| till |
ok, i frequently read ferrite beads should be used in IC power supply line.
some questions: why not use small inductors like http://www.reichelt.de/bilder/web/B400/!SMCC.jpg
?
and for the beads, i got some AMIDON beads http://www.reichelt.de/bilder/web/B400/T20.jpg
diameter 6,48 mm
inner diameter 3,05 mm
thick 2,44 mm
AL (ĩH/100Wdg): 34
MHz max.: 30
and enamelled copper wire 0,6 mm diameter.
So how many windings of wire should i make with the wire trough the bead? Or what values of those inductors of the kind package looking like resistors? Shoul the ferrite bead near to IC like the capacitors, or should it be near to power supply to filter HF coming from powerline, regulators etc. ? What about a Filter chain like 1000uF, 100R, 220uF, 100uH, 220uF,ferrite bead, 10uF+100nf at IC? |
|
|
| Jocko Homo |
It looks like what you have there is not a ferrite, but a powdered iron toroid, used for winding inductors.
Ferrites beads are usually specified as to what ferrite mix they are made of. Toroids for inductors are usually specified as to "max useful frequency", and uH/x turns.
A ferrite bead should be either a type 43 or 73 mix. There are a few other mixes that they are made with, but are not as common. Unless the supplier can tel you which mix, do not buy them unless you have lots of experience.
Jocko |
|
|
| till |
Thanks for information. Its no problem, i got 10 of those "wrong" beads/torroids for free. Whatever i will do with them... I have no problem to get some ferrites with type 43,61,or 77 .
But please give me information how many windings i need, or what values for inductors, to make the power supply line for a DAC / AD1865NK.
My problem is not to get the right parts if iīm told which ones, but i have no information about values to use or a schematic to looks at.
In other words, i lack the experience you mention. |
|
|
| Bricolo |
Jocko, I haven't found tha material used in the chokes I use
are they any good? |
|
|
| Elso Kwak |
| quote: | Originally posted by Bricolo
Jocko, I haven't found tha material used in the chokes I use
are they any good? |
Bricolo,
If look at the graphs in the paper you submitted you will see that the BL02RN2 offer excellent attenuation at "CD frequencies".
Guido Tent is suggesting BLM31A J601SN1 also from Murata. These are SMD beads. (Farnell part# 581094)
No more hints from me, remember you are the student in electronics, I am just the bloody chemist.:angel: |
|
|
| till |
| Anybody who could give me some answers? |
|
|
| Bricolo |
| quote: | Originally posted by Elso Kwak
No more hints from me, remember you are the student in electronics, I am just the bloody chemist.:angel: |
You know I also studied chemistry ;)
I hope my teachers won't read this, but for the momment I learnt more from you and other people here on diyaudio than from them |
|
|
| Elso Kwak |
| quote: | Originally posted by till
Anybody who could give me some answers? |
Hello Till,
Your coils are useless at high frequencies.:idea: |
|
|
| Elso Kwak |
| quote: | Originally posted by Bricolo
I hope my teachers won't read this, but for the moment I learnt more from you and other people here on diyaudio than from them |
Now, is this a judgement of value on you or your teachers.....?
:confused: :eek: |
|
|
| till |
which coils, the SMCC ones, or the toroids Jocko allready told me are useless? What frequencies are we talking about?
- So how many windings of wire should i make with the wire trough the bead?
- Or what values of inductors is needed?
- Should the ferrite bead near to IC like the capacitors, or should it be near to power supply to filter HF coming from powerline, regulators etc.?
- What about a Filter chain like 1000uF, 100R, 220uF, 100uH, 220uF,ferrite bead, 10uF+100nf at IC? |
|
|
| Bricolo |
| quote: | Originally posted by Elso Kwak
Now, is this a judgement of value on you or your teachers.....?
:confused: :eek: |
Absolutely not!
