Hi,
First I am unaware of any issues with matching of rectifiers. Only one halve of a bridge conducts at any time. Differences may cause some extra 50Hz but it is bound to be minuscule compared to all else.
BUT, for power amplifiers I tend to use more complex transformers and simple 40A dual schottky diodes in a full wave circuit reminiscent of a tube power supply. The parallels extend then with the CLC...CLC filtering, even if the L/R are build into the PCB and the cap's are high value low voltage. That is, if I can design from scratch.
As far as fast/slow, they have different kinds of problems at turn-off. So called fast diodes may have glitches at higher frequency but more energy.
Only tube rectifiers and schottky diodes have no problems on turn off. The same principle as shown in the prior tube link (add a low/no glitch diode in series with a normal bridge) can be applied.
Past that, sticking an extra cap into a resonant circuit will not eliminate the resonance, it will merely adjust it's frequency.
Doing so however may allow a more effective damping of the resonance, for which RC circuits with around 3 - 10 times the capacitance of the "filter" cap and a suitable value resistor (value usually needs empirical derivation). Ideally properly snubbed filter cap's should have as large a value as can be supported.
I recently bought a job-lot of low ESL Japanese 4uF polyprop cap's for bypass duties (physically very small) in some project that parallels yours and Electrocube 15uF mylar cap's of nearly equally small size to use as snubbers, two in paralleled for each main bypass cap, their ESR is high enough to allow me not to use a separate resistor.
A lot of this stuff needs first simulation and then using a network analyser to get right. Anyway, something like that on the secondary windings may be actually of some real use.
I also bought a job lot of superfast IR 40A TO247 diodes and 40A Schottkies to make into a suitable "enhanced bridge".
As a further rule, four smaller value capacitors are a better choice than two big ones IF (big if) some resistance or inductance may be added between them (I like designing it into the PCB print), resulting in a lot less high frequency rubbish reaching the actual amp.
Robert Cordell recommends low value resistors, they are better than nothing, personally I suggest equal DCR Magnan/Ferrite cored inductors, you can get 1mH/0.1R/10A DC in a quite compact package and with a resonance in the MHz range for less money than audiophool bridge rectifier.
This we agree upon. One problem is that instrumentation needed is not trivial.
If you look at the rail, you will see a lot of 100/120Hz ripple. It will obscure the small stuff, so you may add a highpass to tune the ripple and extra gain to your 'scope and so on.
There are good reasons why I would like a 50-100K USD Tek DPO/DSO, it would make much such work a doozie...
I have no problems with the flames as such. I am always happy to provide a better reasoned, more ironic, subtle and pointed reply. What I do feel is a problem is in a different domain of debate.
Ciao T
First I am unaware of any issues with matching of rectifiers. Only one halve of a bridge conducts at any time. Differences may cause some extra 50Hz but it is bound to be minuscule compared to all else.
BUT, for power amplifiers I tend to use more complex transformers and simple 40A dual schottky diodes in a full wave circuit reminiscent of a tube power supply. The parallels extend then with the CLC...CLC filtering, even if the L/R are build into the PCB and the cap's are high value low voltage. That is, if I can design from scratch.
As far as fast/slow, they have different kinds of problems at turn-off. So called fast diodes may have glitches at higher frequency but more energy.
Only tube rectifiers and schottky diodes have no problems on turn off. The same principle as shown in the prior tube link (add a low/no glitch diode in series with a normal bridge) can be applied.
Past that, sticking an extra cap into a resonant circuit will not eliminate the resonance, it will merely adjust it's frequency.
Doing so however may allow a more effective damping of the resonance, for which RC circuits with around 3 - 10 times the capacitance of the "filter" cap and a suitable value resistor (value usually needs empirical derivation). Ideally properly snubbed filter cap's should have as large a value as can be supported.
I recently bought a job-lot of low ESL Japanese 4uF polyprop cap's for bypass duties (physically very small) in some project that parallels yours and Electrocube 15uF mylar cap's of nearly equally small size to use as snubbers, two in paralleled for each main bypass cap, their ESR is high enough to allow me not to use a separate resistor.
A lot of this stuff needs first simulation and then using a network analyser to get right. Anyway, something like that on the secondary windings may be actually of some real use.
I also bought a job lot of superfast IR 40A TO247 diodes and 40A Schottkies to make into a suitable "enhanced bridge".
As a further rule, four smaller value capacitors are a better choice than two big ones IF (big if) some resistance or inductance may be added between them (I like designing it into the PCB print), resulting in a lot less high frequency rubbish reaching the actual amp.
Robert Cordell recommends low value resistors, they are better than nothing, personally I suggest equal DCR Magnan/Ferrite cored inductors, you can get 1mH/0.1R/10A DC in a quite compact package and with a resonance in the MHz range for less money than audiophool bridge rectifier.
This we agree upon. One problem is that instrumentation needed is not trivial.
If you look at the rail, you will see a lot of 100/120Hz ripple. It will obscure the small stuff, so you may add a highpass to tune the ripple and extra gain to your 'scope and so on.
There are good reasons why I would like a 50-100K USD Tek DPO/DSO, it would make much such work a doozie...
I have no problems with the flames as such. I am always happy to provide a better reasoned, more ironic, subtle and pointed reply. What I do feel is a problem is in a different domain of debate.
Ciao T
I criticized approach that creates problems instead of solving them. For example, before filter was installed dirty current through the ground wire from it to outlet did not exist. After installation of the filter it does exist. I.e. instead of understanding of problems and finding solution like TL does people when follow Rules of Dumbs create problems in places where they did not exist.
I knew only a couple of people there quite distantly, and I don't think any of them are still there, so I can't help with explaining motivation for a move to switchmode supplies.
However, when I finally met Bruce Hofer, I asked him about the power supplies in the new line of Audio Precision equipment, as he had mentioned in his presentation that they were switchmode, about which he'd had his initial doubts. Bruce btw is one of the all-time great analog engineers, despite lately also running the company he founded.
I said someone on a forum had claimed that SMPSs were achieving few-microvolt noise on the outputs, which I very much doubted (in fact I said "Bulls^&t", at which Bruce visibly cringed --- note to self about language in the future). Bruce said No, they worked hard to find a good supplier and paid a premium for what they use, and it manages low-millivolts which is adequate for most of the circuitry thus powered. However he added that one of the crucial components in the new instruments is a very good common-mode choke; I believe he said the inductance was 1mH. I have no idea what the self-resonant frequency is, but I gather it's a pretty fancy part as well.
And he added, the payoff of eliminating the hitherto-ubiquitous toroidal transformer was that he now sees no mains frequency-related energy in the spectra, which is there at a very low level in units using toroids like the System Twos.
Brad
Brad, I think that explanation is simple: peak rectifiers are not allowed anymore on power above 200 Watt, and using third party certified PSs is more convenient for businesses. I know guys who invested in certifications and are properly insured, and supply power supplies to many companies.
That may well explain it. Also size-weight, universal mains voltage compatible. On the other hand, I think they would be well below the 200W limit as a test instrument.
But the agency certifications and insurance are a huge factor to be sure.
As I don't have an old Tek 4900 series mainframe, just an old 2235 or not very good software on my PC, and no noise generation rig, I am at the mercy of experienced designers.
I am not there in simulation yet. In modeling the 120's input stage, when I increased the diff pair degeneration, the gain went up. I have one heck of a lot to learn about SIMetrix. Fortunately Bob Cordell came to the rescue with better models for the old transistors. Had to do a tad bit of editing, but they imported fine. Now I know the file format, it is pretty easy.
Does the audiofool bridge work better than a $2 one? Seems so. Better than four $0.65 diodes, doubt it as that is what I think is inside. I'll buy some next order and try them in one of my other amps. Direct test, standard with a single ceramic cap, to fast, then full snubbers.
I was thinking if using chokes instead of/additionally to the resistors in the rails would be viable. Chokes are expensive but they don't have a problem of drop-out a regulator would. Chassis space is an issue though.
In the basic prebuilt line filters, some have a choke in the ground, some do not. The more costly, have over 1mH chokes, the little $2 ones, more like .25. Quality and construction of the internal caps is unknown. Company reputation, country of manufacture, buyer be ware.
