Kevin Weddle

Remember that the current specified on a fuse is the peak current. You can have a 3amp average transformer and still get higher peaks. The wire amperage chart is the average current. In other words it's different than the fuse. The current for a 22AWG is about 1A average. Could somebody corroborate this. I did an experiment one time and stumbled across this observation. I used bare copper wire for the ouptput of a bridge rectifier and it blew. I used two kynar wire and it was ok. This tells me I was around the current I had calculated. I know I calculated average current so the peak must have been higher. I must mention also the way I calculate current. I break down the sine wave into equal parts and manually determine the current for each part. Then I get the average by adding the total current and dividing by the number of parts. This is most accurate I believe.

I am in the process of a project and I am concerned that I may need many different supply voltages. I want to revamp this circuit I have to include the correct supply voltage. I really want to maximize this circuit and the constraints of a few voltage levels is not acceptable. How difficult is it going to be to step down, filter, and regulate many voltage levels? I am also looking for a shortcut. Maybe I could tap a low resistor value divider to get the voltages. The current might be a little high and wasteful. Maybe a row of zeners would work. Surely this is how it is done in serious electronics applications.

Do you think a series transistor would work? Set the current you will need when the voltage is 24V. Bias the transistor with the input voltage. Use a second transistor to reverse the junction. Drive that transistor with PWM clock. You can step down the voltage with resistors to run the PWM clock. A PWM clock can be constructed from an inverter oscillator with diodes and resistors to dissipate the voltage from the capacitor. There is an example of this in the theory section called "pulse width modulation".

I tried to build an autotransformer, but it turned out to be a high current inductor. It doesn't quite obey the 20db per decade law. The reason I could not make it an autotransformer is because the reactance wasn't high enough to connect it to the mains. I needed the thing to consume 20A peak with the load connected. But it blows the 20A peak fuse of the mains even unconnected. I am considering purchasing a 20A autotransformer, but they are expensive. To make the autotransformer, I used about 40AWG magnet wire that I bundled together so that it would handle the current. I wrapped this onto a transformer core. That magnet wire is difficult to solder to. Can anybody guess what guage of magnet wire is used in a 20A autotransformer?I don't know if they bundle smaller wire together or just use a big wire. It would have to be a large wire to handle 20A peak current.

A pulse counter will continue to count the pulses indefinitely. Let's say you have a high frequency and the pulses you count are many for a certain duration. So you set the duration with the enable line. You still have a bunch of frequencies that need to be separated so that you have one frequency at the input to the counter. Use a bandpass filter. Amplify the frequency until it becomes a pulse. Now you can count the pulses and determine that you have that frequency. If it's not there, then you won't have any pulses to count. Do this with each frequency you want to make sure is there.

I don't think switch bounce is the problem. I can reset it and it still triggers because there is a pulse on the set. Does this sound like switch bounce? I am not really sure what switch bounce amounts to. I know that it is an irregularity in the pulse caused by the bouncing contact which would cause a ripple in the pulse.

I want to be certain that the latch is not in the unknown region when I try to trigger it. Can I skimp on the voltage to keep it from triggering?

I am using the Borland compiler and the examples don't seem to work. I make a call to a standard function in the correct header file and it says it doesn't exist. What could the problem be? I have had similar problems with serial communcations functions. I have a theory as to why it does not work. I think my windowsXP does not support these functions. Could this be true?

Whatever happened to the sine wave oscillator that used a square wave with an inserted DC level on the transistion. I think it was called a modified sine wave. The article you have listed does not include this tidbit of information. Do you think this is too high tech of a topic.

Sec has the exact quote that I intended to be illustrated. The emitter impedance has to be low because of the the two base resistors. You can't just choose high value resistors and expect to get the high impedance. The transistor dictates the value of these two resistors. This is why a need a special type of transistor. I can in effect make the change in voltage over the change in current independent of the base resistors as far as the impedance. Let's say the change in voltage over the change in current is 1kohm. The impedance is then 1kohm. But the change in voltage causes a greater change in current at the other end. This way, I have a 1kohm impedance with a change in voltage that causes a greater change in current at the other end of the device.

