Power Conditioning / Surge Protection

I only stated that when charge time gets higher - ripple is higher (capacitor is charged from the bottom of the ripple to the next peak)

Which is wrong. The next peak (voltage) will get smaller, and the total ripple will be lower. The high frequency ripple (which most impacts THD) in most amplifiers and can induct IMD will get much lower. Don’t believe me then feel free to build an amp and test it and/or simulate one.

I only argue that at the moment when net capacitor current is zero (peak of the wave) voltage on capacitor depends on source voltage and source impedance. That’s the peak supply voltage for the output stage. How much it will drop depends on total source impedance including house wiring, power cord, fuse, transformer windings resistance and added impedance of the filter in conditioner.

Here you are totally ignoring the load the is not synchronous to the source. You can’t do that. That is where the error in your logic is. This only applies, to some degree, at much less than 2x line voltage. Dynamics we tend to associate with mid-bass.

If this filter is poor then voltage drop, especially on inductive reactance, can be high. Even if we assume only 10% it will result in 20% loss of max power - equivalent to about 6% of drop in perceived loudness.

You are treating the "inductor" as a resistor effectively, and the load as a constant current load, both in the DC domain with this argument. Again, that is incorrect logic. Even with 0 resistance, any frequency beyond 2x line frequency in the load (and effectively less depending on timing) mostly eliminates any benefit of charging in the short term.

Large linear supplies have a lot of filter capacitance reducing voltage ripple to very small resulting in even narrower and higher charging current pulses and much higher voltage drops on conditioner’s filter impedance.

And again, if you add resistance or inductance, the size of those charging currents gets less, meaning the high frequency harmonics in them gets less (less noise / THD and potentially less IMD), and the length of time of charging gets longer.

I use Furman Elite 20PFi. They call it "Power Factor Correction", but in reality it is huge inductor in series with large capacitor (in parallel to load) that stores energy delivering up to 55A current for spikes (it presents resistive load to mains).

Let’s look at your Furman. Do you really think it delivers 55A peak currents from that capacitor in your system with a linear supply? It does not, not practically at least. The only time it can supply anything into a linear supply is when the voltage on that capacitor is above the transformer reflected voltage on your linear power supply capacitors. The problem is the voltage on the cap in the Furman and the reflected voltage of the capacitors in the linear supply in the amplifier will be exactly the same as it progresses through the charge (AC) cycle. Effectively, the capacitor in the Furman ends up in parallel to the capacitors in the linear supply (reflected through the transformer) once the diodes start conducting and it does not end up transferring any power at the start of the charge cycle, and only a small amount at the end of the charge cycle when the AC voltage decays. It is likely a film capacitor and hence able to store much less energy than the caps in the amplifier power supply. The only time it will supply a large peak current (very short duration and not much energy), is if the amplifier has a large load right after peak of the AC cycle.


For an LC PFC to be effective as PFC, it must be exactly tuned to the load including the reactive components in the load (i.e. capacitors). Odds are your Furman is rarely presenting something that looks like a resistor to the mains. The inductor does improve power factor and reduce THD on the AC lines, but effectively the capacitor is doing very little to help that due to the nature of the load.


I will guarantee you that under peak loading, due to the inductor in the Furman, the rail voltage on your supply is slightly dropping, and that is a good thing, means the ripple/noise is being kept under control. I would expect taking out the AC capacitor in the Furman would have almost 0 impact on the DC rail voltage in the amplifier under loading but I would not suggest it. What it will be good at is suppressing high frequency noise coming in from the AC line.

So you think large spikes every 120th or 100th of second causing large voltage spikes with high harmonic content is going to create less noise and distortion than a slowly changing voltage at the same frequency? What is the music peak is at 180 Hz or above?

I go with cool meaningless name for "balanced surge protection". Whole house is the way to go. Better to have the surge protection as far from the equipment as possible.

Does it really kill dynamics? Unless you are clipping what is the mechanism by why "dynamics" are reduced? How would it be different from the normal line voltage variation from say 120-127?

Wouldn't that resistance make the spikes smaller and reduce noise and distortion?

The problem is that filtering places resistance in series that causes big voltage drops and kills dynamics.

Sorry but your example is too simplistic to be realistic. That 10V drop assumes the capacitors never charge. They do of course.


As well if there was 1 ohm resistance the spikes would not longer be 10A since current would be limited by voltage drop across the resistor (see charged capacitors above).

