The 3914 seems to be the natural choice for the KITT kit. It can operate in "Dot" mode, has up to ten outputs, and can run from voltages up to 18V. Most automotive power systems can drop as low as 9V (really dead battery) and have as much as 14.4V when the alternator is charging (a faulty regulator can easily output over 15V, so design for worst-case scenarios!). Factor in the possibility of reverse-voltage (swapped battery cables or jump-starting) and the inherently noisy enviornment (ignition and switching spikes), and you wind up with some pretty hefty design problems. So, the wider the operating voltage and the less the current draw makes for easier work. The 555 also has, mostly, these same parameters, so it looks, at first, like a match made in heaven. Or Silicon Valley.
My first design was to use the 555 in astable mode, with an integrator on the output of pin 3 to shape the square wave to a triangle. Feed this to the input of the 3914 (pin 5), and you have a sweeping effect. Or so I thought. The problem popped up when I looked at the datasheet for the 3914. The input is designed to "see" 0-1.2V, and light up one of ten LEDs based on that. Since there are ten steps, and 1.2V max, each LED represents 0.12V (it's different with the 3915/16 chips, but the range is still 0-1.2V).
That wouldn't work, as the output from the integrator would ramp from 0-to-gosh-knows-what and back down. There would be a time where the voltage was out-of-range, and you'd get either a full-bar or no display. Not what we want. We (I) want a consistent range, one that will track with the 555 and the varying Vcc of the car's power. This was the point where a cup of coffee, the drone of Phineas and Ferb in the background, and the 555's datasheet snapped on the Klieglight in my diabolical mind (Hey, Ferb! I know what I'm gonna do today!).
Even when you think you know a chip, there's invariably something you don't see in the datasheet until you really, really need it. Fortunately, it was the fact that I'm looking for a better way to send a triangle wave to the 3914 that the little graph showing what the charge/discharge waveform on pin 2 of the 555 (in astable mode) looks like - a triangle! Another look at the internal block diagram shows the divider network of the 555's comparators, which is a series of three 5K resistors. That makes the waveform hover between 1/3 and 2/3 Vcc. Perfect! All we need to do is match that over on the 3914, and we solve both problems. It won't matter what the voltage is, as we're always dividing it in thirds, and we keep the waveform in that middle third.
Sometimes, I just amaze myself.
Now, the only problem is to get the 3914 to play nice with the 555. Another peek at the datasheet, and a familiar sight was seen in the block diagram: the same type of divider network as the 555. The "top" of the network is at pin 6 (Rhi), and the bottom is pin 4 (Rlo). In the case of the 3914, there are ten 1K resistors making up the "ladder". So, if one were to put a 10K resistor from Rhi to Vcc, and another 10K from Rlo to ground, we'd have a "Vcc --> 10K --> ladder --> 10K --> GND" configuration, and the 3914 will now track the inputs right along with the "Vcc --> 5K --> CHG/DSCHG --> 5K --> GND" of the 555's tank section. Normally, the 1.2V internal reference of the 3914 is run to the Rhi pin, and that is why the input sees just the 0-1.2V range. Another quick check of the datasheet shows the input can easily handle the full power range, so we're just going to skip the internal reference for the ladder (but not the LED current!), and use Vcc. I don't know who designed the 3914 this way, but I'd like to buy them a cup of coffee.
Now, about that internal 1.2V reference: it's also used to set the output current limit for the LEDs. Putting a 1K between pins 2 and 7 allows 1.2V/1000R = 0.0012A, or 1.2mA. This is then factored by ten to get 12mA of current for the LEDs. Don't let this resistor drop too low, as the max current is about 15mA (which would be about 810 ohms - a 2K2 will give about 5.5mA). In any event, this value is why we won't need limiting resistors between the outputs and the first transistor.
After a quick breadboarding, the circuit worked as expected (a first!), with the exception of the fact that my resistor's manufacturer took the whole "10% tolerance" a little to the extremes. My high LED (LED10) wouldn't light up. Swapped the 10K pair around, and now my lower (LED1) was dark. My DMM shows that one 10K is 9912 ohms, the other is right at 9997. This was enough to throw things off. I'd recommend measuring your resistors to ensure that the two you put in there are as close to 10K as you can get, and that they're matched to each other (see Dave's video on resistor testing rigs). By the way, since 1/3 of 12V is 4V, and there are ten steps, that's 0.4V per LED. A trivial bit of info, but it may help someone when trying to diagnose a problem.
Lastly, we now have to drive high current bulbs from a 10mA output. An active-low output. Great. Experience has taught me that an NPN transistor likes to switch with a positive base input, and a PNP is just the opposite. So, a direct line to the 2N3906's base, a limit resistor on the collector to ground, and tie the emitter to Vcc, and we're in business. Like the 2N2222 or 2N3904 NPNs, the current capability of the TO-92 packages is about 500mA. Way too close to the T-194s, so we're going to add another stage. This time, since we now have an active-high, we'll just use some TIP-31s. This'll give you a good 3A if you want to run brighter bulbs. I didn't seem to need any base resistors on the TIP-31s, but shoving a 100R in there may be a good idea. Yes, I could've used two NPNs creatively wired up, but that would mean that one transistor would be on when the other was off, meaning that current would be flowing somewhere even when we don't need it. Sure, a couple of mA is nothing for a car, especially when there's typically 600A of reserve in the battery, but it's the principle of the thing. Using a PNP means that both transistors are either off or on. There. Much better.
If you've read this far, go collect your hazard pay.
For those wondering, the program used was gEDA, and I am still trying to cope with the damn thing. There's no easy way to edit components. The 555 shown has pin 2 opposite of where I need it, and the 3914 needs to shift some pins around for my particular application. There's no way to shrink overly-large devices (like those big Darlingtons) to fit, and it keeps attaching traces to points I don't want them. Those annoyances aside, I'm starting to like the (FREE!) program.
Disclaimer: I avow that KITT-A, KITT-B, and KITT-D are of my own designs, and that KITT-C was modified from an online collection (somewhere!) that I modified. I also avow that anyone, anywhere, and anyhow can use and abuse these circuits as they please, provided any neat revelations, improvements, or corrections of glaring mistakes make their way back to EEV Blog or other similar online electronics sites. EEV Blog and myself are not to be held liable if this circuit violates your local laws, ordinances, or female relatives.
nop
Now go out and build something.
Home
Help
Search
Login
Register
