| Christer |
There are many BJTs that come in different Vce ratings but seem
otherwise to be the same eg. BD135/137/139, BC546/547.
Except for the Vce rating, there usually seems to be no other
differences and prices differ only marginally or not at all. A higher
Vce should imply a different doping of the BC junction so one
could expect other parameters to be affected. While most
datasheets, like for those BJTs above, are often rudimentary, I
did managed to find some japanese BJTs with more detailed
datasheets and that come in different Vce ratings. Even though
further info like Cob and fT was specified, there was no hint at
any differences apart from the Vce rating.
If nothing else differes and prices are the same, why have
different models? There's got to be a catch somewhere.
Is there any reason to prefer high or low Vce over the opposite
when the low Vce version is sufficient?? |
|
|
| Netlist |
Christer,
good question.
I guess it might have something to do with production costs. Also, the price we pay when buying a small amount of devices is totally different from what big companies pay for large amounts.
Looking at the BD135-series spice parameters, as an example, the values are all the same for the higher Vce rated BJT’s which confirm your findings with the datasheets.
When buying transistors for repairing electronics I usually select the one with the highest Vce.
One fits all is the keyword. |
|
|
| Christer |
Yes, probably prices differ much more in large quantities, but the
question then still remains, are there any differences, or do the
manufacturers just test the decvices and label them according to
manufacturing variations? If so it might explain why they would
charge more for higher Vce devices. It still doesn't make much
sense for retailers to stock all versions if they anyway charge
almost the same price for all types. It should be cheaper for
them then to stock just the highest rated type.
Also, supposing it is just about manufacturing variations in
breakdown voltage, wouldn't this still have implications also
for other parameters? |
|
|
| Netlist |
Again, taking the BD-series as example, the datasheet reveals a difference in max Hfe between 135 and 137/139 serie.
Looking at the BC556-series, a difference in noise figure is noticed.
Take a look at the 2N5550 and 2N5551. Basic parameters only show a difference in Vce.
The datasheet however list other minor variations between them.
So, in critical design environments these small changes other the Vce will probably be taken into account.
All datasheets I compared were from Fairchild. |
|
|
| Christer |
Hugo,
it seems you have found more detailed datasheets than I have
managed. Which brands were you looking at? I checked the
datasheets for at least three or four brands for the BD135/137/139. |
|
|
| sreten |
As far as I'm aware high beta type transistors have low Vce.
So I'd guess typical hfe for the higher voltages is lower.
:) sreten. |
|
|
| Christer |
| quote: | Originally posted by sreten
As far as I'm aware high beta type transistors have low Vce.
So I'd guess typical hfe for the higher voltages is lower.
:) sreten. |
Yes, I thought too that might be one reasonable consequence,
but one would think the datasheets should make a note on that
difference.
Hm, actually when rechecking the datasheets for BC546/547
I now note there is no A version of 547 and no C version of 546,
and for the unqualified versions the hfe ranges are different. Still,
there is both 546B and 547B, both with the same hfe range, and
I have not found any other difference between 546 and 547 than
the voltage rating.
Maybe it is reasonable to assume that lower-Vce devices have
higher hfe in general? |
|
|
| Christer |
I have been thinking some more about this and looked in my
semi. physics book and it still puzzles me.
It seem not at all clear that there should be any relationship
between hfe and breakdown voltage. My book does actually
give a relationship between Vcbo and Vceo which depends on
hfe but it is rather brief and unclear about this. Further, I think
all datasheets I have seen usually have a simple relationship
where Vcbo is a few volts higher than Vceo, typically 5V, which
goes against that hfe should be involved, at least to any
degree worth bothering about.
Vcbo depends on the doping concentration in the collector only
as far as I understand, with the exception of some transistors
where there is an effect called punch-through, which seem more
of an exception than a rule.
Hfe, on the other hand, does not depend on the collector doping
at all, but only on the dopings of the base and the emitter and
the base width. There seems to be no relationship at all between
Vcbo and hfe, as far as I can see.
Remains the case with punch-trhough then. This is said to occur
in transistors with low base doping and a small base width,
and causes a breakdown to occur for other reasons and at a
lower voltage than in the usual case. In this case there could
be a relationship since the base width is involved also in the
breakdown voltage. However, this seems not to be applicable
to most BJTs.
One might also speculate about the manufacturing tolerances
in doping concentrations etc. and whether this may have some
effect. It is difficult to control the exacat doping concentrations
and also the base width, I suppose. However, althought this
causes some interdependence it seems not obvious it should
lead to any relationship between hfe and Vcbo except when
punch-through is involved.
Surely there must be some people who know the answer??? |
|
|
| Christer |
It is strange, people often consider me to be too theoretical, but
when I occasionally ask very simple and practical questions, nobody
seem to have an answer. :confused: ;)
Let me rephrase my original question then:
Suppose you have the choice of two BJTs which are described
in the same datasheet and the only piece of information that
differs between them in the datasheet (apart from the parts
number) is the max. rating for Vce adn Vcb. Also suppose the
Vce rating of both these devices is more than sufficient for
your purpose and that the price is the same. Which of these
two devices do you buy and why?
