|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
82.156.128.160
In Reply to: RE: cMP - the open source high-end Memory Player posted by cics on December 30, 2007 at 04:42:01
Improving Hybrid Power Supply through filtering. Part 2.Hi All,
Building an all linear power supply unit for my cMP-setup is beyond my DIY capabilities. So for a quite a while I’ve been looking for ways too improve the Hybrid Power Supply situation in ways that are in reach of my DIY capabilities. (A Hybrid Power Supply is: a Linear PSU -> P4 & linear PSU -> Pico PSU (160 XT) -> P24 ).
A linear PSU may not produce switching noise, but the rectifiers diodes and regulators used inside linear PSU’s still produce noise. That’s is why guys who build there own high quality DIY linear PSU stress the importance of using high quality low noise rectifiers diodes and high quality low noise regulators (Noise that isn’t produced, doesn’t need to be filtered out). Also they use circuit designs in which much attention is given too smoothing ripple and noise filtering through use of: large chokes, large capacitance, capacitors with (very) good ((V)HF qualities, gyrators, or whatever means.
In my cMP setup I use mass fabricated linear bench PSU’s (Velleman, Peaktech, TTi) for powering the P4 and Pico on the P24. Although mass fabricated Linear PSU’s are a big improvement over using a standard (noisy swithing) ATX computer PSU, I tried too see if some extra filtering on the output of these standard linear PSU’s, would improve sound quality.
Some results where nothing less but astonishing. This is what I tried:
1. Ferrites
I first tried the most simple way of filtering I could think of. Clamping ferrites on the output. Ferrites simply can be clamped on the power lines feeding the P24 (coming from the Pico) and the P4 (coming from a linear PSU). When done correctly (!!) ferrites on the P4 and P24 effectively filter frequency’s from 100 mHz to 500 mHz. The positive effect on sound quality can be easily heard.2. Cap - ferrits – cap (pi-filter).
When reading papers on suppressing conducted EMI I red in the murata paper (C39E) that when ferrites are combined with caps, the filter band can be lowered down too 1 mHz. See paper.
So I tried to combine ferrites with caps (with good VHF qualities) to create a pi-filter.
Some creativity and ingenuity is needed when applying a pi-filter from ferrites and caps on the P24. All (!!) P24 wires which are connected too the Mobo need filtering.
But when done correctly, a pi-filter from ferrites + caps on the P4 and P24, greatly (!!) improves sound quality.3. Messing around with pi-filters from ferrite based coils/chokes + caps.
Since the pi-filter improved sound quality significantly, I decided too see if there would also be an simple way too filter the 300 kHz switching frequency of the Pico (Which is well beyond 1 mHz)
I tried too see if using a simple ferrite based choke had any effect.
I had some mains filters lying around with 40 mH ferrite based chokes inside.
http://www.schurter.com/var/schurter/storage/ilcatalogue/files/document/datasheet/en/pdf/typ_FPP-01.pdfI also had such a data transmissionline filter from TDK lying around.
http://www.tdk.co.jp/tefe02/e971_zjys.pdf
These little ferrite based chokes are used for noise removal on data transmission lines.I first tried them on the P4 as this is most easy and simple too do and also just too see if there is any effect on sound quality at all.
Too my surprise these ferrite based chokes have equal effect (!!) on sound quality as the pi-filter made of ferrite+caps. However I find implementing much more easy. It only takes about 15 minutes too solder such a filter onto a P4 extension cable. Just cut a P4 extension cable in half and solder the ferrite based chokes onto the P4 extension cable.
Since it is so easy, I constructed all kinds of variety’s. See pictures
Van on filtering, de-coupling, demping, ect All varieties roughly had the same (big) effect on sound quality.
Really amazing !!
It is also much less work than constructing a pi-filter from ferrites + caps4. Creating a filter through coupled inductors.
Techguy on DIYaudio pointed at a very special use of coupled inductors.
http://www.hamill.co.uk/pdfs/ciabfbb_.pdf. (see link at bottom)
In this paper a filter is presented which works through cancellation.
AC ripple and noise is passed through 2 coupled coils, in such a way that the AC ripple and noise will cancel out each other.
See picture
Van on filtering, de-coupling, demping, ect The ferrite cokes in these pre-fabricated filters are also coupled.
So it was easy too try and see what effect they would have on sound quality if used this way.
Since construction was again very easy, I constructed several variaties. See picture.
Van on filtering, de-coupling, demping, ect These coupled inductor filters sound best !! Not by a mile. But these coupled inductor filters noticeably sound better than the filter techniques used under 1, 2 and 3.
