convert 150 - 300mV into 5Vdc

[Don Pearce]

Terry Given wrote...

After 3-4 years goes by and it no longer works, you'll be thinking
about that off button as you make the trek to Radio Shack for the
hearing-aid battery. After 6-8 years has gone by you'll be sorely
missing that off button when it fails again. After 9-12 years...

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

3-4 years sounds pretty much like shelf life numbers to me. Would an
on-off switch really make much of a difference?

d
Pearce Consulting
http://www.pearce.uk.com

 

[Terry Given]

Kevin Aylward wrote...

Yep, the J105 is better, you should add it to your library. $1.22 qty
25 at Newark, #38C7317, http://www.fairchildsemi.com/ds/J1/J105.pdf


I suspect your transformer model. Real transformers don't have the
square loops I remember from your model, IIRC, and their saturation
is gradual. A slow 30mV/sec ramp wouldn't create enough secondary
voltage with a real transformer to negatively charge the 1.0nF cap
in Damaschke's circuit above, providing a way to turn off the JFET.


But the relay circuit would have to operate on 300mV. :>(


Indeed, one wakes up at night worrying about the extra non-operating
state. But if circuitry was added to detect this state and kick the
oscillator, then the benefit of simplicity is lost.

Thanks,
- Win

As is so often the case with "clever" circuits, things which seem to good to
be true usually are Very rarely is an engineering decision win-win (pun
unintentional

Cheers
Terry

 

[Terry Given]

Terry Given wrote...

After 3-4 years goes by and it no longer works, you'll be thinking
about that off button as you make the trek to Radio Shack for the
hearing-aid battery. After 6-8 years has gone by you'll be sorely
missing that off button when it fails again. After 9-12 years...

Thanks,
- Win

kinda - I keep a small stock of batteries for my various wodgies. Lest
Murphy ensure the damn thing wont go just when I need it (and 5 minutes
after Dick Smith (antipodean RS) closes

Perhaps I should put a new battery in my vernier (which has an on-off switch
I often forget to use now (both LR44s) so that I replace one battery
every 2 years, and always have a half-full battery around....

BTW I recently bought a little vernier for a company I contract to - 1200mm
+/- 0.1mm. Its a 2-man job to use it No batteries either

Cheers
Terry

 

[John Larkin]

kinda - I keep a small stock of batteries for my various wodgies. Lest
Murphy ensure the damn thing wont go just when I need it (and 5 minutes
after Dick Smith (antipodean RS) closes

Is DS still around down there? He opened a bunch of stores here,
including one in Burlingame (honorary Silicon Valley) but they all
shut down. His flyers were funny.

Perhaps I should put a new battery in my vernier (which has an on-off switch
I often forget to use now (both LR44s) so that I replace one battery
every 2 years, and always have a half-full battery around....

BTW I recently bought a little vernier for a company I contract to - 1200mm
+/- 0.1mm. Its a 2-man job to use it No batteries either

I prefer a mechanical dial caliper. All these digital gadgets seem to
die in a few years. I have a cooking timer, the twist-tick-bing sort
of thing, that has outlasted a half-dozen LCD things, and is far
easier to use.

I went to buy windshield wiper blades last Sunday. That's always an
ordeal. They have an LCD pushbutton thing that's supposed to find the
part number, given your car type. It's a real pain to use, but on
Sunday it was stuck in a loop running memory diagnostics! Luckily,
they also have a grubby book hanging on a chain, easier to use and
much more reliable.

John

 

[Jim Thompson]

On Wed, 02 Jun 2004 08:13:52 -0700, John Larkin

[snip]

I went to buy windshield wiper blades last Sunday. That's always an
ordeal. They have an LCD pushbutton thing that's supposed to find the
part number, given your car type. It's a real pain to use, but on
Sunday it was stuck in a loop running memory diagnostics! Luckily,
they also have a grubby book hanging on a chain, easier to use and
much more reliable.

John

I had to replace the wiper blades on my pickup truck a few weeks ago,
they'd turned rock hard from sun exposure.

