99Hawk120

Got an interesting problem... and I don't have any catalogs or spec sheets any more to know what chips are out there to do what I want.

I have two AC voltage signals. They're probably at most 15V peak, varying between 15V and 0V (no negatives). The information is carried within the frequency portion, so the voltage could be stepped down without affecting things.

I need to create a circuit that will do the following:

Switch 1 must turn on when Signal 1 > (frequency threshold 1)
Switch 2 must turn on when Signal 2 > (frequency threshold 2)
Output of the circuit must turn on when Switch 1 AND Switch 2 are both on.
Output of the circuit must be capable of sourcing enough current to turn on a relay or switching power transistor.
The frequency thresholds MUST be adjustable after the circuit is built. The ideal method would be with a potentiometer.
The frequencies being used will be less than 1kHz (preliminary calculations suggest 313Hz and 77Hz).
The circuit must isolate the input signal, as other electronic devices are using them and I don't know what they're doing with them.

Ideally, if there was a digital chip that acted as a frequency switch (Logic 1 above some frequency, logic 0 below) that would be perfect. A simple and gate could then be used to tie the two together and drive the output.

Anyone got any ideas or know of such a chip?

MikeS

It sounds like you want to use something like an adjustable hi pass filter for each signal and combine the filter outputs with a logic gate. I don't really have any resources right now except for the web, so really the best I could do is look for schematics of circuits that might help and try to make sure I get what you are trying to do. Just what are you doing anyways?

99Hawk120

You basically nailed it, but I don't think a standard RC filter will work. I need something a lot more "on/off"--RC filters drop down too slowly.

As to what exactly I'm doing, I'm trying something clever with the skip shift solenoid. The two frequency signals are the speedometer and the tachometer.

Spifz

Ahh
yes,
see my post asking what the signal is coming out of the tach?

Is that the answer? is it AC 15v? Thats cool
ok, here's basically what you need:
A 'missing pulse detection circuit' so we chop the AC into an approximate square wave, and then, like you said, using a pot, adjust our pulse detection circuit to 'turn on' when it isn't getting pulses from your input as regular as it wants them.

For example, if your input is only producing 500hz, and the circuit is adjusted for 600, then it will expect a pulse every .0017 seconds, but one will only arrive every .002 seconds, so the circuit will provide an output of '1'. Anyway, I gotta go, I'll give you a better rundown, including the circuit ;-p later.

99Hawk120

As far as I can tell about the tachometer, it's a square wave of unknown on time and unknown off time, with a frequency of RPM*4/60 Hz. I think if you just count the leading edge and measure the time between pulses, you'll be set. The magnitude of the square wave will have a peak of Vsource and minimum of 0. (Vsource = the voltage at the battery, +12V when the alternator is not charging, +13.5V to +15.5V when it is).

The speedometer appears to be an AC sine wave of unknown voltage, with a frequency of 4000ppm (pulses per mile) for the electronic speedometers.

Spifz

Here is a Missing Pulse Detector circuit from the good 'ol radioshack 555 timer book

Mims, Forrest M. III, Engineer's Mini-Notebook , Fort Worth TX, 1984

the input is the square wave, or anything close to a square wave, as long as the voltage goes above and below 5v I think. The output is on as long as the input pulses are happening as fast or faster than what the circuit is set for. If the frequency input drops below the setting, the output will go low. So I think two of these circuits back to back, sharing a 556 if you want to get tricky, and feed the output into a NAND, then the output will be what you want.

99Hawk120

Hmm.

Unless I'm misreading that circuit, it won't do what I want.

Let's say I want to trigger on a frequency of 100Hz/10ms (because that makes the math easy).

If the incoming frequency is greater than 100Hz, great, the output stays high.

If it's, say, 80Hz (12.5ms), here's what you get:

When the pulse comes in, it kicks the output high. After 10ms, it expects another pulse and doesn't get it, so the output goes low. Problem is, the next pulse 2.5ms later kicks the output back to high, and then 10ms later it drops low again.

That ain't gonna do the trick if that's how I'm reading it. I can't have oscillations in the ouptut signal, it either has to be DC HIGH, or DC LOW, and at 99Hz it needs to read a constant low, and at 100Hz it needs to read a constant high.

jwscab

OK, it looks like you need a frequency dependant switch. The best thing I can think of off the top of my head is a frequency to voltage converter, followed by a comparator, with the threshold set to the voltage converted from a specific frequency. The converter is a National Semi Part, and you're in luck, it is versatile enough to do exactly what you need all in one chip(comparator is on board). The part number is either LM2907 or LM2917, one has a built in zener diode for better supply regulation, which probably isn't a concern. Search Nat's website for the data sheets, they have application circuits, much like you need. Hope that helps.

