Cheap, low power, isolated DC-DC converter design

[Michael Rigby-Jones-2]

I have a requirement for monitoring multiple RS232 lines (19600bps)
connected between some instruments and a PLC in a factory environment.
I have to ensure full isolation between all the channels, I must not add
any significant load to the RS232 line and I must stick as closely as
possible to the standard RS232 logic level thresholds

 

My first though is to power an RS232 line driver (MAX232 etc.) from an
isolated dc-dc converter and then isolate the TTL logic from the micro
using opto-couplers.  Obviously I would need a separate supply (or at
least separate isolated outputs) and separate RS232 line drivers for
each channel.  I only need to receive data, no data will be transmitted
over the isolated interface.  so I don't think I can get away with
kludges like driving an opto-coupler directly.

 

Firstly, is this is a reasonable solution?  Are there alternatives?
Assuming it is, I have looked at using the potted PCB mounted DC-DC
converters that are available from several manufacturers, but they are
rather expensive (>5UKP each).  Since the current requirements are so
low (driver and opto-coupler will be under 10mA), efficiency is not an
issue and I've been looking for a cheaper solution.  I found the
following application note:

 

http://www.exar.com/Files/Documents/sipex/ApplicationNotes/ANI-11_Isolat
edRS232_120406_B.pdf

 

This shows an ingenious method of using the switched capacitor supply
within a line driver to drive a small transformer (which are $1.60 each
in the low quantities I will be needing).  Now since I don't need  to
regenerate the RS232 line voltages this extra driver would be redundant
in my design, but I was considering driving one of these these small
transformers from a 50% PWM output from the micro, with a suitable
transistor driver.  Would there be any potential issues driving multiple
transformers from a single driver, or should I have a separate driver
for each transformer?  My main concern is if any high voltage transients
appear on one line, they could be more easily coupled into the others if
the primaries of all the transformers were coupled together.  A low
driver output impedance would help to suppress this I guess.  Thoughts?


 

Cheers

 

Mike


=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Olin Lathrop]

Michael Rigby-Jones wrote:
> I have a requirement for monitoring multiple RS232 lines (19600bps)
> connected between some instruments and a PLC in a factory environment.
> I have to ensure full isolation between all the channels, I must not
> add any significant load to the RS232 line and I must stick as
> closely as possible to the standard RS232 logic level thresholds

How much load is "not significant"?  Is there any power at all available on
the driving side?  If these are truly RS-232 lines, then they should be
capable of a few mA at least.

You can make opto couplers work at a few mA.  One trick is to use two of
them, each with a floating phototransistor output.  A high input level
lights one LED and a low the other.  Then arrange the two phototransistors
in a totem pole on your side.  That will be good enough to drive a CMOS
logic input, and you can go from there.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
[phone].  Gold level PIC consultants since 2000.
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Dwayne Reid]

At 04:26 AM 10/6/2009, Michael Rigby-Jones wrote:
>I have to ensure full isolation between all the channels, I must not add
>any significant load to the RS232 line and I must stick as closely as
>possible to the standard RS232 logic level thresholds
>
>My first though is to power an RS232 line driver (MAX232 etc.) from an
>isolated dc-dc converter and then isolate the TTL logic from the micro
>using opto-couplers.

I have been doing something similar for several years now to receive
DMX lighting data.  In my case, I use a tricky little homebrew DC-DC
converter to generate 5V @ 50 mA which is then used to run the
receiver chip and an opto.

I can share the DC-DC converter with you but would ask that you don't
release it into the wild, at least until I finally get around to
submitting it to one of the design mags.  I had actually planned to
do that several years ago but just never got around to it.

Best estimate is that the total parts cost of the converter is about
US $0.25 in hundreds, not including magnetics.  That's after building
several hundred of them.

Converter runs from 6-30 Vdc and is probably about 60%
efficient.  I've used it with a variety of transformers, even gone so
far as to use it with a common-mode choke that I purchased surplus
(several thousand pieces).  The only component that runs warm is the
resistor in the snubber network.

dwayne


--
Dwayne Reid   <dwayner@...>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
[phone] voice          [phone] fax
www.trinity-electronics.com
Custom Electronics Design and Manufacturing

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Wouter van Ooijen]

> I can share the DC-DC converter with you but would ask that you don't
> release it into the wild, at least until I finally get around to
> submitting it to one of the design mags.  I had actually planned to
> do that several years ago but just never got around to it.

I's like to use that 'in the wild', so let us know when and where you
publish it!

--

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: www.voti.nl/hvu

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[M.L.-2]

It's pretty typical to use a gate-drive transformer in forward-mode
when you need a loosely regulated low power isolated supply.

You can easily drive the primary of a gate drive transformer with an
H-bridge type circuit .. you can use 74xx buffers, emitter followers,
etc.
Rectify the secondary and you have a low current isolated loosely
regulated supply. It's a forward mode converter so Vout is
proportional to Vin.

A look at Coilcraft shows gate drive transformers for $1 in low quantity.


--
Martin K.
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Olin Lathrop]

Dwayne Reid wrote:

> I have been doing something similar for several years now to receive
> DMX lighting data.  In my case, I use a tricky little homebrew DC-DC
> converter to generate 5V @ 50 mA which is then used to run the
> receiver chip and an opto.
>
> I can share the DC-DC converter with you but would ask that you don't
> release it into the wild, at least until I finally get around to
> submitting it to one of the design mags.  I had actually planned to
> do that several years ago but just never got around to it.
>
> Best estimate is that the total parts cost of the converter is about
> US $0.25 in hundreds, not including magnetics.

"Other than that, Mrs Lincoln, how did you like the play?"

Wouter van Ooijen wrote:
> I's like to use that 'in the wild', so let us know when and where you
> publish it!

C'mon guys, how hard can this be!?  A little oscillator driving a
transformer with the feedback killing the oscillations would do.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
[phone].  Gold level PIC consultants since 2000.
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Wouter van Ooijen]

> C'mon guys, how hard can this be!?  A little oscillator driving a
> transformer with the feedback killing the oscillations would do.

Maybe, but the $0.25 BOM sounds interesting.

--

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: www.voti.nl/hvu

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Bob Blick-4]

On Tue, 06 Oct 2009 20:13:14 +0200, "Wouter van Ooijen" <wouter@...>
said:
> > C'mon guys, how hard can this be!?  A little oscillator driving a
> > transformer with the feedback killing the oscillations would do.
>
> Maybe, but the $0.25 BOM sounds interesting.

You left out the "Mrs. Lincoln" bit, which could also be referred to as
the "elephant in the room" - the transformer.

Cheerful regards,

Bob


--
http://www.fastmail.fm - A fast, anti-spam email service.

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Dwayne Reid]

At 11:29 AM 10/6/2009, Olin Lathrop wrote:
>Dwayne Reid wrote:
> > Best estimate is that the total parts cost of the converter is about
> > US $0.25 in hundreds, not including magnetics.
>
>C'mon guys, how hard can this be!?  A little oscillator driving a
>transformer with the feedback killing the oscillations would do.

For two bits?

dwayne

--
Dwayne Reid   <dwayner@...>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
[phone] voice          [phone] fax
www.trinity-electronics.com
Custom Electronics Design and Manufacturing

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[M.L.-2]

On Tue, Oct 6, 2009 at 2:42 PM, Dwayne Reid <dwayner@...> wrote:

> At 11:29 AM 10/6/2009, Olin Lathrop wrote:
>>Dwayne Reid wrote:
>> > Best estimate is that the total parts cost of the converter is about
>> > US $0.25 in hundreds, not including magnetics.
>>
>>C'mon guys, how hard can this be!?  A little oscillator driving a
>>transformer with the feedback killing the oscillations would do.
>
> For two bits?
>
> dwayne
>

555 oscillator or dual opamp.

--
Martin K.

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Olin Lathrop]

Wouter van Ooijen wrote:
>> C'mon guys, how hard can this be!?  A little oscillator driving a
>> transformer with the feedback killing the oscillations would do.
>
> Maybe, but the $0.25 BOM sounds interesting.

But that was without the magnetics, which will be the expensive part.  $.25
can buy a fair number of jellybean transistors, resistors, and capacitors in
reasonable quantities.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
[phone].  Gold level PIC consultants since 2000.
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Marcel Duchamp]

Dwayne Reid wrote:

> At 11:29 AM 10/6/2009, Olin Lathrop wrote:
>> Dwayne Reid wrote:
>>> Best estimate is that the total parts cost of the converter is about
>>> US $0.25 in hundreds, not including magnetics.
>> C'mon guys, how hard can this be!?  A little oscillator driving a
>> transformer with the feedback killing the oscillations would do.
>
> For two bits?
>
> dwayne
>

It all depends on the transformer.  You can build a Royer oscillator for
less than 10 cents... not counting the transformer.  How much does the
transformer cost in your design?
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[M. Adam Davis-2]

On Tue, Oct 6, 2009 at 1:29 PM, Olin Lathrop <olin_piclist@...> wrote:
> C'mon guys, how hard can this be!?  A little oscillator driving a
> transformer with the feedback killing the oscillations would do.

I, for one, still look at 'switching' power supply design as a black
art, such that I'm not comfortable trusting something I design even
though it works and I mostly understand it.

So if you have or can design a solution almost effortlessly then it
sounds like you'd have an eager audience here.

I know I would love a very cheap, poorly regulated, isolated 5v-5v ps.
 Especially if the same general design can scale from 5mA to 500mA
with minor changes or component choices.

-Adam

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Mike Harrison]

On Tue, 6 Oct 2009 13:29:55 -0400, you wrote:

>Dwayne Reid wrote:
>> I have been doing something similar for several years now to receive
>> DMX lighting data.  In my case, I use a tricky little homebrew DC-DC
>> converter to generate 5V @ 50 mA which is then used to run the
>> receiver chip and an opto.

Funnily enough I've recently been thinking about how you might be able to do isolated DMX receive
with no power on the RX side.
A few ideas I had - not tried any yet....

As DMX is RS422, you potentially always have power regardless of data content & duty cycle

a) A low-power oscillator, maybe low tens of MHz, simply powered from the data lines, coupled
inductively or maybe capacitively through the PCB, so no transformer. The RS422 spec states a
maximum number of receivers, so as long as you stayed within that limit it might be doable in-spec.

b) most DMX devices (especially those you want to save every penny on) only use a few channels, so
the data rate out of the interface can be somewhat lower than the 250K line rate - this may have
possibilities to reduce the bandwidth of the isolation components to save power/cost if you can do
the 250k receive at low enough power. Not sure how practical this would be as you'd still have to
receive at 250K and do the byte counting for addressing. .

c) I saw an interesting circuit idea in a mag recently for isolating an SPI port - driving a pulse
transformer with very short pulses (40ns or so), derived by driving each side of the primary with a
slight delay, using a couple of HC04 gates. This converted the input pulse edges into short +ve and
-ve pulses  The receive used a bus-hold latch arrangement, again with a couple of hc04's, withits
state flipped each way by  the short pulses.  As the pulses are very short, the avarage power draw
is very low, and may be doable with power leeched from the RS422 lines & some capacitance.




--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Olin Lathrop]

M. Adam Davis wrote:
> I, for one, still look at 'switching' power supply design as a black
> art, such that I'm not comfortable trusting something I design even
> though it works and I mostly understand it.
>
> So if you have or can design a solution almost effortlessly then it
> sounds like you'd have an eager audience here.
>
> I know I would love a very cheap, poorly regulated, isolated 5v-5v ps.

If the input voltage is known well, then it becomes particularly easy.  All
you need is a free running oscillator that can turn a pass element on and
off, or two of them out of phase.  You don't even need feedback since the
output will be the input (minus some losses) times the transformer ratio.

> Especially if the same general design can scale from 5mA to 500mA
> with minor changes or component choices.

Perhaps the topology can be similar, but some things are going to change
when you change the current by 100 times.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
[phone].  Gold level PIC consultants since 2000.
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[William "Chops" Westfield]

On Oct 6, 2009, at 12:25 PM, Olin Lathrop wrote:

> that was without the magnetics, which will be the expensive part.

The problem with most isolated DC-DC converters is that they end up  
using some custom-made transformer designed for optimal efficiency in  
exactly that circuit.  Surely there has to be SOME standard low-cost  
magnetic component suitable for <50mA worth of power isolation?  
(thinking of: "Joule thief" bifilar ferrite bead, 600ohm telco  
transformers, ethernet magnetics, DSL magnetics, etc...)

BillW

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Dwayne Reid]

At 03:11 PM 10/6/2009, William \"Chops\" Westfield wrote:

>The problem with most isolated DC-DC converters is that they end up
>using some custom-made transformer designed for optimal efficiency in
>exactly that circuit.  Surely there has to be SOME standard low-cost
>magnetic component suitable for <50mA worth of power isolation?
>(thinking of: "Joule thief" bifilar ferrite bead, 600ohm telco
>transformers, ethernet magnetics, DSL magnetics, etc...)

Exactly.

The version of this that I have built the most of (multi-hundreds)
uses a small common-mode choke that I bought for perhaps US $0.20
each or so.  Not exactly designed for the purpose but works well enough.

There's a pretty good chance that it will work with Ethernet
transformers (I haven't characterized one of those yet) and it for
sure will work with cheapo audio transformers.  I've run the
converter as low as 10 KHz and as high as 500 KHz.

