For a 100 mA load how big an inductor do you need? When I looked at this in the past if the inductor was not big enough the circuit reverts to cap input. The transformer needs to be scaled for a choke input supply as well. I like the results but even getting suitable parts is difficult. Here is an interesting discussion in the context of a 1KW amp for Ham Radio: Resonant-choke Power Supply I will dig into the resonant choke input idea more.
So you think it is more than just differing PSNR ratios on different designs ?
I found the SIC to change the bass character in a negative way especially on low level stuffMeasuring soft fast versus conventional on power amps was quite dramatic. Strange that in some circuits it is audible and others it isn't.
Right now I am looking at low current preamp power supplies. Previous measurements indicate high current may behave quite differently.
Please note even at low current there are differences, just not as large as expected.
Turn off noise as you know creates EMI that affects circuitry differently than simple power rail noise. My feel is that most current measurement metrics do not adequately show the influence of noise.
Last edited:
I for one would think so.
The base line is initially different. For example, if you have separate electronically regulated PSU lines for the input stage and VAS, then chances are you power lines are already much better filtered than otherwise.
If you add to that a power line filter ("conditioner") which preceds the whole amp, then again, the level of your power line muck will be still lower.
And I seriously doubt anyone will ever actually hear the difference between -120 and -140 dB of attenuation of power line disturbances. I could be wrong.
think there's some talking at cross purpose here:
For a long time I have used extensive CLC filtration after the diodes throughout my system in addition to usual regulation methods and it made a big difference to the sound. More recently based on John's advice here about adding a cap across the mains to see if it made an audible difference, I started working on mains filtration, and somewhat to my surprise in light of the measures I had already taken, this also brought clear additional improvements - particularly when I added CLRCLRCR filtration to mains / line of my digital equipment. So I don't doubt that these measures can be very effective.
My question to Wayne was only with regard to how much using different diodes impacts differently on different designs.
For example one might expect broadband high FB designs to have much higher PSNRR so might be more immune to . . . power supply noise and I was wondering if wayne thought that his observations went beyond this obvious conclusion.
For a long time I have used extensive CLC filtration after the diodes throughout my system in addition to usual regulation methods and it made a big difference to the sound. More recently based on John's advice here about adding a cap across the mains to see if it made an audible difference, I started working on mains filtration, and somewhat to my surprise in light of the measures I had already taken, this also brought clear additional improvements - particularly when I added CLRCLRCR filtration to mains / line of my digital equipment. So I don't doubt that these measures can be very effective.
My question to Wayne was only with regard to how much using different diodes impacts differently on different designs.
For example one might expect broadband high FB designs to have much higher PSNRR so might be more immune to . . . power supply noise and I was wondering if wayne thought that his observations went beyond this obvious conclusion.
Last edited:
Demian,
You may want to put the inductor on the AC side of the bridge. For a complete list of passive current shaping rectifier circuits see:
http://www.sdaengineering.com/mepcon10/papers/154.pdf
Please note that the best looking results of the circuit in 3.1.5 are not practical to achieve. The parallel traps will draw the harmonic currents of any non-linear load connected to the mains. Proper dimensioning for power dissipation thus become a big question mark. The solution has its merits in closed systems like the 400Hz power distribution of avionics systems for instance.
Giorgio
You may want to put the inductor on the AC side of the bridge. For a complete list of passive current shaping rectifier circuits see:
http://www.sdaengineering.com/mepcon10/papers/154.pdf
Please note that the best looking results of the circuit in 3.1.5 are not practical to achieve. The parallel traps will draw the harmonic currents of any non-linear load connected to the mains. Proper dimensioning for power dissipation thus become a big question mark. The solution has its merits in closed systems like the 400Hz power distribution of avionics systems for instance.
Giorgio
Interesting - I will think on trying a higher value resistor but adding 1R, in my case to 6.8uF, although it may not be ideal still may be about 200 times better than having no added resistance at all and if the HF noise source is low impedance the amount of noise reduction we achieve begins to become less and less as the damping resistance gets higher. I think this might end up being a trade off between two conflicting ideals
Did you get good measured improvement with 15R and / or good subjective improvement ?
Last edited:
Seems we're running this same cirle for the n-th time.
Everybody is talking about removing line disturbances at or after the filter caps. Nobody seems to realize that those line disturbances are also affecting the power transformers.
Why do people not realize that it's MUCH more effective to remove as much of those disturbances coming from the grid BEFORE they ever get to the power transformers?
That way, the transformer is also offloaded and can be more efficient, true, not by too much, something like 3...5%, but it's that much less energy used under normal circumstances and that much more available headroom. And less wasted heat.
One needs to look at what one is getting from the grid. Because this is so variable and depends on one's specific circumstances, simply plug an oscilloscope and check out your incoming power. See those lovely peaks hitting as much as 900V, and you will find that most of the noise we object to is done by 80 kHz, including 5th harmonic. After that, the reduction of their relative power will make little difference.
It's better to prevent it from ever getting in than to cure it afterwards. A true full filter system will have a separate filter for each device, so that it protects each and every one of them from its neighboring devices as well as from the grid borne muck.
Everybody is talking about removing line disturbances at or after the filter caps. Nobody seems to realize that those line disturbances are also affecting the power transformers.
Why do people not realize that it's MUCH more effective to remove as much of those disturbances coming from the grid BEFORE they ever get to the power transformers?
That way, the transformer is also offloaded and can be more efficient, true, not by too much, something like 3...5%, but it's that much less energy used under normal circumstances and that much more available headroom. And less wasted heat.
One needs to look at what one is getting from the grid. Because this is so variable and depends on one's specific circumstances, simply plug an oscilloscope and check out your incoming power. See those lovely peaks hitting as much as 900V, and you will find that most of the noise we object to is done by 80 kHz, including 5th harmonic. After that, the reduction of their relative power will make little difference.
It's better to prevent it from ever getting in than to cure it afterwards. A true full filter system will have a separate filter for each device, so that it protects each and every one of them from its neighboring devices as well as from the grid borne muck.
Last edited:
Yes, the logic you expound is good.
My logic up until recently was "why clean up before the bridge when the diodes create a whole new lot of noise which needs cleaning up after the bridge so lets's clean it all up in one go"
However, now experience has shown that cleaning up both before & after the bridge is necessary if we want the best results.
My logic up until recently was "why clean up before the bridge when the diodes create a whole new lot of noise which needs cleaning up after the bridge so lets's clean it all up in one go"
However, now experience has shown that cleaning up both before & after the bridge is necessary if we want the best results.
As I understand the near infinite impedance at the secondary while no diodes are conducting is one thing that makes a transformer susceptible to saturation - and thus more susceptible to line noise. A slight load directly at the secondary such as a large film cap helps with this. A cap is a good load here rather than a resistor because the cap won't draw real power or dissipate heat inside the chassis.
I may be out of my depth here but I thought that when a core saturated, the magnetic field fans out from the core resulting in greater coupling to nearby circuits, thus increasing magnetic noise.
So in many ways a large RC at the secondary seems like a good thing from my limited knowledge.
A large cap here also absorbs the rectifier transients that will otherwise enter and be radiated by the transformer - this is better than a primary-side filter at containing rectifier noise I would think.
Line noise should clearly be filtered before it gets into the transformer. In my experiments I was already using a large filter at the input.
I may be out of my depth here but I thought that when a core saturated, the magnetic field fans out from the core resulting in greater coupling to nearby circuits, thus increasing magnetic noise.
So in many ways a large RC at the secondary seems like a good thing from my limited knowledge.
A large cap here also absorbs the rectifier transients that will otherwise enter and be radiated by the transformer - this is better than a primary-side filter at containing rectifier noise I would think.
Line noise should clearly be filtered before it gets into the transformer. In my experiments I was already using a large filter at the input.
What I liked best was actually to just put a large cap directly across the secondary with no series resistor, and to snub the primary instead. I had an EMI filter at the input that had a 330nF cap, so I snubbed the primary with a 1.5uF RC.
Having the snubber at the secondary seemed like more of a compromise, and it seemed like a waste to have a big film cap with a resistor in series, keeping the cap from absorbing HF noise from the rectifiers. I think a very small resistor may be a good idea to damp the capacitor's self-inductance so it doesn't resonate with the rectifiers, but I never tried that.
I don't think I ever got perfect suppression of resonance in any configuration, but this was what eventually worked the best for me.
I didn't have a good way of quantifying line noise or rectifier noise, so I payed most attention to resonances that I could measure and the rest evolved out of testing by listening and playing on the simulator.
Since then my bench has been torn down and relocated, and I won't be able to make any more tests with the same setup. It could be my experience will change once I get to it again.
I aplogize if it came out a bit too strong, that was not my intention.
But I am frustrated a little by the fact that people with impeccable design reputations seem to have a rather negative attitude towards power line filters.
I realize that power filters, or in modern communis euphemism "power conditioners", are second only to cables in snake oil peddling, but among the myriad offered there are bound to be a few gems, here and there.
It's really much like us ourselves - keep yourself on a good diet and you are much less likely to need a doctor.
As for before and after, a slight modicifactio here from me: that depends on how well the job was done before. There are far too many variables involved to be able to GUARANTEE top results, I should know, I've been making and selling them for the last 13 years.
As I believe you've found out for yourself, the ONLY way to know if it works for your circumstances is to try it. That's the reason why I offer a 30 day money back guarantee, I know first hand how fickle they can be, not satisfy when you are sure they will, and amaze you when you never ever expect it.
That's also why I advocate a single device with its own dedicated filter. That's the only way to be really sure it's doing all it can. You need to protect each and every devices in a system from phase manipulations of its system bretheren.
The capacitor doesn't absorb the noise, it just provides a path for it. The resistor dissipates the noise. A reactive element cannot dissipate power,
only a resistive element.
For people who would rather just look at the pictures, rather than read all the words.
Armstrong, K. “Key knowledge for the efficient design of electronic products and their EMC – that we were never taught at university”, Keith Armstrong, ANSYS Seminar “Next Generation Signal Integrity and EMI Simulation”, 23rd March 2011, Oxford, UK.
www.ansys.com/staticassets/ANSYS UK...rmstrong_Presentation_ANSYS_March_23 2011.pdf -
See more at:
Fundamentals of EMC Design: Our Products Are Trying To Help Us
The S/N ratio of the Blowtorch thread is exceptionally low for such a masterpiece of low noise design
Going back to low noise power supplies for low power applications...
I'd like to make a distinction between topological issues and implementation or engineering issues.
The most suitable topology, IMNSHO, is the choke input power supply with the choke on the AC side of the bridge. For a complete list of alternatives I would suggest to review the paper at:
http://www.sdaengineering.com/mepcon10/papers/154.pdf
A word of CAUTION: the best looking results of the circuit in 3.1.5 are not practical to achieve. The parallel traps will draw the harmonic currents of any non-linear load connected to the mains. Proper dimensioning for power dissipation becomes a big question mark. The solution has its merits in closed systems like the 400Hz power distribution of an aircraft, for instance.
The series resonant circuit is probably overkill and, in any case, an headache to design and engineer.
Proper selection of the switches can range from Schottkys (best) to SiC (exotic) to fast recovery plus snubbers (cheap). Not rocket science.
The EMI mains filter is yet another subject. The most important feature is to have minimum leakage current. A so called "medical" EMI filter works best here.
It would be interesting to discuss each point from an engineering standpoint...
Giorgio
Going back to low noise power supplies for low power applications...
I'd like to make a distinction between topological issues and implementation or engineering issues.
The most suitable topology, IMNSHO, is the choke input power supply with the choke on the AC side of the bridge. For a complete list of alternatives I would suggest to review the paper at:
http://www.sdaengineering.com/mepcon10/papers/154.pdf
A word of CAUTION: the best looking results of the circuit in 3.1.5 are not practical to achieve. The parallel traps will draw the harmonic currents of any non-linear load connected to the mains. Proper dimensioning for power dissipation becomes a big question mark. The solution has its merits in closed systems like the 400Hz power distribution of an aircraft, for instance.
The series resonant circuit is probably overkill and, in any case, an headache to design and engineer.
Proper selection of the switches can range from Schottkys (best) to SiC (exotic) to fast recovery plus snubbers (cheap). Not rocket science.
The EMI mains filter is yet another subject. The most important feature is to have minimum leakage current. A so called "medical" EMI filter works best here.
It would be interesting to discuss each point from an engineering standpoint...
Giorgio
Hear! Hear!
Giorgio, if you would take a 'tiny' transformer with no choke, instead of a big transformer and an AC side choke, would you not get the same effect?
Jan
That GREATLY depends on its cutoff point. Most are 99.9% useless as they start operating at around 600 kHZ, when 99.9% of the junk to be eliminated has already passed through. The IEC-cum-line-filter is one of those practically useless types.
You get to hear how problems come from mobile cell phones, wireless telephones, raio, TV and so forth, and while I will certainly not deny that they can have serious issues, especially with very widebandwidth amps, you should also consider that their relative power is very low.
Also, one's best chances of catching something from the air, like an AM station, is biggest with very high gain circuits, such as RIAA eq amps, or microphone amps. For line sources, your chances are rather small to nil. I haven't heard of any such problems with line input for at least 20 years.
I don't know about cosmic rays and sun storms.
dvv,
After your post earlier today I am thinking the best way to start with a PSU would be a 1:1 EI isolation transformer in a steel box with with 10uF & 1R before & 10uF after.
This would clean up the mains quite a bit AND remove any DC elements that might be present. For power amplifiers this might be enough in itself but for low power, fairly constant current devices we could add LRCLRC filters before going to toroidal step down transformer.
Does anyone see any drawbacks with this ?
mike
After your post earlier today I am thinking the best way to start with a PSU would be a 1:1 EI isolation transformer in a steel box with with 10uF & 1R before & 10uF after.
This would clean up the mains quite a bit AND remove any DC elements that might be present. For power amplifiers this might be enough in itself but for low power, fairly constant current devices we could add LRCLRC filters before going to toroidal step down transformer.
Does anyone see any drawbacks with this ?
mike
The trouble with any transformer is that it always has its own sonic signature, even when it is not in saturation. Isolating transformers pass this signature on to everything after them.
During the last 13 years plus, I have never heard a power line filter using a transformer which did not, more or less subtly, change the sound of my system. This is why I stay away from them.
Just my 2 pennies' worth.
- Status
- Not open for further replies.