Looking for small and medium signal transistor with low Vbe.

[Alan0354]

Hi

I need to design a bias spreader for 2Vbe only. I need to find a NPN transistor in TO-126 that has very low Vbe at a few mA. Detail reasoning is described in this post: http://www.diyaudio.com/forums/soli...tempco-between-spreader-power-transistor.html

I am using KSC3503, it's over 0.6V. So far, I am using a big MJW as spreader transistor to get Vbe of about 0.51V. I need to find a smaller transistor with low Vbe.


Also, I need to find a small signal PNP that has very low Vbe also. Please give me some suggestions.

Thanks

 

[DF96]

I suspect you have already answered your own question in the other thread. Low Vbe means high power, other things being equal. About the best you could do is find one which has a high power chip in a medium power encapsulation - presumably on the grounds that it will only be used for switching so won't get too hot.

 

[Alan0354]

I suspect you have already answered your own question in the other thread. Low Vbe means high power, other things being equal. About the best you could do is find one which has a high power chip in a medium power encapsulation - presumably on the grounds that it will only be used for switching so won't get too hot.

Ha ha, not the answer I want. But I kind of suspect that already, I immediately picked the MJW the moment I observed the problem. I was hoping for miracle that there is one small transistor with low Vbe.

Thanks

 

[AndrewT]

as stated in that other Thread, I think you are looking for a combination of lowish Vbe and highish hFE.

A transistor that is very good at near saturation may offer some advantage.
But the bc550/560 would be too high Vbe at typical multiplier currents.

What is the uncle of this device that is also good at near Vsat. Keantoken knows. I seem to recall it is a higher power version and may have a lower Vbe at the same currents.

 

[Elvee]

I was hoping for miracle that there is one small transistor with low Vbe.

Everything else being equal, a transistor having a higher Hfe will have a smaller Vbe for a given collector current, because the base current, hence the base current density, will be smaller.
You could opt for superbeta transistors, like 2SC1984 or ZTX1047 for example (all manufacturers have some similar types).
Although not strictly a superbeta, the BD433 is a low voltage, high gain, relatively powerful transistor and it should have a reasonably low Vbe.

 

[DF96]

Isn't Vbe set by the emitter current density, not the base current density?

 

[Alan0354]

I just thought of something. The Vbe is related to the current capability of the transistor. I am thinking about looking for low voltage transistors that has high current capability. KSC is high voltage and low current, maybe if I look for a low voltage device in TO-126, I might have better luck both in hfe and Vbe. I don't need high voltage at all here. Same as the PNP that I use BC560.

Thanks

 

[Alan0354]

Everything else being equal, a transistor having a higher Hfe will have a smaller Vbe for a given collector current, because the base current, hence the base current density, will be smaller.
You could opt for superbeta transistors, like 2SC1984 or ZTX1047 for example (all manufacturers have some similar types).
Although not strictly a superbeta, the BD433 is a low voltage, high gain, relatively powerful transistor and it should have a reasonably low Vbe.

Thanks

Any suggestion on PNP in TO-92. I am just going to order a few of 5 each.

 

[Mark Johnson]

KSD1691G (link) is rated 5A, 60V in the TO-126 package. If you buy the G suffix device (1709 parts in stock at DigiKey) you will get Beta=200.

High max current means large emitter area means lower Vbe at a given collector current. High beta means lower Ibe at a given collector current, which means lower Vbe.

Or you could just use a power diode in a TO-220 package, selected for the Vfwd at Ifwd point where you wish to operate. Diode datasheets are very good about presenting their I-V curve, as long as you stay away from the price-is-my-only-requirement 1N400x jellybeans. They're also good about presenting their I-versus-V-versus-Temperature characteristic.

 

[Mark Johnson]

This one looks good too. 10A, 50V, Beta=400. No it's not a Darlington. It's a Sanyo NPN device after the technology transfer to ON Semiconductor.


_

 

[Elvee]

Isn't Vbe set by the emitter current density, not the base current density?

Since you mention it, I now have some doubts, but even with afterthoughts I still think it is mostly linked to the absolute base current density. This would need to be verified, and second or higher order effects certainly come into play, but I think the base current will still be dominant for normal devices, operated at normal current levels.

If sufficiently detailed datasheets exist for Hfe-selected devices, this could be checked.

I'd have to look into my paper databooks, since the electronic versions are generally much more succinct.

 

[john curl]

Vbe (I am pretty sure) comes from basic physics and it depends on material (silicon) the amount of doping with impurities, AND the physical size of the chip. A large chip is like many smaller devices in parallel, so each 'device' is operating at a lower current than the total current would initially imply. You can find devices with high current and in a TO-92 case if you look around. Perhaps there are special devices (silicon) that are doped differently, but I don't know what they are.

