Isn't a bipolar transistor considered to be fully saturated once the collector voltage falls basically one diode drop below the base voltage? If so, that seems like a much simpler metric to go by.

One diode drop below the BaseoOn voltage is 0V. Maybe I do not understand.
There are more than one defecation. I don't want to get into which is best.

 

Isn't a bipolar transistor considered to be fully saturated once the collector voltage falls basically one diode drop below the base voltage? If so, that seems like a much simpler metric to go by.

The collector will almost never drop a diode below the base voltage as that would put Vce at 0 V.

One seemingly well-defined metric is that the boundary between active and saturation is when the Vcb = 0 V. But while this looks nice on paper, the transition between active and saturation is a smooth one and, from a practical standpoint, the device is still behaving much more like it's in the active mode than saturation at this point and continues to do so for some point further, usually until the base-collector junction is forward-biased by a couple hundred millivolts, at which point the behavior starts changing quickly. But even this is not definite and if the device is being pushed hard the transition can noticeably start even while the collector-base junction is still distinctly reverse-biased.

 

There are more than one defecation.

You gotta love spelling correctors!

 

What basis do you have for concluding that it probably stays flat after Ic/Ib = 10?

There is nothing magical about a forced beta of 10. Power transistors are typically speced in saturation at a value of 5 or even less.

It is a convention for spec comparison that dates back to the earliest days of commercially-available discrete small-signal transistors and is likely do to one manufacturer happening to choose it for their spec'ed behavior and other manufacturers following suit just so they could compare and contrast their transistors to the others.

Hi,

Did you miss the line after that:
"At some point however it (Vce) may rise with increased base current simply because the BE voltage will rise more and more."

The magical part came in from the graph you posted. Notice they halted the plot exactly when Ic/Ib was equal to 10.
That does not mean that other transistor data sheets can't be different.

For my own purposes, I only rely on measurements and safety margins. If I see it saturate at 20 I'll probably shoot for 10. If I see it just going into saturation at 10, i'll go lower, maybe even that 5 you were quoting.

 

I could teach for days on this topic.
Ib1 is Base on current, say 5A. Then I have Ib2 Base turn off current = about Collector current -25A.

Here is a Base drive circuit that I have used many times. "Proportional base drive." When Q1 is on current flows C-E, around the transformer 1 turn. There are 10T E to B. The 0 to 25A in the C will force o to 2.5A in the Base. 10/1 (current transformer)
The turn off is done on the 100 turn winding. It takes 2.5A there to discharge the Base charge at 25A.
View attachment 361434
Note the high voltage transistors have a B-E voltage of about 1V not 0.65V like low voltage transistors.

It sounds like you are talking about something else though when you talk about sweeping the charge out of the base region.
Although that's a good thing to discuss as well, it's not the same as trying to get the transistor INTO saturation in the static sense. Reverse bias is to get the transistor OUT of saturation as fast as possible, which can be done with a short pulse sometimes referred to as "a snap off pulse".

Have you ever used a Baker Clamp or similar?

 

Have you ever used a Baker Clamp or similar?

I have used Baker Clamps but much of my work was 800V through 1500V. It can be hard to find good diodes that match the transistor. Many HV diodes have more than one die inside. Over the 40 years I did this there were times when there were no HV diodes with one diode drop. I went to using B-C junction of a transistor to make a diode. Those were desperate times. We went to Motorola transistor group, and they made us diodes on the transistor line. We also had a good relationship with Hitachi Power Semi and ST Micro for HV transistor and diodes.

I love Baker Clamps at low voltage.

We made things at very high volumes and thus need circuits that work when the date codes change on the transistors. For decades we changed the Base drive every time the transistors changed date codes. (changed a resistor or add a pot)
For a while I watched the power loss and adjusted the Base Drive for least power loss. While that was wonderful it cost money. Readjusted the Base current every 30 seconds as the unite worked. (micro controller)
Next, I did the adjustment at the factory not in the field, via a micro controller. Once set there was no need to readjust.

Another reason I did not use Baker Clamps at high voltage is the C-E voltage drop. My current might ramp from 0 to 25A and the clamp holds the C-E to 0.5V for the entire cycle. The Proportional Base Drive holds the transistor at a fixed gain not a fixed Vc-e. The Vce starts out at 0 then ramps up to 0.5V. There is some power savings.

 

Hi,

Did you miss the line after that:
"At some point however it (Vce) may rise with increased base current simply because the BE voltage will rise more and more."

The magical part came in from the graph you posted. Notice they halted the plot exactly when Ic/Ib was equal to 10.
That does not mean that other transistor data sheets can't be different.

