I would suspect that breakdown voltage is a 'designed in' characteristic, and will depend on physical dimensions of the transistor structure. The "cherry picking" that the manufacturers do is usually related to gain and gives something like an A, B or C suffix like BC108A or BC109B. How would the manufacturer actually test a batch for breakdown voltage, as if you take the device to breakdown, then it is destroyed. Would the manufacturer be prepared to stake the breakdown voltage of a production batch, on the basis of a few tested examples ? I'm not sure that he would. It's a bit like having tested fuses really, isn't it ?
You are right to a certain extent. However, Mfg's do sort by breakdownvoltage. They do this with a constant current supply set at a very low current and measure the voltage.
There are a BUNCH of issues here. What's the actual voltage applied in the application?
Given volume discounts and inventory reduction, it's not unusual for a vendor to use a part in an application that would do fine with a MUCH lesser part.
Do you really care about Bvceo? That's relevant if e is really o. Excellent metric, but sometimes not the metric you want.
If you're running it anywhere near 80V, you might wanna think about secondary breakdown. Devices with similar text specs can behave quite differently at high voltages.
Worry about temperature. If you're selecting breakdown voltages, make sure you measure at temperature extremes.
One simple way to test breakdown is to put your voltmeter in series with the power supply. IF the meter is 10megs and you have the reading and the power supply voltage, you have enough info to calculate the current at various voltages. If you need more current shunt the meter with a resistor. If you try to use a current meter and the device goes into avalanche, you've just toasted your transistor and maybe your meter.
I don't know how they do it today, but back in the day, there were a few "formulae" for transistors. For a given die, you ran it across the test bed and put it in whatever bin the specs met. If it didn't meet any, it went into the 2n3904 bin.
Whether you can up-spec a part by testing depends a lot on the yield of that part. Some parts have high yield and they have to ship better parts than the spec requires. But that's not always the case. Back when I didn't know better, I tried to select 1% resistors out of 20% for a test jig. Turned out there was a BIG hole in the distribution that was
10% wide. 10% resistors had a big hole in the distribution
5% wide and so on. Somebody got there before me ;-) But I fooled them, just recalculated the ratios to be
10-20% off the standard value.
mike
--
Return address is VALID!
Bunch-O-Stuff Forsale Here:
http://mike.liveline.de/sale.html
Ah, ok. I can see how that would work. Do manufacturers ( rather than distributors like SEME) really still do this ? Seems like a lot of trouble to go to. There are so many (hundreds of thousands ??) of transistor types available now, I would have thought that it was possible to select one for whatever breakdown voltages, gain, and whatever other parameters you might require, without difficulty ? Maybe not. Thanks for the info.
Absolutely they do ! Semelab manufactures transistors too btw. They and Zetex probably represent the entire UK semiconductor manufacturing sector.
The Semi makers have to test the devices for function and things like gain, so they test for loads of stuff at the same time. It's all automated.
Here's an example of what sometimes goes on with device selection.
Decades ago we were using a Motorola part, an MJ410 in a high power audio amplifiers as a driver. The purchasing dept found the RCA410 'equivalent' at a lower price and initial samples worked fine so we used them.
Some months later we started getting stability problems with *some* of the RCA410s. An alert test engineer noticed that the problem parts had a specific code stamped into the TO-3 header. It turned out that were 3 such codes in total; C, CN, and CNX.
I sawed the lid off one of each and even a quick inspection by eye showed that each one had a different die fitted.
RCA initially denied everything but I eventually got an admission form their Belgian European Headquarters.
Motorola's spec on the MJ410 was quite minimal. It didn't matter to them because they made the part all from the same line. RCA however looked at the skimpy spec and reckoned they could fit lots of devices that failed to meet better and more financailly worthwhile specs into their eqivalent RCA410 device spec. So they made RCA410s from the fall-outs from 3 other product lines with widely varying other specs (that weren't on the data sheet). Those varying 'other specs' were what was giving us trouble.
I want an NPN T039 or T05 driver transistor for use in an audio amplifier. The supply rail is 75volt. Quiesent current is about 10ma I think. 1 Watt dissipation at 25 C. The original was an RCA 39252 (which I have no data on) or a 2N2102. They may have been selected to to work at 75volt.
You need to give more details on the circuit. If this is the first stage of the amplifier, it's a lot easier than if it's the "driver" for the output stage. If this thing goes between ground and 75V, you want WAY more than an 80V transistor. If it goes between +-75V, you need way more than a 150V transistor. Unless this is a special transistor, you can't expect to get anywhere the maximum power dissipation at anywhere near the breakdown voltage. Take a serious look at the dynamic characteristics, peak voltages and currents driving the capacitive load and the secondary breakdown curves. The RCA Power Circuits manual SP-51 has a whole chapter on this.
And worry about the power numbers too. That 25C number is a metric that gives you some idea about what's going on. What you really care about is the die temperature. And it's WAY more than ambient in a small package like that. mike
--
Return address is VALID!
Bunch-O-Stuff Forsale Here:
http://mike.liveline.de/sale.html
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.