I have been in this game for 50 years plus and I am at a loss for the expression Quasi saturation? What do you mean by this ! I know what a quasi output stage is and I know what a quasi complimentary pair is
Regards Trev
Regards Trev
Saturation is as I am sure you know when a transistor's collector (we're talking BJT's here) voltage reduces to a point where it does not go (much) lower. This may be in the region of 0.2V depending on the current, and current rating of a transistor.
If you now take a look at the output characteristic of a modern high voltage power transistor (and other devices not just high power) then its output current is relatively flat at voltages over 5V. But as the voltage falls below about 5V the collector current reduces. For example, if the current is 4A at 5V then at 2V it might have fallen to 2A. It is because the local base-emitter voltage has reduced, causing the effective gain to fall. You can boost the current back up to 4A by increasing the bias and base current. So this effect seems like a saturation effect, but is not the "hard" saturation voltage which may be in this case about 0.5-1V.
It is called quasi-saturation as a result, but its origin is in the high resistance epitaxial layer used in the collector to achieve high voltages (and high second breakdown voltages) which de-biases the base at higher currents, reducing the gain at lower voltages.
To see this, use a dual-channel oscilloscope set up to measure the IV characteristics of a transistor, put a device on a heatsink and observe the I:V results for a fixed base current over a collector voltage swing of 0-10V.
A family of curves produces a locus of inflexion points from the origin to maybe 2 to 4 V as though a resistor, which is called the quasi-saturation resistance.
This did not happen, BTW, with the original RCA hometaxial 2N3055. That was one of the advantages of that transistor, but today's high speed devices have higher second breakdown performance so even with their quasi saturation, out-perform it.
If you now take a look at the output characteristic of a modern high voltage power transistor (and other devices not just high power) then its output current is relatively flat at voltages over 5V. But as the voltage falls below about 5V the collector current reduces. For example, if the current is 4A at 5V then at 2V it might have fallen to 2A. It is because the local base-emitter voltage has reduced, causing the effective gain to fall. You can boost the current back up to 4A by increasing the bias and base current. So this effect seems like a saturation effect, but is not the "hard" saturation voltage which may be in this case about 0.5-1V.
It is called quasi-saturation as a result, but its origin is in the high resistance epitaxial layer used in the collector to achieve high voltages (and high second breakdown voltages) which de-biases the base at higher currents, reducing the gain at lower voltages.
To see this, use a dual-channel oscilloscope set up to measure the IV characteristics of a transistor, put a device on a heatsink and observe the I:V results for a fixed base current over a collector voltage swing of 0-10V.
A family of curves produces a locus of inflexion points from the origin to maybe 2 to 4 V as though a resistor, which is called the quasi-saturation resistance.
This did not happen, BTW, with the original RCA hometaxial 2N3055. That was one of the advantages of that transistor, but today's high speed devices have higher second breakdown performance so even with their quasi saturation, out-perform it.
Last edited:
Hi Trev,
I don’t consider myself to be a device physics guy, but here is my take on quasi saturation in BJTs. I believe the term quasi saturation, as being used here, is a bit of a misnomer. In rough terms, it is beta droop due to a combination of high current and low collector-base voltage that is exacerbated by light collector doping. It is better described as beta degradation caused by base push-out. In essence, the physics and effective structure of the device change as a function of current and base-collector voltage so as to change the nature of the collector-base depletion region.
Most understand hard saturation, where the base-collector junction becomes significantly forward biased, significant current flows from base to collector, effective beta goes very low and collector-emitter voltage goes as low as it can.
I consider quasi saturation to be in the range where there begins to be a forward bias of the base-collector junction such that some current begins to flow from base to collector, reducing beta. This region is a matter of degree. It may begin where there is a relatively small forward bias of the junction, where the base-emitter voltage exceeds the collector emitter voltage by a few tenths of a volt, and maybe microamperes will begin to flow from base to collector, detracting from beta. High collector resistance due to a lightly doped collector will cause an increased internal voltage drop in the collector region at higher collector current levels that will hasten the forward biasing of the collector base junction. I associate saturation or quasi-saturation with forward biasing of the collector-base junction that will cause some, perhaps small, current to flow from base to collector. Some will understandably disagree with this definition of quasi-saturation. I would say it is convenient slang to associate reduction of beta at low Vce and high current with quasi saturation.
However, what we are talking about here begins to occur at higher collector-base voltages where the collector-base junction has not yet become forward biased in traditional terms. Rather, beta has decreased (drooped) due to changes in the effective base-collector junction geometry and depletion region. Think of it as the base width becoming larger, which is usually associated with a decrease in beta, completely apart from base current flowing into the collector. That is why the term base push-out comes into play and that the term quasi-saturation is a bit of a misnomer. High current corresponds loosely with the phenomenon called high injection, and that in turn relates to a high concentration of minority carriers in the base region. The Kirk effect, which we often associate with ft droop in power transistors at high current, is also involved here.
At the same time, light doping of the collector (to achieve high breakdown voltage and low capacitances) makes it easier for the base to push out into the collector region under high injection conditions, making it easier for the effective width of the base to increase. Higher doping of the collector will reduce these effects, but will usually result in lower breakdown voltage and higher capacitances.