For my own purposes, I only rely on measurements and safety margins. If I see it saturate at 20 I'll probably shoot for 10. If I see it just going into saturation at 10, i'll go lower, maybe even that 5 you were quoting.

So, because that data sheet chose to stop at a forced beta of 10, that makes 10 a magical value???

 

So, because that data sheet chose to stop at a forced beta of 10, that makes 10 a magical value???

Hi,

Well what reason do you give for them stopping at the exact value of 10 ?
Can it really be an arbitrary value, given that ALL of the graphs are like that. If even one stopped at another value I might have doubts, but I don't think they were stupid or anything like that.
It ends up being a compromise that apparently many designers are aware of, hence the data sheet rendering. There's also a switching time concern deeper in saturation.

The operation gets a lot more complicated at saturation, so it's hard to suggest that every single transistor under the sun is going to observe this rule of thumb. It we drive it too hard we waste power, not enough and we don't get adequate Vsat levels. If Vbe rises too much, which actually raises the Vsat level.


Who knows, maybe they used 'ai'

 

Hi,

Well what reason do you give for them stopping at the exact value of 10 ?
Can it really be an arbitrary value, given that ALL of the graphs are like that. If even one stopped at another value I might have doubts, but I don't think they were stupid or anything like that.
It ends up being a compromise that apparently many designers are aware of, hence the data sheet rendering. There's also a switching time concern deeper in saturation.

The operation gets a lot more complicated at saturation, so it's hard to suggest that every single transistor under the sun is going to observe this rule of thumb. It we drive it too hard we waste power, not enough and we don't get adequate Vsat levels. If Vbe rises too much, which actually raises the Vsat level.


Who knows, maybe they used 'ai'

I have already explained where the value of 10 originated and why and how it became the defacto specification.

 

I have already explained where the value of 10 originated and why and how it became the defacto specification.

Oh I am sorry, I asked the question wrong maybe?

I meant to ask:
What reason do you give for them possibly to NOT be using the gain of 10 idea (instead of them giving that gain of 10) in the data sheet.

If the answer is because sometimes we want to go lower, I can't argue with that, but the gain of 10 idea can't possibly be a hard-and-fast rule it's just a rule of thumb. It does have a feeling of magic that seems to be associated with it, but when we use the word magic we are naturally going to have to define that in reference to technology, and in doing that we come up with the rule of thumb. We know that rules of thumb are just that when it comes to almost everything because they are heuristics, and heuristics work in many cases but could fail in some, with the 'some' considered to be a small number of failure cases contrasted with a large number of success cases ... just ask any chess grandmaster.

So, the better description of the gain of 10 rule of thumb is that it is a 'heuristic', which sometimes translates roughly to 'magic'. It's magical that it works in so many cases.

 

It's magical that it works in so many cases.

How is it magical? As I understand it, raising the ratio gives you a higher Vcesat, lowering it gives you a lower Vcestat. Are you claiming that it is somehow magically optimal in many cases? Because I gave seen nothing that supports that. Not magical, just a reasonable compromise, where Vcesat will be at a level that is acceptable for most applications.

It is like saying there is something magical about boiling macaroni for 7 minutes because most people will find it acceptable at that time, even though I prefer it more al dente at 5 minutes, and my wife prefers it softer at 9 minutes.

 

It's magical that it works in so many cases.

What is “magical ” about this?
The BJT is in saturation as soon as the collector voltage VC falls below the base voltage VB.
This opens the base-collector junction and thus drastically increases the current into the base node.
The "enlargement factor" depends on the “normal” B factor and the base-collector doping and can only be estimated.
The series resistor RS, which is common in switching operation, must now be dimensioned so that - with a known switching voltage - it ensures the necessary voltage VB at the base node - and thus also the required condition VB>VC.
Since the value of the increased base current is not known without prior measurement, it is common practice to take a safety factor into account to ensure saturation.
This consideration leads to the rule of thumb that IB is considered equal or even somewhat larger than IC/10 (to be copared with IB=IC/B for linear operation and B >>10 in most cases).

 

How is it magical? As I understand it, raising the ratio gives you a higher Vcesat, lowering it gives you a lower Vcestat. Are you claiming that it is somehow magically optimal in many cases? Because I gave seen nothing that supports that. Not magical, just a reasonable compromise, where Vcesat will be at a level that is acceptable for most applications.

It is like saying there is something magical about boiling macaroni for 7 minutes because most people will find it acceptable at that time, even though I prefer it more al dente at 5 minutes, and my wife prefers it softer at 9 minutes.

