I received a Direct Message from a member, who asks the question in title of this thread:
It occurs to me that the combined experience and wisdom of the entire diyAudio membership, is vastly greater than my own. So instead of trying to answer this question alone, I offer it here to everyone at diyAudio. "Two heads are better than one" -- C. S. Lewis (link)
Thank you for your kind replies, Mark Johnson
I'm solving drivers for end transistors where 2SA1930/2SC5171 will be used and I don't know which parameter is more important, VBE or Beta. My meter (LCR T7) shows me the hFE and Ube values. I assume the first is the current gain Beta and the second is Vbe. I have a very good match for Beta, but the Ube values are at least 30mV different between polarities. They are original TOSHIBA parts from old stocks. And then I have copies from Aliexpress and there the situation is the opposite. Ube differs between PNP/NPN by only 2mV, but hFE has a difference of 35 values. And I don't know what is more important for this application. I would be grateful for any advice.
It occurs to me that the combined experience and wisdom of the entire diyAudio membership, is vastly greater than my own. So instead of trying to answer this question alone, I offer it here to everyone at diyAudio. "Two heads are better than one" -- C. S. Lewis (link)
Thank you for your kind replies, Mark Johnson
I would say Beta (hFE) should be in the same ballpark. But in the end, any amplifier with decently specified parts (i.e. not driven to the edge of spec) with negative feedback will compensate for the small difference in gain. That's probably just fine for a stage with just one pair of transistors.
More importantly, you need your transistor's current gain and base-emitter voltage to be in the same ballpark at the same operating current; that's collector-emitter current in this case. the collector-emitter current affects both the gain and the base-emitter voltage drop.
If you want to measure Vbe and hFE accurately, I recommend you ditch the T7 LCR-guesstimator and lend a real LCR meter from a friend or someone else. You could also set up a circuit for measuring it manually, using a CCS/CCL.
More importantly, you need your transistor's current gain and base-emitter voltage to be in the same ballpark at the same operating current; that's collector-emitter current in this case. the collector-emitter current affects both the gain and the base-emitter voltage drop.
If you want to measure Vbe and hFE accurately, I recommend you ditch the T7 LCR-guesstimator and lend a real LCR meter from a friend or someone else. You could also set up a circuit for measuring it manually, using a CCS/CCL.
If the OS is an EF2 i would use the devices with the higher beta whatever the gain difference between N and P,
for an EF3 gains do not matter much.
for an EF3 gains do not matter much.
From the comments, I get the impression that the question isn't fully understood. I think the question is sufficiently well defined.
The key issue is whether the current gain (Beta or hFE) or the base-emitter voltage (Vbe) is more critical when matching complementary PNP and NPN BJTs used as drivers for the output transistors in a push-pull amplifier."
The key issue is whether the current gain (Beta or hFE) or the base-emitter voltage (Vbe) is more critical when matching complementary PNP and NPN BJTs used as drivers for the output transistors in a push-pull amplifier."
Vbe matters when you’re paralleling outputs. But it’s vbe @ic, which is dependent on base current. So beta matters. Vbe’s that are the same in “diode test mode” won’t be the same in operation if the betas aren’t really close.
Vbe between NPN and PNP are usually different even for supposedly complementary types. “Complementariness” at low current can be pretty bad for power types, as the beta vs Ic curves often dont even have the same shape.
The question was if VBE or Beta was important for selecting NPN/PNP driver pairs.
It is not so common to parallell drivers but if you should do it VBE will be important NPN/NPN or PNP/PNP.
It is not so common to parallell drivers but if you should do it VBE will be important NPN/NPN or PNP/PNP.
Applies to outputs or drivers. Paralleling drivers has become more common in recent years since ruggedized TO-220s like the C2238, C3298 have gone AWOL and all that’s left are the fragile TO-126’s. So what do you see now in 200 wpc amps? Either blown drivers or two pair. When you do that, you’ve got to match them close, because of the lack of degeneration that the outputs have. Luckily, they come in bags of at least 100 and the processes are pretty tight these days. Chances of getting matches are pretty good.
I have found that Vbe matching between the N and P channel devices reduces DC Offset in some complementary topologies. So, like a lot of the broad questions we get here, the answer is often "it depends".
For input stages vbe matching between complementary types is possible and practical, especially with old-school types that have modest gain (100 to 200). Not as easy when the gains are up above 500, but then the input bias (and offset) currents are lower to begin with.