Matching Vbe of Surface mount transistors.

[moffy]

I need to match some MMBT3640 transistors for Vbe but unsure how to go about it for surface mount. Anyone done this? Would prefer the TO-92 version, PN3640/MBS3640 but they don't seem available at a reasonable price (ebay but expensive).

[Alex Eisenhut]

I assume you're asking "how" physically?

http://www.ebay.ca/itm/SOT6-SOT-23-TO-DIP6-A-Wells-IC-Test-Burn-in-Socket-Adapter-With-Circuit-Board-/121096387827?pt=LH_DefaultDomain_0&hash=item1c31e83cf3

[c4757p]

What application needs both the high switching speed of the {PN,MMBT}3640 and matched Vbe?

That socket looks handy. Though, when I've matched small numbers of SOT-23s for one-offs, I just stuck sharp probes directly across base and emitter, which worked well enough if it did send probably about 5% of them onto the floor...

[Rupunzell]

Depending on circuit needs, this could be a LOT easier with much better parts match and thermal tracking.
http://www.thatcorp.com/300-series_Matched_Transistor_Array_ICs.shtml

There was a time when matched BJTs in a package was pretty common.


Bernice

[Alex Eisenhut]

What application needs both the high switching speed of the {PN,MMBT}3640 and matched Vbe?

That socket looks handy. Though, when I've matched small numbers of SOT-23s for one-offs, I just stuck sharp probes directly across base and emitter, which worked well enough if it did send probably about 5% of them onto the floor...

I am a bit miffed at the price. The bare sockets sell for even more. 

[c4757p]

Depending on circuit needs, this could be a LOT easier with much better parts match and thermal tracking.
http://www.thatcorp.com/300-series_Matched_Transistor_Array_ICs.shtml

There was a time when matched BJTs in a package was pretty common.


Bernice

I doubt those will do. The PN/MMBT3640 is a very specific transistor, a very low storage time type (bit slower than the NPN 2N/PN/MMBT2369) for high speed switching. Not something a normal linear BJT can usually substitute for, and not the sort of thing you'd often see in matched pairs...

[Marco]

PN3640/MBS3640 but they don't seem available at a reasonable price (ebay but expensive).

80 cents for single on mouser is expensive?

[Rupunzell]

Indeed, question is why VBE match, how close does the VBE match needs to be, does the VBE match need to track over temperature, how many matched devices and how are they used?

Still, small geometry BJTs were offered back in the day.


Curious minds wanna know?


Bernice

[quote author=c4757p link=topic=46977.msg662431#msg662431 date=1430276389
I doubt those will do. The PN/MMBT3640 is a very specific transistor, a very low storage time type (bit slower than the NPN 2N/PN/MMBT2369) for high speed switching. Not something a normal linear BJT can usually substitute for, and not the sort of thing you'd often see in matched pairs...
[/quote]

[moffy]

PN3640/MBS3640 but they don't seem available at a reasonable price (ebay but expensive).

80 cents for single on mouser is expensive?

The shipping was much more than the cost of the transistors. sigh.

I assume you're asking "how" physically?

http://www.ebay.ca/itm/SOT6-SOT-23-TO-DIP6-A-Wells-IC-Test-Burn-in-Socket-Adapter-With-Circuit-Board-/121096387827?pt=LH_DefaultDomain_0&hash=item1c31e83cf3

Thank you, that looks like exactly what I want.

Indeed, question is why VBE match, how close does the VBE match needs to be, does the VBE match need to track over temperature, how many matched devices and how are they used?

Still, small geometry BJTs were offered back in the day.


Curious minds wanna know?


Bernice

[quote author=c4757p link=topic=46977.msg662431#msg662431 date=1430276389
I doubt those will do. The PN/MMBT3640 is a very specific transistor, a very low storage time type (bit slower than the NPN 2N/PN/MMBT2369) for high speed switching. Not something a normal linear BJT can usually substitute for, and not the sort of thing you'd often see in matched pairs...

[/quote]

I am creating a 10MHz better than 1% linear triangle wave. I used matched NPN and PNP pairs (differential) to switch the currents. Sims look very good so far and the MMBT3640 has some unique characteristics, especially low Ccb nonlinearity as long as biased with a few volts. The nice thing about the NPN and PNP differential configuration is that the same square wave but level shifted drives both inputs. It is much faster than current mirrors or any other configuration. It's really not essential to match that closely but I would feel more confident with them matched. Also came up with a simple fast comparator to go with it.

Thank you all for your input. It's great.

[moffy]

Just ordered a couple.

[Rupunzell]

Might try MPSH10 = NPN, MPSH81 = PNP. These are small die RF devices with low capacitance (less than 1pF).

There are small die matched diff pairs in various flavors made by Linear Systems. Pricy for what they are.
http://www.linearsystems.com/products.php?pr=transistors


Bernice

I am creating a 10MHz better than 1% linear triangle wave. I used matched NPN and PNP pairs (differential) to switch the currents. Sims look very good so far and the MMBT3640 has some unique characteristics, especially low Ccb nonlinearity as long as biased with a few volts. The nice thing about the NPN and PNP differential configuration is that the same square wave but level shifted drives both inputs. It is much faster than current mirrors or any other configuration. It's really not essential to match that closely but I would feel more confident with them matched. Also came up with a simple fast comparator to go with it.

Thank you all for your input. It's great.

[moffy]

Might try MPSH10 = NPN, MPSH81 = PNP. These are small die RF devices with low capacitance (less than 1pF).

There are small die matched diff pairs in various flavors made by Linear Systems. Pricy for what they are.
http://www.linearsystems.com/products.php?pr=transistors


Bernice

The MMBTH81/MPSH81 looks interesting, looks quite similar to the PN3640. Thanks.