"PPP" transistor

Rather confusing... What appears to be clear is that you have this signal that goes from

0V to 5V and you want to drive four outputs each going from 0V to 10V. You do not specify that the output signal should be some kind of replica of the input, you go off what seems to be tangests (1) voltage divider, (2) some kind of mythical (as there is no design to analyze) current surge, (3) some kind of smoothing for no perceived reason, (4) a false 1Mhz CMOS limit, (4) something about biasing resstors (again, no design to analyze). At least we know that ther is some kind of 12V supply, but there is no info other than "approximate": regulated, unergulated, current capability, etc? Well, a simple amplifier can be set for a gain of +2; if you want some isolation between 4 output connectors one could use resistors in series withthe opamp output and each connector (say 100 ohms). One opamp, two gain-setting resistors and maybe 4 output "isolation" resistors. There are a lot of mid to low power opamps that have over 10Mhz GBW that would fill the bill.

I would be thinking of something like a 10 volt regulator and CMOS buffers that convert the 5 volt logic to a 10 volt output driven by this regulator.

I don't understand what you mean by "smoothed".

What about the 5 volt logic signals isn't smooth?

I wasn't clear - it's PWM. The idea is to get a voltage that I can control from moment to moment, between 0 and +10v. (This is to drive a dimmer).

It's easy to get a voltage between 0 and about 5v, by switching a logic signal on and off rapidly in a given pattern, and then smoothing the Ons and offs to an average, across a cap. If the pattern is "on half the time, off half the time", you get +2.5v out, and so on.

Except I have 4 independent sources of PWM, and I need to use them to generate 4 independently varying voltages from 0 to +10v, all sharing a common ground. The simple way is to take the on and off sequence and use it to switch a +12v line on and off, voltage divide that down to

+10v, put it across the cap, and I'm there.

What I have been doing is using a norton op amp to switch the 12v on and off, but that's cumbersome and takes up board real estate. And a CMOS buffer IC, which would be happy to handle 12v for me, appears to be limited to about 10ma output, which is just a little too close to my worst case current usage. I'm hoping there's something out there like a ULN2003, except able to source voltage instead of sinking it.

Okay, thanks for the additional information. Whit this in mind, I suggest you look at active low pass filters (perhaps 2 or 3 pole) made with LM324 quad opamps, all 4 filters with one pack. If the filters have a gain of 2, the original 5 volt PWM waveform will be converted to a smooth 0 to 10 volt signal.

You might download a copy of Filter pro, to explore some possibilities.

Actually, I was missing the obvious. A single NPN transistor, used as emitter follower, will get me exactly what I want. Wouldn't a ULN2076B do the trick? Or a CA3802?

It won't give you exactly what you asked for.

So how can I know what you want?

I'll try to simplify one more time.

I want a single component that will take in a square wave with a varying duty cycle of +5v and 0v (standard output of a logic pin on a PIC), varying as fast as 1Mhz. I want it to output the same waveform, except at 0v and +12v (or, allowing for voltage drop, maybe 0 to 11v). I need about 12ma of current available on the output. I have about 10ma available to drive the input These numbers (5v in, 12v out, 12ma, 1Mhz) put it outside the reach of a CMOS buffer. I can do it with a Norton Op Amp (I know, because I have), but I want a lower component count.

But I think a simple NPN transistor set up as emitter follower will do the trick. I just need to pick an NPN that is happy to be off at logic

0 and on at logic 1, and can handle 12v, which sounds like almost any medium power NPN with a 4.7k resistor between the emitter and ground, and the collector tied to a +12v rail.

What happened for your requirement that the pulses be smoothed to an average DC voltage? An opamp active filter would take very few parts, do the averaging, and provide the output current as a buffered copy of the smoothed voltage.

I don't think the 0/5V logic will switch the NPN follower when the emitter is at 10V. Perhaps you need a common emitter NPN switch driving the base of a PNP with emitter connected to 12V (perhaps with the necessary base resistors contained in the transistor packages to save board space). Output and smoothing cap connected to the PNP collector. You may or may not get the correct sense of voltage versus PWM duty cycle, if this matters.

Steve

John, something else I've been waiting for someone to mention since the first post by the OP (I'm a hobbyist and only have rather narrow viewpoints on the subject) follows from this:

From this, I focused on the "in frequencies ranging from really slow to about 1Mhz" and the "PWM" parts. Frankly, my own limited experience with PWM is that I select a frequency and then adjust a duty cycle within it. So I was left wondering what "really slow" might mean!

So then I read:

and,

I noted the "smoothed" part in the first line and the "run the waveform across a cap to smooth it" after the voltage divider, which I naturally took to imply a source impedance that would couple with the cap to yield a filtering of the PWM.

So all this meant that an analog voltage was desired, developed from a PWM. But this brought me back to this "really slow" to MHz thing. How can a simple RC filter work here? There has to be a basic PWM rate going on and, so far, I haven't seen any comment about what that is.

Anyway, I'm confused.

Jon

Why? What's the criteria for "really slow"? Is it actually PFM? In any case, you'll need a smoothing filter - you don't hang the cap right on the CMOS output, you put a resistor there - that makes a first-order RC low-pass filter.

You will, however, have a problem driving CMOS at 12V from a 5V signal, so you'd need level shifters. Actually, a ULN2X03 could do the job -

+12V | R1 | 5V PWM --[ULN2003]o----+---R2---+--- 0-10VDC out | C1 | GND

If you have control of the PWM, you can adjust your duty cycle in S/W to limit the positive excursion to 10V. Since the ULN is only an NPN darlington, your pulses will be inverted, but that can be fixed in S/W, or you could use a 2803 (8 sections) and double up on them to get another stage of inversion.

Of course, selection of R1, R2, and C1 values are left as an exercise for the reader. ;-)

Good Luck! Rich

That's two of us.

I have seen PWM implementations that included variable frequency, as well as variable duty cycle (thermal controls that have a minimum on or off time of a line cycle, that gain extended dynamic range by extending the timing cycle once the minimum on or off time approaches a line cycle). But this poster has not been consistent in his description of what he is trying to do.

Why are you confused? He has some kind of vague input signal that he can't define properly, wants some kind of vague output, and expects you to read his mind, sort it out and devise a circuit. :-)

Ed