Again, by no means am I an expert on device physics. This is just my way of understanding the phenomenon that is loosely termed quasi saturation. I’m sure there are others here who could explain it much better and with more detail and precision.
Cheers,
Bob
Thanks Lads !
From both your replies I now understand what you are on about I came across this effect some 40 years ago while working on moving coil head amps where we were in fact using very low and I mean low collector voltages But I have never heard it referred to quasi saturation
its been an honour Me Ellis/ Mr Cordell
Trev
From both your replies I now understand what you are on about I came across this effect some 40 years ago while working on moving coil head amps where we were in fact using very low and I mean low collector voltages But I have never heard it referred to quasi saturation
its been an honour Me Ellis/ Mr Cordell
Trev
The person who started this thread wrote "I've got an application in mind for an NPN bipolar, running about 10mA, at low VCE" right at the beginning of post #1. Low collector voltage circuits benefit from transistors without quasi saturation.
Post #1 shows some oft-recommended transistors, many of which (surprisingly!) exhibit quasi saturation at the desired collector current. Post #15 shows one that has so little quasi saturation, the plot legend says "wow!"
Post #1 shows some oft-recommended transistors, many of which (surprisingly!) exhibit quasi saturation at the desired collector current. Post #15 shows one that has so little quasi saturation, the plot legend says "wow!"
Sure it would be great to have them curve traced; I'll PM you.
There is a thread here about problems getting the comp. I think someone bought the
lot from Rochester that I was talking about. A few people on that thread were talking
about buying them, there were close to 1000.
There is already an abudant rather than scarce supply of terms used to imply the same thing. Who coined the dopy expression "quasi saturation"? Must have been someone not well-versed in theory, hence not having the skills needed to make a meaningful contribution to the subject. Very unserious.
Rochester has O-rank and they have Y-rank in tubes. I would prefer Y-rank in a bulk bag.
With tubes, characteristics are tightly bunched. So, you may get zero matched pairs with your complements.
I have found with bulk bags, the characteristics are spread out more and you have more likelihood of finding matched pairs with the batch of complements.
Just my observation from curve tracing multiple hundreds of parts. When I am tracing a batch, I record every measurement and store them individually so that I can sort matched pairs.
I was told, probably 30 years ago, that semi companies "set up" the fab line to do
a run of wafers for a given part type. I don't know how many they make but I'd
assume it is based on some minimum and perhaps anticipated large orders. Those
parts go out to distributors and it could be several years before it comes around
again to make that particular part number. Obviously, parts that are discontinued
are not in the lineup for production.
N101- Quasi saturation term has been around for at least 30 years, though it might have been described as base push-out I guess. As Bob pointed out the Kirk effect is different but related.
It may not have been so commonly understood as that as it took a while for device simulation tools to catch up. AFAIK the first simulator to include this was when Colin McAndrew published his VBIC model which I think was in 1995.
What other terms have you heard to describe quasi-saturation?
It may not have been so commonly understood as that as it took a while for device simulation tools to catch up. AFAIK the first simulator to include this was when Colin McAndrew published his VBIC model which I think was in 1995.
What other terms have you heard to describe quasi-saturation?
Last edited:
Why invent smart names. Name it "Low Vce".
Saturation is only important in switching when you look for low on voltage or fast turn off.
Saturation is only important in switching when you look for low on voltage or fast turn off.
Not really. Low Vce does not necessarily mean that a transistor is in quasi-saturation or Kirk effect regime!
The problem with quasi saturation is that it affects what would be normal operating region and reduces the gain, increasing the distortion as the output voltage reaches towards the supply rail. Especially, therefore, in circuits running the transistors at perhaps 12V supply rails for example.
It is not a "hard" saturation effect that slows down switching a lot.
The problem with quasi saturation is that it affects what would be normal operating region and reduces the gain, increasing the distortion as the output voltage reaches towards the supply rail. Especially, therefore, in circuits running the transistors at perhaps 12V supply rails for example.
It is not a "hard" saturation effect that slows down switching a lot.
Thankfully the curve tracer plots attached to the first seven dozen posts in this thread, have given me some good ideas about which transistors to use and which to avoid, for good behavior during low Vce operation. If I studiously avoid the bad ones then I don't need to work out how to model quasi-saturation: in my little sub-universe of devices, it doesn't exist!
This is playing with words.
Of course there is a distortion increase, of course it doesn't slow down switching. So what.
Yes, in fact, both the beta drop off and ft drop at low Vce and high injection is due to increased transit time, if I recall.
In any case, the physics are not straightforward for those not steeped in device physics.
Cheers,
Bob
That's true, the model also involves a large excess charge in the base. I do know that the models circa 2005 were remarkably good at modeling gross overload recovery, etc.
Example from Mark J's sub universe selection
The On Semiconductor datasheet for 2SC6043 gives an fT figure for Ic of 10 m.A. with Vce equal to 2 V.
This device is one graphed in an attachment to post 15 on page 2.
How many of the others in the "sub universe" selection have datasheets with this sort of detail.
The On Semiconductor datasheet for 2SC6043 gives an fT figure for Ic of 10 m.A. with Vce equal to 2 V.
This device is one graphed in an attachment to post 15 on page 2.
How many of the others in the "sub universe" selection have datasheets with this sort of detail.