# LM339 Dual Supply driving 2N3904

#### markH

Joined Dec 5, 2009
19
The attached circuit is part of a home automation project which determines events and then counts them.
The input to the LM339 (configured as a dual supply zero crossing detector) is variable frequency(0 to 15Hz), variable amplitude(DC to +10/-10V), sinusodial-like.

The circuit as shown, does in fact work pretty well.
However, there is some variability in the results which I can't let go of.

Poking around circuit while no events are being generated (LM339 input is -1.3V), the voltage at point A is -15V, at point B is -12.6V, and at point C is -4.9V. According to LTSpice, these voltages should be -15V,-15V, and 0V.

I suspect, but do not know, that I am driving Q1 (2N3904) outside of its operating range and perhaps have damaged it.

My question then, is this a valid use of a 2N3904 with a dual supply LM339? If not, what design change(s) are called for? TIA

Best Regards

#### Attachments

• 91.9 KB Views: 43

#### ericgibbs

Joined Jan 29, 2010
16,790
hi mark,
Please post your LTS asc file, we can then check the circuit.
E

#### Papabravo

Joined Feb 24, 2006
19,578
it certainly looks sketchy to me. Consult the datasheet for the 2N3904. Pay particular attention the reverse bias limits of the part listed under "Maximum Ratings"

Vebo is listed as 6.0 volts. This means the voltage on the base CANNOT be more than 6.0 volts below the emitter voltage. the emitter is at point C @ -4.9 Volts and the base is at point B @ -12.6V, So:

$$\; - \;4.9\text{ Volts}\;-\;(-12.6\text{ Volts})\;=\;7.70\text{ volts}$$

A clear violation of the absolute maximum rating. Because of the size of the resistors which have a tendency to limit the current the device may continue to function, but I would consider it compromised. If this were a manufactured product I would issue an immediate recall and I would definitely redesign this circuit.

Last edited:

#### BobTPH

Joined Jun 5, 2013
6,077
A schmitt trigger would be better on the input sde. And I have no idea why you need two transistors after the comparator. One to invert the sugnal makes sense, but why two?

#### markH

Joined Dec 5, 2009
19
I have attached the ASC file for the entire project.

Rest assured this is not a commercial product, it only lives on a protoboard in the basement and provides event count data to a HomeSeer HS4 home automation system.

As far as component selection, well, the 2N3904s were in the junk box...

Best Regards

#### Attachments

• 10.2 KB Views: 1

#### markH

Joined Dec 5, 2009
19
A schmitt trigger would be better on the input sde. And I have no idea why you need two transistors after the comparator. One to invert the sugnal makes sense, but why two?
There are two different counters, a stand alone external counter, and the 4N25 which drives an separate counter on a 1-Wire network.

As a total novice in circuit design, having separate circuits for each counter seemed a prudent way to avoid any possible interactions

#### ericgibbs

Joined Jan 29, 2010
16,790
hi mark,
The Vbe d/s values for the Q1, 2N3904 are being exceeded.
E

Last edited:

#### ericgibbs

Joined Jan 29, 2010
16,790
hi mark,
Adding 3, 1N914 clamp diodes to those Vbe's should solve the problem.
E

#### DickCappels

Joined Aug 21, 2008
9,512
The penalty for low current avalanching of the emitter-base junction is loss of low current beta. Should not be a problem with this circuit since the revers bias on Q1 is a few hundred microamps.

I agree with Papabravo - I would not allow the reverse bias in a product to be offered for sale, but at the same time it is unlikely to give you any problems.

#### markH

Joined Dec 5, 2009
19
hi mark,
Adding 3, 1N914 clamp diodes to those Vbe's should solve the problem.
E
View attachment 280467
Fantastic, Thx! Fortunately I still have a real electronics store nearby, so a trip first thing tomorrow.

#### markH

Joined Dec 5, 2009
19
The penalty for low current avalanching of the emitter-base junction is loss of low current beta. Should not be a problem with this circuit since the revers bias on Q1 is a few hundred microamps.

I agree with Papabravo - I would not allow the reverse bias in a product to be offered for sale, but at the same time it is unlikely to give you any problems.
I appreciate your commercial concerns, but this is really a hobby project that I've been messing with for (literally) years. Hopefully with the diode fix, my event detector/counter will be one more step closer to finality.

Thanks to everyone for your participation!