[AMRadio] Class AB and B audio XFMRS

John Lawson jpl15 at panix.com
Mon Mar 13 00:09:48 EST 2006

On Sun, 12 Mar 2006, W5OMR/Geoff wrote:

> John Coleman ARS WA5BXO wrote:
> How difficult would be it be 'sweep' the mod iron?
> The reason I ask, is because so much of the surplus stuff we get is still of 
> Mil-Spec design.  Therefore, if the mil-spec transformer says it's freq 
> response is +/- 1db from 100Hz ~ 3000Hz, how would one determine exactly what 
> this particular peice of iron will be capable of handling, in the 
> 'non-comercial' application?

  An instructive test for most audio transformers is to apply nice, clean 
square-waves to them, and observe the result both in the time 
(oscilloscope) and frequency (spectrum analyzer) domains.

   The 'better' a transformer is, the fewer artifacts it will add, 
especially at the "ends" of it's specified bandwidth.

   Of course you can always sweep a transformer quite easily yourself - a 
stable audio generator and AC voltmeter (that will read accurately over 
the range of frequencies you'll be using) is all that is required.

   First, choose some reference frequncy, often 500Htz or 1KHtz - adjust 
the generator output to some convenient reading on the meter - doesn't 
matter what it is, as long as it's not down in the microvolt range..!

   If you have a dB meter - all the better, set your output level so the 
meter reads '0' or '-10' dB - or .775 VRMS if the transformer is near 
600-ohms output...  but no matter what type of meter you use, the object 
is to lower the frequency until a substantial fall-off in voltage is noted 
(the generator must NOT change it's output at all during this) - then 
raise the frequency until a similar fall-off is observed.  You'll get a 
pretty fair idea of the working bandwidth of that particular tranny. 
This will change under load, especially with large DC currents flowing in 
the windings - but the wider, the flatter, the better. If you want to get 
Fancy - get some (linear) graph paper and plot the output reading vs. 
frequency from, say, 10 Htz to 50 KHtz.

   Then do this again with square waves and watch on a scope for ringing 
and rounding of corners - you can use a distortion analyzer if you have 

   If you really want to get obsessive, you can plot out the magnetic 
curves, too - but this is outside the scope of this post.

   If you're completely insane, you can also calculate the acoustic 
environment of the transformer... speed of sound in steel, acoustic energy 
coupling between Mass 'A' of copper and Mass 'B' of laminations...

   At any rate, soon you'll get a 'feel' for the various transformers at 
your disposal.


John  KB6SCO

More information about the AMRadio mailing list

This page last updated 22 Feb 2018.