I am not there in simulation yet. In modeling the 120's input stage, when I increased the diff pair degeneration, the gain went up. I have one heck of a lot to learn about SIMetrix. Fortunately Bob Cordell came to the rescue with better models for the old transistors. Had to do a tad bit of editing, but they imported fine. Now I know the file format, it is pretty easy.
Does the audiofool bridge work better than a $2 one? Seems so. Better than four $0.65 diodes, doubt it as that is what I think is inside. I'll buy some next order and try them in one of my other amps. Direct test, standard with a single ceramic cap, to fast, then full snubbers.
I was thinking if using chokes instead of/additionally to the resistors in the rails would be viable. Chokes are expensive but they don't have a problem of drop-out a regulator would. Chassis space is an issue though.
In the basic prebuilt line filters, some have a choke in the ground, some do not. The more costly, have over 1mH chokes, the little $2 ones, more like .25. Quality and construction of the internal caps is unknown. Company reputation, country of manufacture, buyer be ware.
What is in something like this? Isolation transformer?
OneAC Power Supply - Line Conditioner CL1102 CL 1102 ~ Take A Look!!! | eBay
OneAC Power Supply - Line Conditioner CL1102 CL 1102 ~ Take A Look!!! | eBay
Thorsten, my onw statistical gathering research over the years indicates that the vast majority of mains noise lies between 10...80 kHz, where about 90% of it resides. These are all household appliances, computers (DAMN big polluters!), tube audio (by far the worst polluter ever, I presume due to thermal noise added to the mix, but I am not at home with tube gear!), etc, sit, and includes up to 4th harmonic.
As a tube affictionado, I suppose you'll take issue with tubes, but it's really a simple test: observe the 1...100 kHz spectrum on your scope with the tube audio off and the with on. Then discuss the problem with the oscilloscope, because the S.O.B. is transistor based.
Above 80 kHz, there's still some muck, as from cell and wireless phones, but bear in view their very low power factor.
My filter is down by 22 dB at 20 kHz, and by 80 kHz, it's down by over 46 dB, reaching a maximum attenuation rate of -72 dB at around 600 kHz. It does have a small comeback hump around 400 kHz, but it's not too bad, around 4 dB.
Oh yes, not to forget, it is an LC filter. While most perhaps do make matters worse, I really can't tell as there are not too many around, I would suggest many do so because they rely on the most unreliable ground for full effect, something I have managed to eliminate as a factor totally.
Nevertheless, as I stated and you reiterated, it is well neigh impossible to KNOW, even if we can guess, but your guess is as good as mine, how a filter will fit in or not fit in a system until we have actually tried it. I do believe it's one of the most heueristic devices ever.
As for the regenerators, they are just as dicey propostion. Theoretically, they should offer the best restults, but practical issues make sure that doesn't happen. In effect, they need to regenerate the sine wave, and it seems many have trouble with that, what they pass for a "sine wave" is notched blur to me; I have no idea why this is to, I never even tried analyzing one, let alone making one.
After that, we need what might be viewed as a power amplifier to boost that regenerated wave. This has its own share of problems, not to mention the required real estate and price issues.
In short, that's a nice theory which is still waiting for its time, in my view. The only such items I have any practical experience with are the PS Audio products, and with all due respect to Paul McGowan, while his is a better fit into many systems than mine, when they both work his regenerators are no match for my passive devices in terms of music quality attainable.
I know of at least 5 cases when his regenerators were replaced by my fiters, and in 3 of those 5 cases, which were local, it takes just minutes to understand why. One is in CERN in Switzerland, and one is in Dubai, property of a Russian gentleman.
Just one more thing - I am very careful when I name names because of Naim. Older generations of their products, from the 80ies and early 90ies, do change for the better with my filter, but newer generations are much more touch and go, some do well, some don't do well at all. Therefore, in all truth, one should not say that Naim products do well with the filter, such a generalization is simply not true, but must be very specific regarding models.
Alas the greenies are in ascendance. You can't even easily obtain simple transformer adapters with dual bobbins and low primary-secondary capacitance --- everything for things like chargers etc. is supposed to be switchers now, with far poorer isolation. Balu Balakrishnan, who used to work for me as a tech while finishing up his EE Masters at UCLA, has helped to make products using his PI chips as something mandated. Energy vampires you know
At higher powers the concern is the power factor. If you look at your local mains you will usually see a significantly clipped quasi-sinusoid, due to the numbers of cap-input power supplies (most of them switchers, but the first thing the AC hits is diodes and a HV bulk cap). So we now have requirements on power factor correction and more switching electronics to manage that (to make the current demanded from mains more sinusoidal).
Brad
The simple truth is that most power line filters do not offer much power handling, and if they do, more often than not it will be a transformer based device. At high power levels, all transformers will eventually start to saturate, so what you'll get out is anybody's guess.
Of course, my filter is no different, the only snag is that the saturation point is well above its nominal rating. Regular model is declared at 10A at 240VAC, or 2,400 VA; this is mandated by the obligatory European safety regulations 10A fuse, which must be found in all home devices (this excludes some home APPLIANCES, such as refrigerators, stoves which run off three separate phases anyway, hot water boilers, etc, but not say air cons). First signs of saturation start to appear at around 13 A, and by 15A, they are quite obvious. Yet, at 10A, they are simply not visible however you look at them. Much less audible.
Super model doubles the lot - 20A at 240VAC, better filtering curve, halved output impedance, but also twice the price of the regular model.
Wayne, there are VERY few amplifiers on this planet which will ever draw 4,800 VA from the mains, probably less than 10 all told. So, any power amp rated up to say 500/1,000 Watts into 8/4 Ohms will have no problem whatsoever.
And here we come to your second point, that filters work best on source equipment. While this is easily explained - source components rarely manag to gobble up even 100 VA from the mains, so almost any filter will accommodate them in terms of power - I find that when an amp is well matched with a filter, it will be a difference very easily heard, and on occasion, it could go as far as stunning.
A US Naim dealer from St. Paul, Minnesota, commented that until he had used my filter, he didn't know Naim could do such bass, and bear in mind, at that time, Naim amps all had fully electronically regulated power supplies, even the small 30 wpc model. That was in 2004.
My biggest surprise, albeit a very pleasant one, was when Krell power amps reacted as they did, most favorably. Those were the late-90ies models, and I managed to get their service schematics and was even more surprised to discover that they had double regulation per channel! How and why, I really have no idea, but the systems they were sitting in in both cases cost well over $100,000 each. I honestly did not expect there to be much of a difference if at all, and to hear such a tremendous difference was a shock to me, literally. I was amazed.
And that's all I am prepared to say on the subject, because I am beginning to feel uncomfortable talking about my own products, and especially the power line filters. I read ads too, and know only too well the kind of BS written there, it's worse than in case of cables. Myth, magic and voodoo gallore!
My last thought on the subject - remember, there are many variables in the whole thing for anyone to KNOW how any filter model, by anyone, will sit in any system, until you actually try it. Only then will you KNOW.
A good approach I strongly support.
However, without myself in the discussion, as it is becoming an obvious conflict of interest. All I can offer are some effects I have looked into myself, but which are true and good only for myself.
My current PC is an HP unit, using an Intel i-5 core chip, running at 3.3 GHz, and backed up by 6 GB of RAM. It's now 7 months old. My monitor is a 20 inch unit from ViewSonic, designated VA2216w, now about 2 years old. Don't know which HDD is inside, but it's a 500 GB unit running at 7,200 rpm. The mammaboard has integrated everything but the kitchen sink, and there are no additional cards inside.
My room is kept at a constant 24 degree centigrade (75 Fahrenheit) by the air con throughout the year.
Without the filter, the temperature probe says the inside air temperature of my PC is 31 (88 deg. F) degrees centigrade afte 2 hours of working with it. When it's run off the filter, inside temp drops to 29 centigrade (80.6 deg. F). HDD operating temp is 45 C (113 F) without and 41 C (106 F) with the filter.
The picture on the LCD screen is a little sahrper, a bit better defined and with visible deeper color.
Take it from here.
