Searching \ for '[OT] Constant-current cap charging' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: piclist.com/techref/power.htm?key=current
Search entire site for: 'Constant-current cap charging'.

Exact match. Not showing close matches.
PICList Thread
'[OT] Constant-current cap charging'
2000\04\04@210711 by Jinx

face picon face
part 0 1556 bytes
This o/p stage is fine when there are only 1 or two solenoids to drive
(each has its own 4050-TIP121-cap) but the revised one has 15. The
recharging of 15 caps causes a huge surge on the PSU, which causes
the AC tap on the transformer to dip below that which can be reliably
measured by the PIC so the clock will/does lose time. The quick fix
was to put a resistor in series with the isolating diode, but as the circuit
also has a fast-forward mode, the slower charging of the o/p caps
makes the f-f stepping unacceptably slow (not to me, to the factory,
even though it's likely to be used only once a year)

A bigger transformer I tried gave much better results, but that had to be
abandoned because of space restrictions. For 29 of the 30 seconds
the tx isn't doing much, but has to deliver the goods straight after the
trip. The IC's in the circuit are diode-isolated + reservoir caps, their
DC supply isn't affected by the surge.

So is there a simple constant-current charger that would lessen the
surge on the PSU that I could put between the PSU and the junction of
all the 1N4001's ? A FET something ? The caps are completely
drained by the solenoids tripping and need to be recharged to at least
12V as quickly as possible (1 second ?) without that initial current
surge to minimise the PSU droop. Also, is this power-dumping harmful
to the cap in the long term ? They're standard PCB 25V 85C types.
The original has been in service for four years with no breakdowns
so far.

TIA


Attachment converted: growth:solenoid.gif (GIFf/JVWR) (0000CBDE)

2000\04\05@103627 by jamesnewton

face picon face
Well, how about
1) a small battery between the rectified transformer output and the power
supply regulator?
2) a nice big super cap after the regulator with a very low value resistor
to the solenoid caps?
3) replace / improve AC wiring to the unit? (probably won't help and would
be expensive...)

---
James Newton spam_OUTjamesnewtonTakeThisOuTspamgeocities.com 1-619-652-0593
http://techref.massmind.org NEW! FINALLY A REAL NAME!
Members can add private/public comments/pages ($0 TANSTAAFL web hosting)


{Original Message removed}

2000\04\05@145355 by dal wheeler

flavicon
face
----- Original Message -----
From: "Jinx" <.....joecolquittKILLspamspam@spam@CLEAR.NET.NZ>
To: <PICLISTspamKILLspamMITVMA.MIT.EDU>
Sent: Tuesday, April 04, 2000 7:05 PM
Subject: [OT] Constant-current cap charging


<snip>
>The recharging of 15 caps causes a huge surge on the PSU, which causes
> the AC tap on the transformer to dip below that which can be reliably
> measured by the PIC so the clock will/does lose time. The quick fix
> was to put a resistor in series with the isolating diode, but as the
circuit
> also has a fast-forward mode, the slower charging of the o/p caps
> makes the f-f stepping unacceptably slow (not to me, to the factory,
> even though it's likely to be used only once a year)

Could you pick off your timing on the primary side of the transformer (if
isolation not an issue)?

2000\04\05@181212 by Michael Rigby-Jones

flavicon
face
part 0 5729 bytes
<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Let's see, Q=It and Q=CV.&nbsp; So It=CV, therefore I=CV/t</FONT>
</P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Pluggin in C=2200uF, V=12 volts and t=1 second gives a current of only 26mA.&nbsp; Assuming that my calcs aren't total horse$hit, then the total current is less than 1/2 amp for all 15 clocks.&nbsp; I would suggest a constant current circuit for each clock would be the prefered solution.&nbsp; Not sure if constant current diodes are available in current ratings that high but a cheap and quick way of doing this would be to use a LM317 voltage reg in constant current mode (as described in the app notes)</FONT></P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Regards</FONT>
</P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Mike Rigby-Jones</FONT>
</P>
<BR>
<UL>
<P><FONT SIZE=2 FACE="Arial">{Original Message removed}

2000\04\05@182913 by paulb

flavicon
face
dal wheeler wrote:

> Could you pick off your timing on the primary side of the transformer
> (if isolation not an issue)?