Let's also assume this is for a new design and that we are not
considering that any left-over devices may be more useful for
future projects if the are higher rated. |
|
|
| jcarr |
Christer:
In my own experience, BJTs that are identical on the data sheet other than the maximum Vce are likely to show some statistical differences if you actually sit down and measure them in large quantities (hundreds or more).
From what I have seen, the tendency is to get more high-hFe devices from the lower Vce BJT. From memory, the capacitance and ft also statistically come out a little better with the lower Vce BJT.
hth, jonathan carr |
|
|
| wrenchone |
| If you can, try to get your hands on a data book that shows the process type for each transistor. Fairchild used to do this in their small signal data books. Siliconix also showed data for each of their process types in their old data books. Depending on the device application, the geometry will change. For high speed saturated devices, gold doping may be added to reduce carrier lifetime, though that usually doesn't show up in the process sheet. |
|
|
| Christer |
Jonathan,
Thankyou, that was one of the type of answers I had hoped for,
either some experience of measuring large quantities of devices,
as in your case, or a theoretical argument about what should
differ and why.
Wrenchone,
Yes, I suppose someone with enough knowledge of semiconductor
physics might be able to make something out from the process
info. However, I didn't intend diving that deep into the matter.
I was rather after whether one can draw some general conlcusion.
At least with the common and easily available BJTs here in Europe
there are many cases of devices which seem to differ only in the
voltage rating, BC546 vs. 547, BD135 vs. 137 etc. Since prices are
usually the same or differ only very little, it is a purely practical
matter when shopping components: Can I make an informed
choice between the alternatives, despite no such information
appearing in the datasheets, or should I jut toss a coin and be
happy? :) |
|
|
| jcarr |
Christer:
Have you already made the pcb layout? If not, my hands-on experience has been that small-signal BJTs from Japanese companies like Toshiba, Hitachi, NEC et al show less statistical deviation regarding the characteristics than small-signal BJTs from European or US companies like Motorola, Philips, Fairchild et al. This is true even when the devices come from a specific Hfe-range group.
However, with European devices frequently the base will be the center pin, while more Japanese devices have the base as one of the outer pins. Hence the pcb layout also needs to be a little different.
hth, jonathan carr |
|
|
| wrenchone |
| The process sheets really don't go into much semiconductor physics mumbo-jumbo. There is generally a picture of the device layout, description of typical application, graphs of important typical device charateristics, and a list of devices in that particular product family. I would think it's just what you're looking for if you want to pigeonhole a device for a given application. However, I haven't seen it done for European devices. That's not to say that the information isn't out somewhere. |
|
|
| millwood |
| quote: | Originally posted by Christer
It is strange, people often consider me to be too theoretical, but
when I occasionally ask very simple and practical questions, nobody
seem to have an answer. :confused: ;) |
simple questions are usually difficult to answer. That's more a rule than an exception.
I think what happened at the factory is that they test transistors (or wafers) and see for that particular transistor/wafer, what the Vce is and then lable accordingly.
Kind like how mil/aero spec ICs are marked, and how CPUs are marked. Intel / AMd didn't set out intentionally to produce slower CPUs just some of them turned out to be slower than others.
Those I would expect the variance on transistors to be far smaller than that for CPUs. but it is one possibility. |
|
|
| Christer |
Jonathan,
Thanks for the added info. My question was not so much prompted
by any specific project, but rather as a general question about
this situation since it tends to arise quite often when buying
from the local suppliers. Otherwise, it seems most, or at least
many, of your experienced collegues on the forum also consider
japanese transistors generally superior. I have also found them
to usually have much more informative datasheets. The sad problem
is that they are so difficult to buy here. Our major local supplier
has been cutting down on those few types they used to have.
I recently bought what apparently was the last three they had
in stock of one type, and their reason for not selling this product
anymore was claimed as frequent and serious delivery problems
from their distributor. Another local distributor gave me Indian
second-source (with unkonwn compatibility) and counterfeits also
seem a common problem. There are still European distributors
of many of the more common devices, but that usually brings
added costs and minimum orders etc, and is not an option if
one needs only a few devices just for a prototype. It is better
then to wait until one has a larger order, and use more easily
available types for first prototypes. Your point on different pin-outs
is worth noting here, of course.
I would imagine many other european DIYers, and probably also
professional engineers, often ask themselves the same question
as I did in this thread.
Wrenchone,
Yes, thanks for your patience and helpfulness. Most of the devices
where this problem arises are not the top-of-the line devices,
but good industry-standard devices, so in a really critical
application it is probably better to eventually try getting one
of the many superior, but harder-to-buy devices. |
|
|
|
