indulis

This is a little misleading in that ALL diodes exhibit this phenomena, forward bias-on, reverse bias-off, not just switching diodes... the "switching" part of the name refers to speed as it relates to reverse recovery, not it's "action"!! There are also diode's which exhibit a "soft" reverse recovery characteristic, and some with a "ultra-fast" recovery characteristic. Diodes like the 1N914 and 1N4148 are lumped into a catagory called "small signal diodes".

The moderator is not wrong!! There are cases where you do want to do this, but IT IS NOT nessessary everywhere!! Guru... how exactly do values in the range of 10,000uF and RF go "together"??

Actually it is better because the ESR will be lower with parallel caps. Who would use a capacitor of that value at those frequencies???

1000 mils is an inch mils are not mm 1 mil is .001" As I recall, .0014" (1.4mils) is 1oz Cu... so 2oz Cu is 2.8mils... etc

As the 431 starts to draw current through R72 it will start to turn on Q13 which will load the 3.3V output in the event "it" has a light load and hold the output within load regulation limits.

It really depends on the application. This is true, but you have to use the RMS value of the ripple current, something that is not always easy to calculate or measure. Sometimes you want ESR to be high, sometimes you want it to be low. For example... in a SMPS, the input filter likes a higher ESR to... 1) kill the "Q" and 2) change the filters "output impedance" relative to the input impedance of the converter (DC-DC conveters can look like a negative input impedance and there can be interaction which can cause oscillations). Output cap's like low ESR to reduce ripple, but the transfer function has a "zero" caused by the output capacitance and its ESR which can be a pain when trying to stabilize the feed back loop.

If you "short" a diode, it will never "turn on". If it never turns on, no current will flow through it and all the current will flow through the short. In other words, no the diode will not be destroyed.

As I said... try this with a battery operated/powered scope this will give you a clue!!!

I believe the answer is "leakage current" in conjuntion with a capacitive divider (as mentioned above). If your hands are moist, touch the tip of the scope probe with one hand and the scope ground wire between two fingers with the other hand. Squeeze the ground wire harder between your fingers, and watch the waveform... it will change in aplitude and will always be of the same freqency as the local line voltage. Guess what happens if you do this with battery operated scope? When I say "moist hands", I don't mean with water, I mean with perspiration, which, because of salt content, is a better conductor than plain water.

How does one select the type of power supply and or topology to use when designing a power supply? Should it be continuous, discontinuous, isolated, non-isolated, voltage mode, current mode, forward, flyback, Cuk, synchronous, non-synchronous, etc., or not even a SMPS at all and a linear instead? The target application and cost as well as other factors dictate/influence the final choice, and each and every one of "them" has advantages and disadvantages... it's all a game of trade-off's!! Consumer electronics tends to focus on cost, hence more profit for the manufacturer, while other areas, telcom, military and hi-rel for example, have different needs… load sharing, or maybe it's low noise, good transient response... etc. Obviously Audioguru has amassed a great deal of practical knowledge over the years in his career in consumer electronics, I believe... at Philips or was it Panasonic up there in the Toronto area? Don't get me wrong, that knowledge is worth it's weight in gold and many on this forum, as well as others, have benefited from it!!! However, on the basis of statements like To which I will once again say... just because you ain't seen it, doesn't make it so... and just because I didn't give a "real world" example of "this" circuit in a product you can buy a Radio Shack, still doesn't make it so. No I NEVER said "...none are made that way...", I said it may not be very practical to do so (If your gonna quote me at least make sure that it's accurate). Basically, that just means you won't see it in any commercial products, but there may be applications where you may want to make it that way. Right back to where we started... V=L*di/dt says it all... this is "THE EQUATION" for the voltage across an inductor. If your up on you calculus, take the limit of the function as "t" approaches zero... there's your answer as to high the voltage can go. Sure there are higher order affects that come into play that limit the absolute number, but as a fist order approximation, it is close enough here. So, unless you can prove this formula is wrong, some of your statements are indeed incorrect. If on the other hand I am wrong, please, by all means prove it, but not by saying... show me a product where it's used, and if you can't, then it's not possible. Show me analytically that I'm wrong. Like I said... I'm no genius

Just because Audioguru hasn't seen one, doesn't mean you can't make one. That's not to say it's "practical to do so"... The theory and equations were given.

It would be fair to say this is a subject that I know a "little bit" about, being a DC-DC converter (SMPS) design engineer. Audioguru's statement... is incorrect. My response was... The point is that you don't have to use a transformer (for flyback's its a coupled inductor). Why do you believe what I said is wrong?

