Level shift + amplify

Let's say I have a signal where 1v is digital 0, and 4v is digital 1.

I am a little lost how can I use PNP+NPN transistor to have +11..12(for 1) or -11..12V(for 0) output, provided that I have +12/-12 PSU?

UPDATE: This is for low-power MOSFET driver, which has to be done in discreet transistors.

• Is it a design constraint that you have exactly 1 PNP and 1 NPN? Or can you have more than one of each? – JustJeff Apr 13 '11 at 1:23
• The less the better, so it if it's necessary, it can be more. – BarsMonster Apr 13 '11 at 4:49

Is this what you mean?

• Very very close, but is that possible to replace output resistor with another transistor somehow to get faster transitions/higher drive capability? – BarsMonster Apr 13 '11 at 11:18
• @BarsMonster - I was just thinking the same myself. I'm still thinking ... – MikeJ-UK Apr 13 '11 at 11:47
• First off, @BarsMonster: good question, +1, I have really been enjoying thinking about it. And @Mike, good answer, too. +1. I have been thinking about this a lot, I think pretty much the right solution to this problem is a +/-12V comparator, discrete or otherwise, right? I have been playing around with a few ideas but I haven't figured it out quite right... never designed a comparator and the only bipolar op amp I have studied is the 741... – Adam P Apr 15 '11 at 2:14
• @AdamP The slew rate and bandwidth of the comparator will a limiting factor, especially with cheap comparator ICs. I am curious how far can you go with an IC+discrete. – jpc Apr 15 '11 at 2:30
• @BarsMonster, MikeJ-UK - To make it a push-pull: the collector of Tr1 is either at +12V (input low) or 0V (input high). You replace the output resistor (pull-down to -12V) with an NPN transistor in common-emitter. Use a resistor divider from collector Tr1 to -12V to drive the base. Choose resistor values such that base voltage is either just below or well above Vbe, depending on collector of Tr1 being 0V or +12V resp. – stevenvh Apr 15 '11 at 11:18

My attempt at a solution.

R5 and R6 are only there to model source resistance and are not part of the circuit. C1 and R4 are used as load for simulation.

V4: input source (10MHz 0.5 duty ratio 0-to-5V square wave for simulation)
V3: reference source (2.5V for simulation)
V1: +12V supply
V2: -12V supply

It's basically an op amp or comparator. I only did some transient simulations to check it out. It might suit your needs. Main drawbacks are complexity, 1:1 output level sensitivity to the supply levels, no output short circuit protection.

You would need to trim up the supplies a little bit if you need to get closer to +/-12V output. I would recommend increasing R1 and R3 to 5k or 10k or 20k. This would lower quiescent power consumption but would decrease output current capability--but you probably would not need to drive a lot of current with this, right? You would want to play with the values to get it right for your needs.

V3 sets the switching threshold. There is no hysteresis. You might be able to set up a positive feedback loop from output to input to make schmitt trigger to add hysteresis.

The output stage could probably also be changed to push-pull common emitter amplifiers instead of followers to get a wider output swing.

Schematic:

Simulation:

• When I see people inventing such crazy transistors schemes - I just feel stupid :-) Unfortunately, this is too complex :-/ – BarsMonster Apr 29 '11 at 22:32

That sounds like a TTL-to-RS232 level shifter.

In particular, you might be interested in 2 cheap MOSFETs and 2 resistors using the Dale N0XAS configuration.

• Unfortunately, this is not for RS232. This is to build fast (>10Mhz) driver of tiny N-mosfets (not these bulky power mosfets, tiny ones, like 2N7002). That's denies the idea of using MOSFET's by itself - it's just going to be too slow. – BarsMonster Apr 13 '11 at 4:51
• A TTL-to-EIA232 level shifter would have the output voltages reversed. – stevenvh Apr 13 '11 at 10:46
• @Bars BJTs are going to be even slower unless you go to great lengths to keep them out of saturation. Small signal MOSFETs are nice, quite cheap and fast. – jpc Apr 14 '11 at 10:39
• @BarsMonster: Even though cube-shaped CM canisters were not designed for your cylinder-socket LM application, perhaps they will still work with a bit of duct tape. :-) – davidcary Apr 16 '11 at 23:35