LC networks in typical audio PSU applications

I am familiar with one reason right off the bat why inductors are not used in SS designs. Note that tube amp PS' still use inductors for linear supplies. The current draw is so very low for tubes....low current, high voltages. SS designs low voltages, high currents. A typical 10H inductor used in a tube PS will "hold" say 800V, but only run at 300mA. This inductor is about fist sized & priced accordingly. Now, how big & expensive do you think a 10H inductor that runs at 200V & 10A would be?? Seek out PSUD from Duncan's Amp Pages for their FREE PS SIM program,...try it out.

_______________________________________________________Rick.........
 
Last edited:
Inductors for low frequency smoothing duty are large, heavy and very dependant on load current. They are still popular in 3 phase power supplies where the ripple frequency is 300Hz (360Hz) and it is important to reduce harmonic content in the supply and improve power factor. In most audio applications the load current varies widely and the amplifiers have good power supply rejection meaning an inductor will not work well all the time and the amplifier does not need such smooth DC any way. For example if you have 70V rails with 0.5V ripple at full load with an amplifier which has 60dB ripple rejection the ripple on the output will be almost -100dB at full load unless the amplifier clips. Single ended amplifiers present a constant load and have poor power supply rejection making inductors historically attractive in this application but with these efficiency is hardly a consideration.

Also modern capacitors are much more compact for a given value but inductors have hardly shrunk.

SMPS's cheat by raising the frequency so inductors become quite small and PWM takes care of a lot of the ripple.
 
I am familiar with one reason right off the bat why inductors are not used in SS designs. Note that tube amp PS' still use inductors for linear supplies. The current draw is so very low for tubes....low current, high voltages. SS designs low voltages, high currents. A typical 10H inductor used in a tube PS will "hold" say 800V, but only run at 300mA. This inductor is about fist sized & priced accordingly. Now, how big & expensive do you think a 10H inductor that runs at 200V & 10A would be?? Seek out PSUD from Duncan's Amp Pages for their FREE PS SIM program,...try it out.

_______________________________________________________Rick.........

Except that you don't need a 10H input choke for a typical SS amp because higher current lower voltage means critical inductance is lower. The main reason, cost and weight aside, is that most SS amps are not Class A and therefore designing a choke input power supply is difficult. The current drawn from the supply varies and therefore the value needed for critical inductance changes.

Rob.
 
Ayre Anyone knows its PSU?

PSU's are designed by formula, you can design one to get whatever performance you want. Many electronics texts from the 50's and 60's have examples which go into great detail about the design of the power supply.

I have had several amplifiers with choke filters, they were all single ended and they all had audible hum but in the 30's to 50s when they were built hum was normal. Nowadays a semiconductor series pass regulator will cost less and outperform an LC filter by orders of magnitude both in terms of impedance and ripple.
 
Choke input filters are rarely used with loads which a exhibit a wide variation in load current like class AB and B audio amplifiers because the choke only provides good filtering over a narrow range of load current. They are well suited to class A and class C (CW) loads.

As for the sound a PSU shouldn't have a sound, DC has no sound, you can mix in hum and noise early in the signal chain if that is what you desire and turn it off again if the desire passes, it makes no sense to design a power supply with a sound. When I listened to valve amplifiers on a daily basis I got used to the hum and noise and it sounded odd without it If I still owned the nice 60's DIY Williamson ultralinear amplifier I used in my teens bigger filter capacitors and a series pass regulator would have been installed because I am used to silent amplifiers now.
 

CLS

Member
2005-06-17 6:58 am
Taiwan
Hmmmm...

I have an impression that I've read somewhere stating that "LC in PSU provides better regulation than only C or RC.... ", or I read it wrong, or bad memory:eek:

As to the term 'DC', I think it's not that simple. The voltage is (or should be) constant, but the current is changing. Current is changing! Ah ha! that's AC :D Even in SE class A circuit, as long as there's signal, there'd be current change. I think that's the main reason we hear the PSU ;)

The only circuit with constant current is perfectly balanced class A PP, or SE with shunt regulator.
 
