Using the B-E junction of a transistor as a diode

[ChrisP58]

So, the Base-Emitter junction of a BJT is essentially a diode.

My interest is purely economic. I have a circuit that uses a PNP (2N3906) that is also connected to a diode (1N4148).

I can get a dual PNP in a SOT-363 package. By using half of it for the diode, there is only one part to buy and populate.

It simulates just fine in Ltspice. But are there any non obvious potential issues with using it as such?

 

[dougy83]

The VBE of 2n3906 is <5V; this may be an issue for you. If you don't mind it acting as a low voltage zener diode, then you may be OK.
Also, I believe the base and collector are often connected together for the diode function, to reduce the base current.

 

[crutschow]

I believe the base and collector are often connected together for the diode function, to reduce the base current.

Yes, that reduces the effect of the base-emitter resistance and makes the diode appear more ideal.
Ltspice simulation comparison below (straight line is ideal with the log current scale):

If you need more than 5V reverse breakdown, you can use the collector-emitter junction with the base-emitter tied together.
That will give nearly as ideal a curve as connecting the base-emitter, but will have the breakdown voltage of the transistor collector-base junction (40V for the 2N3906).

 

[Tony Stewart]

My interest is purely economic. I have a circuit that uses a PNP (2N3906) that is also connected to a diode (1N4148).

what are your economic and performance expectations? In volume these parts are 1 or 2 pennie’s.

 

[ChrisP58]

Thanks all. I didn't know about the base collector connection.

The dual PNP I'm looking at is the MBT3906DW1T1G. DigiKey full real price of $0.043.
The single PNP MBT3906DW1T1G is $0.03 plus a 1N4148WX-TP at $0.018
So the bom cost difference is less than a penny.

The two parts do take about twice the real estate.

But the big savings comes in the assembly costs. My current CEM doesn't break the placement cost out but, if memory serves, my previous one bid on a placement cost of about $0.15 per component. As I use this circuit in 3 places, I might save about $0.45 per board.

 

[Nigel Goodwin]

As others have said, reverse Vbe breakdown is VERY low - most transistors die if used in an unprotected multivibrator circuit from a 9V battery, and many don't survive from a 6V battery either. What voltage are you going to have across the diode?.

I've no idea what the 'spec' of the collector/base diode junction is?, as it's not something that ever comes up?.

 

[AnalogKid]

I've no idea what the 'spec' of the collector/base diode junction is?, as it's not something that ever comes up?.

Interestingly, I've seen it come up at least three times in the last year. The reverse-breakdown voltage of a common small-signal transistor works to form the world's most simple transistor oscillator.

Simplest_LED_Flasher_Circuit

ak

 

[ronsimpson]

I've no idea what the 'spec' of the collector/base diode junction is?, as it's not something that ever comes up?.

The B-C diode is good. On the data sheet there is often a B-C breakdown voltage. If not, there may be a C-E (with B_E shorted) voltage. If not use the C-E voltage rating. I have used B-C as a diode many times, for power parts. There were years when I could not get 1500V power diodes and we used the C-B junction.

Motorola's P2N2222a transistor has a C-E break down voltage of 40V. A C-B max voltage of 75V. B-E max is 6V.
The C-B current is not rated but I know the Base will take 600mA. (in the data sheet)

 

[Tony Stewart]

Thanks all. I didn't know about the base collector connection.

The dual PNP I'm looking at is the MBT3906DW1T1G. DigiKey full real price of $0.043.
The single PNP MBT3906DW1T1G is $0.03 plus a 1N4148WX-TP at $0.018
So the bom cost difference is less than a penny.

The two parts do take about twice the real estate.

But the big savings comes in the assembly costs. My current CEM doesn't break the placement cost out but, if memory serves, my previous one bid on a placement cost of about $0.15 per component. As I use this circuit in 3 places, I might save about $0.45 per board.

Your CEM ought to use loaded costs based on $75 to $100/hr and 0.15$/pc is about 10x too much. Your estimate would make resistors prohibitive. It seems like your volumes are too low.

