High Q cored inductors?

This may seem like an oxymoron but I have basically got to create a very good LPF with a corner around 500kHz. The filters that come out of the design are all around the 50uH mark which means they'll have to include a ferrite core probably. Obviously I can use air core but this is for production and, as well as the space on the PCB being an issue, we don't want to have to wind them in their hundreds.The spec is very tight and simulation suggests a Q of 100 would still not be enough.

At the moment I don't know how I'm going to do this. If you look around the usual places (Coilcraft, Wurth etc...) you see a small number of RF inductors in this range with Qs of around 40. Then they have a load of power inductors which are basically chokes and therefore likely low Q and they don't even bother to give the Q, it's not a priority for the applications they are used in.

But surely it is possible to make a high Q inductor that is not air core and not enormous. If you pick the right core material you can make your own. but are there no manufacturers for this kind of thing? Is there anywhere I need to look that I don't know about?

Thanks
James

To reach high Q on 500kHz you need to use ferrite or better ferrite-iron core. Check e.g. for EPCOS products.

Pool

It is not so easy to make a coil with practical useable Q above 100 and non air core. A high Q value >100 is probably measured under relative ideal conditions. If something in the surrounding disturbs/cooperate with the magnetic field will probably also Q be reduced and inductance increased. As higher the Q value becomes, the more sensitive for non air in surrounding. Also variations in inductance can cause problem if it is a frequency critical design. 500 kHz is not an extreme high frequency, why not make an active filter which not require this type of coil? Compare with old AM radios with internal ferrite antenna, they needed to be regenerative as else would the Q be to low.
If you know specification for a coil that is possible to produce, can custom coils be ordered at many places, here is one: SMD Power Inductors, High Current Inductors, LAN Transformers, Power Chokes, LAN Magnetics, POE Transformers Manufacturer.
Have you tried to search for 125 kHz RFID antenna with ferrite core?
Check with https://www.fair-rite.com or TDK, good sources for ferrite materials.

Thanks for those links, I'll check them out now.

Afraid I can't use an active filter, it has to be passive. I have a tight spec on noise figure and intermodulation which won't be achievable if I go for active filter. It's a very sharp response that I need and this would likely involve a lot of amplifiers!

Thanks
James

See also Ferroxcube - ferrite cores, bobbins & accessories. They have several MnZn ferrite materials that can be useable at these frequencies. Be prepared to have good isolation to other magnetic material in your circuit to avoid reduced coil performance, I doubt that a such critical design is possible in production. Maybe two coils perpendicular to each other can improve results.

If I were you, I would either do an active filter without inductors, or use a fixed toroid inductor with tunable caps for tuning.

With ferrite core you can easily get Q of several hundreds. I remember designed and built a bandpass filter at 150 kHz with coil Q of around 600 to 800.

Look on the TDK book you will find the ferrite core material for you purpose. Good luck!

The "bible" of ferrite related design, Snelling, Soft Ferrites indicates feasible Q numbers up to 500 for 500 kHz with pot cores. Core material, air gap and winding have to be carefully selected, it's unlikely, that you'll find suitable catalog parts. Filters in telecommunication applications have been custom designed this way for at least 50 years.

I agree, that active filters are a more convenient solution, if they fit the requirements. Otherwise, get the ferrite data books, do the calculations and preferably make a test inductor. There are also inductive component manufacturers, who know about the stuff and are able to design ferrite inductors from the scratch and make them according to your specifications.

For example in a 5 pole 500kHz LPF (whatever topology) if you use inductors with Q=50 or inductors with Q=500 there is not significant difference in rejection performance.
Do a simple simulation to prove this.

vfone said:

For example in a 5 pole 500kHz LPF (whatever topology) if you use inductors with Q=50 or inductors with Q=500 there is not significant difference in rejection performance.
Do a simple simulation to prove this.

Click to expand...

That may be true, but a critical part of the design spec is the sharpness of the knee of the LPF, and the insertion loss in band, which are both reduced by a poor Q, right? A simulation suggested a Q of 100 would not be good enough so I'm looking at something better. Thanks.

---------- Post added at 23:01 ---------- Previous post was at 22:58 ----------

FvM said:

The "bible" of ferrite related design, Snelling, Soft Ferrites indicates feasible Q numbers up to 500 for 500 kHz with pot cores. Core material, air gap and winding have to be carefully selected, it's unlikely, that you'll find suitable catalog parts. Filters in telecommunication applications have been custom designed this way for at least 50 years.

I agree, that active filters are a more convenient solution, if they fit the requirements. Otherwise, get the ferrite data books, do the calculations and preferably make a test inductor. There are also inductive component manufacturers, who know about the stuff and are able to design ferrite inductors from the scratch and make them according to your specifications.

Click to expand...

Thanks, that's useful. I agree, I'm getting the feeling I won't find catalog parts good enough, which is no surprise. But, I have found a local business that specialises in winding inductors for you, if you specify the core, turn etc... So, if I come up with a design I may get them to wind them for us in production.

---------- Post added at 23:03 ---------- Previous post was at 23:01 ----------

Phytech said:

With ferrite core you can easily get Q of several hundreds. I remember designed and built a bandpass filter at 150 kHz with coil Q of around 600 to 800.

Look on the TDK book you will find the ferrite core material for you purpose. Good luck!

Click to expand...

That's very impressive! If I can get that I think I'll be happy. I might have to get that TDK book out....

---------- Post added at 23:06 ---------- Previous post was at 23:03 ----------

E Kafeman said:

See also Ferroxcube - ferrite cores, bobbins & accessories. They have several MnZn ferrite materials that can be useable at these frequencies. Be prepared to have good isolation to other magnetic material in your circuit to avoid reduced coil performance, I doubt that a such critical design is possible in production. Maybe two coils perpendicular to each other can improve results.

Click to expand...

Yeah, I've used Ferroxcube, they're pretty good. I actually have one of their RM10 cores in another part of the circuit. I've had to insert a gap to improve linearity, although I can't seem to get hold of their gapped versions. Is MnZn what I'm looking for for this kind of application?

---------- Post added at 23:07 ---------- Previous post was at 23:06 ----------

Thanks for all the replies! I've been beavering away on another part of this design for the past couple of weeks and after not having many replies at first figured that was it. But everything has been really helpful so thanks. I'll be back on the filter design next week...

MnZn is available in several variants and by several manufactures.
You have to calculate which variant that suits your application best and if it is good enough.
Some ferrite data can be found at magnet-tech.com.
More about ferrite material: www.fair-rite.com/newfair/materials.htm
Observe that a ferrite core is a bit temperature depending. Can cause problem if required inductance have very tight value.

bit of an update here as I have had time to put some ferrite configurations together and do some simulations.

Basically, I'm a long way off at the moment. Simulation suggests I am going to need a 20th order low pass filter with inductor Qs of around 300 to be able to meet my spec! I have tried other configurations like Elliptic (which may reduce my inductor count by 1if I'm lucky) but it seems no matter what you try you end up pretty much where you started.

I have tested a few cores around the RM10 size and so far the highest Q I have measured (by series resonating it with a high Q cap to ground to created a notch) is 100. An air core measurement on the same bobbin yielded very low Q, around 11, most likely due to the proximity effect and having around 5 layers of windings to reach by target inductance.

Basically it seems the aim of the game is to get as close to an air core as possible (i.e. get your Al down by using gaps), whilst maintaining minimal interaction between windings. The only way to do that is to have a large diameter winding. It's starting to look pretty huge!

As I am interesting in up to 500kHz I have tried 3C90, N87 and N97 cores. They seem to be good choices to me.

So, at the moment not having much luck and I'm a bit out of ideas after a day of testing.If anyone had any input I'd be grateful. Even if it is to say "yes you're doing everything right and everything you've found makes sense"!

Thanks
James

If you have several layers is it important how you are placing these layer. Do not wind a complete layer and then start on next.
Simplest is to cut small plastic isolators. Place them along the ferrite rod so you get several sub-sections with and wind one section at the time.
These isolators can be removed afterwards.
If you want to reach a Q above 100, use silver-plated wire. Higher Q, 200 or more, use litz wire.
Wind only a single layer, with one wire diameter air between each turn, equally spread the winding along whole ferrite core. If a coil-carrier is used, make it as thin as possible.

As I am interesting in up to 500 kHz I have tried 3C90, N87 and N97 cores.

Click to expand...

All these are "power" ferrites. For high Q, temperature stable filters other materials are usually preferred. Air gap is a must, in addition litz wire will be most likely needed to achieve highest Q values at 500 kHz. See some comparative figures from an Epcos data sheet:

Micrometals advertises that they have RF inductor cores that yield Qs of around 300 at lower frequencies. I just got a few in I'm hoping to use at 10MHz. Check this page out: **broken link removed** and also their app notes on RF cores.

They're all powdered iron toroids, which are a pain to wind though...

FvM said:

All these are "power" ferrites. For high Q, temperature stable filters other materials are usually preferred. Air gap is a must, in addition litz wire will be most likely needed to achieve highest Q values at 500 kHz. See some comparative figures from an Epcos data sheet:

Click to expand...

Thanks, that's really useful. It does look as though M33 would be my best choice. It's possible we discounted it due to it claiming it is best above 100kHz and we are trying to go down to 10kHz. But, it is probably safe to say we are more concerned with Q at 500kHz as this is where sharp filter response is required. I will just have to see how good it is at 10kHz when I try it. I also need to find some Litz wire it seems.

One thing though, I can't seem t view the pdf you have attached. I've just scoured the EPCOS website and can't find it either. Could you post a link please?

Thanks
James

---------- Post added at 10:15 ---------- Previous post was at 10:10 ----------

mtwieg said:

Micrometals advertises that they have RF inductor cores that yield Qs of around 300 at lower frequencies. I just got a few in I'm hoping to use at 10MHz. Check this page out: **broken link removed** and also their app notes on RF cores.

They're all powdered iron toroids, which are a pain to wind though...

Click to expand...

Thanks also for this. Micrometals look like they do a few materials that could be suitable. I'll try and get some samples and put them to the test.

The application example apparently has been stripped off from the recent Epcos RM6 datasheet. I copied the page from a 1996 databook, it's about 12 MB, I fear too large to post it Edaboard.

No problem, thanks for letting me know