Hi Bob,

I explained that already in my previous post. A heuristic is sometimes viewed as a magic formula because it skips a lot of detail and works 9 times out of 10 or even better. If you had 99 cases that worked and only 1 that didn't using that heuristic, that might be deemed magical in a sense.
This does not mean that the fairy Godmother of transistors had a hand in it, it's not that kind of magical.
Oh wait, to be more PC I better rephrase that:
This does not mean that the fairy Godperson of transistors had a hand in it, it's not that kind of magical

However, you are free to disagree, but in doing so I would hope you could explain why you disagree based on some of the examples we had been looking at, such as that data sheet that shows EVERY curve halting at EXACTLY Ic/Ib=10.

I am not saying you are wrong, at least not just yet, but why then do you think they stopped every single curve at Ic/Ib=10?
I am also not saying that you have a good reason for this, but just that I'd like to hear what it is.

 

What is “magical ” about this?
The BJT is in saturation as soon as the collector voltage VC falls below the base voltage VB.
This opens the base-collector junction and thus drastically increases the current into the base node.
The "enlargement factor" depends on the “normal” B factor and the base-collector doping and can only be estimated.
The series resistor RS, which is common in switching operation, must now be dimensioned so that - with a known switching voltage - it ensures the necessary voltage VB at the base node - and thus also the required condition VB>VC.
Since the value of the increased base current is not known without prior measurement, it is common practice to take a safety factor into account to ensure saturation.
This consideration leads to the rule of thumb that IB is considered equal or even somewhat larger than IC/10 (to be copared with IB=IC/B for linear operation and B >>10 in most cases).

Hi there,

Check out post #33 so I don't have to rewrite it all again. If you prefer though, I can copy and paste it into another reply to you.
It could be simply that we have a different idea about what 'magical' means.

If you look at the data sheet that was posted previously, you can see they stopped each and every curve at Ic/Ib=10. Why do you think they did that?

 

If you look at the data sheet that was posted previously, you can see they stopped each and every curve at Ic/Ib=10. Why do you think they did that?

I think the answer is simple:
Because - from many experiments - one knows that Ic/Ib=10 is a good and well-proven rule of thumb.
Of course, it would not a problem to include curves for some other (additional) parameters like IC/IB=6, 7, 8.....20.
The question is if such an increased set of curves would have any practical value.

More than that, I think that a similar consideration applies for all functions which are given in the data sheets.
For example, the input characteristics IB=f(VBE) or the transfer characteristics IC=f(VBE) are also given for some (arbritrary) typical VCE values only (as a parameter).

 

one knows that Ic/Ib=10 is a good and well-proven rule of thumb.

I have use high voltage high current transistors that are well known for their low gain.
I have used Galamian Germanium transistors that are known for their low gain. I used a transistor at a gain of 2 because that is what it took. (40 years ago)
Every time someone states a rule, some donkey finds an exception.
edited

 

Galamian??
You mean Germanium?

Ahh… those old germanium power transistors. Brings back fond memories. They had leakage currents measured in milliamps at high temperatures.

 

Hi there,

Check out post #33 so I don't have to rewrite it all again. If you prefer though, I can copy and paste it into another reply to you.
It could be simply that we have a different idea about what 'magical' means.

If you look at the data sheet that was posted previously, you can see they stopped each and every curve at Ic/Ib=10. Why do you think they did that?

Yes, different ideas of what would make a value "magical". Magical implies something unique or special about that number such that you would expect unrelated parties to come up with it independently.

An example would be the zener diode voltage of about 4.7 V at which the temperature coefficients from zener and avalanche breakdown cancel out. There is something magical about that particular value, something that people that had no idea of what the underlying mechanism is would still be expected to discover the behavior.

There is no such magic for a forced beta of ten. It is merely a comparison point that evolved historically, probably mostly driven by Motorola's choice to us it in some of the first data sheets that specified test conditions for saturation parameters.

Had they chosen a value of 12, or 15, or 8, there's a fair chance that that would have become the number that so many people think has some magical property.

As for why that particular data sheet stopped at a value of ten -- the fact that they showed data for any values other than ten is what makes them stand out. Device manufacturers don't like to provide any more specification data than they feel they have to. Characterizing a device is not cheap and the more data they publish increases the chanced of lawsuits down the road if devices that don't meet all of them cause a problem. They went to ten because that is the value that they needed to include in order to be able to compare their specs to everyone else's. That they included curves for 20 and 50 indicates that they believe there is sufficient interest amongst their customers for designs that don't push so hard into saturation, either for power or for speed considerations.

