 |
Search Sci.Electronics.Basics |
|
|
|
|
|
|
 |
|
|
Sci.Electronics.Basics -> Power Supply Rectification and Smoothing
There are 94 messages in this thread.
You are currently looking at messages 60 to 80.
|
Author: EeyoreDate: 01:13 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >spamfree@spam.heaven wrote:
> >> Eeyore wrote:
> >> >spamfree@spam.heaven wrote:
> >> >
> >> >Eh ? They have a 50V rating. Why not use it ?
> >>
> >> Yes, and I've seen quoted a 63V surge capability, but what I wanted
> >> assurance of is the need or not of some multiple of a safety factor.
> >> Thanks for that, jack
> >
> >The 50V (63V surge) rating will be more than adequate, despite what some
weirdos
> >seem to think. Ignore the voodoo.
>
> I wonder why the Evox Rifa folk worry about transients and spikes.
I'm confident you've misunderstood or misread their application note.
There are no transients such as you imagime on a PSU reservoir cap.
Graham
|
|
|
|
Author: ehsjrDate: 01:14 16-10-07
|
|
spamfree@spam.heaven wrote:
> On Mon, 15 Oct 2007 15:22:26 -0700, Tim Wescott <tim@seemywebsite.com>
> wrote:
>
>
>>Yes, you can use the itty bitty caps. Verify their current ratings if
>>you can -- some of what makes the spendy caps spendy is the AC ripple
>>current they can deliver without overheating.
>
>
> I'm confused. This ripple current is surely what I want the caps to
> absorb, not to deliver, or am I on the wrong tack as usual? :)
>
> Anyways, it seems that the increased surface area of many small caps
> over one big one would favour heat dissipation. Just depends on what
> you can get your hands on for reasonable moolah. Seventy-five cents
> for 4700uF seems pretty good value?
>
>
>>You can parallel as many caps as you want, for more capacity and
>>current. 50V isn't enough for a 36V supply.
>
>
> Someone jumped on me for suspecting this :)
>
>
>>You can connect caps in
>>series as long as you use some balance resistors. You ought to have a
>>ballast resistor on the supply anyway, to make sure the caps discharge
>>by themselves
>
>
> Is this really necessary for 36V? Or is safety not the only reason to
> discharge caps?
>
>
>>-- you can combine the ballast and balance functions by
>>ballasting each capacitor bank separately.
>
>
> How does the balance work? What value and where would you put these
> resistors?
>
>
+ ----+----+------+----+----+----+----+--->
| | | | | | |
[C1] [4.7K] [C3] [C5] [C7] [C9] [C11]
| | | | | | |
+----+------+----+----+----+----+
| | | | | | |
[C2] [4.7K] [C4] [C6] [C8] [C10][C12]
| | | | | | |
Gnd ---+----+------+----+----+----+----+--->
How it works: imagine that the leakage current through
C1 is 500 uA when connected to 36V and through C2 is 100 uA
if it is connected to 36 volts. That means that C1 has an
equivalent resistance of 72,000 ohms, while C2 has an equivalent
resistance of 360,000 ohms. Now, if you install them in series
without the parallel 4.7K resistors in the diagram and apply 36
volts, the voltage across them will divide with 6 volts across
C1 and 30 volts across C2.
Now add 4.7K resistors as shown in the diagram above.
The voltage will divide with ~17.54 across C1 and ~18.46
across C2. That is because the effect of the 4.7K resistor
is far greater than the effect of the 72000 and 360000
ohm equivalent resistances of C1 and C2.
The exact value of the equalizing resistors in this circuit
is not critical, as long as they are all the same. You want
a minimum of 100 ohms per volt, and it could go as high as
10 times that.
Ed
>>Have fun.
>
>
> I am, thanks, Tim.
>
> jack
|
|
|
|
Author: EeyoreDate: 01:16 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >spamfree@spam.heaven wrote:
> >
> >> I'm confused. This ripple current is surely what I want the caps to
> >> absorb, not to deliver, or am I on the wrong tack as usual? :)
> >
> >The caps do both. They 'absorb' the current as the cap is recharged at the peak
> >of the input voltage from the transformer/rectifier and then deliver it to the
> >load when that input is providing no more charge. You see the
> >transformer/rectifier deleivers the input power in 'blips' every 10ms and the
> >caps job ios to smooth it out.
>
> Yes, I understand the buffering effect, but had not thought of the
> caps as producing their own ripple.
Absolutely not. They are entirely passive components. They get charged up every 10ms
and then they release some of their charge during the rest of the mains half cycle.