99Hawk120

I found it like 2 minutes before you posted even though I've been searching the web for days with no luck.

You're right, that's exactly what I need. Now to get one and build it...

Spifz

Oops, yes, I agree, not exactly what you want. If you are going to try to approximate the frequency with the time period of one cycle, then you will have to choose between two cases:

1) you always assume that the input freq. is higher than the target in which case you will falsely pulse the output when the input frequency is too low

2) you always assume that the input freq. is lower than the target in which case you will falsely pulse the output when the input frequency is too high

This is because when you see the rising edge, you have no idea whether this current wave will be shorter or longer than the target, so you have to 'guess' by having a default state until the circuit has timed the frequency and determined (correctly, but late) whether the frequency is higher or lower.

I guess this kind of circuit won't quite do what you want, I'll keep looking

99Hawk120

OK, I've figured out how to design the circuit I want using the LM2907.

However, that chip can only output 50mA max. I know the load I want to use is gonna need more than that. Can't tell you off the top of my head how much, but I'm guessing 1-5 amps.

So clearly I need to add a driver circuit. A simple relay would work, but I'd like to do it all in solid state. The constant ticking of the relay turning on and off would get annoying, especially for the reverse lockout solenoid. I'm sure there's a way to do it using a power BJT (transistor), but... I've forgotten how. Too much time spent stuck in front of the keyboard and not enough in front of a bench anymore.

Anyone care to help me out here?

Tim Burgess

You may want to use a MOSFET to control the load. You will be needing a voltage comparator stage with hysteresis on the output of the LM2907 & the comparator may be used to drive the MOSFET directly in your application, since fast switching time is not critical. If you are sourcing current from +12 volts, use a P channel MOSFET, and if you are sinking current, use an N chnnel MOSFET. With the P channel MOSFET, the device is turned on when the gate to source voltage is negative. With the N channel MOSFET, the device is turned on when the gate to source voltage is positive. If using the P channel device, connect the source to +12 volts, and the drain to the load. If using the N channel device, connect the source to ground and the drain to the load. Check out www.digikey.com for the MOSFETs available. Id is the maximum drain current, Pd is the maximin power dissipation (Idmax^2 * Rds on), and Vds is the maximum drain to source voltage. Parts are available to handle up to Id = 75 amperes, easily.

99Hawk120

Tim,

According to my tech specs on the LM2907, it is not just a frequency to voltage converter. The IC itself already contains a comparator. Why would I need another one after it?

In addition, the circuit I have already has a hysterisis in it, so that shouldn't be a problem.

The problem I foresee is the "load" is placed between +12V and the collector of the built in "50mA max" BJT transistor. I can't seem to find a way to hook up the MOSFET to make it work.

Take a look at the following two spec sheets:

http://www.national.com/ds/LM/LM2907.pdf
http://www.national.com/an/AN/AN-162.pdf

The circuit I want to use is in Figure 9 (page 8) of the application notes (second file).

What would happen if I place a resistor in place of the load, and another resistor between pin 4 (the emitter of the built in transistor) and the ground? Then I should be able to use the voltage across the resistor to turn an N-channel MOSFET on and off, right? Connect the gate directly to pin 4, connect the source to ground, and the drain to one side of the load? Unless I'm missing something...

jwscab

You can use either a mosfet or a power transistor, a darlington type, which can handle alot of power, and don't take much to drive into saturation('on') The mosfet will work, but I've had more luck with the BJT's. I've used 2N6045(NPN) before with good luck, they handle alot of power. Onsemi is the manufacturer, the also offer PNP devices. Good luck.

99Hawk120

I spent some time doodling circuit diagrams yesterday, and I think I have something that will work. Now I just need to get me some chips and try it out.

Thanks for the help everyone!

Tim Burgess

To drive an N chnnel MOSFET from the output of the 2907, do the following:

Connect pin 4 (emitter) to ground. Connect pin 5 (open collector) to a 1K resistor. Connect the other end of the resistor to Vcc (pin 6). Connect pin 5 to the gate of the MOSFET. Connect the source of the MOSFET to ground. Conect the drain of the MOSFET to he high current load. The other side of the high current load should be connected to +12 volts. When the output of the comparator is low, the BJT is off, and the MOSFET is on.