 

Yes, different ideas of what would make a value "magical". Magical implies something unique or special about that number such that you would expect unrelated parties to come up with it independently.

An example would be the zener diode voltage of about 4.7 V at which the temperature coefficients from zener and avalanche breakdown cancel out. There is something magical about that particular value, something that people that had no idea of what the underlying mechanism is would still be expected to discover the behavior.

There is no such magic for a forced beta of ten. It is merely a comparison point that evolved historically, probably mostly driven by Motorola's choice to us it in some of the first data sheets that specified test conditions for saturation parameters.

Had they chosen a value of 12, or 15, or 8, there's a fair chance that that would have become the number that so many people think has some magical property.

As for why that particular data sheet stopped at a value of ten -- the fact that they showed data for any values other than ten is what makes them stand out. Device manufacturers don't like to provide any more specification data than they feel they have to. Characterizing a device is not cheap and the more data they publish increases the chanced of lawsuits down the road if devices that don't meet all of them cause a problem. They went to ten because that is the value that they needed to include in order to be able to compare their specs to everyone else's. That they included curves for 20 and 50 indicates that they believe there is sufficient interest amongst their customers for designs that don't push so hard into saturation, either for power or for speed considerations.

Hi again,

I agree somewhat with that, but in the first paragraph you seem to be asserting a criterion that I don't see as being very valid. Namely, "unrelated parties to come up with independently". I don't see that as a valid assertion because we see a lot that we might call magical just because a large number of people feel the same way. For example, celebrities. There's no natural selection process for that, it's just that a lot of people like certain individuals more than others. And we have to keep in mind that 'magical' is not a technical specification of any kind.

I can see by your discussion of the 4.7v zener voltage that you are looking for a magic parameter that comes from a natural processe while I don't hold to that stringent rule. I do agree that when a natural process does in fact produce such an interesting parameter or relationship that it is highly magical, but I just can't limit my own personal view that much. As I said though, I feel more comfortable now calling it a heuristic because the word magical is not as descriptive as I would like and subject to too much interpretation.

As to your last paragraph, I think that is very extravagant of you to suggest that the only reason they stopped at a value of 10 for every single plot was only because they felt like it, as if there was no logical reason for that, or wanted to save space on a written page. They want people to know how to use their devices and a data sheet is a way to advertise their product. It's interesting that you said that they went to the value of 10 in order to be able to compare their specs to everyone else's, but then the question becomes why did 'everyone' else choose the value of 10 as well. They could have went to 5 and that would still include 10 so there would still be a way to compare different devices. And just the fact that they all did it that way is kind of magical in a way.

The reason we disagree somewhat is because you are looking for a natural process that defines what 'magical' is while I don't need that stringent of a criterion. I believe that it is magical just because everyone is using it and find it works in so many cases, and it appears in a lot of literature.

I don't believe the value of 10 will hold for all cases. Temperature extremes, higher currents, things like that could affect the behavior as well. That's not what the value of 10 is for though and most designers know that. I guess I find it interesting that everyone uses that value.

So it seems to be a matter of natural process vs cultural behaviors, and that behavior may even have been born through the data sheets themselves so that manufacturers had a way to say that their transistor is saturated with Ib=Ic/10. That's the real reason we see the halt at a value of 10 for the Beta, they don't have to go any lower, unless of course they want to show more info for some devices. It's not only that data sheet either. If we look at the 2N2222A, we see the following:

CE SAT VOLTAGE:
Ic=150ma, Ib=15ma, 0.3v (the magical 10)
Ic=500ma, Ib=50ma, 1.0v (the magical 10 again)
BE SAT VOLTAGE: Ic=150ma, Ib=15ma, 0.6v to 1.2v
Ic=500ma, Ib=50ma, 2.0v

The test conditions show that the Ic/Ib ratio is 10 in order to perform the test. This is the same for the 2N4400 and similar, and also for the 2N5301 they also show those two test points.

So it also seems magical that it appears on a lot of data sheets as a test condition.

 

I think the answer is simple:
Because - from many experiments - one knows that Ic/Ib=10 is a good and well-proven rule of thumb.
Of course, it would not a problem to include curves for some other (additional) parameters like IC/IB=6, 7, 8.....20.
The question is if such an increased set of curves would have any practical value.

More than that, I think that a similar consideration applies for all functions which are given in the data sheets.
For example, the input characteristics IB=f(VBE) or the transfer characteristics IC=f(VBE) are also given for some (arbritrary) typical VCE values only (as a parameter).

Hi,

Yes, and it also seems to be an industry standard. A lot of data sheets show that value of forced Beta.