One often sees people ask for help in choosing a ferrite mix for a particular application. A recent thread on social media asks for help designing a unun for the 1.8MHz amateur band, and it has provided the opportunity for participation, even if the content was not good.
An important early step in designing a ferrite cored transformer is to find a combination of ferrite material, core geometry, and number of turns to deliver acceptable core loss at the lowest desired frequency.
Design of a transformer to cover just the 1.8MHz ham band is a relatively simple exercise.
Let’s examine a 50Ω:50Ω autotransformer on a common FT240 core using some popular ferrite mixes (even if the mix is not available in that core size).
The mixes chosen are 31, 43, 46, 52, 61, 73, 77.
First step is to calculate the minimum number of primary turns that delivers InsertionVSWR closest to 1.1 based on published material characteristics. We will then look at expected core loss. That has been done with a Simsmith model that I have published for other projects. The model results are summarised in the following table.
Note that a different number of turns are required for each mix.* means that the mix is not readily available in a FT240 core size. The table is sorted in order of ascending core loss.
Note that a large number of turns and low permeability are not conducive to the low leakage inductance that is needed for broadband performance (though in this specific case, the required bandwidth is quite small).
Above is a column chart of the seven options.
Summary:
- mix 61 is the lowest, relatively low cost and good availability, but way too many turns for this core;
- mix 52 is low loss, moderately low cost and good availability for limited range of cores, but way too many turns for this core;
- mix 46 is low and practical, low cost from the manufacturer, in a limited range of core sizes, and not widely stocked;
- mix 43 is next, low cost, readily available in a large range of core sizes;
- mix 31 has around three times the loss of mix 43, limited range of core sizes and not so readily available;
- mix 73 has higher core loss and again not readily available in a large range of core sizes; and
- mix 77 has even higher core loss and again not readily available in a large range of core sizes.
The darling of the bunch is mix 43, if you are objective you can see why it has been so popular for so long.
Whilst 46 is a good material for a lot of applications, and I have used it in some designs, I tend to prefer 43 in published designs as others may find the cores more practical to source. Mixes 61 and 52 are just too low in permeability for a broadband 50Ω transformer at this frequency.
More turns could be used to drive core loss down for some cores, but more turns increases leakage inductance which is the enemy of broadband high ratio transformers.
There are further steps to the design process. The results above are a good start to design of a higher ratio autotransformer, essentially you are unlikely to develop a good design on a core and turns combination that does not deliver a good 1:1 ratio transformer.
These results apply specifically to the scenario described.
Model files
Attached is FT240-1-1-160m.7z containing a model and material files. The model is an autotransformer model used and described in several recent blog articles, and in this instance is configured for a 1:1 unun which means there is zero leakage inductance, and InsertionVSWR and CoreLoss is due to the shunt magnetising admittance of the transformer primary. Whilst a 1:1 unun might not seem to require the core, this is a first step to designing a different ratio transformer.
Be aware that ferrite tolerances are relatively wide, so measurement of a prototype might not reconcile exactly with the prediction based on the datasheet.