My teachers are nice, I only said that I've learnt more things here for the things I'm interested in (audio related electronics) |
|
|
| till |
i have nothing against disussing the quality of Elso or others as a teacher in this thread, but i would much prefer to learn if these ferrites/inductors in power supply line are snakeoil or valid physics. In second case there must be some numbers in Herz we want supress, Henry the devices should have , etc.
http://www.diyaudio.com/forums/show...1765#post381765 |
|
|
| Bricolo |
This has already been answeared
the coil is OK, the toroid isn't (use a ferrite bead instead) |
|
|
| Elso Kwak |
| quote: | Originally posted by Bricolo
This has already been answeared
the coil is OK, the toroid isn't (use a ferrite bead instead) |
Sorry, wrong answer. |
|
|
| till |
Yes, but:
- So how many windings of wire should i make with the wire trough the bead?
- Or what values of inductors is needed?
- Should the ferrite bead near to IC like the capacitors, or should it be near to power supply to filter HF coming from powerline, regulators etc.?
- What about a Filter chain like 1000uF, 100R, 220uF, 100uH, 220uF,ferrite bead, 10uF+100nf at IC? |
|
|
| jcx |
for good high frequency attenuation you only use 1 pass thru the bead, not "turns"
when the ends of the wire (or worse, adjacent turns) get close together they have capacitive coupling that shorts out the inductive/resistive impedance from the ferrite - check out "self-resonant frequency" for multiturn inductors
its better to use more (or larger) beads (with a hole that barely passes your wire) on a straight wire than to try to increase inductance with extra turns - some multi hole beads are made to get higher loss in a small space but the rule is still one pass of the wire thru each hole |
|
|
| till |
| ok, if they have no values i will give up and not use them, i use some 100uH coils instead. |
|
|
| till |
| Oh, thanks, some hard information, so i use a large bead with small hole , material 43, no turns, some of them on a wire. |
|
|
| jcx |
finding hard specs for lossy beads is frustrating, the best you will find is impedance vs frequency plots at varying current levels, you can sometimes extrapolate the low frequency end of the graphs for a "L" value - it will typically be low, fractions of uH
lossy beads are used for attenuation of >> 1 MHz frequencies, inductors and chokes are better described by a finite "L" value < 1 MHz but are limited by their self resonant frequency
http://www.steward.com/Technical_Info.asp may help |
|
|
| Jocko Homo |
You need to stop thinking of ferrites as coils, chokes, or inductors.
They are lossy materials that exhibit a resistive effect when inserted into a circuit. If you know the frequency range that you are trying to eliminate, then you look for a bead with the right mix.
You most likely want a 43 type material for what I assume you trying to do. Maybe a type 29, but they are not common. Type 73 is common, but probably not what you want. (They work in some types of output stages....................)
Fair-rite probably has web site explaining all of this.
Jocko |
|
|
| till |
oh thanks, now i start to understand something about this.
What i want is a power supply as good as possible without much effort in multilayer PCB and this, so i get some of those 18Bit resolution of the AD1865, and not only 10 or 12...
As i understand i need a clean power supply as else the noise may be higher than the lower Bits. Also i thought with HF on the power line some IMD may disturb my audio signal, and this may be a reson to filter HF away ( from were ever the HF may come)
so i could get material 43 like small toroids, inner diameter 0.05 mm and outer 5.80 by 1.52mm, or material "ferroxube 3b" small cylindrical beads diameter 3.5mm, lenght 3; 5 or 7,5mm and a hole of 1,3mm.
I assume the last are right? |
|
|
| Jocko Homo |
Is another source to check out. Assuming that someone like Vishay hasn't bought them out.
Good explanations on how they work.......goofy way of specifying material.
Jocko |
|
|
| Charles Hansen |
Till,
I'm not sure if others have made it clear. At these very high frequencies there is enough capacitive coupling between adjacent turns in any multi-turn device to make it ineffective.
Regarding ferrites, I would avoid them at all costs. In my experience they do more harm than good sonically. This is hard to get a handle on because when you first put them in the circuit, they do sound good. But over time they become magnetized and harm the sound. So they have the unusual property of making your circuit sound better when you put them in, and again sound better when you take them out. This effect is easily demonstrated by demagnetizing any ferrites in your audio system with a bulk tape eraser.
