemi.com/pdf/datasheet/2n3055a-d.pdf)
-
Lenovo and Dell use jellybean transistors as thermal sensors on their desktop PC's
-
Lenovo and Dell use jellybean transistors as thermal sensors on their desktop PC's
Yes, the physics does not change at all for the part being cheap. Going to be accurate to under 5C straight out the box with only the reference voltage of the ADC chip, good enough for a PC, and with a little bit of correction per batch you can get to 1C easily.
-
Lenovo and Dell use jellybean transistors as thermal sensors on their desktop PC's
Yes, the physics does not change at all for the part being cheap. Going to be accurate to under 5C straight out the box with only the reference voltage of the ADC chip, good enough for a PC, and with a little bit of correction per batch you can get to 1C easily.
It's got nothing to do with the hFE though.
https://www.sensortips.com/featured/get-temperature-sensor-transistor/ (https://www.sensortips.com/featured/get-temperature-sensor-transistor/)
-
Yes, the physics does not change at all for the part being cheap. Going to be accurate to under 5C straight out the box with only the reference voltage of the ADC chip, good enough for a PC, and with a little bit of correction per batch you can get to 1C easily.
A good delta Vbe temperature measurement circuit can achieve better than 1C with even a cheap transistor and no calibration. Accuracy is primarily limited by the accuracy of the current ratio used for measurement, and series resistance including base spreading resistance. The error contribution from series resistance can be removed with a Vbe temperature measurement using three currents but this is rarely done because keeping the measurement current low has the same effect.