DC-DC Converters, conventional vs electron flow, and the reference ground

[dknguyen]

I was taking a look at switching converters again. Specifically PMOS buck converters that can work at 100% duty cycle, but suffer from the defficiencies of PMOS (relative to NMOS). It seems that there's quite a bit of effort that goes into using NMOS's in buck converters because of their better characterstics (which also means better availability), but then it requires added complexity because of high-side boostraping, etc, and as a result they still can't work at 100% duty cycle (not without massive costs and technical problems associated with a floating switching gate drive supply which is silly in a buck converter since it's more complicated than the buck converter itself).

It seems that this problem has it's root in the legacy mistake of how charges were defined as flowing from positive to negative (conventional flow), and so this "conventional charge sink" was defined as ground. Since then all electronics have been built using the conventional charge sink as ground.

But to my understanding, electronics could have just been as easily designed so that the sounrce of conventional charges/holes (ie. the electron sink) is the reference. And it would have probably been the case if things were defined correctly at the start. And if that did happen, then the NMOS could easily be used on the of a buck converter without any high-side drive complications.

Is this reasoning correct? Particularily the part in bold. It's just so pervasive now that it's hard to be sure if you could resdesign all electronics so that (I will describe the following in terms of conventional current flow), the highest +V in the system is the reference, and and the various voltage supplies are derived by being at varying potentials below +V. Basically flipping things around so the electron sink is actually the reference voltage of the system.

Analogy: Instead of dropping a different balls from different altitudes so that it falls onto level concrete, we now drop all the balls from the same altitude so that they fall into holes of various depths.

 

[Hero999]

You'll be fine, until it came to cars and boats which need a negative earth to prevent problems with corrosion so by your definition, they'd need a positive earth meaning lots of negative numbers and still they'd still need P type devices.

EDIT:
If the polarity and common grounding between the primary and secondary aren't important then you could just pick +V as the reference, for the sake of the buck.

 

[dknguyen]

lol this terminology is getting confusing. Negative earth = traditional earth? But yes, I am coming to see the case you are talking about. It's still somewhat blurry to me though I have to think it through. To prevent corrosion you want your materials to have a store of electrons to draw from right?

Hehe, unfortunately that won't work if the buck converter is being controlled by any semiconductor device or is being used to power any semiconductor device. Which almost always is the case.

 

[Hero999]

Yes, I'm talking about negative earth in the traditional sense of the word and it's true that there needs to be excess electrons to prevent corrosion. I've heard that some old cars used a positive earth system which caused all sorts of corrosion problems.

 

[dknguyen]

Hmmm so maybe I'll be able to reason my way around this lol. It was done for a reason! Since corrosion is much more widespread and general than the one specifici case of a PMOS buck converter. Almost everything else uses a half-bridge type scheme which means you run into gate-drive complexities anyways. The only other case I can think of other than that is using a "high-side NMOS" to switch on and off circuits and loads so the circuit can remain connected to the reference, but that's not realy a big deal.

 

[crutschow]

The point you call ground or common is arbitrary. Corrosion is a concern in only a few special cases. If you have an isolated source (such as a transformer-rectifier supply), then you could use a NMOSFET as the switch to control the negative side of the voltage. You could still call that the common point, it won't make any difference to the regulator (as long as the source is isolated). But many PWM chips are designed for driving PMOSFETs for controlling the positive output, so you would have to select a PMW chip that can drive an NMOSFET in the negative line.

 

[Nigel Goodwin]

I see no reason whatsoever to ever consider either conventional current or electron flow, it has no bearing on any design process - it's complicating things whilst achieving nothing. Personally I consider current flows from high to low (top of the hill to the bottom), it makes no difference if the 'hill' is positive or negative.

 

[Hero999]

Exactly Nigel.

What about AC circuits?

For single phase normally one side of the supply is neutral, the other is phase, all current flows from phase to neutral, if it's split phase, (centre tapped transformer) there's phase and anti-phase, I see phase as positive and anti-phase as negaitve.

If there's more than one phase it's P1, P2 and P3, I still see neutral as 0V but I used complex numbers to calculate the current and voltages.

 

[dknguyen]

The issue I am talking about is not really electron vs hole flow. It's actually about why the electron source ended up being used as the reference for almos tall circuits rather than the electron sink. However it draws in electron flow vs hole flow because it seems that was a determining factor into why the reference was chosen to be what it is (simply, that what was thought to be the sink was chosen to be the reference). Then electron vs hole flow does matter because if it directly plays into how things ended up and if it were actually chosen correctly then it would have beneficial implications for actual circuits.

It makes a difference when you consider all the effort that goes into making NMOS transistors work in a non-synchronous buck converter. If reference had been chosen to be the electron source then it using the NMOS would entail no extra gate drive requirements, none of the disadvantages of PMOS, and allow 100% duty cycle. Sure you could choose the electron sink to be the reference in your buck converter since it's technically arbitrary, but it's not much use when the regulator's controller, as well as everything it's supposed to power is using the electron source as the reference. And making it floating completely defeats the purpose.

 

[Nigel Goodwin]

The issue I am talking about is not really electron vs hole flow. It's actually about why the electron source ended up being used as the reference for almos tall circuits rather than the electron sink. However it draws in electron flow vs hole flow because it seems that was a determining factor into why the reference was chosen to be what it is (simply, that what was thought to be the sink was chosen to be the reference). Then electron vs hole flow does matter because if it directly plays into how things ended up and if it were actually chosen correctly then it would have beneficial implications for actual circuits.

It makes a difference when you consider all the effort that goes into making NMOS transistors work in a non-synchronous buck converter. If reference had been chosen to be the electron source then it using the NMOS would entail no extra gate drive requirements, none of the disadvantages of PMOS, and allow 100% duty cycle. Sure you could choose the electron sink to be the reference in your buck converter since it's technically arbitrary, but it's not much use when the regulator's controller, as well as everything it's supposed to power is using the electron source as the reference. And making it floating completely defeats the purpose.

Sorry, I don't see what any of that has to do with anything?.

 

[dknguyen]

Sorry, I don't see what any of that has to do with anything?.

I don't see what it is you don't understand. You have constantly pointed out in the past that for the deifnition is largely arbitrary and does not really affect things. THings are the way they are largely due to legacy.

But what I'm trying to point out is that in this case is that the definition has very real negative implications. It causes the buck converter circuit to be more complex while having less performance than could otherwise be realized if the reference had been defined as the electron sink. Electron vs hole flow is a different issue, but somewhat related. Sure the reference definition might be somewhat arbitrary but that ends when all other circuits you are trying to power follow a different definition.

I hear lots about how the definition of charge flow or which is the reference is arbitrary. But very little about the actual technical advantages that just happened to come with how the electron source was defined as the reference. I've only ever run into one which was just mentioned- corrosion. For the most part, all the issues all seem to be disadvantages or irrelevent.

 

[mneary]

I, too, totally miss the point. The definition was cast in stone long before the discovery that N-channel MOSFETs were easier to make than P-channel.

Is there a campaign afoot to get the electronics industry to swap plus and minus? How would this impact physics? Would protons become negative?

 

[dknguyen]

Just pointing out that the issue is not always as arbitrary as many people make it seem and it actually does cause real "needless" technical disadvantages and complications at an application level. Usually the most that is mentioned about it is that it is a bit backwards from a physics perspective but ultimately arbitrary until you get into semiconductor physics.

ANd I also wanted confirmation that it would be possible to build electronics all using the electron sink as the reference because using the electron source IS so pervasive that it gets hard to think about whether it's possible or not. SUre it's possible with negative voltage regulators, etc, but with more complicated circuits like RF, other analog, and VLSI digital it gets a bit more difficult to think about.

 

[Nigel Goodwin]

I still don't have the faintest idea what you're on about?, use whichever convention you like, it has no effect on circuit design or anything else.

 

[dknguyen]

My point is that the whether the reference chosen to be electron sink or source it's not as arbitrary as you keep saying it is and DOES matter.

Please see the attached PDF schematic for what I am talking about.

 

[Nigel Goodwin]

My point is that the whether the reference chosen to be electron sink or source it's not as arbitrary as you keep saying it is and DOES matter.

Please see the attached PDF schematic for what I am talking about.

It's still got nothing whatsoever to do with conventional current or electron flow?, which makes no difference to anything.

Certainly switching high or low, and negative or positive, affects your circuit design process (as different configurations are required), but it's got nothing to do with conventional current or electron flow.

 

[mneary]

So it's inconvenient to build a buck converter with NMOS transistors?

For this we should convert to positive ground electronics? What about all those boost converters?

 

[Speakerguy]

In the second and third parts of your diagram, you flipped the e- source and sinks but didn't switch the N channel FET accordingly.

 

[mneary]

I don't think it was arbitrary. Vacuum tubes would have needed a floating heater supply for each stage. Early transistors had to coexist with tubes. It seemed to me that most of the popular germanium transistors were PNP, and they were very difficult to use in small signal stages because of the negative ground. We had to jump through hoops to build small signal amplifiers and logic, because common emitter stages required unrealistic filtering on the supply. When NPN silicon transistors became available at low cost, it made it easier to build common emitter amplifiers. If we had a positive ground then all of the amplifier stages would have to use PNP or P-channel.

Take any bipolar circuit and convert it to positive ground. You will find yourself using tons of PNP/P-channel transistors except for one or two in the final stages.

Aside from the occasional high side switch in a power drive or buck converter, electronics is best off by a large margin, with negative ground.

 

[Nigel Goodwin]

I don't think it was arbitrary. Vacuum tubes would have needed a floating heater supply for each stage. Early transistors had to coexist with tubes. It seemed to me that most of the popular germanium transistors were PNP, and they were very difficult to use in small signal stages because of the negative ground.

Again, you're completely ignoring how things work, and are stuck in an incorrect mindset.

There were no problems with using PNP transistors in valve circuits, or anything else - it's simply a question of your reference point. If you want to run with a negative chassis and a positive HT rail, then it's just a matter of 'flipping' the entire circuit, so the emitters are at the top, instead of at the bottom.

I don't know if it's just me who doesn't have problems with this simple and obvious concept?, but I grew up through the valve/transistor days, and in the early days it was common to have chassis rails drawn as either negative or positive, so you got used to the fact it makes no difference, it's only how it's drawn.