By the way, perhaps the reversed diode in the AC circuit for the LED was actually a zener!
Then, the circuit is exactly correct, given that he computes the zener voltage correctly and the resistor correctly. He also has to account for the fact that only 1/2 cycle will flow of the AC waveform, when he computes the power consumption. For the power computation, see my post in this topic for a hint about how to get the right answer! [For the 2v & 1ohm,
I don't often get involved in these basics things as much as I should, but sometimes I feel compelled to give a hint or two...
For computing power on a given circuit, it helps to think of locating all the forcing functions that produce a voltage or current, and then identify the loads of interest. If you have a voltage source, calculate the RMS voltage (which is defined to be the integrated voltage for POWER purposes). See the integral below, which applies for calculating the RMS value for one periodic piece of a waveform. In this case, T1 and T2 are the start and stop times of one period of the switch being on and off. To evaluate, break the integral into 2 pieces, one for the time of the switch being on, and then one for the time of the switch being off (T1 to Y for the first where the switch is on, then Y to T2 for the 2nd where the switch is off).
When I graduated it was the era of 8008/8080. ADCs were pricey and every design (except maybe military) used a mux ahead of the ADC. I once spoke to a guy who graduated with me and so should have followed nominally the same growth path. He was designing an autoclave (sterilise medical instruments) and used about 6 individual ADCs that chnaged slowly, because he hadn't followed the technical journals and app notes.
Today things have changed, not completely, but it is now far more common to see multiple ADCs in a system. You can even get multiple ADCs inside a micro.
Over my 30 + yrs career I could write a book of these ' lack of basic knowledge ' stories It is incredible how the more education engineers obtain the more they totally loose they basic EE knowledge ( if they ever had any ? ). I find it beyond comprehension that a PhD candidate can get a PhD degree and never once really do any real hands on electronics design.
I have a good friend who is a prof at a well know military school, who related such a story to me. Another EE professor who taught power systems engineering ( he has a MSEE in power systems ) was watching my friend prototype a circuit design. He ask my fiend which parts were the capacitors and which were the resistors ???
Gee I would not like to challenge him with a question as to what a drive motor looks like ??
"Why do some people (usually PhD's) use a high speed 16bit ADC with no anti-alias filter to measure something that could be done with 12bits a few times per second?"
I feel for you, mate. I have MORE THAN ONCE been told "why would we need a filter? We're only sampling at once per second" Once, when the data turned-out bogus, the response was to discard all "out of range" values (leaving sparse data, and completely disguising a serious problem that later cost us >>£1E7). On THIS project, the decreed plan is to go to 16-bit, 200kS/s to solve the noise problem. The thought is that bigger spending gives better results, no? :-(
Incidentally, I've been in this game a long time, and the (a) correct way to do this stuff seems obvious to me, but maybe it isn't to everyone, so just let me describe a good default way to go when doing data acquisition:
Sample as fast as possible. If the achievable sample rate varies because of variable number of channels, use the lowest fs you are stuck with.
Use a correct anti-aliasing filter, that gives adequate attenuation at fNY (=fs/2) for your application. Its corner frequency must be lower than fNY; with a single-pole attenuation curve and high number of A/D bits, it must be surprizingly much lower. (If you want high (16-bit) accuracy, you have to trade-off frequency or filter complexity.) This is why you want fs as high as possible. Another reason is so you don't have to mess-around with this analog design once you have it working.
In the digital realm, perform any necessary calculations (offset cal, product [to get I*V=>W]), and then digitally filter the data on-the-fly down far enough that you can then decimate it to the needed final sample rate. This is done in every ΔΣ converter and is not weird or hard. With easy-to-deploy library "decimation" or "dual-rate" filters, the computations are very very efficient. "Averaging" is a kind of naïve filtering, but not as good as a proper Bessel, Butterworth Cauer, etc. design.
Of course there are lots of optimizations and compromises to be considered, but I think this is good as a basic approach, and I can cite a mountain of white-papers, appnotes, manuals and textbooks for further reference.
If you are stuck with a stream of bad data, there are only so many things to do. Heavy filtering, and relying on the gods of Noncorrelated Noise and the Law of Large Numbers are about it.
One guy I knew, who knew enough simple math for this, once got a 110V soldering iron in a set of tools from the USA, but he was in a 220V location. He figured that all he had to do was put a diode in series with the iron element and he'd be OK. The soldering iron burned out and he could not figure out why. I made him do the calculations, working out the resistance of the iron from the power at the rated 110v, then the power at 220v. Then halve that with a diode, and it's still twice as much as you should have. You could almost see the light bulb going on over his head :-)
Surprising how many engineers don't get the PWM = linear variation of power concept.
One engineer I knew, that did understand the PWM concept, and then finely tuned his heater control loop, failed to appreciate the effect of "24v" supply voltage on a vehicle that could vary from 20v up to 29v (that's a power variation of 2.1 to 1).
Same guy failed to understand the concept of bouyancy, so putting the aforesaid heaters in top of the cabinet, and putting fans in the bottom with a chilling coil, assumed the air would circulate up one side and down the other. OK, With a small amount of heat , yes, the air did circulate, but put in a lot of heat , and you are left with a puddle of cold air at the bottom, and very bouyant hot air at the top, and the fan would not be strong enough to offset the bouyancy. Of course it's all in a feedback loop and basically latches up.
PS also agree on your other points, Why do some people (usually PhD's) use a high speed 16bit ADC with no anti-alias filter to measure something that could be done with 12bits a few times per second?
NASA's Orion Flight Software Production Systems Manager Darrel G. Raines joins Planet Analog Editor Steve Taranovich and Embedded.com Editor Max Maxfield to talk about embedded flight software used in Orion Spacecraft, part of NASA's Mars mission. Live radio show and live chat. Get your questions ready.
Brought to you by