capacitors for a voltage divider circuit

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am

Post by rshayes » Wed Feb 28, 2007 10:11 pm

The capacitors that Nippon Chemicon makes with those sizes and values have ripple currrent ratings on the order of 1 amp. Each series string of capacitors could carry 1 amp, so the four parallel strings that you propose could carry a maximum of 4 amps. The actual current would be more than 10 amps, so the capacitors would be severly overstressed. Even at the maximum rated current, the lifetime of the capacitors would be in the thousands of haurs at best. With that level of stress, the lifetime might be reduced to hours or minutes.

A motor rumming (not starting) capacitor might hold up in this application. These are usually oil impregnated paper or plastic rather than electrolytic. If you can get these surplus they might be a better choice.

The best solution is probably a surplus transformer with a 28 volt, 10 amp secondary.

An alternate solution would be a surplus 28 volt switching supply.

User avatar
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey

Post by MrAl » Thu Mar 01, 2007 12:12 pm

Hi again Robert,

In addition to rshayes post, i want to caution you that 360uf is
too large of a capacitance to use in series with a 2.4 ohm load
which must operate at 24 volts (which would be 10 amps).
360uf would put 37 volts across the 2.4 ohm resistance, which
would be way too high for something that should run at 24 volts.
The cable could burn up and even catch something on fire.

Here is a short table listing cap values and the voltage they will
produce across a 2.4 ohm resistor where the string is run at
120vac 60Hz...

360uf => 37.1v
270uf => 28.5v
220uf => 23.4v

From the above you can see why i chose 220uf.
You can get 270uf from connecting two sets of caps in series,
where each set is made from two 180uf in parallel, but then
the current rating will only be around 3 amps (according to rshayes
report of 1 amp max per cap). This wont be enough to run
the 11 or so amps continuous that will result once the unit is
plugged into the wall. There is a chance it will run for a little
while, but i wouldnt expect it to run for too long.

360uf can burn the cable up unless you get lucky and the cable
resistance increases enough to limit current once it heats up.
Now i wonder if that 2.4 ohm spec you gave was cold ohms or
hot ohms.
LEDs vs Bulbs, LEDs are winning.

User avatar
Posts: 681
Joined: Mon Dec 10, 2001 1:01 am
Location: USA,World

Post by Joseph » Fri Mar 02, 2007 10:18 am

Ideally, the capacitor way does not waste energy. But it depends on the power supply being able to losslessy charge and discharge the capacitor.

A better solution is to place an inductor in series with the load instead of a capacitor. X(subL)=ωL. ω=2π<i>f</i>
You want the impedance, X(subL), of the inductor at 60 Hz to be about 10 ohms.

Solving for L yields 28mH.

The main problem with this way is that the magnetic core of the inductor will saturate if it is not large enough to distribute the magnetic flux through enough volume of magnetic material.

The best answer for you may to be to use a triac or SCR. Even though an SCR only works on half the ac cycle, since you need only 24 volts, it still requires adjusting down the conduction angle by means of a resistor or two and a capacitor. Either device should be rated for at least ten amps. Be sure to keep safety in mind.

Robert Gotts
Posts: 3
Joined: Mon Feb 26, 2007 4:07 pm

capacitor voltage divider

Post by Robert Gotts » Fri Mar 02, 2007 6:29 pm

Thank you all for the electronics wisdom that I lacked.
The plan has been laid to rest.
I'll take the transformer solution.
Back to the books.
Thank you
Bob Gotts

User avatar
Posts: 3037
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA

Post by haklesup » Mon Mar 05, 2007 1:00 pm

It is really hard to beat a transformer when it comes to efficiency and reliablilty and simplicity. While it may cost a lot for a 10A version, it will waste less energy than other means like a series cap and being less complex than an SCR based circuit it would be more reliable. Over several years, it may end up being cheaper to maintain.

In a single use situation, the transformer is a no brainer unless a 120V heater core was available instead. If you were designing a product for sale in any kind of volume, then you should look closer at an SCR type circuit.

With an SCR based circuit you would still apply 120V to the 2.4ohm heater you just wouldn't do it continuously. The circuit would chop the AC waveform and only apply portions of it for a short period of time. Such a circuit supplying sufficient current and reliablilty as compared to the transformer would have a similar cost unless you made a bunch of them.

User avatar
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey

Post by MrAl » Mon Mar 05, 2007 4:21 pm

Hi again,

Yeah a transformer is probably the best way to go.

I would say a triac instead of an SCR (to get equal plus and minus
loading) but even with that i would have to think about it a bit more
as some practical issues come up. For example, using a triac where
it is triggered near the end of each half cycle (for equal loading)
the angle where it triggers would be 180-33.6 degrees in order to
produce 240 watts in a 2.4 ohm resistance, and this correlates to
a sine wave voltage of 93.95 volts. In other words, for each half
cycle the load is switched on by a source of about 94 volts, which
gives rise to a current spike of over 39 amps. Im not sure i like
that idea as it might cause some problems with noise for one thing,
putting a nice size dip in the sine wave for a short time.
The other practical thing that comes up is getting the firing angle
just right. The output power is fairly sensitive to a change in angle.
For only a 5 percent increase in firing angle the output increases up
to 275 watts (from 240 watts nominal). This is an increase in power
of 15 percent. This means some way of controlling the firing angle
fairly accurately has to be incorporated into the design. This mean
a cap and resistor is probably not a good idea, nor any other approach
like this where the angle can change over time. Starts to sound like
a micro controller type approach is really called for here, where the
timing can be controlled via a crystal. Of course this doesnt help the
noise problem however without costly line filtering.
Just for reference, a change in voltage of 5 percent will cause a change
in output power of 10 percent, which is true for a transformer design,
but there's no chance of long term drift up to 20 percent, 30 percent,
etc. With a transformer there is also no 'doubling up' of evils, where
you get an unexpected long term increase in firing angle AND the
line voltage increases, causing maybe a large increase in power output.

The inductor idea sounds interesting, but i would hesitate because
as we all know turning off an inductor is always a hard thing to do.
A large inductor can put out a hefty voltage spike without some sort
of suppression, and it's not that easy to suppress a spike on an AC
line. Maybe a trans long as it doesnt burn out.
Still, the magnetics will probably cost here too.
LEDs vs Bulbs, LEDs are winning.

User avatar
Posts: 681
Joined: Mon Dec 10, 2001 1:01 am
Location: USA,World

Post by Joseph » Wed Mar 07, 2007 6:37 am

That point was good. The inductor would be placed in series with the load in a way that the windings of a transformer with two primary windings are placed in series with each other. Since the 60Hz is sinusoidal, a voltage spike will occur only when the power is disconnected, for the most part. I did forget about that problem, but the same thing happens maybe to a lesser degree on the windings of a transformer, too. It would help to place a snubber circuit
usually consisting of a series resistor and capacitor across the switch contacts. The
capacitor should probably be about .22uf and the resistor
may be around 100 ohms.

User avatar
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey

Post by MrAl » Wed Mar 07, 2007 10:17 am

Hi again,

Also a good point about the tranformer having some leakage
inductance which would normally produce some, although small,
This situation is much lessened however by making sure the switch
is on the primary side (a good idea anyway, so that when the unit
is turned 'off' the transformer is also 'off'). In this case, the load
is always connected to the transformer secondary side so there
is always going to be a good sized energy absorber connected in
the circuit. With a simple inductor that goes open circuit of course the
problem is going to be much worse.
If we look at the open circuit of the primary we can also adopt the
view that any leakage inductance, even though small, that is in series with
the load will be operating into the secondary, and with the primary suddenly
open circuited the secondary all of a sudden goes to a high impedance,
which means a large voltage can be produced across it (at least for a
short time) which then means it gets stepped 'up' by 5 times to the
primary, which can then possibly produce an arc across the switch
thus limiting it's useful life. Luckily, most line operated transformers
are designed close to the BMax of their core rating so that a voltage
that develops much above their voltage rating (primary is 120v here)
gets clamped and absorbed by the core after a time period roughly
equal to maybe a half cycle or so. I guess this is why transformers
operated in this manner usually dont require any special spike energy
absorber circuits. I cant say that is would be bad to add a suppressor
across the switch though, as it may just make the switch last longer.
I know lots of things like this including motors, that are turned on and
off with just a switch though and they last for years, so i guess this
suppressor is more or less optional.

I am happy that most others are also agreeing now that the transformer
solution seems to be the best long term solution, and also the safest.

The only other issue that comes up then that sometimes comes up with
transformers operated over long time periods is transformer overheating.
The scenario is that the transformer overheats, either because it is
mounted in a confined area (ie inside a wall where there is no free air
flow) or develops a shorted turn. In either case sometimes the
thing can get hot enough to cause wood or paper to start flaming,
which of course means a fire. A fire of course means property damage,
or even worse, death.
The solution developed over the years is to connect at least one thermal
fuse in series with the primary, where the thermal fuse is mounted
so that it touches the windings through at most maybe one layer of
tape. If the transformer overheats the thermal fuse blows disconnecting
the transformer from the mains and everything cools down.
To be extra sure, a few thermal fuses mounted in various places on
the transformer would detect heat spots as well. They could all be
wired in series. If any one blows the mains gets disconnected. This
would be a very safe thing to do.
Of course the transformer has winding resistance and core losses, which
contribute to heat rise. If the transformer is mounted in a confined area
the heat continues to rise until something burns up, or in the case where
there is a thermal fuse, it blows the fuse, so it goes without saying now
that it should be mounted in a place where it has free air flow, and that
is it unlikely that this will change in the future (ie dont mount in the
basement ceiling between floor joists because if someone else
modernizes the basement by sheetrocking the ceiling, suddenly the
transformer is in a confined area! )
Maybe i am being a bit over cautious, but hey it pays in the long run.

Picking the temperature rating of the thermal fuse(s) might be
a little tricky, but something around 80 deg C is probably ok.
Mythbusters did a show on trying to start a Christmas tree on
fire with light bulbs. They used a kiln dried tree and something like
3000 light bulbs on that one little tree, and even after many hours
the temperature rose but didnt get hot enough to start wood on fire.
It takes a lot to start a wood fire really, but unlike a light bulb, a
transformer can actually do it. Limiting the temperature to 80 or
even 100 degree C should protect it though.
LEDs vs Bulbs, LEDs are winning.

User avatar
Posts: 681
Joined: Mon Dec 10, 2001 1:01 am
Location: USA,World

Post by Joseph » Wed Mar 07, 2007 5:57 pm

AL, considering leakage inductance was good. occasionally we had to repair power switches in stereo amplifiers and receivers whose contacts had gone bad despite the capacitor soldered across the contacts.

I think the values I gave for a snubber in such a case were a bit weak. Maybe a 2.2uF in series with a 10 ohm 2w resistor would be better.

The inductive impedance method has been used in florescent lamps for a long time. But, I agree, the transformer is the safest way to go. That way shock hazard is reduced because of the isolated secondary. Also, they are readily available which is easer than trying to obtain or design a suitable inductor.

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 36 guests