Agreed. We simply can't expect all Hfe measurements of a random lot of parts to be consistent from 2mA through to >10A when the indicative Hfe curve drops off like a roller coaster. The comparison of Nigel's MJ2955 with 2N3055 in these graphs gives you an idea of the problems this transistor has when used as a complement to 2N3055. When the test current is within the specified range though, I'd still expect all genuine MJ2955 transistors to give normal range results.
Attachments
@Nigel, was that part one of their "Multicomp" brand ones? Im guessing so since as they no longer stock the ST ones, and everyone else discontinued them.
Personally i'd just fit MJ15003/4 in that case.
edit: Hah, they do have the Onsemi ones - just with a TO-3P package shown rather than a TO-3 !!
Personally i'd just fit MJ15003/4 in that case.
edit: Hah, they do have the Onsemi ones - just with a TO-3P package shown rather than a TO-3 !!
Last edited:
Heh,heh...I wonder if Jez Siddons had a look at old PNPs like the MJ2955 in the development stage? I don't doubt that the instrument is correct in what it reads at the current shown but the question is rather about how useful that measurement is for power transistor matching or acceptance tests, perhaps as Nigel was doing. Beyond a go/no-go test, you'd have to say the tester just isn't capable, regardless of whether the reading seems to be in the right range or not 
As I see it, the datasheets cut off at a minimum 100mA current but the Hfe trend @25C is definitely falling with current below that for the NPN type but still rising for the PNP type, which ties in with Jez's measurement of 90 at 2.5 mA for 2N3055 and Nigel's measurement of 273 for MJ2955, presumably also at 2.5mA.
It seems all is fine with the tester but not with the meaning of the test.
As I see it, the datasheets cut off at a minimum 100mA current but the Hfe trend @25C is definitely falling with current below that for the NPN type but still rising for the PNP type, which ties in with Jez's measurement of 90 at 2.5 mA for 2N3055 and Nigel's measurement of 273 for MJ2955, presumably also at 2.5mA.
It seems all is fine with the tester but not with the meaning of the test.
Maybe they could upgrade those meters with pulse measuring techniques, it's not that difficult to source up to 10A for 10us, or do pulsed voltage breakdown tests up to 100V. A dot matrix screen can show a chart.
On the other hand, almost all old bipolar power transistors have a hFE that falls almost 10:1 at high current. In some cases the opposite effect can be observed too, hFE can fall 10:1 at very low current <1mA, like a "lossy" b-e junction, as in case of built-in pull-down resistor, or to increase open-base Vce breakdown, or just low quality clones (that give unsurpassed performance in analog noise generators, like 10dB higher noise figure than usual). Those meters could measure small signal transistor noise too, as it is a good parameter to compare quality.
On the other hand, almost all old bipolar power transistors have a hFE that falls almost 10:1 at high current. In some cases the opposite effect can be observed too, hFE can fall 10:1 at very low current <1mA, like a "lossy" b-e junction, as in case of built-in pull-down resistor, or to increase open-base Vce breakdown, or just low quality clones (that give unsurpassed performance in analog noise generators, like 10dB higher noise figure than usual). Those meters could measure small signal transistor noise too, as it is a good parameter to compare quality.
Genuine ON Semi device BTW. From the typical characteristics in post 21 it seems that 500mA is a suitable current to match the devices, as they won't be a particularly good match anywhere else, though of course you may prefer, as I try to, match at around 2A.
Pulse testing is great if you have the kit. I strap the devices to a big heatsink (directly, no mica) and run the test for hfe measurement at 4A (the spec. current), using a 5V 20A ex-computer PSU, as quickly as possible (take the reading and disconnect). It is also possible to measure Vbe but it actually takes longer to take two readings, or three, to see if Vbe changes, by which time the transistor is showing signs of warming up a little.
Pulse testing is great if you have the kit. I strap the devices to a big heatsink (directly, no mica) and run the test for hfe measurement at 4A (the spec. current), using a 5V 20A ex-computer PSU, as quickly as possible (take the reading and disconnect). It is also possible to measure Vbe but it actually takes longer to take two readings, or three, to see if Vbe changes, by which time the transistor is showing signs of warming up a little.
Anything from Farnell will almost certainly be genuine.
The problem is that the spec of a xx2955 is so loose, that almost any pnp power device will meet the letter of the spec but not the typical values. A 30 MHz ft part would meet the datasheet guaranteed parameters, but will oscillate in many circuits
The problem is that the spec of a xx2955 is so loose, that almost any pnp power device will meet the letter of the spec but not the typical values. A 30 MHz ft part would meet the datasheet guaranteed parameters, but will oscillate in many circuits
I have an ON MJ15002 measuring hfe 240 using the DCA75....
I just put it on the power supply at 10.8 volts. 1K resistor to base so 10 mA base current and power supply current says 1.49A initially.
So still HFE around 140.
Using 330 ohms resistor (connecting leads less accurate) it goes to 3 amps.
So it is around 100 when 30 mA base current. I tried few seconds only as it gets hot.
This transistor was bought some 8 years ago from Farnell or RS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.