Energy Star appears similar to UL certification (except of course UL is a private entity whereas Energy Star is a US Government certification) in that any product that is submitted and meets certain standards gets to use the label:
How a Product Earns the ENERGY STAR Label : ENERGY STAR
Class D Home Theater receivers would surely pass, but Class A tube amplifiers need not apply.
DVV,
What impedance were the Krell's being operate at, on High sensitivity 8 ohm load speakers it will not matter much, drive big ESl's or magnostats and you will hear a difference by going to the wall. Again i have not heard your setup , so I'm not doubting you, maybe one day i will give one a twrill again ....
Sometimes things are different in revision ....
What impedance were the Krell's being operate at, on High sensitivity 8 ohm load speakers it will not matter much, drive big ESl's or magnostats and you will hear a difference by going to the wall. Again i have not heard your setup , so I'm not doubting you, maybe one day i will give one a twrill again ....
Sometimes things are different in revision ....
Hi,
Basically, my advise, understand the problems.
Most engineers do not.
When I see a transformer symbol, I see a raft of "parasitic" components superimposed over the simple symbol and I prefer to control these as much as possible (means custom designs). If I cannot control their values, I like to at least know them.
The same goes for passives, I see a capacitor symbol, I see a complex RLC network. Where I can, I even prefer custom stuff here, because I can shift values in ways that suit me, but for electrolytics I'd rather good off the shelf made in japan (like really made there) than most MIC stuff, even if they will give me any custom design I like. And again, know the parasitics actual value.
One of my regular suppliers for small stuff has an originally very expensive 2nd hand RLC meter (and an automated Elcap Reformer), usually every time I buyt 50 Euro's worth of stuff from him I turn up with a bag full of stuff I want to measure on this meter or want to reform. He seems happy enough to indulge me...
If you cannot follow my example, there are somewhat indirect ways to get a pretty good (like within 10% which is good enough for most work) idea of parasitic values, though it means making measurement jigs.
SMPS moves the problems from one domain into another. The high frequency means that the components needed to deal with the noise can be cheap and small, but the application is considerably difficult. Given the truly dramatic saving from a SMPS over a conventional supply it makes good sense to spend much engineering time to get it right.
Ciao T
Basically, my advise, understand the problems.
Most engineers do not.
When I see a transformer symbol, I see a raft of "parasitic" components superimposed over the simple symbol and I prefer to control these as much as possible (means custom designs). If I cannot control their values, I like to at least know them.
The same goes for passives, I see a capacitor symbol, I see a complex RLC network. Where I can, I even prefer custom stuff here, because I can shift values in ways that suit me, but for electrolytics I'd rather good off the shelf made in japan (like really made there) than most MIC stuff, even if they will give me any custom design I like. And again, know the parasitics actual value.
One of my regular suppliers for small stuff has an originally very expensive 2nd hand RLC meter (and an automated Elcap Reformer), usually every time I buyt 50 Euro's worth of stuff from him I turn up with a bag full of stuff I want to measure on this meter or want to reform. He seems happy enough to indulge me...
If you cannot follow my example, there are somewhat indirect ways to get a pretty good (like within 10% which is good enough for most work) idea of parasitic values, though it means making measurement jigs.
SMPS moves the problems from one domain into another. The high frequency means that the components needed to deal with the noise can be cheap and small, but the application is considerably difficult. Given the truly dramatic saving from a SMPS over a conventional supply it makes good sense to spend much engineering time to get it right.
Ciao T
Hi,
Well, looking at these filters I see two issues.
1) Being designed for EMI Testing compliance their filtering ability where it matters is, as me mate chalky used to say, "Feck All" (this is apparently a technical term in Ireland).
2) Noise leakage into the chassis is increased.
Ciao T
Well, looking at these filters I see two issues.
1) Being designed for EMI Testing compliance their filtering ability where it matters is, as me mate chalky used to say, "Feck All" (this is apparently a technical term in Ireland).
2) Noise leakage into the chassis is increased.
Ciao T
What if they look like this .... ?
General Purpose Transformers - Products - Quality Transformer and Electronics
General Purpose Transformers - Products - Quality Transformer and Electronics
- Status
- Not open for further replies.