I'm finally getting off my butt and have written Electronic Design
about submitting it.  I'm not sure what the rules are, but I'd think
that I can post it after they have printed it.  I also don't mind
sharing it privately even before that.

dwayne

--
Dwayne Reid   <dwayner@...>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
[phone] voice          [phone] fax
www.trinity-electronics.com
Custom Electronics Design and Manufacturing

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[Jesse Lackey]

For what it is worth to people discussing DMX/RS485... I solved the DMX
isolation problem with a Maxim MAX1480ACPI chip.  28 pin DIP,
transformer inside of it, and at $11 for qty 100 it is always the most
expensive part of the design.

I've been wanting to do something about this, as in, trade a large
expensive single chip for a few of something cheaper.  However, all the
other isolated RS485 stuff I've seen is data only.  Some datasheets give
sample mini isolated dc/dc circuits using unavailable magnetics.
Nothing compelling.

In my designs there will always be a micro that can generate a PWM to
run a transformer (or whatnot), and there could be some sort of isolated
feedback to regulate the PWM on/off so that the isolated side generates
a voltage within a decent range.

Dwayne - would love to see your circuit as well...
(<http://www.celestialaudio.com/ca_dmx_32_V2_nfet/index.html> is the
board I use the maxim part on, just FYI, and have used it in other
designs as well.)

Cheers,
J





Mike Harrison wrote:

> On Tue, 6 Oct 2009 13:29:55 -0400, you wrote:
>
>> Dwayne Reid wrote:
>>> I have been doing something similar for several years now to receive
>>> DMX lighting data.  In my case, I use a tricky little homebrew DC-DC
>>> converter to generate 5V @ 50 mA which is then used to run the
>>> receiver chip and an opto.
>
> Funnily enough I've recently been thinking about how you might be able to do isolated DMX receive
> with no power on the RX side.
> A few ideas I had - not tried any yet....
>
> As DMX is RS422, you potentially always have power regardless of data content & duty cycle
>
> a) A low-power oscillator, maybe low tens of MHz, simply powered from the data lines, coupled
> inductively or maybe capacitively through the PCB, so no transformer. The RS422 spec states a
> maximum number of receivers, so as long as you stayed within that limit it might be doable in-spec.
>
> b) most DMX devices (especially those you want to save every penny on) only use a few channels, so
> the data rate out of the interface can be somewhat lower than the 250K line rate - this may have
> possibilities to reduce the bandwidth of the isolation components to save power/cost if you can do
> the 250k receive at low enough power. Not sure how practical this would be as you'd still have to
> receive at 250K and do the byte counting for addressing. .
>
> c) I saw an interesting circuit idea in a mag recently for isolating an SPI port - driving a pulse
> transformer with very short pulses (40ns or so), derived by driving each side of the primary with a
> slight delay, using a couple of HC04 gates. This converted the input pulse edges into short +ve and
> -ve pulses  The receive used a bus-hold latch arrangement, again with a couple of hc04's, withits
> state flipped each way by  the short pulses.  As the pulses are very short, the avarage power draw
> is very low, and may be doable with power leeched from the RS422 lines & some capacitance.
>
>
>
>

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[M.L.-2]

On Tue, Oct 6, 2009 at 5:11 PM, William "Chops" Westfield
<westfw@...> wrote:

>
> On Oct 6, 2009, at 12:25 PM, Olin Lathrop wrote:
>
>> that was without the magnetics, which will be the expensive part.
>
> The problem with most isolated DC-DC converters is that they end up
> using some custom-made transformer designed for optimal efficiency in
> exactly that circuit.  Surely there has to be SOME standard low-cost
> magnetic component suitable for <50mA worth of power isolation?
> (thinking of: "Joule thief" bifilar ferrite bead, 600ohm telco
> transformers, ethernet magnetics, DSL magnetics, etc...)
>
> BillW

gate drive transformers

--
Martin K.

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

[David Duffy (AVD)]

As you say, DMX is still 250K even if you're sending less than a whole
universe. How does the number of channels change the bandwidth
requirement of the isolation components?

Mike Harrison wrote:
> b) most DMX devices (especially those you want to save every penny on) only use a few channels, so
> the data rate out of the interface can be somewhat lower than the 250K line rate - this may have
> possibilities to reduce the bandwidth of the isolation components to save power/cost if you can do
> the 250k receive at low enough power. Not sure how practical this would be as you'd still have to
> receive at 250K and do the byte counting for addressing. .
--

___________________________________________
David Duffy        Audio Visual Devices P/L
Unit 8, 10 Hook St, Capalaba 4157 Australia
Ph: [phone]      Fax: [phone]
Our Web Site: www.audiovisualdevices.com.au
___________________________________________

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist