IMO, it's a case by case thing. Class AB stages, where the load varies with signal level, are particularly good candidates, especially triode and ultralinear.
I'm not following the argument about cost. Good chokes and caps are not cheap; high voltage regulators like the Maida circuit are not very expensive. For most circuit requirements, a Maida reg with suitable ratings might involve $15-20 worth of parts.
I'm not following the argument about cost. Good chokes and caps are not cheap; high voltage regulators like the Maida circuit are not very expensive. For most circuit requirements, a Maida reg with suitable ratings might involve $15-20 worth of parts.
SY’s point about low impedance down to DC is so very important. A problem with variations on the “LC…LC” theme at low frequencies is this: The impedance of the power supply will not only increase, but worse, it will show the resonant effects of those Ls and Cs. Whenever Ls and Cs cohabitate, a resonant circuit is created. With multiple Ls and C’s, multiple resonances are created. The goal is to make sure that these resonances are at very low frequencies and are well damped by resistances. I have built some nice sounding circuits using LCR variations, so please know that I’m not dissing the genre. The nature of the circuits being powered by this kind of PS does make a big difference. Some circuit types are more susceptible to low frequency PS anomies; others are relatively immune.
Attached is my simulation of an LCLCLC supply that I came across once. This is a real world, fairly typical, case, although not an optimum design in my book. A lot of DIY’ers live with these effects, perhaps unwittingly. The Y-axis is magnitude of impedance (the phase of the impedance varies too, but that’s a whole other story). The exact values of the components are immaterial; this is only an example! What you want to see is a very low impedance at all frequencies. Obviously you don’t get that. For most of the audio range the final cap determines the PS impedance, and it goes pretty low at the upper end, as expected and hoped. But in the sub-bass range, the resonances show up as nasty impedance peaks. How much these would bother the supplied circuit depends on that circuit, as I said earlier. Furthermore, if the supplied audio circuit use plate chokes, transformers, or cathode bypass caps, these reactances can interact with the supply reactances in even more complicated ways.
If you were ever tempted to wrap feedback around an amp that was powered by such a supply, you might soon learn the definition of the word “motor boating”. Again, I’m not against this time-honored genre of supplies; I’m just urging caution, and the consideration of an alternative: regulation.
A well-implemented regulated supply can drop the PS impedance to below 10 ohms, perhaps even below 1 ohm, from DC to daylight. Some people eschew regulated power supplies because many of them use feedback. I think a lot of this is simply a stigma carried over from the experience of feedback in audio stage, and, yes I know the PS is part of the overall audio circuit. Many people prefer the simpler regulation attained via shunt elements like gas regulators and even zeners. So there are many options to consider besides LCLCRCLC...
And cost? SY is right again. How much do a couple of big chokes cost?
Attached is my simulation of an LCLCLC supply that I came across once. This is a real world, fairly typical, case, although not an optimum design in my book. A lot of DIY’ers live with these effects, perhaps unwittingly. The Y-axis is magnitude of impedance (the phase of the impedance varies too, but that’s a whole other story). The exact values of the components are immaterial; this is only an example! What you want to see is a very low impedance at all frequencies. Obviously you don’t get that. For most of the audio range the final cap determines the PS impedance, and it goes pretty low at the upper end, as expected and hoped. But in the sub-bass range, the resonances show up as nasty impedance peaks. How much these would bother the supplied circuit depends on that circuit, as I said earlier. Furthermore, if the supplied audio circuit use plate chokes, transformers, or cathode bypass caps, these reactances can interact with the supply reactances in even more complicated ways.
If you were ever tempted to wrap feedback around an amp that was powered by such a supply, you might soon learn the definition of the word “motor boating”. Again, I’m not against this time-honored genre of supplies; I’m just urging caution, and the consideration of an alternative: regulation.
A well-implemented regulated supply can drop the PS impedance to below 10 ohms, perhaps even below 1 ohm, from DC to daylight. Some people eschew regulated power supplies because many of them use feedback. I think a lot of this is simply a stigma carried over from the experience of feedback in audio stage, and, yes I know the PS is part of the overall audio circuit. Many people prefer the simpler regulation attained via shunt elements like gas regulators and even zeners. So there are many options to consider besides LCLCRCLC...
And cost? SY is right again. How much do a couple of big chokes cost?
Attachments
Hi,
I know that some people will have other opinions but for myself I don't see any advantage of using choke based supply rather than active regulation in a preamplifier. In most cases a preamplifier have constant current draw and the main purpose of the power supply is to deliver clean DC from a impedance that is low from DC up to any frequency that can be handled by the amplifier. The easiest way to achieve that in a power supply is by active regulation and it is not that difficult to achieve good enough regulation and low enough impedance. The difficulties with designing an active regulator should not be ignored however and some effort is needed when it comes to chose design with regard to load and line regulation, noise level, transient response and also how to avoid high current pulses through the rectifiers.
In the preamplifier I use today I use a relatively simple active regulator based on a 6BM8 triode/pentode with a VR tube as a reference, it works very well and give me noise levels that are much less than what is coming from the amplifier tubes themselves and the hum level is something like >110dB down in the line amplifier and is magnetically coupled from AC wiring, not from the regulator.
BTW even a tube based active regulator will be cheaper than a choke based alternative with the same performance.
Regards Hans
I know that some people will have other opinions but for myself I don't see any advantage of using choke based supply rather than active regulation in a preamplifier. In most cases a preamplifier have constant current draw and the main purpose of the power supply is to deliver clean DC from a impedance that is low from DC up to any frequency that can be handled by the amplifier. The easiest way to achieve that in a power supply is by active regulation and it is not that difficult to achieve good enough regulation and low enough impedance. The difficulties with designing an active regulator should not be ignored however and some effort is needed when it comes to chose design with regard to load and line regulation, noise level, transient response and also how to avoid high current pulses through the rectifiers.
In the preamplifier I use today I use a relatively simple active regulator based on a 6BM8 triode/pentode with a VR tube as a reference, it works very well and give me noise levels that are much less than what is coming from the amplifier tubes themselves and the hum level is something like >110dB down in the line amplifier and is magnetically coupled from AC wiring, not from the regulator.
BTW even a tube based active regulator will be cheaper than a choke based alternative with the same performance.
Regards Hans
A good configuration is to combine a choke-input raw DC supply with a regulator of some kind, active NFB or shunt. You can use either fast-recovery SS diodes or tube rectifiers. The choke-input raw supply smoothes out the current spikes and transients which are a problem with capacitor-input raw supplies.
The question then becomes, why build a supply that has that stuff on it in the first place? It is not at all difficult to build a regulated supply that is tight and has ridiculously low noise.
IME, most of the audible problems with line rejection come from less-than-optimal grounding, but I haven't tried to get too fancy with raw supplies.
IME, most of the audible problems with line rejection come from less-than-optimal grounding, but I haven't tried to get too fancy with raw supplies.
It's a good question. I think raw power supply audibility “through” regulation may be largely ground related, and should in theory be entirely conquerable. But I think a lot of times the sharp current spikes characteristic of capacitor input power supplies may indeed cause the power transformer to approach saturation, even if it supposed be operating within specs or indeed well below specs. The resulting leakage flux can induce current in nearby conductors, magnetically. This is a problem with DC heater supplies too, where I’ve ended up with much larger transformers than I originally thought would be required, to vanquish induced hum, and worse, higher frequency hash. Lately I’ve tried using those remote in-line laptop switching charger/supplies to provide raw DC to on-chassis linear regulators for heaters.
Even if this hash is many dB down, it might still be unsettling the listening experience, reducing listening “ease”. Perhaps this is one reason why the low-DCR crowd seeks such monstrous power transformers - without knowing it, perhaps what they are really selecting for is low saturation flux. I don’t really know all this for sure, but, more than once, I’ve experienced better audible results with choke-input raw supplies for regulated B+. In these cases, BTW, I’m less convinced of the purported advantages of tube rectifiers over SS, except for the desirable gradual warm up time.
Even if this hash is many dB down, it might still be unsettling the listening experience, reducing listening “ease”. Perhaps this is one reason why the low-DCR crowd seeks such monstrous power transformers - without knowing it, perhaps what they are really selecting for is low saturation flux. I don’t really know all this for sure, but, more than once, I’ve experienced better audible results with choke-input raw supplies for regulated B+. In these cases, BTW, I’m less convinced of the purported advantages of tube rectifiers over SS, except for the desirable gradual warm up time.
Agreed and I think that it is easy to overlook this problem. For my B+ supply I use relatively small capacitors and with a series resistor to reduce the problem, as I use tube rectifiers the problem is anyway not that difficult as the tube rectifier will reduce the maximum current in the peaks.
Building DC supply for heaters is more difficult from this point of view as we dont want to waste more power than necessary so trying to avoid any series resistor but also here it pays of to be careful not to use too large capacitors. Using common mode chokes on the AC side take care of any common mode noise and using snubbers in parallell of the rectifier diodes will remove any risk of reverse recovery induced ringing to reach the B+ supply side through the transfomer.
Using methods like these we can manage well without any chokes and achieve very low noise levels and stable supply voltages.
Regards Hans
I expressed that in a bad way, what I meant is that when using tube rectifiers you need to ensure high enough secondary resistance so that allowed peak current is not exceeded, this has the additional benefit that any problems due to high peak currents will be less. Most data sheets for tube rectifiers have a value of min series resistance, if the transformer resistance is not high enough you need to add series resistors.
Adding resistance of course make the whole power supply less efficient but it is still a good idea to add resistance and/or not to use so high capacitance also in power supplies with solid state diodes if possible.
Regards Hans
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.