How would this potential loss be audible in a "poorly regulated" amplifier?  That would require you justifying that there would be more noise with higher resistance. You have not done that yet.

Charging pulse starts at the bottom of the ripple and ends at the peak of the waveform. When ripple is larger this time gets longer. When you comprehend this we can go forward

.
It is impossible for you to discuss this as you don't understand what is happening. It is unfortunately obvious to me you have never had something like this on the bench, never had something like this in a simulator. You are just trying to run it through your head and are doing it wrong. You repeatedly assume the charging current is the same in all circumstances. If the circuit changes, i.e. you add resistance or inductance to the charging circuit, then everything changes. The waveform of the charging current changes, the total conduction angle changes, etc.

Amount of ripple depends on capacitance and load current and does not depend how capacitor is charged since ripple is an effect of voltage drop during capacitor discharge cycle hence has nothing to do with capacitor charging.

You are painfully showing here that you have no idea what is actually happening in the circuit. Sorry can't call it any way but that. Unless you learn this is not the case, you are unable and uneducated enough to discuss this. That is simple reality.  It is not my job to teach you about the basic concept of choke regulation in a power supply, but it is your job to learn it if you want to call other people "low in knowledge" and not have egg on your face.

Amount of this voltage drop (amplitude of ripple) defines width of charging pulse, since capacitor is charged only from the bottom point to next peak. This bottom point was defined by discharge cycle thus charging has nothing to do with it. 

Again, you are showing that you simply don't understand what is happening in enough detail to form any coherent view on what is happening. Amount of voltage drop does not define the width of the charging pulse. The amount of time the input voltage (rectified) is above the output voltage (rectified) defines the time of the charging pulse, literally by definition and by any coherent view of how the circuit operates. We have already determined that the average rectified rail voltage will be lower. Guess what, that means that the amount of time the rectified input waveform will be higher that the output increases.

PFC does not work by by extending conduction angle but rather eliminating (shifting) phase between voltage and current to present resistive load to mains. 

I said the LC circuit in the Furman works by extending conduction angle, I did not say that was what PFC in general was. I can't post pictures here, but perhaps this will help you understand it a bit better as this is exactly what an inductor will do in a linear power supply.  The C in the Furman as pointed out earlier, when combined with most audio amplifiers, will be near useless in impacting PF or THD for that matter.

https://www.allaboutcircuits.com/technical-articles/power-factor-thd-why-linear-power-supplies-fail-...
The fact this article shows the inductor in the output circuit is meaningless. It would work exactly the same in the input circuit, obviously with the inductor value adjusted based on the transformer turns ratio.  The L in the Furman absolutely increases power factor in a linear amplifier. It absolutely is power factor correction for a linear amplifier, however, its effectiveness will be highly dependent on amplifier capacitor bank size and odds are it will rarely result in a high power factor.


Based on what you have written here, by understanding of electronics and especially power supplies and power electronics is obviously way more extensive and accurate compared to yours.


Here, maybe you can increase your knowledge so we can have a proper discussion:

https://sound-au.com/lamps/pfc-passive.html#acc

Oh, and P.S., even if your capacitor bank is larger once you hit a practical size, the peak current and the charging waveform starts to look the same.   I will let you figure out why, but this may give you some hints, see section 5.3.

https://sound-au.com/power-supplies.htm

I am talking about noise and distortion because that is what we hear. The lower the resistance the bigger those spikes are, the bigger the harmonics of the AC and the bigger potential for THD due to AC harmonics.

We don’t hear power loss.

Those spikes are also only at 100/120 Hz so no "benefit" to frequencies above that.

Unless the amp is voltage clipping you have not made a good argument for true loss of dynamics though the loss would not be much different than line voltage variation. That resistance limits the peak current but extends the charge time so I your 10A example the peak may drop to say 5A but charge time increases and the voltage drop becomes small overall.

Garth Powell writes some words but offers no proof or even solid technical discussion. This is word smithing for consumer marketing. The higher current the peaks the larger the instantaneous voltage change which is much harder to filter out for subsequent electronics. That so called compression only happens if you are driving the amp into clipping.  If you are not driving into clipping no solid technical explanation has been presented for loss of dynamics. More likely seems a decrease in the THD unless clipping.