Never been rained on in three years... just used to wash the
windshield ;-)

...Jim Thompson

 

[Kevin Aylward]

Kevin Aylward wrote...

Yep, the J105 is better, you should add it to your library. $1.22
qty 25 at Newark, #38C7317,
http://www.fairchildsemi.com/ds/J1/J105.pdf

Well, apparently, this is in my library. It was laziness to go through
them all an look at their specs.

I suspect your transformer model.
Maybe.

Real transformers don't have the
square loops I remember from your model, IIRC, and their saturation
is gradual.

All I did was knocked up a few typical values for Hc, Bs, and Br (See
LTSpice help for a description), again for proof of concept.

I did a few plots of the basic ramp of current (constant V) in the
inductor and although, it ramps up fast on saturation, its not a step.
Its a "typical" shape sort of thing.

I would have to do some more playing with it to see if the values I
chose were really reasonable, but off hand it don't seem that far off.

A slow 30mV/sec ramp wouldn't create enough secondary
voltage with a real transformer to negatively charge the 1.0nF cap
in Damaschke's circuit above, providing a way to turn off the JFET.

I agree. As I noted, the output voltage in my circuit was essentially
zero, until the saturation point which caused the negative spike to turn
the fet off.

But the relay circuit would have to operate on 300mV. :>(
Indeed.


Indeed, one wakes up at night worrying about the extra non-operating
state. But if circuitry was added to detect this state and kick the
oscillator, then the benefit of simplicity is lost.

And still not foolproof. I think the lithium battery is quite a good
idea.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Watson A.Name \Watt Sun - the Dark Remover\]

for perhaps a decade without maintenance, whereas with a battery you'd have
to replace it every 2-3 years if rechargeable.

Although you might convince me that with a lithium battery you could still
get ten years...

I would think that a pair Li coin cells like a CR2032 would give enough
voltage at low current to start up, and as soon as the big power
converter starts to put out, then the coin cells are no longer
furnishing current. And a small current thru a bleeder resistor could
sort of trickle charge them, or maybe that wouldn't be a good idea. In
any case, with the low current demands, the coin cells should last for a
very long time.

 

[Watson A.Name \Watt Sun - the Dark Remover\]

Terry Given wrote...

After 3-4 years goes by and it no longer works, you'll be thinking
about that off button as you make the trek to Radio Shack for the
hearing-aid battery. After 6-8 years has gone by you'll be sorely
missing that off button when it fails again. After 9-12 years...

Brings up another question. I seem to get the impression that most
hearing aid batteries are zinc-air (at least it seems so at Rat Shack),
which means that once you pull the tab off it, then it has a limited
lifetime due to the chemical reaction.

How long do you have to 'use up' the zinc-air cell once the tab is
pulled? With a hearing aid, it obviously won't be but a few days since
it's on all the time.

 

[Terry Given]

Is DS still around down there? He opened a bunch of stores here,
including one in Burlingame (honorary Silicon Valley) but they all
shut down. His flyers were funny.

very funny indeed - have you noticed the logo is Dick (head) Smith

I prefer a mechanical dial caliper. All these digital gadgets seem to
die in a few years. I have a cooking timer, the twist-tick-bing sort
of thing, that has outlasted a half-dozen LCD things, and is far
easier to use.

spoilsport. everyone knows you need DSP & bluetooth to build an egg timer

I went to buy windshield wiper blades last Sunday. That's always an
ordeal. They have an LCD pushbutton thing that's supposed to find the
part number, given your car type. It's a real pain to use, but on
Sunday it was stuck in a loop running memory diagnostics! Luckily,
they also have a grubby book hanging on a chain, easier to use and
much more reliable.

John

especially when the power fails

Cheers
Terry

 

[John Larkin]

Although you might convince me that with a lithium battery you could still
get ten years...

We use lithium coin cells for uP memory backup, connected directly to
the Vstby pin of a 68332. We've tested units after up to about 10
years in the field, and battery mAh capacity measures just as good as
new ones, usually better for some strange reason.

Lately I've been using serial EEPROMs. They are smaller and a lot
cheaper, but need four pages of driver code!

