Why BJT is used as diode?

[cheenu2002]

Hi,
I have seen in many circuits that a BJT is used in place of a diode by shorting the base and collector. Is there any particular reason behind this? What are the problems faced by designers if a PN Jn. diode is used in place if diode connected transistor?

 

[FvM]

You would use the CB rather than the EB diode because of it's higher reverse voltage. A simple reason can be minimizing the amount of different parts in a design.

In IC design, you have a transitor in diode configuration in current mirrors to achieve best matching.

 

[erikl]

What are the problems faced by designers if a PN Jn. diode is used in place if diode connected transistor?

In bandgap circuits, the ΔVbe difference of the forward BE junctions of 2 diode-connected BJTs carrying different current densities are used to compensate for the Vgs - resp. Vth - differences of 2 MOSFETs carrying the same different current densities. Due to the BJTs' current gain B, the BE junctions carry (by a factor B) lower currents, thereby "work" in current ranges where their Vbe=f(I) stick closer to the exponential relation than it would be the case if there was no current gain (i.e. series resistance is less relevant).
Due to their current gain, the usage of diode-connected BJTs in bandgap circuits is - concerning accuracy - superior to simple diodes. On the other way round, if one would choose to use simple diodes, they would need to have B-times the area of the correspondent transistor in order to achieve the same accuracy.

 

[dick_freebird]

The gain of the transistor "linearizes" the diode log-I in the
data I have taken. This is key to operating the diodes at
different current densities yet getting a constant delta-V.
The current gain also nearly eliminates series resistance
(at least, until you approach saturation).

Saturation is bad news, it will mess up the storage time and
capacitance. You have to watch that, using transdiodes.

 

[E-design]

They have lower reverse leakage.

 

[cheenu2002]

Thanks a lot experts!.. for your answers.

 

[erikl]

They have lower reverse leakage.

Why should they?
A BJT taking up the same area as a diode in equally doped material (i.e. using the same junction(s)) always will have a (bit) larger leakage because of its inherent (small) current gain, which means that a small EB current provides an additional (though small) contribution to the total leakage current.

 

[E-design]

I have used this fact to my advantage often where I needed a discrete diode with low leakage and the PIV only needs to be around 5V or so.

 

[erikl]

I have used this fact to my advantage often where I needed a discrete diode with low leakage and the PIV only needs to be around 5V or so.

Correct. But this is comparing apples and oranges: The 1N914 is a 1A diode, whereas the 2N3904 is a 200mA BJT (both max. ratings), i.e. their junction areas (and doping profiles, probably) are quite different. Your measurement results emphasize these differences.

In this forum, mostly IC design & layout questions are discussed. Hence for comparisons, comparable layout scenes should be used. General comparison statements on non-comparable devices aren't meaningful.

Merry Christmas!

 

[E-design]

Repetitive peak forward current of a 1N914 is about 220mA not 1A

 

[FvM]

Funny simulation. Unfortunately it's too good to be true. You may want to repeat the simulation e.g. with V1 of 10 and 20 V
and also try with a real transistor. It would be less fun to reveal the result in advance.

P.S.: You're right, 1N914 is a 1N4148 equivalent, a signal diode.

 

[E-design]

I suppose you never read Bob Pease articles.
https://en.wikipedia.org/wiki/Bob_Pease

 

[FvM]

No, but I knew before. See my above statement

You would use the CB rather than the EB diode because of it's higher reverse voltage

. You should do the same.

Regarding your simulation, Multisim is apparently applying a linear rather than a correct exponential relation in reverse
current calculation. It doesn't consider at all the EB-diode breakdown voltage of 5 to 10 V. I also won't trust it in comparing
reverse currents of different parts. (It can be correct of course).

 

[E-design]

I use both CB and EB. I knew that the EB has lower breakdown, found slightly more leakage but switch faster that the CB junction. The simulation was just for illustration of this fact, but I have verified this in practice. I am surprised the simulator result agreed so well.

 

[FvM]

I agree, that an EB reverse voltage of 5V is safe with most general purpose silicone BJT, although the reverse current can be considerably higher than the claimed 10 pA.

 

[E-design]

As always performance must be tested with the parts at hand as it can vary widely between manufacturers.

 

[erikl]

I suppose you never read Bob Pease articles.

That's a good one (and Bob Pease anyway). So you should also respect the text below your red-framed paragraph, it tells you the essential reason for low leakage current: very small junctions (from JFETs in this case) usually guarantee for very small leakage currents. Not the device type, the junction area is responsible.

Additionally you compare the 2N3904 with the gold-doped 1N914, which makes its junction essentially leakier (s. your red-framed text) due to the Au generation-recombination centers near the band gap center.

 

[E-design]

So what is your point? I indicated that using the transistor junction as a diode is better for low leakage than a real (general purpose) signal diode.

Added after 57 minutes:

I agree, that an EB reverse voltage of 5V is safe with most general purpose silicone BJT, although the reverse current can be considerably higher than the claimed 10 pA.

Quoting from Bob's writing "Other strange things that diodes can do"

A typical 2N3904 emitter diode can turn ON or OFF in 0.1nS with negligible overshoot and less than 1pA of leakage at 1V, or less than 10pA at 4V (This diode does, of course have the base tied to collector)

 

[erikl]

So what is your point? I indicated that using the transistor junction as a diode is better for low leakage than a real (general purpose) signal diode.

Of course you are right: In terms of low leakage current it's always better to use a small transistor instead of a big diode. This is even true in IC layout. An equally sized diode on the same real estate, however, will carry a bit less reverse current than the diode-connected BJT. That's simple electronics: no current gain in a diode.