If you want to achieve good isolation, one of the easiest, cheapest, and best sounding things is to use C-R-C pi filters. That assumes that you are working with low-current circuits (not power amps) and that the added R is tolerated.
Good luck, and don't forget to listen to what you build. |
|
|
| Jocko Homo |
And there are some problems that only ferrites will fix.
Sorry, Charlie.........
Isn't that what the Star-kist guys say?????
Jocko |
|
|
| jewilson |
Charles,
You are saying that the magnetic properties of a FERRITE changes do to DC and some small HF energy. :xeye: |
|
|
| Jocko Homo |
He just hates everything magnetic. Not a bad phobia, but some things can't be avoided.
Jocko |
|
|
| till |
ok, so i have a dozend of beads. Cylindrical, hole 1,3mm, D= 3,5mm height some 3, some 5 and some 7,5mm . Material is " ferroxube 3b ".
Should is use a lot of shorter or less of the longer ones on each line? Should i use them on a) only + and - voltage line, b) on ground also, c) on analog and digital supply or only analog, d) only on the power lines or also on signal between DAC chip and I/V converter stage ?
for the special kind audiophile gurus: e) do they sound better if painted green? |
|
|
| jean-paul |
| quote: | | for the special kind audiophile gurus, do they sound better if painted green? |
Leave them one night in olive oil and put them in the oven for one hour at 200 degrees Celsius the next day. Totally different then.
You can combine this action together with baking a pizza if you like. Don't eat the bead however. |
|
|
| Charles Hansen |
| Actually they sound best when you leave them in your drawer, far away from your audio system... |
|
|
| Fred Dieckmann |
"You are saying that the magnetic properties of a FERRITE changes do to DC"
The last time I looked they did ............
From: http://hyperphysics.phy-astr.gsu.ed...olids/hyst.html
" When a ferromagnetic material is magnetized in one direction, it will not relax back to zero magnetization when the imposed magnetizing field is removed. It must be driven back to zero by a field in the opposite direction."
From: http://www.coilws.com/magneticandhow.html
"So, it is very important in a choke or inductor design, not to drive the core into saturation by increasing the current (AC or DC). Usually it is the DC current that saturate the cores since it is a constant current, and puts the cores to a certain flux level."
A another description of the Hysteresis Loop at:
http://www.ndt-ed.org/EducationReso...teresisLoop.htm |
|
|
| Fred Dieckmann |
"Actually they sound best when you leave them in your drawer, far away from your audio system..."
Except for my Versalab Red Rollers.........
Better take those pulse transformers out of the digital interface.
My point is there are places for ferrite materials in audio systems. You have to know the characteristics of the materials and how they are used. I thought everyone owned a bulk tape eraser for demagnetizing the materials in their audio circuits. Mine has has cooling holes drilled in the case to use it longer before the thermal breaker kicks in. |
|
|
| till |
| quote: | | Actually they sound best when you leave them in your drawer, far away from your audio system... |
as nobody wants to answer my question this may be the best suggestion until now. But do you really say they affect sound in a bad way when used in the spdif recivers power supply? or only on analog circuit?
Anyway, i would prefer to try it out for scientific reasons, but better to know the right way how to use them before. |
|
|
| Jocko Homo |
Why on earth would you want to put them on your data lines?????
Where you use them depends on how you want to use them.
IOW, just what do you expect to accomplish by using them. You need to decide that before you get answers that you can understand.
Jocko |
|
|
| till |
| quote: | | Where you use them depends on how you want to use them. |
i did not knew they have more than one puprose. The only real case of ferrite bead use i know uses one big bead around all lines and is used at computer interconnects (monitor cables etc). I want to use because if you post on this board you built a DAC and it works, people do replay :"use ferrite beads" and do reply " do not ever use ferrite beads" , so there is one solution to use them and not use them to see what happens. As the ferrite bead is not obvios scum like some other things people here tell you to do it may serve a real purpose, to keep away noise from my circuit. Mainly to keep it away from traveling from the digital part where i expect a lot of noise to be as a normal fact into the analog section of the DAC and into the following amplifing and "probably mixing" stages where i donīt want it.. A lot of postings on this board contain the word ferrirtes, but there is not much hard information. I find the information they to work but not how to use them, sorry. Maybe i do not really need them and better use this approach to separate wanted from unwanted signals in the different parts of the DAC:
http://www.diyaudio.com/forums/show...5618#post395618 |
|
|
| Jocko Homo |
Why do you insist on making life diffiicult???? Get rid of those silly isolators. Expensive part that you don't have a problem that they are a solution for.
If you are tyring ot eliminate EMI.......which seems to be what you are aiming for.........even though you don't seem to grasp why......it is best to......
Keep the loop as small as possible.
That will solve a lot of your problems, without the need for stuff that you are not proficient in.
Yet. Learn the basics, before you try to conquer the mountain.
Jocko |
|
|
| till |
| quote: | | Keep the loop as small as possible. |
ok, will do in the final version. Not easy while using spdif.
| quote: | | you don't have a problem that they are a solution for. |
nice to hear the problem is not there, why did jewilson tell me in the other thread it is?
I canīt estimate if the AC i see on the scope does harm the DACs precision or not. But i know the noise i see on the scope does not get less with other/different/better decoupling, more or less filters in the power supply lines or what else i did try. May be there is no way to measure less noise as it strays in every piece of metal i connect the pobe to. What i could do is measure THD/IMD via FFT what is a stable kind of meausurement and test with different approaches against noise. Such as complete isolation of power supplys. I donīt know if it will cause any diffrence, as nobdy tells me i can only try.
by the way, as nobody seems to be able to show valid measurements of all those fancy clock circuits around on this board: may they also be solutions for a problem not there? |
|
|
| Elso Kwak |
| quote: | Originally posted by till
ok, will do in the final version. Not easy while using spdif.
nice to hear the problem is not there, why did jewilson tell me in the other thread it is?
I canīt estimate if the AC i see on the scope does harm the DACs precision or not. But i know the noise i see on the scope does not get less with other/different/better decoupling, more or less filters in the power supply lines or what else i did try. May be there is no way to measure less noise as it strays in every piece of metal i connect the pobe to. What i could do is measure THD/IMD via FFT what is a stable kind of meausurement and test with different approaches against noise. Such as complete isolation of power supplys. I donīt know if it will cause any diffrence, as nobdy tells me i can only try.
by the way, as nobody seems to be able to show valid measurements of all those fancy clock circuits around on this board: may they also be solutions for a problem not there? |
Hi Till, the scope may be part of the loop. Beware with scope grounded to the safety earth of the mains
:att'n: |
|
|
| carlosfm |
| quote: | Originally posted by jean-paul
Leave them one night in olive oil and put them in the oven for one hour at 200 degrees Celsius the next day. Totally different then.
|
:D
And then freeze them at -180šC for 48 hours.
Take them to room temperature very quickly under fire.
If they don't crack after this, they're good.:clown:
Microdynamics much better.:yikes: |
|
|
| till |
| quote: | | Hi Till, the scope may be part of the loop. Beware with scope grounded to the safety earth of the mains |
Iīm pretty sure it is part of a loop and it is of corse grounded as it īhas a 3 pin power plug. The noise i measure is not 50Hz but really HF like about 40 to 60Mhz / each day different distributions.
Also my impression is the power mains/null/ground lines are full of miracolous noise, null shows about 2V RMS against ground also they are connected at the house power inlet box.
What should i do? belive there is no problem? ok. If this is the way to build 18Bit DACs. |
|
|
| Jocko Homo |
When you lay out a PCB, keep the loops small as to minimse EMI.
Look..........go out and buy some DAC kit......any kit. Build the silly thing. Get a good 'scope......>100 MHz. Learn to use it. (You are a student, right?)
Read Guido's article on grounding. Read it again. And once more. Lay out a simple board.......similar to your simple kit, using his techniques. Measure the two of them. Learn to identify what the differences are, and why they are there. Then decide if you need ferrite beads everywhere, and silly isolators.