 

[john curl]

The fundamental equations:

 

[Mark Johnson]

John, perhaps you refer to gold-doped devices (Fairchild "Process 21")? The gold kills the carrier lifetime and makes very fast saturated switches. Perhaps it alters the VBE too (?) About the only gold doped silicon transistor still made today is the PN2369A (actually the MMBT2369A) which does have a lowish VBE. However since it's designed to be used in high speed digital logic gates, it's a small die that switches small currents, and you don't get the further VBE reduction that would come from large emitter area.

The 1N4148 high speed switching diode is gold doped, and it definitely does have a different Vfwd than other silicon, non-gold-doped diodes. That's why external compensation diodes in IC app notes are never 1N4148's; they don't match the junctions inside the IC.

_

 

[Alan0354]

Thank you Mark.

This is the list of transistors I have so far that I am going to order 4 each to test from Digikey:

For TO126 NPN:

1)https://www.fairchildsemi.com/datasheets/KS/KSD882.pdf

2)https://www.fairchildsemi.com/datasheets/KS/KSD1691.pdf

3)https://www.fairchildsemi.com/datasheets/BD/BD433.pdf


For TO92 PNP:

1)https://www.fairchildsemi.com/datasheets/BC/BC327.pdf


For TO220 NPN:

https://www.fairchildsemi.com/datasheets/KS/KSE44H.pdf

 

[Elvee]

Isn't Vbe set by the emitter current density, not the base current density?

It is indeed the emitter current, thus I was wrong, but!.... perhaps not completely, at least for the end result:
For technological reasons, there is a correlation between most of the parameters, in particular the Vbe and the Hfe, and if a device from a process has a higher Hfe, it will also have a lower Vbe than "regular" devices.

This is illustrated in the sim below, based on NXP models which are pretty detailed and accurate.
Q3 is a low Hfe selection of the BC847 (β=182). At Ie=2mA, its Vbe is 671mV. If the β is artificially increased to 524, same value as the BC847C, the Vbe remains unchanged (yellow and green traces). This BC847CM is chimerical, because none of the other correlated parameters has been modified.

If a real BC847C is compared to this chimerical transistor, the Vbe is lowered to 649mV for the same conditions (red trace).

It is interesting to note that 182*exp((671-649)/26) is not that far from 524

 

[Alan0354]

For a given Ie, Vbe is lower for larger transistor for very simple reason, The large transistor is like a bunch of small transistors in parallel. It is literally like that. If you look under the microscope, the power transistors have emitter like fingers extending out to cover more area. It's like many transistors in parallel together.

If you think of it that way, for two transistors, each will contribute only half the current.

From equation \delta Vbe= VT ln(I1/I2). If you calculate for two transistor where each contribute half, the equation gives 18mV reduction in Vbe. If you have 10 transistors in parallel, using the same equation will give you a reduction of 60mV. 100 transistor will reduce by 120mV....and so on.

So by definition, larger the transistor, the lower the Vbe.

As for hfe, think of it this way, there is always a parasitic base resistance in series from the pin to the base inside. Any current through the base resistance with create a voltage that adds to the Vbe. The higher the hfe, the smaller the current at a given Ie. so the higher the hfe, the lower the Vbe. Also, the large the base, the lower the base resistance, so less voltage drop across the base resistance.

 

[Elvee]

As for hfe, think of it this way, there is always a parasitic base resistance in series from the pin to the base inside. Any current through the base resistance with create a voltage that adds to the Vbe..

No, the purely ohmic part of the base resistance is very small, and has no significant effect. It can be represented by the terminal slope on a Ib/Vbe graph, and this slope is very steep (= low resistance) compared to normal, linear operating conditions.
It can play a role for low Hfe/high voltage transistors operated in deep saturation.
The correlation low-Vbe to high-Hfe is caused by semiconductor parameters, doping levels, geometry, etc

 

[Elvee]

No, the purely ohmic part of the base resistance is very small, and has no significant effect.

For the BC847 examples I took, this parameter (called RBM in spice) is 4Ω for the BC847A and 5Ω for the BC847C. With ~10µA base current, this amounts to ~50µV, much too small to make a difference

 

[Alan0354]

I think the hfe effect is a lot smaller. if you use a bigger transistor. Most of the effect is from the size of the emitter as explained. But theoretically the hfe does affect the total Vbe PARTICULARLY in the output transistor where base current runs into 50mA.