John

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

On a sunny day (Tue, 01 Jun 2004 07:09:35 -0700) it happened "Watson A.Name

"Watson A.Name - "Watt Sun, the Dark Remover""

in message news:[email protected]...

news:[email protected]... [snip]

| 1:45:65
| +in ,------,
| O--|----, | ,-----|>|----+---- +8.5V
| | # # # |
| | # # # |+
| | # # # ---
| | | | | 1.0nF --- 470uF
| | |-' '-|-+--||--, |
| '->|-, | | | |
| | | '-/\/\-+ |
| J105 | | 1.0M | |
| O-------+---+--------+---+---- gnd

Sadly, Damaschke doesn't have much to say about the critical aspects
of this circuit, e.g. transformer properties or JFET

requirements.

It says 1:45:65 but I would assume that there's more than one turn on
the primary winding. So if it had ten turns, then the others would have
450 and 650 turns respectively. But then if it had 20 turns...

:-O

.... your transformer manufacturer will hunt you down and beat you with a
stick

Besides, why does he use such a high ratio of turns on the feedback
winding? I realize that the input is only a few hundred mV, but why
should the JFET need more than a few volts to switch? They shouldn't be
as demanding as power FETs are for gate voltage.

45 x 150mV = 6.75V, you want the FET to 100% cut off.
JP

It takes that much voltage to cutoff a JFET? That's more than a power
FET, and much more than a logic level FET. I figured a couple volts
should do it. ??

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

"Watson A.Name - "Watt Sun, the Dark Remover""

Watson A.Name - \"Watt Sun, the Dark Remover\" wrote...

I putzed around with an old germanium transistor ... [ snip ]

Here's a URL for a converter that takes an input of .3VDC and
converts
it to 5VDC. It uses FETs to get around the .6V problem with
silicon BJTs.
http://www.ece.uvic.ca/~jbornema/Journals/064a-97ia-jmd.pdf

Can anyone tell off the top of their head, what the output voltage
is for those thermocouples that are used in furnaces? This might
run off one of them.

Perhaps someone else can answer the furnace-thermopile question.

John Damaschke's 100mW 300mV-to-5V dc-dc converter is interesting.

In the article he touts the importance of MOSFETs in performing the
work of converting as little as 150mV of source voltage into a nice
5V supply, but in truth this is a simple task given say 8 to 10V to
run the MOSFETs. Damaschke created a 8.5V source for the job, and
that circuit is the real innovation. The tough task of working
with as little as 300mV at startup is provided by a nifty JFET
blocking oscillator, which creates sufficient voltage to start
running the 5V converter. Once the +5V output voltage is well on
its way, a MAX630 (RC4193) step-up converter takes over, providing
8.5V for the FETs If high load conditions on the 5V output drag
the input thermopile down to as low as 150mV, the full converter
continues to work.

Damaschke says his blocking oscillator is similar to those used in
pacemakers, which are credited to Wilson Greatbatch's 1958
invention, http://www.engology.com/eng5greatbatch.htm In
Greatbatch's own
words,

http://www.winstonbrill.com/bril001/html/article_index/articles/151-200/
Ensure that Idss > I saturation of the transformer. This will result in
the transformer requiring quite a lot of turns when this is in the few
ma range.

Lots of turns so that it satuates at low I.

.... your transformer manufacturer will hunt you down and beat you
with a stick

Well, I was quite intrigued by this, so I have now updated SS with an
example oscillator of this type (LVBlockingOsc.sss,
http://www.anasoft.co.uk/LVBlockingOsc.GIF
, ignore the update date). It runs from about 150mv upwards.

The "design" is pretty simple. I only used a transformer with 2 windings
to try and minimise space. It uses my newly added hysteretic non-linear
core model (thanks for the core idea Mike).

Its quite subtle really, it just looks like a normal oscillator

circuit.

It looks like the D1 diode would short out the input voltage if it
wasn't such a low voltage.

 

[Winfield Hill]

Kevin Aylward wrote...

As I noted, the output voltage in my circuit was essentially zero,
until the saturation point which caused the negative spike to turn
the fet off.

I don't understand this part, for very slow ramps (low frequencies)
there'd be no magnetizing inductive voltage drop, hence nothing to
"collapse" at saturation. Unless it had an incredibly-large number
of turns, a real transformer primary would appear to simply be a
length of wire before and after saturation. Damaschke designed his
converter to run at about 1200Hz, and isn't it likely that any real
transformer design suitable for that would not operate conventionally
(i.e. as a transformer) with a 10-second ramp? Except perhaps for a
too-theoretically-perfect Spice transformer?

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Jan Panteltje]

On a sunny day (Tue, 01 Jun 2004 07:09:35 -0700) it happened "Watson A.Name

Terry Given wrote:
"Watson A.Name - "Watt Sun, the Dark Remover""

in message
[snip]

| 1:45:65
| +in ,------,
| O--|----, | ,-----|>|----+---- +8.5V
| | # # # |
| | # # # |+
| | # # # ---
| | | | | 1.0nF --- 470uF
| | |-' '-|-+--||--, |
| '->|-, | | | |
| | | '-/\/\-+ |
| J105 | | 1.0M | |
| O-------+---+--------+---+---- gnd

Sadly, Damaschke doesn't have much to say about the critical aspects
of this circuit, e.g. transformer properties or JFET requirements.

It says 1:45:65 but I would assume that there's more than one turn on
the primary winding. So if it had ten turns, then the others would have
450 and 650 turns respectively. But then if it had 20 turns... :-O

.... your transformer manufacturer will hunt you down and beat you with a
stick

Besides, why does he use such a high ratio of turns on the feedback
winding? I realize that the input is only a few hundred mV, but why
should the JFET need more than a few volts to switch? They shouldn't be
as demanding as power FETs are for gate voltage.

45 x 150mV = 6.75V, you want the FET to 100% cut off.
JP

It takes that much voltage to cutoff a JFET? That's more than a power
FET, and much more than a logic level FET. I figured a couple volts
should do it. ??

I think the idea is here that the RC network charges enough to create
a blocking effect?
More voltage, longer blocked period?
JP

 

[Kevin Aylward]

Kevin Aylward wrote...

I don't understand this part, for very slow ramps (low frequencies)
there'd be no magnetizing inductive voltage drop, hence nothing to
"collapse" at saturation.

There is still flux in the core even though the output is essentially
zero during the ramp up. Just prior to saturation there must be energy
stored in the magnetic field of the inductor. This energy *must* go
*somewhere* when the core saturates.

Unless it had an incredibly-large number
of turns, a real transformer primary would appear to simply be a
length of wire before and after saturation.

Not quite. It still stores energy.

Damaschke designed his
converter to run at about 1200Hz, and isn't it likely that any real
transformer design suitable for that would not operate conventionally
(i.e. as a transformer) with a 10-second ramp?

I agree. If the design relies on there being a voltage on the output
during the ramp up, the design will fail. That is not what is occurring
in my err... "design". All that matters in my circuit is that there is a
transient negative step to switch off the fet when the core saturates.

Install SS and have a play with the example.

Except perhaps for a
too-theoretically-perfect Spice transformer?

Well, I have omitted the capacitances, but I don't think that's a major
issue.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Kevin Aylward]

"Watson A.Name - "Watt Sun, the Dark Remover""
Watson A.Name - \"Watt Sun, the Dark Remover\" wrote...

I putzed around with an old germanium transistor ... [ snip ]

Here's a URL for a converter that takes an input of .3VDC and
converts it to 5VDC. It uses FETs to get around the .6V problem
with silicon BJTs.
http://www.ece.uvic.ca/~jbornema/Journals/064a-97ia-jmd.pdf

Can anyone tell off the top of their head, what the output
voltage is for those thermocouples that are used in furnaces?
This might run off one of them.

Perhaps someone else can answer the furnace-thermopile question.

John Damaschke's 100mW 300mV-to-5V dc-dc converter is
interesting.

In the article he touts the importance of MOSFETs in performing
the work of converting as little as 150mV of source voltage into
a nice 5V supply, but in truth this is a simple task given say 8
to 10V to run the MOSFETs. Damaschke created a 8.5V source for
the job, and that circuit is the real innovation. The tough
task of working with as little as 300mV at startup is provided
by a nifty JFET blocking oscillator, which creates sufficient
voltage to start running the 5V converter. Once the +5V output
voltage is well on its way, a MAX630 (RC4193) step-up converter
takes over, providing
8.5V for the FETs If high load conditions on the 5V output drag
the input thermopile down to as low as 150mV, the full converter
continues to work.

Damaschke says his blocking oscillator is similar to those used
in pacemakers, which are credited to Wilson Greatbatch's 1958
invention, http://www.engology.com/eng5greatbatch.htm In
Greatbatch's own
words,

http://www.winstonbrill.com/bril001/html/article_index/articles/151-200/

Ensure that Idss > I saturation of the transformer. This will result
in the transformer requiring quite a lot of turns when this is in
the few ma range.


Lots of turns so that it satuates at low I.


Well, I was quite intrigued by this, so I have now updated SS with an
example oscillator of this type (LVBlockingOsc.sss,
http://www.anasoft.co.uk/LVBlockingOsc.GIF
, ignore the update date). It runs from about 150mv upwards.

The "design" is pretty simple. I only used a transformer with 2
windings to try and minimise space. It uses my newly added
hysteretic non-linear core model (thanks for the core idea Mike).

Its quite subtle really, it just looks like a normal oscillator
circuit.

It looks like the D1 diode would short out the input voltage if it
wasn't such a low voltage.

Ah....you noticed. Subtle isn't it

When one is doing proof of concept, quick and dirty simulations, one
uses whatever is to hand.

The node wants to be clamped to around twice the supply to get "nice"
waveforms. I'm not suggesting that such a method is the way to go for
production. However, for V < 0.5, its not a bad way to go.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Kevin Aylward]

There is still flux in the core even though the output is essentially
zero during the ramp up. Just prior to saturation there must be energy
stored in the magnetic field of the inductor. This energy *must* go
*somewhere* when the core saturates.


Not quite. It still stores energy.


I agree. If the design relies on there being a voltage on the output
during the ramp up, the design will fail. That is not what is
occurring in my err... "design". All that matters in my circuit is
that there is a transient negative step to switch off the fet when
the core saturates.

Install SS and have a play with the example.

I have added a plot of the basic saturation characteristics with 300mV
constant applied

http://www.anasoft.co.uk/core_ramp_up.GIF

It is an imaginary core, but the basic shape is reasonable to my mind.
Aparantly, spice tells us, that such a ramp up will result in the
relevent voltage spike.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Terry Given]

I have added a plot of the basic saturation characteristics with 300mV
constant applied

http://www.anasoft.co.uk/core_ramp_up.GIF

It is an imaginary core, but the basic shape is reasonable to my mind.
Aparantly, spice tells us, that such a ramp up will result in the
relevent voltage spike.

Kevin Aylward

what if winding resistance is such that Vin/Rwind < Isat?

Then the core wont saturate, ever.

cheers
Terry

 

[Kevin Aylward]

what if winding resistance is such that Vin/Rwind < Isat?

Then the core wont saturate, ever.

Well, err... obviously...

However, I must admit, the circuit does seem a bit subtle in how it
still works with a very slow ramp input...

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Terry Given]

Well, err... obviously...

However, I must admit, the circuit does seem a bit subtle in how it
still works with a very slow ramp input...

Kevin Aylward

If I was simulating it (im not) I'd bung in a resistor, and step its value
to see how much is needed to wreck the circuit. And I would curse that I
cant step a component in a logarithmic fashion, forcing me do do multiple
iterations by hand.

Murphy suggests that the critical winding resistance is 20% below that of
the transformer you choose

Without bothering to estimate its actual value, but if say Isat of 1mA is
required then 150mV/1mA = 150 ohms, so Rwinding probably wants to be < 50
ohms. The winding resistance varies in proportion to the square of the
number of turns (for constant winding area utilisation), so it would be
possible to design such a non-starting transformer.

Cheers
Terry