# Peltier modules - Different voltages in parallel?

Discussion in 'Electronic Basics' started by Rob Salkin, Mar 30, 2005.

1. ### Rob SalkinGuest

I'm a computer scientist and know very little about electronics. After
a while of google searching, I haven't been able to find a solution to
my problem.

I'm building a simple DC device that uses 1 to n Peltier modules to
generate electricity to power a motor. I heat one side of the module
and cool the other to get ~.3-.6v and 60-120mA output. I've found that
using 1 peltier module is more efficient than multiple ones in parallel
(or series). From my research, I think that the dynamic nature of the
output due to the uneven heating/cooling of the modules makes it a
different voltage in parallel problem (or different current in series).
I think that the highest voltage module is leaking power into lower
voltage modules and therefore reducing the output to the motor. I
understand that I could add a diode in series after each peltier module
in order to stop that backflow, but that would effectively be
eliminating the usefulness of all but the highest output module. So,
what can I do to utilize all of the power being generated from these n
modules?

Using the water analogy -- I'd like pour power into the top of a
reservior after each module to allow the pressure to build up enough so
that there would be enough force to overcome the pressure of the
backflow through a pipe at the bottom of the reservoir. Would I use a
capacitor for that? How could I use one without a real ground?

Thanks!

-Rob

2. ### Guest

Commercial Peltier modules are built up of lots of elements, apparently
connected in series.

http://www.marlow.com/TechnicalInfo/introduction_to_themoelectric_co.htm

which suggests that you should be able to connect several modules in
series.

If this generates more volts than you want at a lower current than you
want, you could always use a DC-to-DC converter to transform the high
voltage to a lower, or vice versa.

The Farnell catalogue lists a couple of pages of such devices -
admittedly, mostly aimed at converting 5V, 12V, 24V and 48V DC into
something in the same ballpark.

All of them use some kind of switch - these days it is usually a power
MOSFET or two - to chop the DC input to produce AC, which is then put
through a transformer to get a higher or lower AC voltage, which can
then be rectified to give the DC output you want.

This can be quite efficient - 90% or better.

There are ways of performing similar tricks with capacitors, but
transformers can shift a lot more power per unit volume.

With a bit more ingenuity you can build a circuit which can handle a
moderately wide range of input voltages - modern "universal"
switch-mode power supplies accept mains voltages from 85V rms to 265V
rms.

Your voltages sound more like the output from a solar cell, and you
might find that you could use something intended for that application.

3. ### MarkGuest

interesting... sounds like a similar problem with solar cells, perhaps
you can look for some info on how that is done...

in order to get the most power out of each module and combine them if
each has different V and I, you would need a small switching power
supply for each module then coordinate and combine the output of the
switchers, not simple.

Mark

4. ### Luhan MonatGuest

Mount all of them on the same pair of heatsinks (one on each side).
That will tend to equalize the voltage/current outputs and allow for
either series or parallel operation.

5. ### Rob SalkinGuest

Mark,

What sort of switching power supply do you mean? I don't have much
knowledge in the area...

-Rob

6. ### Rob SalkinGuest

Bill,

The area I'm working with is a few sq inches, so each module doesn't
produce the same V or mA output, due to differences in temperature.
Series and parallel circuits work worse than 1 alone. Futhermore, the
heat/cold aren't constant, so I can't even guarantee that the
individual modules will have a different, yet constant output.
Basically, all I know is that I have n modules that will put out a
range of voltages and currents, at any given time. I *wish* my problem
was that I have too much voltage!

-Rob

7. ### Rob SalkinGuest

Luhan,

I thought about that, but the area is too big. I have about a 5"x2". I
already have thick sheet metal on either side that seems to be evening
things out more than direct heating/cooling. I doubt I'd be able to get
it much more even. Even if the temps were similar, wouldn't even minor
differences cause a loss in useful power?

-Rob

8. ### Guest

The temperature difference across each Peltier junction sets up a
potential difference across that junction. If you stack up the
junctions in series, the potential differences just add up.

If the junctions are in series, you have to have the same current
running through all of them. Because each junction has an associated
resistance you lose some of that potential difference as a resistive
drop in each junction. This also just adds up over all the junctions,
but it doesn't have to distributed in the same way as the thermally
generated potential differences.

This won't be as efficient as drawing the optimum current from each
junction, but if the junctions aren't seeing vastly different thermal
environments, such that you'd be better off by-passing some of them, my
guess is that a single series string of juctions driving a single
pulse-width modulated DC-to-DC converter is going to be the most
practical way of handling your job.

Putting modules in parallel would be much more difficult - the high
voltage strings would just drive curren through the low voltage strings

9. ### John WoodgateGuest

I read in sci.electronics.design that wrote (in
I agree entirely, and I was tempted to post an article advising this,
but I wasn't sure that there was some subtlety involved that affected

10. ### MarkGuest

switching power supplies chop the DC input and can convert the voltage
and current by changing the duty cycle of the switching. They are also
called DC to DC converters. Also called flyback, boost or buck
converters.

I don't know if there is something off the shelf for what you need.

I'm thinking a bunch small boost type supplies whose outputs are
similar to current sources so the outputs can simply be put in
parallel. You need a power supply designer to help you with this.

Mark

11. ### Fred BloggsGuest

In the usual application, the thermoelectric generators are arranged
into identical series strings that are then paralleled so as to achieve
an equivalent resistance equal to the load for maximum power transfer.
Anything else is either less than optimal in power delivery or a waste
of thermoelectric modules. Note that deviation from less than identical
parallel string voltages is not so important when they are all driving a
node that is at 1/2 their open circuit potential. The OP is quoting
ridiculously low output voltages and unconcerned with telling us whether
this is open circuit or loaded, so either he has a gross resistance
mismatch or a pathetically unworkable temperature differential. Either
way- as usual- the thread is a boring waste of time.