Ferrite Bead, Is This A Good Way To Go For A Clean

[xavier]

I need the power input to be super clean. I've read up and found suggestions for a circuit that consist of a 5v supply to a .1UF and a .01UF then a Ferrite Bead in series with the 5v supply then another .1uf ceramic as close to the power pin of the device.

Is this ferrite bead ok for this purpose? HZ0805E601R-10

Datasheet link : http://www.lairdtech.com/WorkArea/linkit.a...=id&ItemID=4884

Any opinions will be greatly appreciated.

[Sch3mat1c]

What noise are you combating, and what noise level do you have to achieve? How efficient does it have to be?

If you need the absolute purest power possible, use a battery. There is no substitute when true nanovolt levels are required.

Anything even *touching* a switching supply will carry noise. And I mean literally, physically touching -- common mode ground currents will gladly flow along your project, even if you aren't connected to the same supply voltage, only ground. This can make coupling signals in and out of a project difficult.

If milivolts are acceptable, a combination of bypass caps and filter inductors will suffice. Pay attention to capacitor ESR and ESL (in order from best to worst: ceramic chip, aluminum polymer, poly film, tantalum, aluminum electrolytic). If a lot of ripple voltage is present (<100kHz), you may need an excessive amount of L and C to achieve enough attenuation; an LDO may be a cheaper solution in this case. Middle RF is fairly easy to filter (100k ~ 20MHz), with a combination of small chokes and ceramic caps usually. Higher RF is more difficult because it doesn't like staying inside wires; the usual approach is lots of ferrite beads and common mode chokes to isolate it.

Be careful that your filter doesn't happen to resonate at an unlucky frequency. This can be difficult to ensure over 20MHz because parasitic resonances are easy to create and hard to track.

Tim

[xavier]

Thank you for the reply Tim,

You always seems to chime in on my questions with a nice explanation I can easily digest. It is truly appreciated. One question, am I ok with that ferrite bead? I'm just concerned it might have some negative effect. The DC resistance is very low so I'm assuming despite the benefits it actually provides, the circuit should be functional?

[CWB]

as tim mentioned : what is going on ?
what is the pre-filter power source and what is it powering up ?

[xavier]

It's a DS2450S 1-wire ADC. A paper called the 1-wire design guide says that the power source for this part needs to be absolutely clean and suggests a ferrite bead in series with the power source. The power source is 12v from a fairly long cable to the board and then a 5v regulator circuit. I'm not too concerned if there will be a benefit more concerned if that is a good bead choice for this. The circuit seems to work fine, I'm just wondering if using this bead in this configuration could have negative effects?

[Sch3mat1c]

http://datasheets.maxim-ic.com/en/ds/DS2450.pdf
Well what the heck, they only data this thing specifies about the actual conversion itself is two lines: "input noise: TBD" and "conversion error 1/2 LSB" (at 8 bits). It goes up to 16 bits, but why, are any of those bits even well-defined?

I don't see how they could possibly think parasite power is stable enough to run more than 8 bits anyway. Probably that's why it's not even rated.

As a digital comms block it's an interesting chip (Maxim has lots of one-wire devices it seems), but an ADC is the last thing it is. One glance at the datasheet tells you where the design engineer's priorities were -- laying out one-wire state machines, NOT analog to digital conversion.

I say pick a better one, if you still can. Two reasons: one, most obvious: there's only three lines of specs about what it's actually intended to do, the rest is communication. Two, Maxim in general: they make a lot of interesting chips, and they sample generously, but good luck buying them in any quantity. Even for a hobbyist, this can be a problem, because maybe your design isn't going to sell to consumers or anything, but you show it to someone else, and they build it, and someone else, and so on, and pretty soon no one has the chip anymore and your project is useless as a source of development.

Offhand, for generic ADC purposes, I've had good experience with Microchip MCP320x series serial ADCs:
http://ww1.microchip.com/downloads/en/DeviceDoc/21290e.pdf
12 bit, good to an LSB or so, available in single up to octal muxed inputs, wide supply range. If you need 16 bit, there are lots of audio range ADCs available; if you need more bandwidth, TI/National and AD are the big brands in higher speed ADCs. Expect to spend on the bigger ones (65MSps 12 bit single channel = about $15 in single quantities; high bandwidth LVCMOS or LVDS interface basically means using an FPGA to receive the data, but you'd probably use one anyway for the DSP blocks you'll need to filter and process it).

Tim