Strange BJT configuration, working as a diode?

[DDT]

I was looking at an old schematic and I came across this strange circuit for low voltage detection section of a larger circuit (not shown). The operation of circuit is straightforward, Q1 is normally conducting and the voltage divider R2/R3 set a threshold voltage, above which, Q2 conducts. When Q2 is normally conducting, the Q3 is off due to the voltage divider R1/R5. Upon depletion of battery, R2/R3 can no longer provide the base voltage for Q2 and it turns off. The Q3 turns on and discharges C2 to provide a pulse. Does anyone know why Q1 is used in this configuration without the emitter connected to anywhere?
R3 is 1.5m ohm.

[tom66]

It's a cheap and simple way of getting a low leakage diode. The disadvantage is that the breakdown voltage is poor, usually only 5V or so if using the B-E junction.

[Cerebus]

It's a cheap and simple way of getting a low leakage diode. The disadvantage is that the breakdown voltage is poor, usually only 5V or so if using the B-E junction.

Yup, typical reverse leakage for an 2N3904 CB junction (with E left open) about 4 pA (about 10 pA if you connect C to E). That's an order of magnitude better then you'll get from any commercial low leakage diode without selecting them individually for leakage - with the exception of some expensive exotica like the JPAD series (which are really diode connected JFETs in disguise).

What I don't see is why one is being used here. You normally do this when you need low leakage in reverse bias and that junction is forward biased and, at first glance, doesn't look like it is ever likely to see reverse bias.

[DDT]

I can assure you the cost was not an issue for the team who designed this (U.S air force). There must be some other reasons. By the way, what is that part number for Q1? there is absolutely no information when I googled it.

[floobydust]

That looks more like a white-noise generator.
But Q2 only allows a VBE across the reverse-biased Q1 C-B.

[TerraHertz]

Are you sure your circuit is right? And the 2N4250 - not misread?

It would make more sense is Q2 & Q3 were NPN, and Q1 is some breakdown device like a UJT.
Then the output labelled "Pulse out" could actually be pulses.

[DDT]

Are you sure your circuit is right? And the 2N4250 - not misread?

It would make more sense is Q2 & Q3 were NPN, and Q1 is some breakdown device like a UJT.
Then the output labelled "Pulse out" could actually be pulses.

Yes they are 2n4250 PNP. Referring to the schematic in the first post, R2 has smaller value than R3, thus voltage developed at the node between these resistors, which is connected to the base of Q2, is smaller than emitter voltage of Q2, therefore, Q2 is conducting. I tried the circuit on a breadboard, change C1 to 1uF and also placed a diode between R4 and Q3 to prevent charge of C2 going back to the circuit when power is disconnected. I also connected a LED to collector of Q3 as a load. After connecting to 9v DC, the C2 charges quickly and if you disconnect the power or decrease it to a point, C2 discharges through Q3 and LED. In the original circuit this pulse was used to activate a SCR.

[TerraHertz]

Aha. Sorry I missed the key phrase "Upon depletion of battery" in your original description.
So it's a battery discharge detection circuit. With adjustable trip level.  I'll shut up now.

[floobydust]

Q1, the strange BJT configuration, is acting as vanilla diode?
I commonly see an reverse E-B junction used as a ~7V zener, which would make more sense for voltage level detection.

X67A I could not find assuming 7734 is the date code.BUX67A but it's a TO-66 power transistor, a bit huge.

[Zero999]

Q1, the strange BJT configuration, is acting as vanilla diode?
I commonly see an reverse E-B junction used as a ~7V zener, which would make more sense for voltage level detection.

X67A I could not find assuming 7734 is the date code.BUX67A but it's a TO-66 power transistor, a bit huge.

The power supply voltage is only 5V, so it must be acting as a diode, unless that transistor has a much lower breakdown voltage, than the average BJT.

[Alex Nikitin]

I think it is actually very simple. The transistor works as a diode to provide a temperature compensation for the voltage threshold. That particular device was probably chosen for the right voltage drop, as the whole circuit works essentially on the difference of voltage between two forward biased PN junctions.

Cheers

Alex

[IanMacdonald]

As there is no positive feedback, I find it hard to see how it could provide a pulse. More likely Q1 is a unijunction or suchlike.

[Alex Nikitin]

As there is no positive feedback, I find it hard to see how it could provide a pulse. More likely Q1 is a unijunction or suchlike.

A "pulse" is a relative term, this circuit looks like a brown-out protection and would produce a reasonable "pulse" in such conditions. We don't know what it is connected to -  could be a Schmitt trigger  .

Cheers

Alex

[theleakydiode]

Are BJTs in diode configuration better thermally matched to other BJT's than regular diodes?

[Cerebus]

Are BJTs in diode configuration better thermally matched to other BJT's than regular diodes?

Not particularly. The only case where they would be is if you used transistors that are packaged as a matched pair. You can also get diodes packaged as matched pairs (and fours too) but they're less common than matched pairs of BJTs, which aren't exactly common in the first place.