 Apart from "get a bigger box to fit the bigger transformer" (there are
obviously common-sense limits to miniaturisation if it impinges on
reliability), this sounds like the best solution.

 Isolation is not an issue, this and *not* level conversion in a
circuit with common input and output ground, is what opto-isolators
are for.
--
 Cheers,
       Paul B.

2000\04\05@184207 by Thomas McGahee

flavicon
face
You might consider sequencing the charging of the capacitors.
Instead of connecting the bottom of the cap direct to ground,
connect the bottom of each cap to it's own transistor. Connect
the bottom of each cap to the collector of the associated
transistor. All emitters go to ground. Each transistor base
requires a current limiting resistor. The resistor then
connects to an I/O pin. The transistors all start out in
the ON state. When you send out the command to fire the
solenoids they all fire. Then you delay however many ms it takes
to ensure that the solenoids have done their job. Now turn
all the transistors off. The caps cannot charge when the
transistor is off. Turn the cap transistors on one at a time,
with a delay between turning the next one on so that the
large initial inrush of current is over on one set before
you activate the next set.

Leave the transistors on. The next time you fire the sets,
repeat the sequence mentioned above.

You might be able to tie two solenoids to a single capacitor.
That would cut down on the number of I/O lines needed.

Or use a shift register to control the transistors.
After firing all solenoids you delay to allow them to do
their job, then you rapidly shift out all zeros to turn off
all the transistors. Now march in "1"s one at a time,
with a delay between each shift to allow for the initial
current surge to subside. Once you have shifted in all
"1"s all the caps will be charged.

This is probably "overkill", but if the customer
insists on a quick recharge of all caps without
an initial overload, at least this is a method that will
work. Transistors need to be able to handle the
initial current surge, so don't go using something like
a 2N2222's !

Fr. Tom McGahee



>{Original Message removed}

2000\04\05@184214 by Thomas McGahee

flavicon
face
part 0 3473 bytes
You have 2,200 uf * 15 = 33,000 ufd. That is .033 Farads.
Let's assume you want to fully charge all 15 capacitors to
12 volts (from a 15 volt supply) within 5 seconds using a
constant current source.  Iconstant= (Emax*Farads)/Seconds
so Iconstant = (12*.033)/5   which comes out .0792 Amps.
To get to 12 volts in 1 second would require .396 amps.

Note that the constant current source circuit goes between
your existing Vcc and the isolation diodes. All isolation
diodes then attach to the collector of the PNP transistor.

As far as power dissipation goes, the constant current
source only provides the constant current until the
capacitors charge up to Vcc-1.8 volts. After that the
current will trickle off to next to nothing. It will
keep the capacitors charged, but that means just supplying
the leakage current.

How often do these things fire? If it is something like once
every 30 seconds, then you can probably use a small heat sink.
You mentioned fast-stepping. The heat sink must be able
to keep the transistor cool for the entire time the fast-
stepping is in effect. Iconstant must be chosen high enough
that the capacitors can recharge back to say 12 volts
within the fast-step time period.

Fr. Thomas McGahee
{Original Message removed}

2000\04\05@184222 by Thomas McGahee

flavicon
face
If your main problem is that the caps charging
mess up your transformer waveform, then go out
and buy the cheapest 12 volt transformer you
can find. It will only have to supply a couple
of ma at most! It can be VERY tiny!

Connect a 100 ohm and a 1k 1/4 watt load resistor
in series across the 12 volt secondary so it has
some loading, and place a .1 uf capacitor across
the 1k load to filter out transient noise.
Connect one side of the 1k resistor to your PIC
ground. Remove the AC sense wire you currently
have on the AC transformer tap, and move it
over to the junction of the 100 ohm and 1k
resistor. TADAAAA! You now have a nice clean,
isolated AC signal that is not affected by the
capacitors recharging.

Cheap, too!

Fr. Tom McGahee

2000\04\06@005257 by Jinx

face picon face
> From: James Newton

> 2) a nice big super cap after the regulator with a very low value resistor
> to the solenoid caps?

> From: dal wheeler

> Could you pick off your timing on the primary side of the transformer (if
> isolation not an issue)?

Thanks, think it'll be a combination of both -- an opto across the mains
as a 240:12 signal step-down, > 0.01F / 16V if I can fit it in somewhere
and R in series with the caps to take the initial sting out of recharging.
Aren't there anti-surge components in monitors/TVs ?

2000\04\06@044339 by Jinx

face picon face
Aha - more good ideas. I'll have to consider the options

Plan A (preferred, if optimistic) -- stop answering phone, hope job
goes away

Plan B (requires the BIG bag of BS) -- try and convince him that the
fast-forward rate he asked for is impractical

Plan C (what'll happen)  --  He gets what he wants. He knows it, I know it.
Resistance is futile, groan spit mumble. Ah customers, bless'em

2000\04\06@051530 by paulb

flavicon
face
Thomas McGahee wrote:

> See the attached PDF file for a very simple constant current source
> that might work for you.

 That or the two-transistor version beat the pants off the LM317 or
constant-current diode suggestions (the latter if only as they are hens
teeth at higher currents, though minimum voltage drop is also higher).

 Tom, why must you post PDFs?  GIFs inline into the mail much easier so
we can see them straight off in the navigator.  Your PDFs appear to
incorporate GIFs, including text, anyway.

 Your other suggestion of the transformer - would be far better off
to use the second transformer to power the PIC as well I should think.
I would think though that the space needed to add *any* second
transformer would be better utilised fitting a decent transformer in the
first place - possibly a C-core or toroidal.
--
 Cheers,
       Paul B.

2000\04\06@074948 by Michael Rigby-Jones

flavicon
face
part 0 3916 bytes
<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Performance wise, the LM317 current source isn't great, however the reason why I suggested it was firstly the minimal component count, i.e. one resistor and the IC, and the fact that they are pretty rugged devices with over temperature protection etc.&nbsp; (although I wonder if that works OK in constant current mode).&nbsp; Absolute accuracy and stability in this application aren't really important, as long as the caps can be charged within a certain time, then it should be fine.</FONT></P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Mike Rigby-Jones</FONT>
</P>
<UL>
<P><FONT SIZE=1 FACE="Arial">{Original Message removed}

2000\04\06@083123 by paulb

flavicon
face
Michael Rigby-Jones wrote:

> Performance wise, the LM317 current source isn't great,

 No, it certainly isn't.  Let's see, did we propose 350 mA charging
current for the caps?  At that current, the dropout voltage is about
1.8V at ambient temperature (but perhaps slightly less if the chip is
HOT) plus the reference voltage is 1.2V, so charging stops at 9V instead
of 12V, does it not?

> however the reason why I suggested it was firstly the minimal
> component count, i.e. one resistor and the IC, and the fact that they
> are pretty rugged devices with over temperature protection etc.
>  (although I wonder if that works OK in constant current mode).

 I should imagine it does.

>  Absolute accuracy and stability in this application aren't really
> important, as long as the caps can be charged within a certain time,
> then it should be fine.

 Well, the caps will be charged to 9V instead of 12.  Is that OK?

 The one-transistor current source has a drop-out of 1.3V, and the two-
transistor version, 0.8V, which makes it my all-time favourite.

 If however you are going to "gang" them, the one-transistor version
can use one resistor and the diodes in common (clever: Use a LED
instead) whilst you need two transistors for *each* source in the
two-transistor version.  Mind you, at least one transistor of those can
be a teeny SMD device.
--
 Cheers,
       Paul B.

2000\04\06@090437 by Michael Rigby-Jones

flavicon
face
part 0 4678 bytes
<P><FONT SIZE=2 FACE="Arial">Michael Rigby-Jones wrote:</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&gt; Performance wise, the LM317 current source isn't great,</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&nbsp; No, it certainly isn't.&nbsp; Let's see, did we propose 350 mA charging</FONT>
<BR><FONT SIZE=2 FACE="Arial">current for the caps?&nbsp; At that current, the dropout voltage is about</FONT>
<BR><FONT SIZE=2 FACE="Arial">1.8V at ambient temperature (but perhaps slightly less if the chip is</FONT>
<BR><FONT SIZE=2 FACE="Arial">HOT) plus the reference voltage is 1.2V, so charging stops at 9V instead</FONT>
<BR><FONT SIZE=2 FACE="Arial">of 12V, does it not?</FONT>
</P>
</UL>
<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">True, but I believe if you read the original post, Jinx said the caps had to charge to at least 12volts to properly operate the solenoids.&nbsp; That kind of implies that the charging voltage is higher, but that you could afford to only let the voltage get to 12.</FONT></P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Actually I proposed a regulator per clock, which is less than 30mA, but on hindsight one regulator probably makes a lot more sense.&nbsp; Unless more clocks will be added in the future there shouldn't be any need to adjust the current source.</FONT></P>
<UL>
<P><FONT SIZE=2 FACE="Arial">&gt; however the reason why I suggested it was firstly the minimal</FONT>
<BR><FONT SIZE=2 FACE="Arial">&gt; component count, i.e. one resistor and the IC, and the fact that they</FONT>
<BR><FONT SIZE=2 FACE="Arial">&gt; are pretty rugged devices with over temperature protection etc.</FONT>
<BR><FONT SIZE=2 FACE="Arial">&gt;&nbsp; (although I wonder if that works OK in constant current mode).</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&nbsp; I should imagine it does.</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&gt;&nbsp; Absolute accuracy and stability in this application aren't really</FONT>
<BR><FONT SIZE=2 FACE="Arial">&gt; important, as long as the caps can be charged within a certain time,</FONT>
<BR><FONT SIZE=2 FACE="Arial">&gt; then it should be fine.</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&nbsp; Well, the caps will be charged to 9V instead of 12.&nbsp; Is that OK?</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&nbsp; The one-transistor current source has a drop-out of 1.3V, and the two-</FONT>
<BR><FONT SIZE=2 FACE="Arial">transistor version, 0.8V, which makes it my all-time favourite.</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">&nbsp; If however you are going to &quot;gang&quot; them, the one-transistor version</FONT>
<BR><FONT SIZE=2 FACE="Arial">can use one resistor and the diodes in common (clever: Use a LED</FONT>
<BR><FONT SIZE=2 FACE="Arial">instead) whilst you need two transistors for *each* source in the</FONT>
<BR><FONT SIZE=2 FACE="Arial">two-transistor version.&nbsp; Mind you, at least one transistor of those can</FONT>
<BR><FONT SIZE=2 FACE="Arial">be a teeny SMD device.</FONT>
<BR><FONT SIZE=2 FACE="Arial">--</FONT>
<BR><FONT SIZE=2 FACE="Arial">&nbsp; Cheers,</FONT>
<BR><FONT SIZE=2 FACE="Arial">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Paul B.</FONT>
</P>
</UL>
<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">I agree totaly, in almost any application that needs reasonable performance I would use a discreet design, or even better an op-amp with current feedback.</FONT></P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Regards</FONT>
</P>

<P><FONT COLOR="#0000FF" SIZE=2 FACE="Arial">Mike</FONT>
</P>

</BODY>
</HTML>
</x-html>

2000\04\06@102709 by Terry

flavicon
face
A couple of suggestions:

1. Is 2200uF the lowest cap that'll work reliably? Reduce it till the
solenoid still works (even when it's warmed up) and use the next higher value.

2. Don't allow the cap to fully discharge. Try putting a small capacitor
between the 4050 and the base of the transistor. The total reactance of the
solenoid and cap should let u retain a couple of volts in the cap. If this
works, the current surge in charging the cap again won't be as bad.

3. Use a switchmode power supply. If 800mA ~ 1A is sufficient you might be
able to use some cell phone battery plug packs. Small but packs a punch.