The AC line only supplies voltage for a small part of the waveform every 1/100 or 1/120th of a second. If you are current starving the AC input the you are clipping. The relatively small loss of voltage on the secondary of the power supply due to some level of inductance on the AC would be fractions of a dB in top end.

Far more likely is inadequate components in the AC filter likely selected for continuous current rating but not the peak currents resulting in saturation of the inductor and increasing the THD not decreasing it like it should.

A universal statement about loss of dynamics without a critical look at the real mechanisms results in universal and likely often wrong statements applied to all equipment performing the same function.

You are saying the voltage ripple gets larger. Prove it. That is a common misconception that is actually not true. The worst case voltage ripple will not get worse with added resistance.

Yes the voltage (average) on the capacitor may be lower. I never discounted that. A lower voltage (which wouldn't be any different from a lower mains voltage) will limit peak output power, but we are talking fractions of a db where clipping will kick in.

Do the actual work. Consider the transfer function. Consider the transfer function in the frequency domain. Voltage ripple will not increase with added input resistance.

It is not a matter of disagreeing of agreeing. This is objective engineering, not subjective listening.  Everything in what I wrote is easily verified through simulations or building and testing.  I will absolutely agree that added resistance whether real resistance or AC impedance will limit /reduce the maximum rail voltage and hence the peak volume, but I cannot agree as it simply is not true, that the effect and operation is as you described and to the to the level you described as that is simply not the case, no more than the real world performance of the Furman will be as they describe. I am not making that statement because "I believe" it to be true, I am making that statement as I know it to be true both from a fundamental engineering standpoint, and a practical, having done it, implemented it, tested it. measured it standpoint.


There may be something perceived as lack of dynamics with some filters and some amplifiers, but any real world loss of peak power using real music is going to be fairly small, and unless you are driving into clipping, loss of dynamics should not be an issue with any competently design amplifier. Instead of just taking the easy answer, which means the problem is never solved, it is better to find out what is really behind the perceived difference.

Bigger ripple means longer charging time - ALWAYS!, but you disagree with this.

No it does not and until you understand this you will never be able to move forward. It only means more ripple IF the charge current is the same, and the charge current will be much different with added resistance.

Ripple is a function of capacitance and load current.

No, it is a function of capacitance, load current, and the way the capacitor is supplied with power.

Yes, additional impedance will make charging pulses smaller, because they depend on source voltage divided by impedance in the charging circuit - but charging time will not get longer!

Of course it will get longer. How do you think an LC based PFC works? It works by essentially extending the conduction angle .... i.e. the time the capacitor is being charged.

Yes, Furman’s output capacitor appears to be connected in parallel to electrolytic caps, but is not. It is on the the other side of the rectifier on the AC side. Voltage on this capacitor follows AC voltage cycle but any loss of charge caused by PS charging pulse is replenished from energy stored in the inductor thru the whole cycle. In addition this (huge!) capacitor has very low ESR

Again totally wrong interpretation of what happens. The diode starts conducting when the AC capacitor that is not at all huge in terms of capacitance compared to the amplifier even adjusted for voltage (it will be a film capacitor versus electrolytics in the amp ....way less storage per volume). And no, the capacitor is not "replenished" the whole cycle. Some of it is being discharged, some it is not. It’s voltage follows the AC line with a lag angle determined by its capacitance and the inductors inductance. Once the diodes start conducting, then the lag angle is a factor of the inductor, the AC capacitor and the dominance DC capacitors due to their much higher reflected capacitance. When the diodes are conducting, the "PFC" capacitor is essentially in parallel with the capacitors on the amp (with the exception of the diode drop). The ESR of the film capacitor is almost meaningless as it does not charge the capacitors of the amplifier.


We are not "arguing". No offence, you appear to have some technical acumen, but you view of how this all works is quite wrong.

Call me skeptical about the environmental protections products as described.  I read the marketing blurb and the patents for the so called "waveform" correction.  Their marketing blurb w.r.t. other surge protectors requiring a good ground to work and theirs does not is just that, marketing. There is no basis for their claim especially. Their waveform correction technology does nothing to protect surges from line or neutral to ground which are important.  Their technology, distilled, is a capacitor across the AC line. They put a inductor in front of the capacitor (and the MOV and gas discharge tube according to the patent), claiming the inductor smooths the clamping. It may do that. It will definitely decrease the effectiveness of the clamping.