Design Challenge - wide voltage input range SMPS
Russell McMahon email (remove spam text)
Requirement summary: Low cost DC-DC converter, 10-200v in, 6 - 10v 5
About 2 years ago I started an interesting and ongoing discussion with my
"design challenge" for a DC-DC converter. I provided my divinely inspired
buck converter as a starting point. This resulted in us hearing about
Richard's relay driver which in due course evolved out of all recognition
into Roman's Black Converter. My original 3+ transistor design best met my
original needs due to the special requirements but a number of other
suggestions proved to be better in many ways for more typical low voltage
I am about to revisit my original design. The original has served well with
thousands in operation. There have been a few failures but the exact number
are unknown to me due to the complex situation and lack of provisions for
feedback from end users. (Made in Taiwan, used in Europe (mainly),
percentage of spares sold with equipment, swap and throw-away on fault, no
provision or desire for feedback to supplier (unless, presumably, failures
get extreme). As far as I can tell the design has worked well and the
failure level is low enough to be easily explained by manufacturing,
installation and operational happenstance. (I'd greatly love better feedback
but have no control over the process.)
SO - it seems a good idea to respec my requirement here (slightly changed
from the original) and see if there are any new ideas which may prove
Power supply for electronic equipment.
Vin is sourced from a highly variable speed alternator.
For a reasonable (but incorrect) mental image imagine the output from a
windmill operated alternator.
Input 0v - 200v DC
Input can and does vary widely.
Most normal range is about 20 - 100v but lower
and higher is common.
Few or no transients of unexpected supply "funnies."
Output 6 to 12 v DC (single voltage, fixed selectable)
(Any given design will use a single fixed voltage).
50 mA to 600 mA capability.
Load may step suddenly.
Cost Low cost in moderate manufacturing volume (500 to a few thousand
at a time).
(Considering all aspects of supply & manufacture)
Reliability Yes please!
As input voltage rises from 0v the output should rise smoothly and once Vin
exceeds Vout_design, Vout should reach Vout_design with minimal extra drop.
(eg if designed for 6v out it would be "nice" if this value was maintained
for all Vin >= say 6.5v. (The present design produces, for a 50mA load,
Vout a few tenths of a volt below Vin, for Vin < Vout_target).
Efficiency is not crucial but should be high enough to allow sensible
component specification and heatsinking. A TO220 (or SMD equivalent) size
main switch is acceptable if necessary. Substantial heatsink is available
(shared with another power device).
As Pout max is say 12v x 600 mA ~= 7 watts then at 50% efficiency at full
power up to 7 watts would be dissipated.
At around 80% efficiency under 2w would be dissipated. In most cases Pout
will be far less than this (say around 1.5w). While the lower the overall
power level the better (for reasons unrelated top power supply issues) power
consumption is not critically important.
Input to Output isolation would be "nice" but is not essential. ie a buck
converter (such as the present design) would be acceptable. Output is
protected by a substantial zener at present - these go nicely short-circuit
on over-voltage. In the very few failures that I have seen nothing beyond
the zener has been damaged.
EMI minimisation is "desirable". Will need to meet CE radiation standards.
(Non mains connected so conduction standards less applicable).
Current construction is single side through-hole due to cost considerations.
This will be compared to Surface mount PTH for the new design to see which
proves more economic. (Main circuitry is not power supply related).
The "tricky" aspects of the design are caused by the high maximum input
voltage and the wide range of Vin.
If switching regulation starts at say 10v for a 9v supply then Vin range is
10v - 200v = 20:1. Requirement for such a wide range is almost unheard of in
The 200 Vin_max is well above the voltage ratings of known commercial ICs -
most are in the 20v - 40v max range. There is a wide range of offline IC's
targeted at the universal power supply market with 110 and 230 VAC input
capability. These either have internal switches rated at typically 600v or
use external FET switches (typically low side N channel). However, the vast
majority are designed for a minimum 85VAC rail and typically require a
reasonably high DC local supply for operation. If one of these proved able
to operate down to around 10V Vin and was overwhelmingly cost competitive in
other respects then linear post regulation may well be acceptable.
Translating simple low voltage designs to high voltage or wide input range
operation may well not be straight forward. Power dissipation in bias or
drive resistors which span from the high side to low side can be
significant. Optimising operation at one Vin is not possible.
Input most welcome. Failing any new revelation or freshly inspired designs
I'll be continuing with the existing basic circuit :-)
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email mitvma.mit.edu with SET PICList DIGEST in the body listserv
See also: www.piclist.com/techref/io/sensors.htm?key=rang
You must be a member of the
piclist mailing list
(not only a www.piclist.com member) to post to the