From: Robert Baer
X-Mailer: Mozilla 4.75 [en] (Win98; U)
Subject: Re: low loss transformer availability?
Date: Sun, 05 Jan 2003 10:29:38 GMT
NNTP-Posting-Date: Sun, 05 Jan 2003 02:29:38 PST
Organization: EarthLink Inc. -- http://www.EarthLink.net
"Smile, it can only get WORSE" wrote:
> Good day all,
> I need help with a transformer design, or perhaps just a nudge in the
> right direction.
> My transformer needs to be small (physically), and have extremely low
> standby losses. It will operate around 3 mhz in a linear AC power
> supply (not switching mode). It's output capability will vary between
> 1 microamp and 150 microamps and it only needs to supply 50 milliwatts
> of power. It should be capable of a 10 to one primary/secondary
> voltage ratio (12 volts in and 120 volts out). Since the output
> current is so small, a lot of secondary wire resistance is acceptable.
> I'm told that toroids are ok, but they still have a lot of stray
> capacitive coupling between the windings.
*** Not true.
Put input winding on left side of core, taking about 60 degrees of
arc, and put the output winding on the right side taking 60 degrees of
There will be 120 degrees of space on top and bottom.
The total capacitance will be under 1pF; if the size of the core is
0.5 inch diameter or more, the capacitance may be under 0.1pF.
> I need to make sure the input current is low when the load current is
> minimal.... hence the interest in ultralow standby current (no
> secondary load current should mean no (minimum) input current to the
> I was wondering about alternative types of transformer
> technology...such as planar. Or, perhaps a ferrite core could be
> machined to allow power transfer into the core while creating
> isolation between the windings?
**I do not suggest the planar configuration - you do not care about
Your major issue is magnetization of the core, the minor issue is core
Do not machine anything to wierd configurations, it is a big waste of
There are a lot of standard, off the shelf configurations.
> I began my research with a signal generator and by plotting the input
> current versus the output current with varying loads on the output. As
> I got down to loads in the mid/high megohm range, I discovered that he
> input was still drawing significant input current, which IS NOT what I
> need. Perhaps I need a torroid based transformer with high input
> impedance so that it doesn't draw much current without a secondary
> load...but, I could not achieve this on the bench test setup.
> Any ideas? Am I missing something?
> PS: Please don't send me any switching power supply design messages.
> My input is AC, my output has to be AC and I can't use switching mode
*** Use either toroid, pot core, E-I or U-I configuratins; they are
simple and efficent.
FerroxCube has a fairly reasonable useage chart to help choose
materials and shapes, but i strongly suggest you contact one of their
engineers, as their info may be out of date and incorrect relative to
what is available *now*.
A toroid seems to be the most efficent, but is a bitch to wind
thousands of turns.
The pot core is also efficent, and one could be nasty and use a three
section bobbin to have extremely low capacitive coupling (primary on
top, nil in middle, secondary on bottom.
This kind of seperated winding design results in a lot of leakage
inductance; that is the tradeoff (this is also true for the 60 degree
Use the smallest core that you can possibly put your windings on; less
material means less losses, since the power is way below what is shown
on the charts.
I suspect your test transformers are larger than 1 inch in any
dimension (L, W, H).
Push towards 0.25 inch as a size target.