CFLs saving energy?

[<me9]

No it's not. It includes their offspring (of whatever age). If I go to the downstairs bog I usually switch off at least 6 lights en route, I may switch only one on.

[Dave Liquorice]

I wouldn't like them as they would switch off when the room was still occupied. Possibly by me cutting up some veggies, so yes a value my fingers. B-)

I have though about a PIR in the utility room. I used to have one in the kitchen of my flat, very useful. Not only would the light go out automagically, it would come on when you entered. If it went out when you were in there you only had to wave to get it back on again.

[Chris J Dixon]

Works very well in my kitchen, I don't often stand still long enough to have to do the arm waving thing.

Chris

[The Natural Philosopher]

well if its primarily resistive, the argument still holds, You wont be getting massive peak inrush currents into your reservoir caps. The diode forward impedance also limits it a fair bit.

Power factor is a meaningless term anyway with non linear impedances, and they don't come much more non linear than a rectifier junction :-)

Ite been a lonmg time since I actually put a scope on a diide leg to measure currents.

And let me make a further point, any mains to DC power supply, SMPS or not, is the same,. They ALL use transformed or not transformed AC into a diode bridge, and a cap afterwards, and they all clip the voltage waveform a bit. Unless its a battery charger in which case its a battery, not a cap, but the same applies.

The only way you can smooth that lot out if it is a really big component of the grid supply, is by shoving in a monster 50HZ tuned circuit at a substation closer to the end user. Or one in every house.

Same as they (used to?) have on the end of the cross channel DC link. An acre of inductors and capacitors to turn chopped DC into a sinewave.;-)

Only tuned circuit I've ever walked round...

[Fredxx]

Inrush? The impedance of a domestic feed is measure in mili-ohms. Short circuit, or as you might call it "inrush" limited by a "primarily resistive" feed will be in the order of 5kA. Diode impedance as you say is low, and makes for an even more peaky current flow.

That's where you are hopelessly wrong. Power factor isn't just leading or lagging, it also comes about through none sinusoidal current flow. You claim to have a degree, but I think a refresher course might be in order. In the mean time have a read of this:

Ite been a lonmg time since I actually put a scope on a diide leg to

I can tell.

Either they do or they don't. What are you trying to say?

Most electronic ballast CFLs diode rectify directly off the mains. The only transformer is on the HF switching side.

They use 3 phase rectification and other techniques to overcome the need for massive capacitor banks.

Just goes to show how out of date you are. I haven't had the pleasure.

[Andrew Gabriel]

That's not the main reason it's done. It's done because the supply infrastructure has to be sized for the VA rating, not the Watts. That's wasteful, because most consumers pay for energy (Watts per second), and not VA per second. Furthermore, the existing supply infrastructure cannot cope with increasing use of low power factor appliances and would need major upgrade to do so.*

This is what the mains waveform ends up looking like at the substation when you get large SMPSU loads without PFC (in this case about 1MW of SMPSU on a 2MVA substation)

it should be a sine wave, but it's miles off, and no longer met required standards. (Substation had to be upgraded, even though it wasn't overloaded.)

What inductive nature?

and then another SMPSU to transform it to the required voltage? Many computers already do that to meet high power factor requirements, but then suffer low PSU efficiency due to going through multiple PSUs before getting to the load. Currently, datacentres are much more worried about PSU efficiency than they are power factor.

*This is not a problem with CFL retrofit because a 100W filament lamp is 100VA load, and when replaced with a 25W CFL with PF=0.5, it's a 50VA load, which is still less than the 100VA load it replaced and the suppply infrastructure was sized for.

[John Rumm]

Source impedance varies enormously depending on how far you are form the substation, and the nature of the LV supply wiring, population density etc. So mili ohms may be correct - but often 100s of them. (my last house had a 0.25 ohm supply impedance). PSCs of over 5kA are certainly found, but are not in the majority of homes IME.

[The Natural Philosopher]

I don't think so. Even a bit of T & E is more than a few milliohms

12 milliohms per meter for 2.5mm sq.

Typical rectifier drops about 1.5v at an amp or so, so roughly 1.5 ohms, with tow in series, 3 ohms in a typical bridge.

But thats not a really useful figure, as it varies ..they dont have a fixed resistance.

But they will blow at ,much over a few tens of amps peak.

The reservoitrs capacitor will have some sort of ESR as well. Probably a few more ohoms.

That's not power factor as such.

Try googling the definition. Power factor only applies to purely reactive terms.

You

Basically says what I am saying?

I am saying that non SMPS do the same thing. ANY DC supply run off the mains introduces distortion. A lightbulb does as well, as its resistance varies with the applied voltage.

So? you have already claimed that transformers are low impedance converters. Ergo what happens on the secondary side is reflected into the primary circuit. Or do you not think that shorting a secondary will blow a fuse in the primary?

Time you went back to school

Not when I was there ..I am not talking about the AC to DC side, I am talking about the up converters back to grid AC from a DC source. Its the biggest f*ck off switched mode power supply you will ver see.

Oh dear oh dear oh dear

[The Natural Philosopher]

Surely that amounts to the same thing?

Yup. seen it many a time and oft on the secondary side of a not especially good 'substation' otherwise known as a cheap nasty mains transformer feeding a rectifier..

Question is, what standards? Standards of waveform, or standards of power loss.. The particular app that I saw worse than that in, became an issue of core saturation in a mains transformer: Now the power loss we could handle. The nasty escaping flux sent a deep hum field throughout the nearby circuitry.

Secondary leakage inductance, wire inductance and the coils in the MCBs etc.

Yes, but being fed from DC those wont suffer the same problems.

Indeed. My B in law reports that they achieve massive power savings in installations by using one large server running up to a dozen virtual servers (VMware) on it. Data centers are where it really does start to get an issue I suppose.

However its an issue right down the line as more and more load everywhere becomes not motors or heating, but electronics needing DC.

As soon as you introduce a rectifier, you are starting to generate odd current waveforms. 3 phase rectifications is a lot better..and its possible to do without reservoir caps, but there still is an effective switching of the load abruptly from one phase to the next at a given point in the cycle.

Ultimately if you want to localise these peak currents and keep them out of the upstream side to prevent extra transmission losses, you need to provide localised LC filtering of some sort. Bloody great tuned circuits whose circulating currents can provide the peaks, Whether that's done at the substation, the premises or where, and at whose cost, is a moot point.

In the end, it boils down to cost benefit - or should. whether it costs more to upgrade the 'backhaul' as it were - the transmission lines, or add in filtering (I don't think 'PF correction' has meaning here, because power factor is a ratio of in phase to out of phase current, and if the current ceases to be sinusoidal, it has no meaning except as a hand waving term) ..

[Fredxx]

The length of a mains circuit is such that the maximum voltage drop must be less that 5% which means that the supply resistance must be less than 0.375 ohms.

Your own argument seems to scupper the idea that the leakage inductance of the substation will be significant in improving power factor.

Diodes are very non-linear, they are open circuit with zero volts and an example of a cheap and noddy diode can be found at

While at 1 Amp, the forward drop is 0.95 V, at 20 Amps the drop is 1.4V. In practice some CFLs have small series resistor to limit the inrush and prevent destruction through overcurrent.

At high overcurrents, your figure of 3 ohms is wide of the mark.

Depends very much on the diode.

I would hope somewhat less than a few ohms!!

Yes it is. They didn't have switched mode power supplies when you did your degree.

What; that non-linear loads create harmonic content and contribute to a reduced power factor?

No. There are SMPS which have two stages of switching, one which takes a current in phase and proportional to voltage in order to give a near unity power factor, and another to produce a constant DC voltage.

I agree, the resistance of a light bulb is dependent on filament temperature, which as you correctly say does have a small modulation at

100Hz, and therefore will have a power factor very slightly less than unity. But that is down to harmonic content, and not reactive power.

?? The switching tranformer is driven by an IC. I can't see what your argument has to do with CFLs, which generally don't have an transformer, or an inductor on the mains side of the bridge or diode rectifiers.

What's that got to do with power factor? I was talking of the AC to DC side, not the DC to AC side.

1 The last time you saw a CFL with a "big inductor"? 2 What "fed from a nice current limited inductive load" means?

[John Rumm]

Traditionally the bulk of the current phase shift has been lagging. As has been mentioned, the leading nature of the shift created from SMPS and CFLs will partly offset that. The extra harmonic noise they contribute (and I don't mean directly - since they are supposed to filter that, but via any deformation they introduce into the supply) however will be "new".

So the thrust of the question was whether that marginal reduction in efficiency resulting from the extra harmonic noise (where a reduction of efficiency would apply to the full network load, not just the bit associated with lighting) added up to anything significant. (I am not suggesting it does - I genuinely don't know, hence the question)

Indeed, about half... (assuming one replaces 60W with 15W, and the 15W has a 0.5 PF)

I would offer the thread suggesting one preserve rather than replace an underrated cable, by attempting to knobble the load (electric shower!) with a transformer.

[Andrew Gabriel]

Not really. (I suspect in many cases that the PFC does actually consume more power than the extra resistive losses if it wasn't there.)

I presume standard of waveform, but I don't know what standard.

All irelevant, in that their limiting factor (PSSC) is of necessity orders of magnitude more than the normal current draw.

I spend much of my time advising customers on Virtualisation. (Usually, we can do better than using VMware, although that is one tool in the arsenal.)

That's not how PFC is done in SMPSU's. There are two ways I know of. The first way uses a capacitor and/or inductor to generate one or two extra lead/lag phases to feed into the bridge rectifier, so it's drawning current at these other points in the mains waveform too. This has a number of disadvantages - the inductor has to be big, and the capacitor path doesn't work too well and is usually not used at all. (It may be one of these inductors you've seen and thought was smoothing the current peaks, but that isn't what it's doing.) I don't think this is much used any more because of the size of the inductor needed, and because the second method gets PF back to 1, near-as-damn-it. Second method is to use a second SMPSU as a front-end boost regulator which is driven from unsmoothed rectified mains and boosts it up to

400VDC from whatever it is at any instant in the mains cycle (except the area where it's too near the zero-crossing point to be viable). It requires inteligent control to match the energy draw at each point in the cycle to the voltage at that point so the current draw is a sine wave matching voltage and the SMPSU looks like a resistive load. These are not as efficient as a non-PFC'ed SMPSU, because you're going through two SMPSU's, but they are pretty much mandated by EU rules today for loads over a certain power. Once you have that level of inteligent control, you can go even further and decide not to draw any current at the peaks, hence compensating for other SMPSUs without PFC in the same piece of equipment. That claws back a little of the inefficiency introduced by the PFC (not much though).

Your definition of PF is wrong. PF has a well defined meaning regardless of the current waveform, PF = W / VA This works perfectly for non-sinusoidal waveforms too, when performing PF calculations such as resistive losses, etc.

Refering to filtering is misleading - that's solving a different problem (and one which is much worse with modern PFC'ed SMPSUs with a front end boost regulator).

[The Natural Philosopher]

Oh dear.

You cant see how it is essentially EXACTLY the same then? You need big iron to compensate for Big C..whether ypu cal it leading bolox, or a sort of tuned circut is...semantics.

That doesn't really work very well without a 3 phase supply, anm ANY rectification and non linearity in the load leads to waveform distortion.

and boosts it up to

Better it may be, but perfect it never will be.By its very nature it will be pulling a series of short pulses: You just hope that they are more or less smoothed out by some choke or other.

Yiup I am sure its all possible, though horrendouysly icky to do.

when is a filter a power factor corrector? it always is in some sense. You ae juts optimising a different part of the characteristic.

Perfect power correction in an LCR circuit occurs when the the thing is TUNED to 50hz. Do the maths.

[Andrew Gabriel]

The inductor simply generates a second phase, and in doing so, a second current peak when the second phase peaks. There's no leading C to be compensated (non-PFC'ed SMPSU's only have a minute leading current shift, which largely isn't responsible for their low power factor), and no mains tuned circuit.

This has been standard (at least in large computer PSU's) for a few years now, with power factors of around 0.96

We're already at 96% perfect. There are probably even higher ones if you go looking, although even 96% is past the point where it matters anymore.

With a front-end filter, which I referred to below. Filtering out 20-50kHz only requires small choke/capacitors. (There's a new generation of SMPSU's operating in the MHz range now, although I don't know if they've worked their way into any mains-fed PSUs yet.)

It's been commonplace at the high end for some years, and has now worked down into datacentre PCs, where people will pay extra for >80% and >90% efficient PSUs.

(May even be in home PC's now - I haven't got a recent PSU to look in.)

Sorry, I thought you knew more about SMPSU's and power factor than you do. Can't really discuss further until you've gone and read up a bit about them, because you have a couple of concepts wrong. For starters, there are no 50Hz tuned circuits involved anywhere.