# 22nH inductor - how to create

Status
Not open for further replies.

#### mik3ca

##### Member
After (nearly) overheating my computer from running LTSpice simulations, I learned that a 0.1uH inductor wouldn't cut it for my needs, but a 22nH inductor will work since I am dealing with around 433Mhz.

I have looked on ebay for some and since I don't do smd, the others I found are air core and tuneable. Just by looking at air core I feel I could make one myself.

Since 22nH is such a small number, is there an equation I can use to calculate inductance if all I used to make the inductor is a flat piece of 22awg shielded (i think steel here since the wire color is silver) wire with maybe 5mm of shielding stripped off each end?

Once I know the perfect length, I will then make a U shape out of it and hook it onto the circuit board through-hole style.

I understand I can make one from a circuit board itself but I want to go against that option because if I need to change the inductor value later, its much cheaper for me to replace the wire with one of different size than to replace an entire circuit board.

#### JimB

##### Super Moderator
flat piece of 22awg shielded
You don't make inductors with "shielded" wire.

(i think steel here since the wire color is silver)
For high frequencies, silver plated wire is often used, because silver has lower resistance than copper, and at high frequencies the current flow mostly on the surface of the conductor (look up "skin effect").
So a plating of silver lowers the RF resistance and increases the Q of the inductor.

JimB

#### JimB

##### Super Moderator
Once I know the perfect length
Have a look here, they have formulae and an online calculator:
http://www.consultrsr.net/resources/eis/induct5.htm
About 30mm of 1mm dia wire will get you in the ball park.

I will then make a U shape out of it and hook it onto the circuit board through-hole style.
Very nice, but bending it into a U shape will change the inductance, the traces on the circuit board will have inductance (and capacitance).

JimB

#### MrAl

##### Well-Known Member
After (nearly) overheating my computer from running LTSpice simulations, I learned that a 0.1uH inductor wouldn't cut it for my needs, but a 22nH inductor will work since I am dealing with around 433Mhz.

I have looked on ebay for some and since I don't do smd, the others I found are air core and tuneable. Just by looking at air core I feel I could make one myself.

Since 22nH is such a small number, is there an equation I can use to calculate inductance if all I used to make the inductor is a flat piece of 22awg shielded (i think steel here since the wire color is silver) wire with maybe 5mm of shielding stripped off each end?

Once I know the perfect length, I will then make a U shape out of it and hook it onto the circuit board through-hole style.

I understand I can make one from a circuit board itself but I want to go against that option because if I need to change the inductor value later, its much cheaper for me to replace the wire with one of different size than to replace an entire circuit board.

Hello there,

For a square shaped loop missing one side (the feed side) the total inductance is the self inductance minus the mutual inductances and this comes out to:
I=4*L*asinh(L/p)-4*sqrt(L^2+p^2)-2*asinh(1)*L+2^(3/2)*L-2*L+4*p+1/2

where p is the radius of the wire, L is the length of one side of the square, and I is the inductance in nanoHenries.
All length dimensions in centimeters.

Since the radius of a 22 AWG wire is 0.03226 if we substitute that in for p and solve for L when I=22 we get something close to 1.6cm
This is the open circuit inductance so the closed circuit inductance will be somewhat less because of the unknown extra mutual inductances of the rest of the circuit but that gives us a starting point.

The shape is 1.6cm up from the board, across 1.6cm, and down 1.6cm, forming a square with one open side (the bottom side).
The construction is fed from both ends, with the joining wires or traces meeting the square loop at right angles.

That formula is a direct result of the application of the Biot Savart Law so in theory it is exact, however as mentioned circuit conditions change it a little, as is usual for small inductances. This has to be dealt with in an application specific manner based on the physical layout.
For information about Biot Savart as well as a solution see my article on that topic right here on ETO.

#### crutschow

##### Well-Known Member
i think steel here since the wire color is silver
That's likely just tinned copper wire, to make it easier to solder.

Steel wire would likely have to much resistance to make a good coil.

#### jpanhalt

##### Well-Known Member
If the source is 22 awg shielded wire, it might be steel. Amazing (to me) how many things have steel (or at least leads that are magnetic) today.

John

#### mik3ca

##### Member
Very nice, but bending it into a U shape will change the inductance, the traces on the circuit board will have inductance (and capacitance).
I would expect PCB traces to have inductance which is why when I build it, I plan to make the holes for the wires that connect together as close together as possible and tracks as thick as possible with large clearance. How else would making a wire a U shape affect inductance?

So I guess I should just use bare wire with no coating on it (not even the plastic) as an inductor?

#### MrAl

##### Well-Known Member
I would expect PCB traces to have inductance which is why when I build it, I plan to make the holes for the wires that connect together as close together as possible and tracks as thick as possible with large clearance. How else would making a wire a U shape affect inductance?

So I guess I should just use bare wire with no coating on it (not even the plastic) as an inductor?

Hi,

Yes i forgot to mention in post #5 that you should not use steel wire just copper or copper plated.
Also, that thicker wire is better because of skin effect.

So for #22 AWG copper or plated copper a square shape with 1.6cm per side, and
for #12AWG 2.2cm per side is a good starting point.

As you can see as the wire diameter goes up the length has to increase.

The shape again is straight up 2.2cm, right angle and straight over 2.2cm, then right angle and straight down.
It looks like an upside down square "U" shape.

Here's a quick drawing...the PC board is colored green, the traces are orange, the wire is brown...

#### Attachments

• 6.1 KB Views: 54

#### Colin

##### Active Member
At 433MHz, the wire becomes a radiator and almost all the energy delivered to the circuit will be radiated as EM. You have to "tap-off" the wire, part-way along its length to get a feedback signal to keep the oscillator working. That's why you make the wire out of a PCB track.
It's much more complex than you think and you need 10 years understanding in this field to know what to do.

#### mik3ca

##### Member
At 433MHz, the wire becomes a radiator and almost all the energy delivered to the circuit will be radiated as EM. You have to "tap-off" the wire, part-way along its length to get a feedback signal to keep the oscillator working. That's why you make the wire out of a PCB track.
It's much more complex than you think and you need 10 years understanding in this field to know what to do.
Ok. would increasing clearance on the PCB track help at least to some degree? because if I'm not mistaken, having two tracks close together, especially ones that carry high frequency is bad?

#### Mosaic

##### Well-Known Member
If the wire is kept to 1/10th the wavelength of the signal it is not much of a radiator. Microstrip or stripline methods of PCB manufacture can help contain the VHF /UHF signals. For a long run of 50 Ω RF signal, you may have to use an SMA connector and a piece of semi rigid coax much like spectrum analyzers and fast scopes do.
I have made a 1Ghz RF ALC power 'head' for an instument grade power level meter that achieves < 0.1dB of flatness. This is done with home etching etc. Just keep circutiry small to avoid excess parasitics. I actual use stub copper traces to tune the flatness with the parasitics.
Simulators like AWRDE etc help with layout and analysis.

The length of the trace and the permittivity of the substrate govern the inductance and capacitance (if double sided).
http://www.sigcon.com/Pubs/news/3_8.htm

A meander inductor trace can be designed quickly using the coil32 s'ware or phone app. Thinner traces = more inductance per unit length. Conversely fatter traces or parallel traces deliver less inductance per unit length. For small signals trace RMS heating should not be an issue.

EDIT:
As you're up in the low UHF range, add lots of vias in the ground traces bordering the signal trace to manage the any microstrip grounding with a doublesided PCB. FR4 @ 0.8mm thickness should work fine.

#### mik3ca

##### Member
I've attempted double sided PCB's once or twice for other small projects without success. I will continue to use single-sided PCB's as they are easier to manufacture and cheaper to order if a 3rd party creates them.

Someone here said about 1.6cm per side for 22awg wire bent in a rectangular U shape is a 22nH inductor. If I measured the whole wire, that would give me 4.8cm which is almost 2 inches which if I'm not mistaken is 1/3 of the wavelength for 433Mhz.

How much percentage of a wavelength should a wire be before it qualifies as a radiator such that it causes EMI or any of that mess to the rest of the circuit? And I'd like to keep my traces close together. If possible, I want to follow the 10mil spacing (clearance) without causing EMI, but if I can get significant improvement over increased spacing then suggest me a new number?

#### MrAl

##### Well-Known Member
I've attempted double sided PCB's once or twice for other small projects without success. I will continue to use single-sided PCB's as they are easier to manufacture and cheaper to order if a 3rd party creates them.

Someone here said about 1.6cm per side for 22awg wire bent in a rectangular U shape is a 22nH inductor. If I measured the whole wire, that would give me 4.8cm which is almost 2 inches which if I'm not mistaken is 1/3 of the wavelength for 433Mhz.

How much percentage of a wavelength should a wire be before it qualifies as a radiator such that it causes EMI or any of that mess to the rest of the circuit? And I'd like to keep my traces close together. If possible, I want to follow the 10mil spacing (clearance) without causing EMI, but if I can get significant improvement over increased spacing then suggest me a new number?
Hi,

That was just one possibility, but let us look at the wavelength.

L=300/433=0.69 meters
0.69 meters=69cm
69/1.6=43
43/3=14

So the whole thing is 1/14 of the wavelength of 433MHz, and each side is just 1/43 of the wavelength of 433MHz.

##### Well-Known Member
After (nearly) overheating my computer from running LTSpice simulations, I learned that a 0.1uH inductor wouldn't cut it for my needs, but a 22nH inductor will work since I am dealing with around 433Mhz.
Inductors aside for a moment I have a dew questions. How does running LtSpice simulations possibly overheat a computer? How will the addition of a single inductor remedy such a situation? Where would one place this inductor? Years ago I tried a little science experiment for a computer forum I was active in involving heat and heat removal. This was maybe ten years ago. The computer was a home brew workstation using dual Intel Xeon processors each processor was dual core. Each core of each processor was run up to 100% processor use.

Here is what I did. The GPU was a 3 D Labs commercial GPU which generated plenty of heat.

The dual Xeon processors generation more heat:

Using thermocouples I measured 4 points with the computer at idle and with processing at 100%:

TC #1 Case Inlet Temperature:

TC #2 Case Exhaust (Outlet) temperature:

TC #3 PSU Inlet Temperature (PSU draw inside the case):

TC #4 PSU Exhaust Temperature:

The system ran at idle for about 30 min and the following screenshot was taken:

CPU Usage was then run up to 100% on each processor core (4 cores total) and was run this way for thirty minutes:

Here are the respective case temperatures at idle and at 100%:
Idle

100%

The ambient room air temperature was about 70 degrees F. I would think unless someone running a computer in an oven the system should not overheat or even come close to overheating running LtSpice and this was a 10 year old experiment. So again my question is how exactly will adding a choke (a singe inductor as described) prevent overheating and how complex are the math functions to generate more heat than can be removed? Maybe I am missing something? Exactly where does this single component get added?

Ron

#### kubeek

##### Well-Known Member
...So again my question is how exactly will adding a choke (a singe inductor as described) prevent overheating and how complex are the math functions to generate more heat than can be removed? Maybe I am missing something? Exactly where does this single component get added?
Supposedly somewhere within the simulation he performed.

##### Well-Known Member
Supposedly somewhere within the simulation he performed.
After (nearly) overheating my computer from running LTSpice simulations, I learned that a 0.1uH inductor wouldn't cut it for my needs, but a 22nH inductor will work since I am dealing with around 433Mhz.
Was that running the LtSpice simulations was taxing the computer enough to overheat which I didn't get. Running the LtSpice simulations should not be all that taxing on the processor which I would guess generates the most heat. That said I do not see how adding a choke or inductor, a single inductor, will prevent overheating? Thus, my question, where would this inductor be placed? How would it reduce heat?

Ron

#### kubeek

##### Well-Known Member
I understand it that he was running a simulation with 100nH, what ever circuit that was, and then with 22nH which he found to be optimal value.
Some simulations like a startup from 0 of a complicated switch mode PSU can take a lot of time to calculate and use the whole available processor, which say on a older notebook with dry thermal paste and dirty heatsink might tax the cooling a lot. I don´t really see any connection between the inductor and the overheating, especially with the 433MHz part.

#### mik3ca

##### Member
...So the whole thing is 1/14 of the wavelength of 433MHz, and each side is just 1/43 of the wavelength of 433MHz...
Mosaic told me...
If the wire is kept to 1/10th the wavelength of the signal it is not much of a radiator.
Thanks for a calculation MrAI but what I'm asking is how long can a wire be before it qualifies as a radiator?

How does running LtSpice simulations possibly overheat a computer?
My computer is old. It only has one CPU.

Status
Not open for further replies.