 

[ChrisP58]

As others have said, reverse Vbe breakdown is VERY low - most transistors die if used in an unprotected multivibrator circuit from a 9V battery, and many don't survive from a 6V battery either. What voltage are you going to have across the diode?.

I've no idea what the 'spec' of the collector/base diode junction is?, as it's not something that ever comes up?.

The 'diode' is connected reverse biased across the B-E pins of the transistor, so the reverser voltage is well below the reverse voltage limit.

 

[crutschow]

I've no idea what the 'spec' of the collector/base diode junction is?,

From the STMicroelectronics data sheet for a 2N3906:

The VCES rating is usually equal to the VCBO rating, so this configuration should work as a diode at up to 60V reverse bias.

Below is the sim for using the collector-base junction as a diode.
Connecting the base to the emitter adds the transistor's reverse current gain (which, of course, is much lower than the forward gain) to reduce the current through the higher resistance base terminal (yellow trace), and thus reduce the forward resistance at the higher current level.
The diode characteristic (green trace) is fairly ideal (straight) up to about 10mA, where the internal resistance causes it to curve upward.

 

[unclejed613]

when i worked for NAD, they had one of the early AM/FM receivers that used electronic tuning, and the original design used 1N4148 diodes to isolate the tuning memory circuit from the power supply when the power was turned off. the "battery" was a 250,000uf supercapacitor. unfortunately, 1N4148 diodes had enough reverse leakage that the supercap would discharge within 2 or 3 weeks, and the owner might come home from vacation and find all of their presets gone. the fix was to replace all of the isolation diodes with the B-C junction of 2N3904 transistors.

 

[audioguru]

1N4148 diodes are in a clear glass case and leak reverse current when exposed to light. Black cased transistors are not affected by light.

 

[Tony Stewart]

When a 1N4148 drives the base, the collector amplifies the shared voltage and current. Using this method not only makes the diode more ideal but lowers the ESR*Ron=T time constant.

The observant will see two diodes in parallel with non-ideal Rs added that are mismatched and result in 50% Pd difference and temperature difference depends on coupling of Rjc-Rjc.

Proof : https://tinyurl.com/2komn5kw using 1N4148 for Base and 1A diodes (1N4004) elsewhere to compare using interactive switch https://tinyurl.com/2q9cw4bk

1st link has a default 5A diode on far left only. Realize this does not include thermal response and would burn out so I lowered the ideal voltage source here (theoretical)

https://tinyurl.com/2mtkx7pb

Conclusion the better diode response uses a transistor to lower the incremental resistance of the external diode by hFE @ 1V.

(sorry for rushed answer)

 

[danadak]

 

[Tony Stewart]

danadak and Audioguru Do you agree with my thermal analysis and solution for the dual diode burnt PCB problem ... oops not here.

 

[Tony Stewart]

One can take it one step further and make an active diode switch with low voltage as in Case A below .
It has a similar threshold but you ca ndrop this with a fixed voltage and R bias to lower the 0.6V threshold at 1mA.

This is extremely useful for CC regulators using a drop of <=100mV instead of >= 0.6V.

But notice how the voltage drop in A for Vce decreases with increasing current from negative feedback.

Comments? Does this make sense to everyone?

sim link

 

[AnalogKid]

View attachment 141591

Very nice. I would have guessed than the Q2 config was higher than the Q1 config. Learned something new.

I wonder if the real world agrees.

ak

 

[danadak]

One can take it one step further and make an active diode switch with low voltage as in Case A below .
It has a similar threshold but you ca ndrop this with a fixed voltage and R bias to lower the 0.6V threshold at 1mA.

This is extremely useful for CC regulators using a drop of <=100mV instead of >= 0.6V.

But notice how the voltage drop in A for Vce decreases with increasing current from negative feedback.

Comments? Does this make sense to everyone?

View attachment 141597 sim link

Am I looking at negative R, great way to make an oscillator ? The beginnings of a tunnel diode ?


Regards, Dana.

 

[Tony Stewart]

no oscillator , its an error amplifer

e.g 1A rough CC with 0.1 drop for Vbe

https://tinyurl.com/2n3qsylc