Not in any of the industrial power electronics books I read. Think about the problem they have at light load, the voltage across the capacitor rises to 1.4 x VRMS then as the load increases it drops to around 0.9 x VRMS with single phase power and full wave rectification. Once they drop to 0.9 x Vrms they have better load regulation than a capacitor input filter but a capacitor input filter PSU usually has about 10% load regulation from 0% to full which is better than the 40% regulation of the choke input filter. The problem is that the choke becomes massive if it has to maintain regulation at low load because the inductance required increases but the windings still have to handle the full load current and have low DC resistance. Swinging chokes help a bit.
But it is well known that choke input supplies have a minimum current draw. Given that, and a sufficiently low resistance choke, they are more stable than a capacitor input supply - unless the cap input can cope with huge charging pulses and the resultant mains supply harmonics.
The test for choke input is not the 40% drop at first, but the stability after that. If necessary, the minimum current can be drawn by a shunt stabiliser which gradually switches itself off for higher output currents.
For many audio circuits the quiescent current is high enough to satisfy the minimum draw for a choke input supply.
The test for choke input is not the 40% drop at first, but the stability after that. If necessary, the minimum current can be drawn by a shunt stabiliser which gradually switches itself off for higher output currents.
For many audio circuits the quiescent current is high enough to satisfy the minimum draw for a choke input supply.
The popularity in valve circuits is purely due to power needs, not voltage. The required energy storage determines the size of the chokes.
Chokes are great for Class A amps, which really hammer capacitor input supplies. Musical Fidelity used choke supplies for their class A transistor amps
Chokes are great for Class A amps, which really hammer capacitor input supplies. Musical Fidelity used choke supplies for their class A transistor amps
Agree with Davidsrsb
CLC filters can work out well for class A audio circuits....a helpful thing to do is pick up a Hammond 159ZC 60mH @ 2A gapped choke, some decent PS caps and try for yourself. These chokes are relatively inexpensive. I took a day off work and rewound a pair last year, in order to get a much more even layering, and run them now in a power supply that draws a steady 600ma. What the incremental inductance is with that load, I do not know, but my circuit "sounds" good, and not a bit of discernable hum. It may work for your circuit as well. A good book that helps in calculatiing ripple voltages of such filters is Building Power Supplies by David Lines, which used to be available from Rat Shack. Build and compare. I encourage experimentation. Enjoy.
Terry
CLC filters can work out well for class A audio circuits....a helpful thing to do is pick up a Hammond 159ZC 60mH @ 2A gapped choke, some decent PS caps and try for yourself. These chokes are relatively inexpensive. I took a day off work and rewound a pair last year, in order to get a much more even layering, and run them now in a power supply that draws a steady 600ma. What the incremental inductance is with that load, I do not know, but my circuit "sounds" good, and not a bit of discernable hum. It may work for your circuit as well. A good book that helps in calculatiing ripple voltages of such filters is Building Power Supplies by David Lines, which used to be available from Rat Shack. Build and compare. I encourage experimentation. Enjoy.
Terry
Are you looking for clever use of inductors in solid-state audio? Think about switching mode power supplies (particularly the ones with zero-current-switching and/or zero-voltage-switching) and class D amplifiers. These are the modern versions of the old choke input power supplies 😀
I agree entirely
The load determines the test for power supply stability if the loads operating range gets below the critical value then the load regulation suffers there is no pretending that it does not count, solution use bigger inductor. There comes a point when it is more beneficial to use a different power supply topology then add iron and copper to the power transformer rather than the input inductor. There is no one design which is best for all applications.
Certainly with valve amplifiers it is possible to get a ratio of maximum to minimum current draw which would result in a practical sized input choke even if the circuit has to be redesigned a bit, in industrial applications a 3:1 ratio of maximum to minimum load is easily accommodated often no filter capacitor is used either. Transistor amplifiers can have a maximum to minimum current demand ratio exceeding 10:1 that is tough to design a reasonable sized choke for. If it becomes necessary to add bleeder resistors then a series pass regulator starts to look attractive
I see a mention of scrounged chokes, here is what I have used with varying success:
Reactors from redundant Phase control and switchmode power supplies, no good for high voltage supplies but great in the 20A + range for low voltage supplies. Welder chokes also fall in this category but stray fields are sizable.
Microwave oven transformers, wire both windings in series ( often secondary is attached to the core at one end).
Discharge lighting ballasts.
