I was wondering if anyone could tell me how much margin should be given to Vceo in practical circuits. I have some boards with MPSA18s on them. My power supply runs a little high, around 43V. Vce on the input pair would be just a couple volts lower than that. Since Vceo for these devices is 45V, I'm concerned they will either eventually fry or they will have higher distortion on the brink of breakdown. Any thoughts would be appreciated.
Thanks.
Henry
Thanks.
Henry
As a rule of the thumb, usually the operating peak voltage would be about 75% of the maximum recommended by the manufacturer. If transistor in question is in class A, the peak will be when Ic goes to the minimum, then it approach Vcc, so, a VCC of 43V for a device rated to 45 is too high.
Good luck.
Good luck.
Thanks for the advice. Yeah, I agree. I'll put in a simple front end regulator to drop the voltage to a safe level.
The power transformers I got were too good a deal to pass up, but about four volts too high...
-Henry
...or try a higher Vceo transistor, there must be other low noise transistor compatible with your system.
Hi,
This is my first post - had to jump-in here before introducing myself properly.
The parameter that matters in this case is Vcbo. Never exceed this, but as long as there is a bias on the base then Vceo is irrelevant. The base current 'shields' the emitter junction, but there is no way to protect the base-collector junction.
This is my first post - had to jump-in here before introducing myself properly.
The parameter that matters in this case is Vcbo. Never exceed this, but as long as there is a bias on the base then Vceo is irrelevant. The base current 'shields' the emitter junction, but there is no way to protect the base-collector junction.
Interesting proposition, and quite logical too, even if it goes against some conventional wisdom.
I had to put it to the test.
Result: it does actually work, but the resistance seen by the base has to be extremely low, lower than 1 ohm in the case of the BD131 I tested.
This somewhat limits the practical usefulness of this trick in existing circuits, but it good to know nevertheless.
Vce0 is the limiting voltage when zero current flows. It is not a reference to whether the base is connected or not.
The "o" from Vceo means open. It would be possible to mimic that condition by connecting the base to a circuit that synthetizes an infinite impedance, but it looks like a self-defeating proposition.
AndrewT, Vceo is measured with a base open-circuit, and is the voltage at which the base region is effectively removed by sweeping the doping from the base region, leaving a resistor. This is why the current rises sharply. If the electrons/holes can be replaced by a current flowing into/from the base then the junctions remain correctly biased and no harm is done. As Elvee correctly states, this can require a low impedance drive to the base, but is a reasonable proposition in most circumstances.
It is very common for RF amplifiers (particularly class-C) to operate with collector voltages 2-3 times higher than Vceo, but as long as there is sufficient base drive no harm is done.
It is very common for RF amplifiers (particularly class-C) to operate with collector voltages 2-3 times higher than Vceo, but as long as there is sufficient base drive no harm is done.
Agree fully - it is a figure of merit, independent of how the device might be used.
If the base of a transistor was connected to the emitter, and a potential applied between the collector and emitter, which junction will fail first? The B-E one can't as it is shorted (the low impedance in Elvee's example), but the C-B one sees the full potential.
Fortunately both breakdown mechanisms are fully reversible as long as there has been no thermal damage.
You need to look at the SOA datasheet curves to really understand the Vce curves and what happens to get the Vceo figure.
i've seen everything from 30% safety margin to 400% safety margin in the selection of Vceo for diff amp transistors. just stay away from KTC/2SC3200's, as they have been known to get leaky quite often, and have bit every manufacturer that uses them with high failure rates. 2SC2240's on the other hand have identical specs (120V Vceo for instance) but are much more reliable.
It is mostly of importance with power devices switching inductive loads.
Try this:
Saturate an NPN device with an inductive collector load, until a defined collector saturation current (eg 100mA) is reached (due to R of inductance plus extra R, or constant current Ic)
Open the base.
Collector will fly to the breakdown voltage, but try and maintain the 100mA current.
This will be the Vceo value. Do it for a "long time", ie not a short controlled pulse and the device gets whacked. SOA limit.
I think this is all a bit academic for low power small sig applications.
Try this:
Saturate an NPN device with an inductive collector load, until a defined collector saturation current (eg 100mA) is reached (due to R of inductance plus extra R, or constant current Ic)
Open the base.
Collector will fly to the breakdown voltage, but try and maintain the 100mA current.
This will be the Vceo value. Do it for a "long time", ie not a short controlled pulse and the device gets whacked. SOA limit.
I think this is all a bit academic for low power small sig applications.
Not quite: for a VAS, or a driver, this means you can operate all the way to Vcb, provided the drive circuit has a low enough impedance.
That's the crux of the problem: the impedance has to be really low for things to remain under control, and a circuit which has not been designed from the start with this requirement in mind will fail to deliver.
Interestingly, having raised the importance of Vcbo I decided to check a few random datasheets. For devices like the BC848 Vcbo is almost the same as Vceo, so the base drive offers limited improvements. I'm rather surprised by this, as for RF devices where Vceo may be as low as 4-5Volts, the Vcbo can be over 10Volts.
I guess that was my point, though not well expressed!
As long as the BE impedance is low, the device cannot be pushed into second breakdown.
http://www.onsemi.com/pub_link/Collateral/AN1628-D.PDF
is a good treatise, the figures on page 4 show the key relationships.
Even when the advertised voltages are close, the actual voltages can be substantially different.
I just measured some BC548 samples:
Philips:
1/ 62V 145V
2/ 59V 144V
3/ 59V 146V
4/ 60V 132V
5/ 65V 131V
Fairchild:
6/ 58V 81V
7/ 54V 73V
TFK:
8/ 77V 143V
Moto:
9/ 68V 142V
I just measured some BC548 samples:
Philips:
1/ 62V 145V
2/ 59V 144V
3/ 59V 146V
4/ 60V 132V
5/ 65V 131V
Fairchild:
6/ 58V 81V
7/ 54V 73V
TFK:
8/ 77V 143V
Moto:
9/ 68V 142V
These parameters are minimum test LIMITS, not design objectives!
When the fabrication process goes well, a large proportion of the product will well exceed the limits, UNLESS it is a very critical and "difficult" parameter, in which case the manufacturer will struggle to make enough.
When the fabrication process goes well, a large proportion of the product will well exceed the limits, UNLESS it is a very critical and "difficult" parameter, in which case the manufacturer will struggle to make enough.
- Status
- Not open for further replies.