Chokes are big and expensive, so people only use them where they are unavoidable or where cost is not an issue. Choke input PSUs have advantages, but they need a minimum current draw otherwise they revert to capacitor input and the voltage rises. You don't need to draw exactly the critical current - anything more will do, until the DC drop of the choke gets too big. They were particularly useful back in the days when big capacitors were not available, for supplies which have to provide a wide range of current. Nowadays we just use big capacitors and SS rectifiers.

A potential problem with LC smoothing is that an L and a C can form a tuned circuit. This will be damped by the choke resistance and the output load, but there is still the risk of low frequency ringing or a rise in output impedance. This can be started either by a change in load current, or a variation in mains voltage input. PSUD2 can show this. You need to ensure that any potential ringing is well away from any critical frequencies (e.g. record warps, arm-cartidge resonance).
 
Hmmmm...

I have an impression that I've read somewhere stating that "LC in PSU provides better regulation than only C or RC.... ", or I read it wrong, or bad memory:eek:

It depends very much on the design of the power supply. Note that there are several LC filter designs, choke input filter, capacitor input filter, pi filter to name a few. RC filters have worse load regulation but chokes can have appreciable resistance as well.
As to the term 'DC', I think it's not that simple. The voltage is (or should be) constant, but the current is changing.
The current into the loudspeaker has to change or you will get no sound but it is the amplifier which should change the current into the speaker not the power supply it is quite easy to design a constant current power supply but the amplifier will be go into clipping as soon as the current reaches its design point.

It is also possible to use a current amplifier or a combination of current and voltage amplifier to drive a loudspeaker

An amplifiers purpose is to change the voltage and hence the current in the speaker, voice coil loudspeakers are current to force transducers. Electrostatic speakers are voltage to force transducers

The only circuit with constant current is perfectly balanced class A PP, or SE with shunt regulator.
You mean constant current from the power supply? A series pass regulator will also be constant current with these amplifiers. Shunt regulators are constant current with any load Shunt regulators are well suited to loads with minimal change in current like the ones you mentioned. Hope that makes sense.
 
Often the hum in SE tube amps comes from the heaters, not the B+ supply. Ironically rectifying a DC supply for high current heaters often makes things sound worse, partly I suspect due to the noise from the rectification process itself.

I think the main advantage of choke input supplies is the suppression of switching artifacts from the rectifiers. This is a big sonic problem with high current C input powersupplies in my opinion. Regulating after a high current C input supply is in some respects like shutting the stable door after the horse has bolted. Yes it will suppress hum, but the uhf noise from the rectifiers will propagate; best to suppress this at source. The current waveform from a choke input supply is very smooth compared to a C input supply.

Rob.
 
Inductor size?

Hi Everyone,
I was about to start a thread when I saw this one top of the list in the Power Supplies section and thought I would post here, hope that is OK!

I am building the altronics 15W Class A amplifer at the moment.
I am building the power supply using LM338 regulators, hopefully I will etch some circuit boards this weekend.
I am interested in trying a CLC filter at the input of the regulators at some stage once the basic setup is working.

My question concerns what size inductor is appropriate?
I am sure there are lots of factors involved but I was wondering if there are any general guidelines.
I reckon I have seen values of about 3mH but I am not familiar with the actual approach to determining the size.

If I use the formula here: Resonant-choke Power Supply
I get a value of approximately 10mH for the Critical Inductance value they mention which seems fairly large.

At the moment I am thinking of using a 3mH air core inductor intended for crossover use which I think should be fine as far as current handling and DC resistance go.

Any advice or comments would be greatly appreciated.
Please excuse my general ignorance, this is my first look at using inductors in power supplies!