I have this latch I want to trigger, but it is very sensitive. So I used a pull down resistor on the input. The input to the latch is the ouput of a gate. This seemed to pull down the voltage so that it would not be so sensitive to trigger with a short duration pulse. Do you think this actually works or am I just in the unknown region of the device? It seem to me that you need both voltage and duration to trigger a latch. If you want to catch a pulse, but not so often, then you can skimp on the voltage and it won't be prone to triggering as easily.

I know that a transistor is biased a it's mid current value. From there the power must be decided on and that will determine the voltage. I suppose half power is reasonable. The real test is to determine the beta and the changes in VCE that will keep the beta constant. I would like to tell of an easy way to find that beta and and voltage. I think there does exist a Q point.

What I am actually looking for is the D flipflop arrangement in integrated form. Is there a more simple integrated circuit. The complexity of these circuits is unwanted.

Please let me know of a simple shift register with a clock input where the data propagates through the device. The one I bought shifts the first bit to the end and requires a shift out function.

You might consider the use of high power resistors, diodes, and capacitors as well as the triac. This way we can avoid the transformer usually associated with electronics controls. I know that current can be determined by the voltage across a resistor. You might throw in an SCR as well.

I am thinking of purchasing an oscilloscope. But I think I want one of those brand new ones from Radioshack. It is has a single probe and it's the handheld type. Could you convince me of a cheap used oscilloscope?

I would not use one of those. My experience has been that many devices are implemented in hopes of prolonging the life of electronics components. You should really not need to do this. I think that the components are hardy enough to withstand negative transients.

You forgot to mention the Bessel characteristic. By the way, these are just characteristics of the filter. Your filter will exhibit something close to one of these. Bessel is the most linear. Butterworth exhibits 20db per decade, which is a little steeper rolloff. The bessel is more like a one to one with frequency. This is desireable because the voltage can be linked to frequency directly. For instance, in the first decade, a drop of 1db might correspond to 100 hz and a drop of 2db will correspond to 200 hz. In the Butterworth, a drop pf 1db will correspond to 80 hz and a drop of 2db will correspond to 120hz.

Should I use half power to get VCE? I want the beta to have maximum range and still be constant.

RI is your base resistor. The emitter resistance of the other transistor sees this base resistor. The impedance, looking at the emitter of a transistor, is resistance of the base divided by beta. RF is a collector resistor as seen by the output stage. You are right, there are no diodes. But there are PN junctions in parallel with the resistors that can be used to calculate the current. You may not believe me when I tell you there is a different version of that circuit you posted. It's missing 2 diodes and a connection point. Anyways, the currents subtract with the upper current being higher and the load gets the rest of the current.

You have to investigate the circuit more closely. The opamp is rated for 20mA, which is what the circuit demands. Look at the ouput of an opamp, the emitter follower. You can easily approximate the current because of the diodes in parallel with the resistors. Notice that the load resistor is in parallel with the circuit before the emitter followers. The other resistor is an emitter element because it is a base resistor when compared to the input stage. Look at the input stage. The base resistor directly affects it's gain.

Why aren't multiplaxars and demultiplexars used for serial to parallel conversions? I understand that the operation of these devices could be used for conversions. Any help as to why would be appreciated. Are they used in chips to help complete these operations? I would like to see a communications project developed from these devices. A multiplexar has select inputs. The select determines which input makes it to the output. If you continuously change the select, you have a parallel to serial conversion. This is what a computer does when it uses the parallel port to create an RS232 condition. Maybe you could use the ouptput of a counter to cycle the select. The input to the counter is a clock. I would be interested in an idea to use the parallel port to send the data and convert it to serial. Serial data was meant to send data at a rate much slower than parallel data. So that the microcontroller could handle other tasks while waiting on the slower data ouput. If you used just parallel on a slower device, you would have to wait anyways. So why not free up the microcontroller by sending the byte loaded into a register then access that byte with it's input to the multiplexar. That way you could send the byte and forget about it. Since the parallel data lines of a computer are always on, you have to send the data and still be able to change the data lines. That is why you load it into a register.

Does the UART require programming? I thought it was just configured. What sort of programming is required?

The input signal as well as the output signal and the DC bias looks correct. What is wrong with 18ma load current? The gain is a little higher than RF over RI but the gain is much lower than open loop.

The 1Kohm to -18 can simulate a load. The current is even close enough. But more importantly, it shows that this is a collector resistor. Whenever you are able to reduce the voltage away from a rail, such as at 0volts, you are reducing the gain. Just apply the signal and notice the gain is much lower.