That's where the ripple comes from, from the regular charge and discharge of the
caps.
Graham
|
|
|
|
Author: EeyoreDate: 01:44 16-10-07
|
|
ehsjr wrote:
> The exact value of the equalizing resistors in this circuit
> is not critical, as long as they are all the same.
And you can avoid the issue entirely by accepting that 50V caps (with a 63V
'surge' rating no less) are absolutely and perfectly totally adequately rated
for a mere 36V supply.
Whoever suggested otherwise ought to be taken out and SHOT !
Graham
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:01:23 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>
>
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>> >Tim Wescott wrote:
>> >
>> >> You can parallel as many caps as you want, for more capacity and
>> >> current. 50V isn't enough for a 36V supply.
>> >
>> >Utter nonsense.
>> >
>> >Why do you think they make caps with a voltage rating ?
>>
>> I was worried about spikes and transients, like the Evox Rifa article
>> warns about.
>
>You won't see any spikes on a PSU reservoir cap. It would require a transient
>with a huge energy to do that.
>
>I suspect you've misread or misunderstood the app note. What page are you
>referring to ?
Page 7. jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:04:21 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>>
>> >> Fair enough, but I assume it would provide extra smoothing which may
>> >> or may not be of advantage here. I still can't find out what the
>> >> smoothness requirements are for these Chinese scooter motor
>> >> controllers.
>> >
>> >Very little I expect. It's just a motor.
>>
>> And a PWM speed controller.
>
>So ? They'll work just fine with a bit of ripple too. You're fretting over things of
>no consequence.
No, I thought I was asking a question about something I know little
about. Experienced people think the controllers (you can virtually
throw anything at motors for a short period) are rather more delicate
vis a vis current requirements. They certainly cannot take three times
the rated voltage that a motor can handle in its sleep. I've heard of
people letting the smoke out of controllers with a 25% increase in
battery DC voltage. Are you sure these controllers are as robust as
motors, as you seem to be implying?
>> >> An guy on another group with experience with these things
>> >> thinks it needs a pretty smooth supply, as there is little power
>> >> conditioning in them as they are designed for batteries.
>> >
>> >Not actually relevant as far as I can see.
>>
>> Well don't most electronic machines designed to plug into the mains
>> have considerable power conditioning in them? I know my computer PS
>> certainly does.
>
>Completely different. Those circuits require a stable voltage for reason way outside
>a simple discussion here. It's chalk and cheese. You're drawing false inferences.
Umm, can you not mention which components in those circuits are more
vulnerable to voltage surges than the components in these controllers?
The controllers generally have components that are just able to cope
with a set voltage of pure DC from a battery. I would have thought
that some of the minimum spec caps might be vulnerable to even minimal
and transitory overvoltages.
jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:07:51 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>>
>> >> BTW, he has given me
>> >> alternative ways to smooth the PS output as his calculations for the
>> >> required smoothness come out with a 3F capacitor which will be
>> >> expensive. jack
>> >
>> >He's a grade one IDIOT if he's talking about 3F capacitors. I'd take any
advice
>> >of his with more than a pinch of salt.
>>
>> He worked it out from first principles using slightly different input
>> data. Would you like to see his workings?
>
>Why not. I'd like to see how he got something wrong by a factor of about 100:1 !
"This produces a pulsing DC. the pulses are at twice the frequency of
the ac or 60 X 2 = 120 times per second in North America or 50 X 2 =
100 times per second elsewhere on the planet. I will assume that you
are "elsewhere" because you said "mains." The voltage supplied
varies
between the peak voltage and 0V and this change in voltage is called
the ripple voltage. To fill in the gaps between these pulses a
capacitor is used to store energy during the pulse and release it
during the gaps. The size of the capacitor is determined by the
current that needs to be supplied and the allowable ripple voltage as
well as the time that the capacitor has to charge up.
The time is determined by t = 1 / 100 = 0.010 seconds
During this time the capacitor must charge up and store enough energy
so that when the pulse ends the capacitor can supply this voltage. But
it can never fill the gap between pulses completely. So the first
compromise is how much of a voltage drop or Ripple is allowed between
pulses. For our example let us use 0.1 V.
This means that the capacitor must charge up to36V - 0.1V = 35.9V and
do this in 0.010 seconds. It must also store enough energy to supply
the required current for a 0.010 second period of time. This value is
supplied by the following formula:
C (in farads) = (Iamps X Tseconds) / ripple Volts
Just as an example let us use 30amps the calculation works out to (30A
X 0.01 second) / 0.1V = 3 farads. That is a big capacitor. In your
case using a 0.004700 farad (4700mfd) capacitor would not be large
enough to filter any of the ripple at all."
>> So who is the grade one idiot between you and the Evox Rifa folk whose
>> formula tells me I need only 10,800 uF - less than half your
>> calculation? :)
>
>Where do you get 10,800 uF from ?
From the formula on page 9 of the Evox Rifa article.
"Needed capacitance to manage a certain level of
ripple voltage is
Cmin= 2P/ [(Umax2 - Umin2) x f rectifier]
P is the power load in watts.
Remember that this is the minimum required capacitance."
>If you actually paid atention to the equation I used you can determine what you
>need for yourself. Let me tell you, 'cookbook' equations don't substitute for
>calculating from first priciples which is what I did.
And tell me, what percentage of ripple suppression were you aiming at
with your first principles calculation? Perhaps we are comparing
apples with cockroaches? jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:11:12 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>> >spamfree@spam.heaven wrote:
>> >> Eeyore wrote:
>> >>
>> >> >If it says '500W' I'd expect it to mean 500W input power.
>> >>
>> >> Well you would disagree with NEMA. jack
>> >
>> >Can you explain what you mean by that ? NEMA has no real relevance in this
matter.
>>
>> NEMA are in charge of electric motor labelling standards, I thought.
>
>You thought WRONG.
>
>NEMA is merely a US trade assocation and it's 'standards' or recomendations have no
authority
>worldwide. Look for *** IEC ** standards instead.
>
>IEC = International Electrotechnical Commission and all your Oz electrical standards
are based
>on IEC ones, not some stupid US trade association like NEMA.
>
>> Surely that's entirely relevant. jack
>
>No it's not.
So you don't like NEMA? I thought it was fairly relevant to US
electric motor manufacturers and suppliers. Is it significantly
different from IEC recommendations?
So could you please quote me where IEC advocate that an XkW motor is
a motor that needs an input of XkW and is not a motor that supplies X
kW of mechanical power for the task required? jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 05:14:03 GMT, ehsjr <ehsjr@bellatlantic.net>
>
> + ----+----+------+----+----+----+----+--->
> | | | | | | |
> [C1] [4.7K] [C3] [C5] [C7] [C9] [C11]
> | | | | | | |
> +----+------+----+----+----+----+
> | | | | | | |
> [C2] [4.7K] [C4] [C6] [C8] [C10][C12]
> | | | | | | |
> Gnd ---+----+------+----+----+----+----+--->
>
>
>How it works: imagine that the leakage current through
>C1 is 500 uA when connected to 36V and through C2 is 100 uA
>if it is connected to 36 volts. That means that C1 has an
>equivalent resistance of 72,000 ohms, while C2 has an equivalent
>resistance of 360,000 ohms. Now, if you install them in series
>without the parallel 4.7K resistors in the diagram and apply 36
>volts, the voltage across them will divide with 6 volts across
>C1 and 30 volts across C2.
>
>Now add 4.7K resistors as shown in the diagram above.
>The voltage will divide with ~17.54 across C1 and ~18.46
>across C2. That is because the effect of the 4.7K resistor
>is far greater than the effect of the 72000 and 360000
>ohm equivalent resistances of C1 and C2.
>
>The exact value of the equalizing resistors in this circuit
>is not critical, as long as they are all the same. You want
>a minimum of 100 ohms per volt, and it could go as high as
>10 times that.
Thanks so much for that, Ed. most useful, jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:16:03 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>> >spamfree@spam.heaven wrote:
>> >
>> >> I'm confused. This ripple current is surely what I want the caps to
>> >> absorb, not to deliver, or am I on the wrong tack as usual? :)
>> >
>> >The caps do both. They 'absorb' the current as the cap is recharged at the
peak
>> >of the input voltage from the transformer/rectifier and then deliver it to
the
>> >load when that input is providing no more charge. You see the
>> >transformer/rectifier deleivers the input power in 'blips' every 10ms and
the
>> >caps job ios to smooth it out.
>>
>> Yes, I understand the buffering effect, but had not thought of the
>> caps as producing their own ripple.
>
>Absolutely not. They are entirely passive components. They get charged up every 10ms
>and then they release some of their charge during the rest of the mains half cycle.
>That's where the ripple comes from, from the regular charge and discharge of the
>caps.
Must be my misunderstanding of the word "deliver". I get the sense of
"manufacture and then pass on", but my dictionary states otherwise,
and that is to merely "pass along". All's clear now, thanks for your
persistence, jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 06:44:20 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>ehsjr wrote:
>
>> The exact value of the equalizing resistors in this circuit
>> is not critical, as long as they are all the same.
>
>And you can avoid the issue entirely by accepting that 50V caps (with a 63V
>'surge' rating no less) are absolutely and perfectly totally adequately rated
>for a mere 36V supply.
>
>Whoever suggested otherwise ought to be taken out and SHOT !
Wow, worse than Sadaam Hussein, perhaps? :)
jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 12:11:47 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>spamfree@spam.heaven wrote:
>
>> Eeyore wrote:
>> >spamfree@spam.heaven wrote:
>> >> Eeyore wrote:
>> >> >Tim Wescott wrote:
>> >> >
>> >> >> You can parallel as many caps as you want, for more capacity
and
>> >> >> current. 50V isn't enough for a 36V supply.
>> >> >
>> >> >Utter nonsense.
>> >> >
>> >> >Why do you think they make caps with a voltage rating ?
>> >>
>> >> I was worried about spikes and transients, like the Evox Rifa article
>> >> warns about.
>> >
>> >You won't see any spikes on a PSU reservoir cap. It would require a
transient
>> >with a huge energy to do that.
>> >
>> >I suspect you've misread or misunderstood the app note. What page are you
>> >referring to ?
>>
>> Page 7. jack
>
>As I thought. It refers to transients on the mains.
Well where else would transients come from in this situation?
>Such transients won't
>appreciably affect the voltage on the reservoir caps, there isn't enough energy in
>them to do so.
Surely that depends on how much energy is in the mains transients.
With a rated surge tolerance of less than twice the nominal working
voltage.... Careful Laddie, ye'll get a bullet! :)
jack
|
|
|
|
|
|
On Tue, 16 Oct 2007 12:13:18 +0100, Eeyore
<rabbitsfriendsandrelations@hotmail.com> wrote:
>
>
>spamfree@spam.heaven wrote:
>
>> During this time the capacitor must charge up and store enough energy
>> so that when the pulse ends the capacitor can supply this voltage. But
>> it can never fill the gap between pulses completely. So the first
>> compromise is how much of a voltage drop or Ripple is allowed between
>> pulses. For our example let us use 0.1 V.
>
>There's the crazy bit. 100mV is an insane target for ripple voltage here.
Why? Perhaps he believes the controllers (which he is familiar with,
and you apparently are not - classing them as being as robust as
motors) might require such smoothing.
What target were you aiming for? Are you sure that would be adequate
for these controllers? Or are you just making a stab at a ripple that
you think would not bother a motor?
What makes your "stab" sane, and others' informed estimates insane?
jack
|
|
|
|
Author: EeyoreDate: 07:11 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >spamfree@spam.heaven wrote:
> >> Eeyore wrote:
> >> >Tim Wescott wrote:
> >> >
> >> >> You can parallel as many caps as you want, for more capacity and
> >> >> current. 50V isn't enough for a 36V supply.
> >> >
> >> >Utter nonsense.
> >> >
> >> >Why do you think they make caps with a voltage rating ?
> >>
> >> I was worried about spikes and transients, like the Evox Rifa article
> >> warns about.
> >
> >You won't see any spikes on a PSU reservoir cap. It would require a transient
> >with a huge energy to do that.
> >
> >I suspect you've misread or misunderstood the app note. What page are you
> >referring to ?
>
> Page 7. jack
As I thought. It refers to transients on the mains. Such transients won't
appreciably affect the voltage on the reservoir caps, there isn't enough energy in
them to do so.
Graham
|
|
|
|
Author: EeyoreDate: 07:13 16-10-07
|
|
spamfree@spam.heaven wrote:
> During this time the capacitor must charge up and store enough energy
> so that when the pulse ends the capacitor can supply this voltage. But
> it can never fill the gap between pulses completely. So the first
> compromise is how much of a voltage drop or Ripple is allowed between
> pulses. For our example let us use 0.1 V.
There's the crazy bit. 100mV is an insane target for ripple voltage here.
Graham
|
|
|
|
Author: EeyoreDate: 07:18 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >
> >Where do you get 10,800 uF from ?
>
> From the formula on page 9 of the Evox Rifa article.
>
> "Needed capacitance to manage a certain level of
> ripple voltage is
>
> Cmin= 2P/ [(Umax2 - Umin2) x f rectifier]
>
> P is the power load in watts.
> Remember that this is the minimum required capacitance."
Yes, a cookbook formula. I prefer working with the raw data.
> >If you actually paid atention to the equation I used you can determine what you
> >need for yourself. Let me tell you, 'cookbook' equations don't substitute for
> >calculating from first priciples which is what I did.
>
> And tell me, what percentage of ripple suppression were you aiming at
> with your first principles calculation? Perhaps we are comparing
> apples with cockroaches? jack
I wasn't working with any 'percentage' that's not a particularly useful measure. I was
targeting a practical ripple voltage. Your 56,000 uF gave a very low value of 2V pk-pk
so I sugegsted a more practical (and cheaper) capacitor size. 22,000 uF should give you
5V pk-pk ripple at 15A load.
Honestly, a bit of ripple won't hurt you.
Graham
|
|
|
|
Author: EeyoreDate: 07:30 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >spamfree@spam.heaven wrote:
> >> Eeyore wrote:
> >> >spamfree@spam.heaven wrote:
> >> >> Eeyore wrote:
> >> >>
> >> >> >If it says '500W' I'd expect it to mean 500W input power.
> >> >>
> >> >> Well you would disagree with NEMA. jack
> >> >
> >> >Can you explain what you mean by that ? NEMA has no real relevance in
this matter.
> >>
> >> NEMA are in charge of electric motor labelling standards, I thought.
> >
> >You thought WRONG.
> >
> >NEMA is merely a US trade assocation and it's 'standards' or recomendations
have no authority
> >worldwide. Look for *** IEC ** standards instead.
> >
> >IEC = International Electrotechnical Commission and all your Oz electrical
standards are based
> >on IEC ones, not some stupid US trade association like NEMA.
> >
> >> Surely that's entirely relevant. jack
> >
> >No it's not.
>
> So you don't like NEMA?
It's not a case of whether I like it or not. It has no authority or relevance outside
the borders
of the USA.
> I thought it was fairly relevant to US
> electric motor manufacturers and suppliers. Is it significantly
> different from IEC recommendations?
Most US regs have typically been rather different to International ones, although UL is
now
harmonising somewhat with IEC.
> So could you please quote me where IEC advocate that an XkW motor is
> a motor that needs an input of XkW and is not a motor that supplies X
> kW of mechanical power for the task required? jack
You'd have to look for yourself but I do know that wherever the IEC rates power in all
the
standards I know it rates by INPUT power.
It wouldn't be a bad thing to know what the input power is for sure.
Graham
|
|
|
|
Author: EeyoreDate: 07:33 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
> >spamfree@spam.heaven wrote:
> >> Eeyore wrote:
> >> >spamfree@spam.heaven wrote:
> >> >
> >> >> I'm confused. This ripple current is surely what I want the caps
to
> >> >> absorb, not to deliver, or am I on the wrong tack as usual? :)
> >> >
> >> >The caps do both. They 'absorb' the current as the cap is recharged at
the peak
> >> >of the input voltage from the transformer/rectifier and then deliver
it to the
> >> >load when that input is providing no more charge. You see the
> >> >transformer/rectifier deleivers the input power in 'blips' every 10ms
and the
> >> >caps job ios to smooth it out.
> >>
> >> Yes, I understand the buffering effect, but had not thought of the
> >> caps as producing their own ripple.
> >
> >Absolutely not. They are entirely passive components. They get charged up every
10ms
> >and then they release some of their charge during the rest of the mains half
cycle.
> >That's where the ripple comes from, from the regular charge and discharge of
the
> >caps.
>
> Must be my misunderstanding of the word "deliver". I get the sense of
> "manufacture and then pass on", but my dictionary states otherwise,
> and that is to merely "pass along". All's clear now, thanks for your
> persistence, jack
The caps do no manufacturing, they simply store charge and deliver it on demand.
Graham
|
|
|
|
Author: EeyoreDate: 07:35 16-10-07
|
|
spamfree@spam.heaven wrote:
> Eeyore wrote:
>
> >Such transients won't
> >appreciably affect the voltage on the reservoir caps, there isn't enough energy
in
> >them to do so.
>
> Surely that depends on how much energy is in the mains transients.
> With a rated surge tolerance of less than twice the nominal working
> voltage.... Careful Laddie, ye'll get a bullet! :)
I suggest you take a look at the voltage on some reservoir caps with a scope and look
for these alleged transients that over-volt them. You'll be looking a long time.
Graham
|
|
|
|
Author: Daniel MandicDate: 07:52 16-10-07
|
|
spamfree@spam.heaven wrote:
> Wow, worse than Sadaam Hussein, perhaps? :)
the One who found him ;-))))?, telling about Stars.
(where is my wet rag...)
> jack
Best regards,
Daniel Mandic
|
|
|
|
|
|
|
|
|
Contact | Electronic Portal
|
|
|
