AC Filter toroids

[rcbuck]

Does anyone know the type of material used for the AC input filter chokes on a typical PC power supply. I see different colors, yellow, green, black, etc. However that doesn't really help.

I pulled a couple out of a PC supply and measured them. One was 6.6 mHy and the other one was 7.2 mHy. There were 30 turns on the cores. I calculated the AL values to be 7222. Looking at the Amidon and Bytemark site I don't find any cores that come close to that AL value at a reasonable cost. I know those inductors are typically specified between 5 and 10 mHy.

I'm not sure my LC meter is reading accurately because it was designed for RF work. The oscillator is running at about 50 KHz.

Any suggestions for which cores to purchase for the filter section?

[Sch3mat1c]

You mean the common mode filter? Those are generally ferrite, sometimes metglas (usually in a two-piece plastic case). They do often have extremely high inductivity, as you discovered.

Colors may or may not mean anything. Lots of things come in black; uncoated black usually means ferrite, but I have one salvaged E-E set which is uncoated black yet clearly not high permeability (ungapped specs: 0.15uH/T^2, 200At saturation -- good for a flyback supply!). Coated black may be ferrite, Cool-Mu or others. Yellow (one side white) is almost always the cheap, lossy powdered iron used for DC chokes. Green may be ferrite. Blue, cyan and gray may be varying grades of powdered iron, MPP and etc.

They also make "B" shaped cores, like an E-I core but one solid piece, which have similarly impressive values. These also use a two-piece plastic bobbin, which is wound in place. I think I measured one at 8uH/T^2. The material is certainly high-permeability ferrite, with unusually high density = very little air gap.

As Amidon goes, your best bet is type 77 ferrite. An average 1" toroid gets something like 2-3uH/T^2, which should be good enough. If not, find something in a catalog -- Mouser stocks common mode chokes.

Tim

[rcbuck]

Tim,

Yes, I'm talking about the common mode chokes. The ones that came out of the PC supply are .7in OD and .4in ID. They have 2 windings of 30 turns each on each core separated by a fiber spacer in the middle. Each winding is in series with one side of the line and they are bypassed with .47uf/250 VAC caps. There are 2 of them so they form a double inline filter.

What are the recommended values for these chokes. I looked on the internet and values seem to be all over the place with 5-10 mHy being the most common.

Ray

[Sch3mat1c]

One possible reference is line filter modules. In those, I see values from 10mH for 2-3A capacity, down to 2mH for 10-15A capacity. The bypass capacitors range from 0.1uF to 0.47uF, although on-board values may differ (I have one power supply, a mere 60W flyback, which has a monster 1uF at its input).

The only real method towards design is 1. guesstimating the actual amount of CM noise by taking parasitic capacitances into account (between transistors and heatsinks, primary to secondary in the transformer, etc.), and 2. using a Line Stabilization Network (LSN), spectrum analyzer and oscilloscope to actually test the real EMI results. Since, after all, EMI is primarily due to parasitic components that are particularly tricky to model, your best solution is testing.

Tim

[rcbuck]

Thanks Tim. I will make a couple of 6 mH and do some testing.