What guys like Wilson and I do, and what you should be trying, are on different levels. You need to master the basics first......and I do not sense where you have by the nature of your questions.
You are not going to learn anything by trying to make something overly complicated.....and for no reason........when you are trying something more than you can handle.
You really don't think that we started out that way, do you?
Jocko
BTW......I see no reason why you can't keep the loops small just because it is SPDIF. The loop has to be small on ANY PCB to keep the EMI down. Read Guido's paper................... |
|
|
| till |
- i donīt need to get a scope, i have two of them, one is TEK7904, what should be fast enough.
- I will not buy any silly DAC kit as the circuits in those kits with TDA1543 etc. i can get working now in about half an hour without need of kit. Did half a dozend of them. Waste of money only good for those who try to sell primitive circuit kits.
- i will not make a pretty silkscreen PCB for every experiment i do to go into competition with you and Per Anders who is able to make the prettyest looking PCB. I prototype and donīt build 10thousands. Also i condider a decoupling cap soldered p2p directly to the IC shorter than whats possible with PCB.
- I did read the article, i did read Horowitz Hill, and i did read some hundred datasheets and application notes from TI and AD
now. Sometimes i find one telling the plain opposite of the other... Of corse i do make shortest possible decoupling pass etc. were possible.
- "Loop" on topic at Jewilsonīs suggestion for ISO150 was the connection of transport and reciver/DAC via spdif.
- I will try the ISO150 circuit as it is an experiment suggested with good reasons behind, in opposite to advice like [buy expensive resistor;buy monster cable; buy black gate cap; paint your nose green; put picture into freezer; lay stone on your head while listening; donīt ever do AB tests.]
- except those i recived via private mail until now the best answer on topic so far is that one from Charles Hansen. Thanks. |
|
|
| jwb |
I just have to say, if you are seeing high frequency millivolt signals with a high-impedance scope probe, that noise is almost certainly coming from the scope itself and/or the probe, not from the circuit under test. I have a really nice Tek analog scope and it has self-generated noise of at least 5mV p-p. Trying to measure anything with the high-impedance probe and the vertical set to 20mV/div is fruitless.
If you want the real story on your circuit's behavior, try designing in some test points. There's several nice ideas for such probes in High Speed Digital Design. Or you could build a nice active differential probe. |
|
|
| till |
Oh thank you so much, that is the statement i waited for. I had this impression. No matter what point i try to measure, if i got evrything possible away with filters and decoupling there stays something looking the same werever i measure.
Do i need an active probe connected to my high impedance plugin, does a high impedance differential amp help (i think not avaiable in high bandwith) or does a 50Ohm plugin do it? |
|
|
| jwb |
Try here: http://www.sigcon.com/Pubs/straight/probes.htm
It describes the construction of a 21:1 attenuating passive probe for > 1GHz operation. For best results use the zero-length grounding (solder the braid directly to your circuit). Attach it to your scope's 50Ω input. |
|
|
| Elso Kwak |
| quote: | Originally posted by till
Iīm pretty sure it is part of a loop and it is of corse grounded as it īhas a 3 pin power plug. The noise i measure is not 50Hz but really HF like about 40 to 60Mhz / each day different distributions.
Also my impression is the power mains/null/ground lines are full of miracolous noise, null shows about 2V RMS against ground also they are connected at the house power inlet box.
What should i do? belive there is no problem? ok. If this is the way to build 18Bit DACs. |
Hi Till,
Maybe this article by Bob Pease explains what I meant.
http://www.national.com/rap/Story/0,1562,18,00.html
If your scope chassis is grounded and the amplifier/circuit under test also you have a huge groundloop not only picking hum but also HF. My scope is usuallly not grounded.
:cool: |
|
|
| jwb |
| You should be aware that an ungrounded, or "floated" scope can be dangerous to the scope, the circuit under test, and the human operating them. Do not remove the scope's ground unless you know exactly what you are doing. |
|
|
| Elso Kwak |
| quote: | Originally posted by jwb
You should be aware that an ungrounded, or "floated" scope can be dangerous to the scope, the circuit under test, and the human operating them. Do not remove the scope's ground unless you know exactly what you are doing. |
Yes I know. |
|
|
| till |
| I did read the scopeÂīs manual and it tells me not to float ground, so i did not. |
|
|
| Bricolo |
But what does give a better measurement?
With the scope ungrounded, and the ground from the probe connected to the device's ground
or with the scope grounded, and a "single wire" probe?
The 2nd solution is more safe, but has obviously a bigger loop |
|
|
| till |
| those links are great reading and iÂīm more and more convinced there is not much real noise at my DAC. I had resisual signals in the scope of about 10mV p2p after playing with decoupling and filtering using a high impedance probe and grounded scope with a gnd clip of ~200mm |
|
|
| Jocko Homo |
In order to look at data lines......or anything HF......where a long ground lead on your 'scope is causing problems, get a HF probe tip. Like the one on this probe.
If you need a ferrite bead on your rails.............the type of bead you would use would look like the one shown. You don't need some overgrown bead that would go on a mains cable.
"Do I need a ferrite bead????"
Maybe.
"How do I know?"
It depends........
If you are suppling a handful of chips from one regulator, you probably do to prevent them from talking to each other. And without resonance problems that an inductor might cause.
One chip per regulator.......probably not. |
|
|
| Jocko Homo |
No.....you don't need a silkscreen.......solder mask.........or even a PCB to make a digital board. All you need is a piece of copper clad. Mount the chips upside down, with the ground pin bent over, and soldered directly to the copper. Works for DIP or SMT chips. You can use axial-, radial leaded or SM parts for bypass.......mounted in all manner of crazy ways......with the ground lead going right to the copper, and hopefully, a short lead to the supply pin.
Tag a piece of wire wrap to the signal pins. You can hold them down to the board with some hot melt glue. Small loop.......low EMI.
You don't think that I run out and make a PCB for every design without making one like this do you?? Maybe if I have made a lot that are very close, and I know that it will work, I might go direct.
OK..........you have a decent 'scope and some ratty DAC kits. You don't have to listen to them to learn something. Get one of the HF probe tips, and start measuring.
Now you have benchmark. Doesn't matter if it is a good or bad one.
Make a board of yours using the above technique. Measure it........compare the two.
If yours is better, now you know that you can do better than they did. Worse.......futz with it until it is better. No difference......now you know how many corners that you can cut and get away with.
You can learn something from taking anything apart. Either an example, or inspiration, for your own design if it is well executed. If it stinks, then you know how not to do it. |
|
|
| till |
ok, thank for that usefull advice.
Iwill need to get a 50 Ohm plugin.
I use separate regulator for each IC.
I think i will not measure much noise and all claims about noise and beads may be more like paranoia, as those 10mV i measure with the normal probe may be picup noise, and with a HF probe i hace some attenuation of 10 or 20 :1 and will also not see more. The 50Ohm plugins are sensitive 10mV/1 and so i will not be able to see much with them in this low volatages.
I admit i build DACs and did not really listen to, like the TDA1543 experiments about inverting the dataline. But i donīt think listening is the right tool to detect noise on the DACs power supply.
Last days i intensively listen the balanced AD1865 DAC and AB with the build in TDA1305 of the transport. |
|
|
| till |
ok, i recived a 50Ohm Y plugin for my scope today. Its a 7A24, i installed it in the 7904. I made some of those diy high speed probes from an 50 Ohm BNC connector, 1m RG174, and a 1k carbon composite resistor.
If moving this probe not connected to anything trough the air, in 1m around the DAC http://home.tu-clausthal.de/~tpa/space/mydac.jpg , i see some signal fom spdif on the CRT.
The power supplys measured at the input connectors of tzhe DAC board look like following, all with the 1:20 passive probe, 5mV/div setting for the 7A24, 50nS/div on the 7A80 timebase:
1st: CS8412CP digital and analog supply at 7805 |
|
|
| till |
| 2nd the 74HC04 supply, 7805 |
|
|
| till |
| 3rd the AD1865s digital supply, 7805 |
|
|
|
