 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
81.56.144.181
In Reply to: A good question... posted by real_jj on April 24, 2006 at 12:55:46:
Things like switching supplies do all of that and .../...
Not really. The only (switching) power supplies above 50W that are legal (in EU at least) have a PFC in frontstage. The current stay approximatively a sine in quite any load condition. So, strong harmonics in PS will soon be a thing of the past.then stuff the switching frequencies and harmonics back up line if the line filtering isn't good.
You mean if the power supply is correctly designed. Regulations do exist (and well, many of them do exist!) to address the problem of EMI/RFI perturbations and susceptibility. I don't know why you can find on the market pieces of gear that do not comply with the standard. It should not be possible.
A correctly designed switching power supply dont pollute the mains above the standards levels (which not only are rather low, but were tweaked in order that it doesn't bother any other piece of equipement that comply to the susceptbility standards.) So...
But I must admit that designing a switching PS is not really easy. You cannot just trust the controller chip's application notes,where the real application engineer, with his hands-on, is pressurized by the liarmen (marketing) to spray rose water over the hard job of using their product.
You need 3 steps: calculation (yes, real, analytic calculations) then simulation, than build not a mock-up, but a real prototype with a real(expensive) board, well laid-out and estimated. Then compare the three, and hunt the reasons of the difference; Then back again. 2 runs is common. Your boss need to be patient. If he is not, fire him. If you don't dare, let him fire you...
Follow Ups:
First, I'm speaking from a US point of view, and there are some very leaky power supplies over here in the colonies.Second, anything with a diode brige creates harmonics, you'd better have a pretty good RFI filter even for that.
Third, I can and do measure crap on the power lines. (this in the USA) It's there, in plain sight, even before the harmonic analysis. One can see glitches and flattening where bridge-rectifier supplies start to operate, etc.
Measurements are measurements. I don't have a jpg on hand or I'd post one for you to try to weasel out of.
So, that's the facts over here. Don't claim something is wrong until you consider the context and the consequences of picking a fight with somebody who does know what they are talking about.
So, ready to apologize now?
First, I'm speaking from a US point of view, and there are some very leaky power supplies over here in the colonies.
So why doesn't the King's colonial administration outlaw power supplies with too bad a current spectrum the way we do in our decadent and fallen continent? Would be good for all parties: power companies, distribution companies, consumers, even employment market since you would have to hire competent engineer to design those new breed power supplies.
Second, anything with a diode brige creates harmonics, you'd better have a pretty good RFI filter even for that.
You're right (I never argue against that fact!). However, a pretty good RFI filter is not enough if you let the wiring around the diode resonate on its stray capacitance and inductance, or the components around the diodes resonates thanks to parasitics components (xformers with their interwiring capacitance etc). The wiring around the diode has to be done in a proper way not to resonate and not to radiate E, H and EMI fields. The way it's done in many so-called high-end audioequipment is just funny in these matters. And also: critically damped (at the VHF impedance of your wiring, which means you have to calculate it, and measure it with a network analyser). When I see the wiring in many so-called high-end equipment with PS at 50/60Hz, I do understand why they cannot use a SPS at 500KHz!
Third, I can and do measure crap on the power lines. (this in the USA) It's there, in plain sight, even before the harmonic analysis. One can see glitches and flattening where bridge-rectifier supplies start to operate, etc.
I know. Recording those events is what power network dataloggers are made for. Not only in the US can the main become crappy: more in the US for 3 reasons:
(1) average final user's lines are longer in the US. The country is large.
(2) the voltage being half than ours, the same perturbation is 3dB higher in Bushland.
(3) the TT interconnexion scheme, with neutrals connected to earth at the xformer and all the users premises, is prone to mix common-mode and differential modes perturbations and to cancel the effect of line impedance in isolating remote perturbation sources. So, you get a "pool" of all the perturbations occuring on the mains connected to your HT transformer, even if/when the source is remote. By contrast, on a TI network, you get perturbations only from your neighborhood, remotes ones are well attenuated. ( but I understand that the size of your country and its high keraunic level in many areas precludes the use of TT networks. Also historical reasons. ).
Don't claim something is wrong until you consider the context and the consequences of picking a fight with somebody who does know what they are talking about. ???
Waooo, ticklish you guy! Can you point out where I "claimed something is wrong"?
I just say, and I say again that a correctly designed switching PS does not inject in the network perturbations above the levels set in the standards. Because it cannot, or won't pass the certification if it does.
I also say that equipment can be built in order to work on very perturbed power networks. As a layman matter of facts, think of devices like self dialysis apparatuses that people use at home. A perturbation-induced event that would bug out the microcontroller or some measure and you're dead and the company is sued down to its roots. However, it works. Reliably. Even in non controlled electrical environment. They just applied the stringent standards for life-sustaining equipment, and went a little farther.
My concern there was not to bash you, which would be childish, but to defend correctly designed PFC'ed switching power supplies from being the culprits for sending crap on your networks, which is way not true...
My other concern was about the fact that it seems so hard to the high-end audiocommunity to switch to PFC'ed switching power supplies from the crappy transformer/rectifier/storage cap scheme.
From proz.com:'The "keraunic number" is the number of days per year on which lightning typically occurs in a given area.'
This number is high in the USA (along with tornadoes) because we have so many trailer parks...
;-)
Being a foreign stranger or a strange foreigner, I cannot distinguish, nor discriminate, nor differentiate nor even discern or descry which words are of current casual regular use from others.
I took the word keraunic from one of Don White's books: Lightning & Lightning Protection , a mandatory reading for my engineers. Our collection (to speak only of those edited by Don White in the '90s) :
- The Aerospace Engineer's handbook of Lighning Protection
- Shielding design, methodology and procedures
- Grounding for the control of EMI
- Electromagnetic Shielding, Materials and Performances
- EMP environment and System Hardness design
- EMI Control in the Design of Printed Circuits Boards and Backplanes
- EMI control in Aerospace Systems
- Lightning & Lightning Protection
On another topic, I agree on your comments in your post on the same thread about the virtues of fast recovery diodes, and SiC among them. The less energy in the inverse conduction pulse, the less energy will energize the stray resonant tanks around the diodes (wire inductance, interwiring capacitance).
But replacing standard diodes by SiC diodes or other fast recovery techies tries to mitigate the consequences (a dv/dt kick making the whole damned PS to resonate with slowly decaying oscillations) instead of the cause: crappy wiring, bad choice of transformer, and no damping of oscillations (see reference to some ancient article by J Hagerman in the same thread) .
As for trailer parks, we don't have them, but Gypsies. Somehow different, I think. They had the reputation to bring plague and bad luck, but not to attract lightning (to increase the keraunic level, to speak PC)
The word "keraunic" comes from (modern) Greek: o keraunos (stress on the last syllable), meaning, as expected, the lightning. Another word for lightning is h astraph (also stress on last syllable, where the h stands for eta, as in zeta, eta, theta in the Greek alphabet).The word for thunder is: h bronth (say: i vrondi).
Ernesto.
...about the RF ringing at rectifier turn-off due to stray capacitance and transformer inductance. He talks about how to design optimum snubber networks to minimize this.
(1) This article is intended to DIYers, who can afford to make impedance mearurements and calculation of the snubber elements for each device they build, because they will only build one or at least very few of them .
In an industrial POV, it is not possible, and only the mix of correct calculation, correct simulation, followed by prototyping and threat of public spanking of the purchase dpt head would he/she dare to modify the BOM (bill of materials) because he/she found a transformer or a rectifier $0.0001 less, only this mixture of actions can avoid the risk of selling a crapy product 6 months after it reached the market.
As he quotes himself: It should come as no surprise that each power supply must be individually tuned for optimum performance.
Fine, but just not industrial. Also make servicing very problematic (try to find out an identical xformer for example after 20 years...)
(2) He keeps silent on the fact that these HF pulses, although well damped by his snubber ( I said "his", but I remember having designed a 100W forward converter in 1981, for which I had trouble, having diodes spikes - I was an ignorant MS youngster with still the smell of my mom's milk on my lips- , and I found at that time a lot of litterature about snuffers, that i applied. ) will propagate along the wiring.
First, this wiring should be done in a way not to act like antenna nor magnetic loops, second, it should be done in a way to avoid any impedance sharp change (at the entrance of the transformer for example) since part of the incident wave will reflect on it, part will go through it and won't "see" anymore the damping snubber...
So, to conclude, I agree with him, but only in a hobby or DIY point of view and although his proposed fix is correct, it does not address the whole problem of spikes on the wiring.
For industrial use, further steps are necessary.
Ah, why is Nature so less gullible than mankind ! Engineering would be as easy as marketing, religion and politics !
In my experience anything that involves semiconductors like diodes that has to be individually tuned ceases being individually tuned after a while.Resistors and transformers are more or less exempt from this, but not capacitors.
These things can vary with load, too, and audio equipment, unlike industrial convertors, runs with a very wide load range, from maybe 5% to 100% of load, depending on the input signal and the level adjustment.
These things can vary with load, too, and audio equipment, unlike industrial convertors, runs with a very wide load range, from maybe 5% to 100% of load, depending on the input signal and the level adjustment.
Which also explain why PFC'ed switching power supplies are quite not used in power amps, despite their big advantages.
It is not easy to design an excellent switching power supply with such a large span of loads.
In technical terms, we want to make the switching transistors commute exactly:
- when the voltage between their pins is null (condition called ZVS, zero voltage switching), or
- when the current through them is null (condition called ZVS, zero voltage switching).
Doing so is called soft swiching , since it avoids high dV/dt and dI/dt when switching.
Those dV/dt and dI/dt are evils since they are the source of strong EMI/RFI in the stray components of wiring (they excite them).
Without them, EMI/RFI is easily managed as long as the board is well laid out.
Another big advantage of soft switching is that less heat is generated since energy stored in parasitic elements is not changed into heat, but rather into another form of energy ( for example, energy stored in the drain-source capacitance of the power transistors, instead of being flatly shorted, is transfered to some inductance in the design, for example the stray inductance from storage caps ).
Since less heat is generated, you can make the PS run faster (500KHz is state of the art for less than one KW), make it smaller too.
So, the aim is to have ZVS condition when the transistors are closing, and ZCS when opening.
For technical reasons above what I have currently the courage to write down, these conditions are very difficult to keep on a large span of load current like seen in a PA, as notes real_jj.
In fact, it is a hot topic in current industrial research and development: I don't know of any IEEE Transactions on Power Electronics since circa 2000 without at least 2 or 3 papers about soft switching.
Many solutions do exist, we are at the time when most of these solutions are yet known, but the best ones have still to emerge from the pool.
enough for today
I wrote
In technical terms, we want to make the switching transistors commute exactly:
when the voltage between their pins is null (condition called ZVS, zero voltage switching), or
when the current through them is null (condition called ZVS, zero voltage switching).
I should have written:
.../...
when the current through them is null (condition called ZCS, zero current switching).
Minor enough that I read what you meant :)
I agree with all of your technical points.As far as the high-end community goes, well, look down the page and you'll find where parts of it relentlessly vilify me.
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