Hey I'm no genius... what did Audioguru say that was correct that I said wasn't.

Even my old Eico 5MHz kit scope I built over 30 years ago could do 1KV, so I doubt and newer scope is limited to 100V. Since you didn't want to give these a go... answer: Yes there will be a spike. The di/dt is from the current going from some level to zero. Again it's... V=L*di/dt As time (t) approches zero, V will approach infinity. Answer: A load will not alow the current to stop flowing quickly, and you want it to be as fast a s possible. If you "plug-in" the numbers, you'll see that 10Hz is way too slow. 10Hz has a period of .1 sec and for a 50% duty cycle, the "on-time" is .05sec If we rearrange the equation... so i=(V*t)/L you get (15*.05)/.1x10^-3=7500A Just a wild guess, but I'd bet that the 15V source can't supply that much current. It's a good bet the inductor is saturated!! To see where the inductor starts to saturate, place a small value resistor in series with the inductor (sub 1 ohm). If you look across the resistor with your scope, you should see a voltage ramp whenever the MOSFET is on. The ramp should be linear. If it starts out linear and then starts to "tail-up" then you are on the edge of saturation. If you have saturated your inductor, you now just have a piece of wire in series with a "on" MOSFET across your power supply... in other words a short!!! The Rds on of your 900V MOSFET is around 1.1ohms, so to a 15V supply that's a pretty good short... around 13.5A worth. I wouldn't think your supply can suport 13.5A!! Increase the switching frequency so that the voltage ramp across the resistor on longer tails up and see what you get then. If you know how much current your 15V supply can source, you can calculate what the minimum frequency is that you have to switch at.

The formula is V=L*di/dt . You need a high di/dt or a very large L. This doesn

I'd be cautious calling it a zener!! What you said was... Sounds like you used the "transformer" word to me. A non-isolated flyback (or if you like "... a step-up power supply that uses a single Mosfet and only a coil...") can generate a voltage just as high as a converter with a transformer (or coupled inductor). Let's do one better, here are the question's... in "theory" (in a ideal electrical world) how high does the inductive voltage spike go??? Why doesn't it get there, and what are some of the things that limit it??

Mr. Klampfer On occasion, some of "your theories", only "loosely" follow fact, and are not 100% correct... for example: This statement is 100% incorrect. A boost converter (what you call a "stepup" power supply or is know more commonly as a non-isolated flyback) and a isolated flyback converter, which has a "transformer" (really a coupled inductor), are considered to be the same, and you can generate the exact same voltage levels from either. You stated it can't step-up the voltage very high with out a transformer... that's just plain wrong. The same applies to your MOSFET response above

AG You may want to review power MOSFET construction and operation (i.e. Miller Effect, and body diodes) and give that explanation another go.

The diode is there on relay coils to protect the switching device. If someone put a R in series with a diode across a relay coil, it was to limit the current thru the diode. To control rise and fall time's, most common are RC and RDC snubbers.

Audioguru Both of those quotes are yours!! There are better ways to add the reference... Vin may not always be constant.

Think about it... what makes a "better short", a diode, or a diode in series with a resistor when it comes to discharging an inductor?

and So which is it?? Adding a TL 431 to this circuit is VERY, VERY SIMPLE!!! A1 can even be used as a buffer. Your joking when you say a non-presision, non-temperature compensated zener will work better than a "real reference" ... right??

You sure you want to stick with this... (hint: a resistor LIMITS current)

Both of these chips are nothing more than CURRENT drivers for inductive loads. I have to go with Audioguru on this one, in that speed, of a DC motor is controlled by PWM or voltage change (resulting in a current change). Just about every "driver", at least the ones I've seen in DC-DC's are TTL compatible, so that don't mean much in terms of control method. A fixed duty cycle, change the frequency approach, I would speculate only works for the very reason Ante stated... your basically starving the inductance of current (L*di/dt ... the applied V is so short, the "i" can't ramp to a high enough level to develop any torque), so it slows down. If this is wrong, I'd love to see the math that supports the "change the frequency approach", as it might be fun to try and apply it somehow to DC-DC's over a limited range.

A lot of them use POL (point of load) converters to further filter/regulate the voltage or even use an intermediate bus architecture. Nowadays, even digital control is used to set the output voltages of POL's in order to set CPU speed. It's come a long way since "the old days"!! You also have to consider, how "crappy" is crappy!!! In an audio application, is 20mV or 200mV too much ripple... I don't know because I've never tried it!! Oh... BTW, the reason the ripple frequency in that scope shot isn't the same as the drain waveform frequency is because it a 2 phase converter (interleaved)