Everything that I build starts to smoke...

March 12, 2008 7:15 PM

I need help designing a simple circuit before I accidentally catch my desk on fire.

In order to test some equipment at work, I need to build a circuit that behaves like the following:

1. Tester pushes a push button
2. LED turns on
3. Circuit then supplies a voltage of 25-30 volts for approx. 10 seconds to one output line
4. Circuit then supplies a voltage of 25-30 volts for approx. 10 seconds to another output line
5. LED indicator turns off

Other considerations:
--I need to use this on a few thousand devices, so anything I can do to maximize battery life would be awesome
--Safety. What should I consider in a circuit like this?

I'm trying to use three 9 volt batteries in series as my power supply, but would be open to any suggestions for a better source.

Can anyone point me in the right direction? I've been tinkering with a 555 timer, but I'm struggling with the alternating output lines as well as the correct output voltages.

posted by jivesoul to technology (8 comments total)

How many amps do you need to deliver?

You should probably be building a low-voltage timer circuit with two outputs. Those two outputs drive two high-voltage relays, which gate your high voltage power supply.

Be careful!
posted by b1tr0t at 7:36 PM on March 12, 2008

B1tr0t brings up a good point -- the amount of current you want to deliver is going to change the design substantially. If it's more than a few mA, you're probably going to want to have your timer circuit switch a relay (physical or a MOSFET).

I think it's possible to do everything you want to do with a few 555s. You may want to get a basic book on timer circuits -- Radio Shack used to make a fairly good one (written by the legendary Forrest M. Mims) but it's been discontinued (AbeBooks has it, though).

What I've always found helpful was to design a flowchart of every action I want to happen in the circuit, and figure out all the time intervals between each state change or action.

There are a bunch of ways you could actually build the circuit. A lot depends on the accuracy you need. It's probably possible to chain a whole bunch of one-shot 555s, each set up to produce the time interval you need between actions, triggering the next timer and the action itself (via appropriate logic gates).

However, for more precise control, you might use one 555 to produce clock pulses and then use counters to actually trigger the various actions. (It helps if you can work it out so that your actions get triggered by nice round numbers of clock pulses, like powers of 2 or 10 ... that way you can use the output pins from binary or BCD counters directly as triggers. E.g., if you want something to happen after 100 clock ticks, you rig it to trigger on the 100s digit of a BCD counter -- the LSB of the 3rd of 3 rippled 4510s -- going high for the first time. This can use a lot of chips but it's fairly easy to diagnose with a 'scope.)
posted by Kadin2048 at 8:40 PM on March 12, 2008 [1 favorite has favorites]

If you need a more precisely regulated voltage, a boost converter is fairly easy to build. You can sample a boost converter chip from a number of manufacturers, and the datasheet will usually contain a typical application circuit and instructions on choosing the values of external components. You'll need to determine what current levels you need, as others have said.
posted by Krrrlson at 8:49 PM on March 12, 2008

Thanks for the input so far, every one.

As far as the amps go, I don't have an exact number but I'm sure that it's pretty low. If it helps, when my boss was describing how he's done this in the past using a manual SPDT instead of automatic timing, he said that he used a 741 opamp, some resistors, and 3 9 volts in series.
posted by jivesoul at 9:12 PM on March 12, 2008

You could do this really easily with a small PIC microcontroller. But that would require writing some software to do exactly what you want and is probably overcomplicating things.

An easier way would be to use a 555 timer as a pulse generator, and hook up its output to the clock input of a 4017 (or similar) decade divider. A decade divider has ten outputs which pulse in sequence as the input pulses. So, first (after a reset...) output 1 will be on, then after the input pulses, output 2 will be on, then output 3, etc...

So now you have a timed sequence of 10 outputs to play around with. You could hook up output 1 to a relay that will connect the 25V to the first UUT (unit under test) port, and output 2 of the divider to a second relay connected to the second UUT port.

Now you've got to worry about voltage and current. The 555 and 4017 devices will run off 3-15V, so you'd have to power them off another battery (or tap one of the three 9V's) or come up with another regulation circuit (maybe a 7805 or something...). Nothing too complicated. There might be a bit more trouble with current limits... you figure maybe 10-50mA to drive a small relay. The 4017 can only supply ~1mA, which won't be enough to drive it directly. You would then have to use something as a more powerful driver, such as a transistor (MOSFET or NPN bipolar, probably) or maybe a 741 opamp.

As a random aside, using 3x 9V batteries is probably a good idea. Although you don't really have to worry about dangerous shocks below 48V, a battery supply is generally safer than one connected to the wall.

Good luck!
posted by lalas at 11:33 PM on March 12, 2008

Easiest:

1. charge two capacitors through two trim pots
   (220uF and 100k trim pots to start, you might have to pad the pots with fixed Rs to get the right time constansts)
2. Use those charging curves to drive logic that will have the on off output you want
   (a single quad NAND will do in this case. The first output is just NAND of the two times - one gate - the second output is T0 AND not_T1 - the other three gates)
3. use the logic outputs to drive transistors which drive relays to control the 30V (you need a 100 ohm base resistor to protect the transistor, any old transistor will do, 2n2221 say -- you might need a trick to increase the current through the relay, but I don't think so)

Use an ATX power supply to get 5V for logic and 12V for relays, and the 9V batteries are fine for the 30V.

Bill of materials: 1 push button switch, 1 LED, 2 100k pots, 2 220uF caps, 1 quad NAND, 2 100R base protection resistors, 2 2n2221 relay control transistors, 2 relays = 13 parts and some wire.

If you're interested, I can draw up a schematic for you tomorrow..
posted by Chuckles at 1:29 AM on March 13, 2008

If the above is not straight forward to you, refer to the description of this time delay circuit.
posted by Chuckles at 1:32 AM on March 13, 2008

What is smoking?

Suggestions for a 555 or 556 timer are good, but they are limited to 15V or so, Vcc. This is the easiest way to get simple, consistent time delays. The internet is full of application notes and this is the obvious solution.

If your drive requirements are low, you can use transistors to make an output stage that can easily accomodate 27 volts using 2n2222's (or 3907s or 3906 transistors), which I think are good to 35 V, Vce. Again, all you are doing is buffering the outputs to allow higher voltage operation.

Relays like Omron G5V-1-DC12 (available at Digikey) would work and are easy to use if you don't understand transistors.

You can use a wall-wart to get the voltages that you need, but for a little test rig that only has to operate for a few thousand units at 20 seconds a shot, monitoring the battery voltage with a bench meter would probably do. The drawback is consistency versus ease, so if you want a repeatable test, use a wall-wart DC supply with a 24 VDC rating, unregulated. With a light load, it'll produce much higher open-circuit voltage which you can regulate downward.

This falls into the category of a 2 on a 1..10 scale. The boss has prior experience using 741's, so why not consider using them? Have him sketch the circuit.
posted by FauxScot at 5:14 AM on March 13, 2008