Thanks very much,
Joe
 
Often the hum in SE tube amps comes from the heaters, not the B+ supply. Ironically rectifying a DC supply for high current heaters often makes things sound worse, partly I suspect due to the noise from the rectification process itself.
Bad wiring layout, even in the 30's designers knew to use twisted pair for heaters. There is no reason a properly constructed DC heater supply will inject rectification artefacts into the signal. The subject of layout is well covered in circuit design texts Check the layout and grounding yourself against good design practice rather than assuming the designer knew what they were doing or it was assembled properly . A spectrum analyser is your friend here. Even loading up the PSU and going over the wiring with a well insulated inductance loop hooked up to an amplifier can find stray fields

I think the main advantage of choke input supplies is the suppression of switching artifacts from the rectifiers. This is a big sonic problem with high current C input powersupplies in my opinion. Regulating after a high current C input supply is in some respects like shutting the stable door after the horse has bolted. Yes it will suppress hum, but the uhf noise from the rectifiers will propagate; best to suppress this at source. The current waveform from a choke input supply is very smooth compared to a C input supply.

Rob.
Again poor layout, as for any switching noise from the diode, The diode is looking into the filter capacitor at one end and looking into the leakage inductance of the transformer at the other end, no prizes for guessing which way any high frequency components are going to go and it is not into the filter capacitors. I have never seen any diode switching artifacts on the rails of a 50Hz capacitor input PSU but poor layout will easily put the charging current pulses into the signal path and a choke input filter might fix that but it is really covering up poor design. Also while choke input filters on single phase PSU's have good power factor they have poor load regulation until the inductor has a certain minimum current flowing, with 3 phase power they are much better but ripple is higher until a minimum current is flowing. One other problem with choke input filters is stray magnetic field, because of the large DC component in the choke the core is usually gapped to prevent saturation, the field around this gap is quite strong and capable of inducing hum in nearby circuits, again layout. Magnetic fields near valves may cause strange behaviour and anode hot spots.

Designing this stuff is no black art there are many fine books on the subject written by people who designed everything from industrial rectifiers to high powered AM modulators. Off the top of my head the RCA transmitting tube manual is a good read, any 70 or 80's vintage ARRL handbook is a good read for design tips.
 
My understanding of the resonant choke PSU is that it works a bit like the choke input PSU but uses a much smaller choke. It gains voltage stability at the cost of much higher harmonics. This is the wrong trade-off for most audio circuits.

If you are stabilising the supply anyway, then CRC smoothing should be good enough. Otherwise conventional CLC with a choke of a few henries. Conventional choke input would reduce charging pulse size, but not necessarily induction into nearby circuits. Like all electronics, you need to balance up the options. Work backwards from the requirements of the audio circuit, via the performance of the stabiliser, to get the requirements for hum at the input of the stabiliser.
 
My understanding of the resonant choke PSU is that it works a bit like the choke input PSU but uses a much smaller choke. It gains voltage stability at the cost of much higher harmonics. This is the wrong trade-off for most audio circuits.

It is not that bad, it has the same harmonic content as a capacitor input filter with all the fundamental and probably a bit of the second and third harmonic removed. The leakage inductance of the transformer attenuates more and more of the higher harmonics as it is effectively in series with the AC side of the rectifier.
It is actually quite a neat circuit for high power use If I had come across it before I would have been tempted to use it somewhere by now. It is overkill for a small audio application but might find use in a large class A room heater.

I am not against choke filters, an inductor is the natural complement of a capacitor and it excels at charging them but there are tradeoffs and they are especially poor at driving widely variable loads, as for using one on an LM3886 this device already has a minimum of 85 dB supply rejection and it is a widely varying load so it doesn't need a very smooth supply and the poor regulation of the supply under a widely variable load will cause the voltage to drop right when the amplifier needs the voltage (maximum signal)
 
OK, I was comparing the resonant choke to the normal choke input supply. I take your point about the transformer limiting higher harmonics.

I am puzzled by people saying that a choke input supply can't supply widely varying loads. This used to be their main selling point, now people say the opposite. Is it that most chokes have too high DC resistance so the voltage drop from that undoes the good done by the inductance? Or is LF resonance the problem? Or a big reservoir capacitor can now do even better than a choke input?