I used a discharge lighting ballast in a CLC Pi filter once and despite my concern about core saturation there was over 200VAC developed across it, final ripple was 50Vrms which was OK for a single phase 4Kv supply with only 36uf total filtering @1A FWIW this was capacitor input filter with a load regulation just over 5% 0 to full load
Discharge lighting ballasts are often fairly well shielded magnetically. Naturally the low loss variety are better. They will be swinging chokes because they have no gap.
My pleasure 😀
Really? While I don't own a spectrum analyser myself, I recall reading respected Engineers' posts on various forums about the noise not only polluting the rails of the subject device but also finding its way into tother equipment via the shared mains feed. I am sure a search on here would find some examples.
I agree that layout and twisting the secondary wires is very important. One way switching noise couples is when a single transformer is used to supply both HT and heaters and there is no electrostatic sheild between windings. I am trying to recall the website where this was measured. It was not trivial. I will post a link when it comes to me.
Last edited:
Here is the link to the article I was thinking of, mentioned in my last post.
RR#002 - DC Filament Supply Test
The linked article contains another link to an article by Lyn Olsen which repays reading too and puts the information presented into context.
Cheers Rob.
RR#002 - DC Filament Supply Test
The linked article contains another link to an article by Lyn Olsen which repays reading too and puts the information presented into context.
Cheers Rob.
Last edited:
Good article it shows a two pole filter is twice as effective as a single pole filter at removing rectification artefacts. a 3 pole filter would have been better still. The article also shows that the fundamental is -48dB compared with an AC supply with the harmonics still lower and that is with a the Cap input filter. The harmonics should have decayed faster with both filters but that was probably a result of benchtop testing. This does not explain how such low level noise found it's way into the signal chain.
FWIW the use of schottky rectifiers and a choke is a sensible idea for a low voltage constant load PSU, the rectifier diodes do not make much difference to the harmonic content in this example but 0.7V less lost power on a 6.3V supply is nothing to be sneezed at. A series pass regulator would give even better noise performance with the added advantages of soft start and stable heater voltage.
The lack of electrostatic sheilding between windings is not a major problem single point earthing minimises any effect from that lack of magnetic shielding is another matter, quality equipment from the valve era always had fully enclosed transformers and or with shorting straps, good practice also kept the power transformer away from the small signal part of the equipment. Instruments often partitioned the chassis to keep fields away from sensitive areas. I can recall valve instruments which would go haywire if you moved while in the same room as them with chassis panels removed.
As for mains noise travelling, line filters (within the PSU) and balanced signal inputs are good practice. Mains noise is considerable, just think of how many off line SMPS's are hooked up to the mains before worrying about relatively quiet transformer based supplies, assume the mains is full of noise and design accordingly.
I once had a problem with buzz, quite low level but enough to be annoying. I noticed that there were 100Hz spikes which concided with the HT rectifier switching on, but these were not present on the HT rail and were not being picked up by the input. I found that the culprit was the elevated heater supply to the input valve. I had used an elevated supply as this was supposed to reduce such problems! No matter how much decoupling I put on this supply the spikes were still getting through. The solution was to ground the heater supply centre-tap. The best decoupler is a piece of wire! Buzz gone.
The mains transformer was a good quality one from Sowter, with a screen between primary and secondaries. It didn't stop spikes hopping from one secondary to another.
The mains transformer was a good quality one from Sowter, with a screen between primary and secondaries. It didn't stop spikes hopping from one secondary to another.
I presume you meant a floating heater supply? Usually valve transformers were wound primary, HT, heater in that order from centre out so the heater winding is directly over the HT winding allowing ample opportunity for capacitive coupling, the noise would have been mostly common mode, grounding the CT removed the common mode capacitively coupled voltage. I am surprised that decoupling didn't work but then again the frequency is quite low.
I am with you on that one the best decoupler is a piece of wire as long you don't create a circuit (loop)
I wouldn't call it a floating supply, as to me that implies no voltage reference. Yes, I was surprised that decoupling didn't work. I can only assume that the spike was too short, so maybe I should have tried a smaller decoupler. I had already carefully worked out the grounds, so I had to think for a while where it would be safe to inject this spike!
Inductor for a 50/60Hz supply and class AB amplifier makes no sense, because it will typically work at discontinous mode and so will cause drop dependant on load. A choke designed for continous mode would on the other hand be gigantic.
- Status
- Not open for further replies.