If all 3 options are workable, you might just get away with series
resistors for the caps and an acceptable fast-forward rate.

Good luck
Terry



At 01:05 PM 4/5/00 +1200, you wrote:
{Quote hidden}

2000\04\06@173151 by Jinx

face picon face
What I'll do at the week-end is drag the 15 clocks outside,  the only
place roomy enough at the moment, and set them all up on planks.
There will of course be brandy standing by for any disorientated
visitor who thinks they've walked into a Dali landscape. Might run a
sweep on how many times neighbours and car park patrons next
door say "What's the time ? ha ha". Haha, we'll see who's feeling
funny on Sunday morning after listening to 15 solenoids clumping
all night (nah, wouldn't really do that).

I'll try an LM317 first, as that's simplest, and perhaps even knock
the caps down to 1000u. The coils are nominally 24V but do work
from a weedy response at 7V to the maximum rating of 30V. 12-
15V is a convenient middle ground. Dropping the caps to 1000u
might mean upping the voltage, and that's a whole new ball game.
Some clocks will have 100ft of cable (figure-8 mains) on them so
that will have to be taken into account.

I saw mention of "ganging" two transistors, but couldn't find a post
where this was explained in detail. Could someone do the honours ?

Thanks for all the help and suggestions so far

2000\04\08@071337 by Jinx

face picon face
Well, I had my day in the sun with a bank of solenoid clocks,
kerlunk, and think the problem is pretty well sort out. I started
with an LM317 as a current limiter, kerlunk, first with a 10 ohm
resistor (1.2/10 = 120mA), and ending up with a 5R6 (214mA). This
charges the caps to 15V in just over a second, kerlunk. Changing
the caps to 1000u was OK with short lengths of wire, but with 50m
on (about 6 ohms), kerlunk, there isn't enough oomph, so it was
back to 2200u. LM317 dropout was only 1V2. Fast-forward could be run
at about 1Hz, or 2 minutes to f-f an hour, with the voltage levelling
out at 8V, kerlunk, enough to drive the longest expected wire. No
significant heating in the LM317, but I put a small h/s on it anyway,
kerlunk.

There maybe something in a fortnight that I can use the discrete
current limiter for, possibly 1.5A, so that part of the thread won't
go to waste

Thanks again

2000\04\09@061838 by paulb

flavicon
face
Jinx wrote:

> LM317 dropout was only 1V2.

 Measured with a CRO, or multimeter?  I'll bet it current limits down
to about 3V, then leaks down to the 1.2V much slower via the resistor.
--
 Cheers,
       Paul B.

2000\04\09@100038 by Jinx

face picon face
> Jinx wrote:
>
> > LM317 dropout was only 1V2.
>
>   Measured with a CRO, or multimeter?  I'll bet it current limits down
> to about 3V, then leaks down to the 1.2V much slower via the resistor.
> --
>  Cheers
>        Paul B.

Unfortunately I don't have a convenient (eg dual scope) way of measuring
the slopes on either side of the LM317. The multimeter I used is a
digital type and the charging is over far too quickly to take any
meaningful measurements. The 1V2 is the settled voltage difference between
input and output. The scope trace I did get showed an S-curve, ie low
gradient to start and low gradient to finish charging. That must be
relieving some of the initial stresses on the PSU, it doesn't droop
anywhere near so dramatically as it used to.

The final drop-out is better than I hoped it would be. As the PSU is
only 17V1, 3V off of that would mean only 14V on the caps. I feel more
confident with 15V9 on the caps. Although as it worked out, the solenoids
seem to be reliable at 8V. Some may be installed in pairs (or more) in
series to save on wiring, so I can't assume that the 8V overhead will
be available for every clock. The limiting factor will be how they
respond to a fast-forward voltage that settles at 8V. My guess is
the installers will have to compromise between f-f speed and wiring.
I'm sure they'd rather it was reliable than being too mean with the
cable. Chances are another of these master units will need to be made
some time in the future so it's good to get a few things sorted out.

More... (looser matching)
- Last day of these posts
- In 2000 , 2001 only
- Today
- New search...