Replacing Ferrite beads by Inductors?

[gildasd]

Ferrite beads of roughly the type I used in a prototype are a bit hard to come by cheaply/fast...
http://befr.rs-online.com/web/p/ferrite-beads/0239589/
It's not so much component price as the postage cost and/or delay (having to use a friend for the TVA number)...

So I was thinking of replacing them by cheap and easy to get axial inductors (chokes):
https://www.gotron.be/en/spoel-axiaal-47-h-205ma-3x10mm.html

This ferrite/choke has a very basic role to play in my circuit;
Take the edge off the noise ICL7660 negative power source, so caps can filter the result, so as not to mess up OP amps amplification further down the line.
In no way is this a high end audio thing (or similar), it's just a sensor amplifier.

[danadak]

Examine f vs z curves, let those be a guide to your choice, but axial
chokes in a PWM application should generally be OK. But cannot
emphasize enough, look at the datasheets.

Regards, Dana.

[gildasd]

Examine f vs z curves, let those be a guide to your choice, but axial
chokes in a PWM application should generally be OK. But cannot
emphasize enough, look at the datasheets.

Regards, Dana.

My problem is that the original ferrites where lifted from old motherboards...
The only info I have is their size; 6mm long by 3.4mm dia...

So how should I interpret the data sheets?
Lets say I have 60Hz noise, 100mV Peak to Peak, how should I choose the inductor?
(I think I know how, but I might be totally wrong)
Thanks!

[gildasd]

Beads can operate up to many GHz, while most inductors are only good up to from a couple of MHz to a couple of GHz.
The ones (inductors) that can allow some practical current for power decoupling are usually not good for over 100MHz.
Therefore, for digital noise, an inductor looks like a capacitor, while a ferrite bead looks like a resistor.
Beads can dissipate RF noise energy, while inductors, even below its self resonant frequency, circulate them.

I'm only filtering power source noise, in the Mhz range, and resulting sample range is in the 100/200hz range.
But from your answer, I get the impression that the inductor route is not going to work..

[wraper]

What I doubt is if that ferrite bead is the right part for your circuit and gives any significant effect at all, as frequency is too low. Normally you should put something like 1uH inductor, not ferrite bead for some additional filtering after switching supply.

[macboy]

If the op-amp is only drawing a small current, then you might find that a series resistor is very effective as well. Just some tens of ohms of series resistance can be a much higher impedance than the shunt impedance of the capacitor, making for an effective filter.

[gildasd]

What I doubt is if that ferrite bead is the right part for your circuit and gives any significant effect at all, as frequency is too low. Normally you should put something like 1uH inductor, not ferrite bead for some additional filtering after switching supply.

The Hzs are low, but the "blip" from the ICL7660 has a vertical leading edge and is very energetic compared to the signal/other noise.
It was pushing through the OP07 and polluting the output significantly.
I tried for more than a week various filtering techniques and all failed or were too power hungry on my limited negative supply.
To my amazement, a ferrite bead "unsteepened" the 7660 blip sufficiently to be treated, to my noise floor, by a couple of electrolytic caps.
Beffore this, I had considered beads as a joke... So a big bite of humble pie for me!

There is a thread "an evening with the 7660" written by someone far more competent, that gave me the basis for this solution.

[Cerebus]

What I doubt is if that ferrite bead is the right part for your circuit and gives any significant effect at all, as frequency is too low. Normally you should put something like 1uH inductor, not ferrite bead for some additional filtering after switching supply.

The Hzs are low, but the "blip" from the ICL7660 has a vertical leading edge and is very energetic compared to the signal/other noise.
It was pushing through the OP07 and polluting the output significantly.
I tried for more than a week various filtering techniques and all failed or were too power hungry on my limited negative supply.
To my amazement, a ferrite bead "unsteepened" the 7660 blip sufficiently to be treated, to my noise floor, by a couple of electrolytic caps.
Beffore this, I had considered beads as a joke... So a big bite of humble pie for me!

There is a thread "an evening with the 7660" written by someone far more competent, that gave me the basis for this solution.

Choosing ferrite beads is a bit of a black art.

You can spend as long as you like looking at datasheets but nothing is a substitute for trying some in your final circuit and seeing what it looks like on the 'scope. Those kits of various values of beads from the manufacturers are a great thing to have but are relatively expensive if you have to buy them but can often be had for free from the manufacturer's reps if you have contact with one. One approach might be to contact a manufacturer and say "I've got THIS problem, what do you recommend and can I please have some samples to try.". With luck, "some samples" might translate into a full kit. Mullard, as was, used to be very generous with kits.

Can anybody suggest manufacturers who tend to be on the more generous side?

[gildasd]

If the op-amp is only drawing a small current, then you might find that a series resistor is very effective as well. Just some tens of ohms of series resistance can be a much higher impedance than the shunt impedance of the capacitor, making for an effective filter.

Tried that, makes it worse by drawing more on the 7660...
But If I only had one Op Amp, this would be the way.

[gildasd]

What I doubt is if that ferrite bead is the right part for your circuit and gives any significant effect at all, as frequency is too low. Normally you should put something like 1uH inductor, not ferrite bead for some additional filtering after switching supply.

The Hzs are low, but the "blip" from the ICL7660 has a vertical leading edge and is very energetic compared to the signal/other noise.
It was pushing through the OP07 and polluting the output significantly.
I tried for more than a week various filtering techniques and all failed or were too power hungry on my limited negative supply.
To my amazement, a ferrite bead "unsteepened" the 7660 blip sufficiently to be treated, to my noise floor, by a couple of electrolytic caps.
Beffore this, I had considered beads as a joke... So a big bite of humble pie for me!

There is a thread "an evening with the 7660" written by someone far more competent, that gave me the basis for this solution.

Choosing ferrite beads is a bit of a black art.

You can spend as long as you like looking at datasheets but nothing is a substitute for trying some in your final circuit and seeing what it looks like on the 'scope. Those kits of various values of beads from the manufacturers are a great thing to have but are relatively expensive if you have to buy them but can often be had for free from the manufacturer's reps if you have contact with one. One approach might be to contact a manufacturer and say "I've got THIS problem, what do you recommend and can I please have some samples to try.". With luck, "some samples" might translate into a full kit. Mullard, as was, used to be very generous with kits.

Can anybody suggest manufacturers who tend to be on the more generous side?

Image 5mV per division:
Top: raw signal from the 7660.
Bottom: After Ferrite + Electrolytic.

[munkeyman1985]

You can try an inductors, they are generally cheap. You can think of that switching signal as the sum of many different signals. When you have a spike on your scope, think of it as a sum of all possible signals including high frequency. Ferrite is ideal for this application, as it will absorb frequency starting the in the MHz and up. If you are not mass producing this project, you might try ordering a bead you can wrap yourself and play with the number of turns. Using Ferrite to block high frequencies from a power supply is probably the most common use for them. They are not a good choice for digital, the square wave of you PWM signal is a sum of odd harmonics, it has high frequency components that will be filtered out by the ferrite.

Being that you are using ferrite with good results, it is probably best to use it.  If you need large transients from your power supply ferrites are a bad idea. 

[T3sl4co1l]

IIRC, last time I played with an ICL7660, the shoot-through was quite substantial both on the input and output sides.  And it's painfully slow, so the shoot-through and ripple puts considerable energy below 10MHz, which is hard to filter: ferrite beads are ineffective.

On the upside, most op-amps have acceptable PSRR in that range, so it shouldn't be too big a problem.  If it is, your first solutions should be:
1. Ditch the 7660. There are faster charge pumps, and much faster inductive switchers -- easier to filter, and some of them will be quieter to begin with (as strange as that sounds, but yes, the 7660 IS just that awful).
2. Cut out the bad-PSRR parts.  (RF sensitivity is rarely documented, unfortunately.)  Choose better op-amps, isolate the sensitive stages, consider local filtering, etc.
3. Consider adding LDOs, or even better, C multipliers.

The tricky range is kind of 10kHz to 1MHz.  Passive filtering is bulky, switchers are noisy, and LDOs and op-amps perform poorly.  Pushing the noise above this range (using a switcher in the >300kHz range, say) makes it easier to filter; pushing it lower (not really an option for inductive switchers, but I suppose you could put electrolytics in a charge pump running this slow) makes it less of an issue, but only if the harmonics don't screw you up in the process (which is where the '7660 is at, unfortunately).

Tim

[gildasd]

Tim, can you recommend a pin compatible option to the 7660?
The boards have allready been spun and I've populated them...

[gildasd]

Tim, can you recommend a pin compatible option to the 7660?
The boards have allready been spun and I've populated them...

LTC1046

Wow, Vo of 50mA compared to 10mA...
But the variable voltage would need a totally different circuit... Using 5 LTC1046 in parallel each driving a channel...
Right now, I drive the 7660 at -9V, for a reference voltage of -2.5V and a max negative output of -3.3V, so I can live with the drop.
-5V is a bit close, but could be worked around.
Thanks, great to know this exists!

[Kalvin]

TC7662B seems to be pin-compatible with the 7660.
http://www.microchip.com/wwwproducts/en/TC7662B

[gildasd]

TC7662B seems to be pin-compatible with the 7660.
http://www.microchip.com/wwwproducts/en/TC7662B

Looking at the datasheet right now... TI has some sweet stuff too, but at a price!

[T3sl4co1l]

Not obvious if TC7662 has reduced noise, but sounds worth a shot.

Tim

[gildasd]

Well that was an unmitigated disaster...
With the 1uH inductors, it works as with ferrites, but uses more amps at  40mA at idle.
But, once a TL084 is seated (anywhere), something happens between it and the inductors... Power goes up to 0.7A with the multimeter measuring, a least 1.2A without (smoking 10 Ohm voltage limiting resistor)...
Bridge all the inductors (image 2): all goes back to normal (but without noise reduction) and 14mA at idle...

I really don't like ferrites, but in this application, they are far simpler and more effective than inductor... One can imagine that with time, the right inductor compromise can be found, but I need to move on!

Sweaty palms moments here...

[gildasd]

I think you are right.
The circuit dis show signs of oscillatting. I'm 90% sure that the power rail inductors are to blame.

[gildasd]

And I had 2 bad TL084CN 
With no load at 12V, all the Ti parts I tested are about 5mW, ST a bit better at 3mW...
The 1st bad one was at 112mW, the other at 900mW - or as much as the smoking 100Ohm current limiting resistor would let by.

[gildasd]

And now I've got weird noise on the scope...   

This is with the probe at x1. There is no other relevant noise on the board, even the 7660 is OK filtered

The problem is created at the LM358 voltage buffer...

The peak to peak is about 140mV and the period at 0.8µS, so about 1.25MHz

Probing around with ceramic caps to ground to try to solve this... Only managed to make it worse so far!

[gildasd]

Fixed.
Needed a lot of prodding, and some caps but solved.
It would have been better with all ferrites, but with just 2 it's ok...
Overall, the output noise is similar to the protoboard, just with lower power usage.