Filter method 4 (coupled inductors) sounds best on the P4 !
It's also very easy too construct.So I ordered 8 of these little TDK data transmission line filters too try them on the P24.
It's very easy too solder them on the P24 lines.
These little filters have a DC resistance of only 0,12 Ohm so there will be hardly any voltage drop.Probably by the end of the week I will have constructed a P24 extension cable with on each line a filter working through coupled inductors.
As always any thoughts, suggestions, remarks and comments are very much welcomed.
Especially what ready made coupled inductors there are on the market which are easy too use for such filters.Too be continued.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 09/05/11 09/05/11 09/05/11 09/05/11 09/05/11Follow Ups:
Very lucid description of your journey, part of which is mine on ferrite/caps so I am very interested. Are you using smt's or others? How costly are these? Let us know how they work on p24 but how do you pair the + and - leads for each voltage supply?
Hi Theo,“Are you using smt’s or others?”
If smt’s stands for: Surface-Mount Technology, than I think the answer is: yes.
The ferrite based coils I use are inside commode mode filters which are designed for being mounted on a PCB.
Their physical appearance, physical size, circuit designs and specs can be fond in these brochures.
For the 230 AC mains filter block see:
http://www.schurter.com/var/schurter/storage/ilcatalogue/files/document/datasheet/en/pdf/typ_FPP-01.pdf
For the TDK data transmission line filter see:
http://www.tdk.co.jp/tefe02/e971_zjys.pdfThere is no real ration or special thought on my side in using these common mode filters.
The only reason I used these filters is: inside them there are (ferrite based) chokes/coils.
Also I had them lying around, thus at hand.These filters are relatively cheap.
The Shurter AC mains filter (with 40 mH choke inside) cost: 4,02 euro.
See link: http://www2.conrad.nl/goto.php?artikel=534625
The TDK common mode data line filters are: 3,07 euro.
See link: http://www2.conrad.nl/goto.php?artikel=443410I don’t advocate use of these filters. I just noticed that the (ferrite based) chokes/coils inside these filters are coupled inductors.
May be, or even probably, there are better coupled inductors available for use in filter with coupled inductors.May be more knowledgeable inmates should comment and point at better suited coupled inductors for this purpose. But unfortunately most knowledgeable inmates around here on the PCaudio asylum are not interested (or don’t see the potential) of some extra filtering on the output of a linear PSU and or some extra filtering of the output of a Pico.
Pitty, because as you experienced yourself: IF DONE CORRECTLY, the sound quality greatly benefits from these filters.The effect of these filters can be heard very clearly when applied ontoo the 3,3 DC volt psu line that is powering the digital part of the Juli@.
Minute changes in these filters on the 3,3 psu line, can be heard immediately in sound quality.I now have applied coupled inductor based filters together with the ferrite+ caps based filters (which where already installed) on the lines that power the: SSD, the digital part of the Juli@ and the P4.
The effect is no less than breath taking. I think I now have some glimp at the ultimate sound quality which guys with an all linear PSU are enjoying.I will let you know the SQ results when I also finished the P24 with coupled inductors filters.
Pairing the + and – wires are not important ihmo.
There is no saturation effect through DC-bias on the ferrites in this situation.
I just will just bundle all ‘-‘ wires and than simply ground all Lac coupled inductor-sides too it.More to come.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 09/05/11
Hi Mark
I noticed that in one of the photo the U1 on the Juli@ had been removed. Have you got around to replacing the LT1117 on th Juli@?
Best Regards
Hi smicyta
I wanted too replace it with a shunt regulator from makes like: Newclassd (the dexa regulator) or from my country fellow man Guido Tent (a Tentlabs Shunt regulator).
Since shipping from Denmark (only 750 kilometers from Rotterdam) would cost 27 euro’s and shipping from Eindhoven (75 kilometers) would cost € 4,70. I choose the Tentlabs Shunt regulator.
Van on filtering, de-coupling, demping, ect
But I could not get it too work. The Julia digital part however works when I power it with my TTi linear PSU. Than I saw that the Juli@ digital part uses 113 mA in my setup.
But being an amateur, I did not check the specs very well before I ordered the Tentlab shunt regulator.
When I connected the shunt regulator too the Juli@ digital part, the shunt regulator did not produce more than 2.3 volts. No matter how I tried. When not connected too the Juli@ digital part, it nicely produces 3.3 volt.
Only after I did read again the specs, I discovered that the Tentlab Shunt regulator can only provide 100mA. So I guess that is might be the reason that the regulator will not output 3.3 V but 2.3 volt.
So right now the bulky and monstrous TTi is still powering the Juli@. But too keep the peace at home, I have too come up with a solution very fast.
So I ordered an IC LT 1086 CT3,3 3,3V POS VR 1,5A T regulator.
My soldering skills are improving. And I also have less fear messing things up. I’m planning too construct my own DIY regulator. I want too feed it with filtered 5 volt DC from the Pico and than feed the resulting 3.3 V too the Juli@ digital part. Everything filtered as much as possible. See how that will sound.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Hi Mark
I am currently feeding the Juli@ with a 5V Belleson regulated supply to Pin 2 & 4 of J1. There is some improvement over an off the shelf linear supply.
I am thinking of bypassing U1 and use a Belleson 3.3V 0.5A regulator for a direct 3.3V feed similar to what you are doing. Just wondering if this is worth doing?
Warm Regards
Hi Smicyta,As also reported by other inmates removing the LT1117 regulator and directly powering the digital part of the Juli@ with 3,3 Volt DC low noise power, it is one of the tweaks/modifications with biggest positive effect on sound quality.
Also in my cMP setup this is the one with biggest impact SQ, compared too al previous power supply changes/modification/tweaks I did.
- ‘smoothing’ caps on the P4 coming from an ATX PSU
- a linear on the P4
- a Pico 160 XT on the P24
- replacing the HDD for an SSD
- various filters on P4 & P24
- various filters on in- & outputs of the linear PSU’s and in- & outputs of the Pico 160 XT.Yesterday I did my first electronic construction and soldering project ever! I constructed a simple voltage regulator based on the LT 1086 CT3,3 3,3V POS VR 1,5A T regulator. It turned out too be not being so hard at all.
Since I like too experiment with filters before and after I again used P4 connectors, so that I can easily swap filters in and out before and after the regulator. Don’t be fooled by the yellow colour of the wires. There is only 5 Volt DC on the input too the regulator. The regulator needs 5 Volt DC input and gives 3,3 Volt output.
Van on filtering, de-coupling, demping, ect I’m not completely surprised that there only was 'some' sound quality improvement in your setup. The noisy elemets in the power supply are still present. You also need too remove the noise sources that produce noise so that noise levels drop as much as possible.
Right now I’min the process of evaluating various home made filters before and after my home made LT 1086 CT3,3 regulator and also what PSU I need for powering the input.
Van on filtering, de-coupling, demping, ect With adequate filtering is it perfectly possible too power the regulator with 5 Volt DC coming from the Pico 160 XT.
Best results so far I get with:
* regulator input: filtering the 5 Volt Pico output with the filter shown in the left lower corner
(a 230 AC filter with 40 mH ferrite cokes inside + caps added)
* regulator output: the simple TDK common mode filter for data transmission lines at the right corner.So do try too clean the power of the Juli@ digital part by providing low noise power but also by removing noisy power sources. Mixing low noise power sources with high noise power sources is obviously giving less noise reduction, thus less sound quality improvement.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 09/08/11
Hi Mark
I do agree with you that a clean power supply is imperative for excellent SQ with a computer based transport.
I really admire your enthusiasm in all those filtering works to clean up the power supply.
Let me digest what you had done so far and see if I replicate what you are doing.
BTW, what wire size do you used from the Juli@ Output & GND of the removed U1?
Warm Regards
Mark: in your comment '...regulator output: the simple TDK common mode filter for data transmission lines at the right corner...' what is the part number of the tdk filter you use?
Hi Theo
The part number for the TDK common mode filters for data transmission lines = ZJYS51R5-2PT-01
But probably there are better solutions out there.
I didn’t do any research yet on if there are alternative common mode filters that better suite the purpose for which I use them (coupled inductors in a filter).
I’m almost sure there are better coupled inductors for this purpose.
I didn’t yet spend some time on research
Until now I just did some experiments with various types of filters too get some first impressions on what effect filtering would have on sound quality in a cMP setup with hybrid PSU situation.
But from these first impressions I conclude filtering definitely should be done in a hybrid PSU situation.
(linear on P4 & Pico 160 XT on P24 & SSD).
It gives a very nice step up in sound quality.
Since I now know that filtering the output of mass produced linear bench PSU’s filtering the input & output of the Pico 160 XT is very rewarding, the next thing too find out is: which filter types have best combination of: ease of construction, costs and effect on SQ. And also too find the construction parts that best fit the job.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Thanks Mark
What´s the specification TDK data transmission filter as you request and from where?
I already included this link in the original text
Please read.
http://www.tdk.co.jp/tefe02/e971_zjys.pdf
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: