Power Supply with Switching Pre-Regulator

[MrAl]

Hello again,

Vic:
Yes, i think it would be good to have a decent switcher
circuit to work with things like wall warts and automobile
supples (for Luxeon and similar). It would be nice
if it didnt get too warm. As you know, a linear will get
warm powering a luxeon from the car 12v battery.

Pablo:
Yes, that could work, but i think something a little more
high powered would be better (100ma or more).
That would help relax the need for two windings, but then
the person building would have to order that part too,
so it's a tradeoff.

Take care,
Al

 

[vicbin]

Hi folks,

Pablo,
That ICL7660 is capable only at 10ma or below, I got it too, and its only suitable for low current needs below 10ma such as your ICL7107 DMM display. I tried before and even driving LEDs at 20ma the Vout dropped about 3 volts ! /ubbthreads/images/graemlins/frown.gif

Remember, you'll need another positive regulator for it since the max Vin only at 10 Volts !

Al,

I know you already provided the way of supplying neg supply but it is from transformer with center tapped.
Please take a look at this pic below, is this ok (and safe) for generating negative supply ?

Any recommendation for the D1,D2 and C2,C3 value especially for 100ma or more ? Or maybe you have other better solution ?

About inductor you put 200uH, could you explain where or how did you get this value ? Because from the datasheet there is only a formula for fixed output (for Adj. version). I'm lost how to use those formula for this kind of circuit since the output can be starting very low to high with very low current to high too ? (Adjustable Volt and Amp)

Your idea separating those modules is very handy too, think of other CPFer that interest or can afford building the linear part only ! /ubbthreads/images/graemlins/grin.gif

Regards,
Vic

 

[MrAl]

Hi Vic,

I'll take a look at your idea of generating the minus
supply and let you know what i find ASAP.
I'll also try to explain the inductor value selection.
I should have this info a little later today ok?

In the mean time, here's the next version of the PS,
so you can take a look and see what you might find.
Note that i'll still be looking at replacing the neg
supply shown here with your idea Vic.

Version 002:

Take care,
Al

 

[vicbin]

Hi Al,

Again, take your time ! /ubbthreads/images/graemlins/smile.gif

This is interesting, I noticed something new here, what are those D6 (1N5226), 2N4401 transistor and the 10K resistor used for ?

The linear C-in, how big for 5 Amps ? and are those Cout for switcher still the same 3 X 300uF even for 3 or 5 Amps ?

Regards,
Vic

 

[MrAl]

Hi again,

Vic:
Your idea seems to work very well, so i included it on
the new schematic for the new version...(see below).

The new parts (incldg. 2N4401) is to swamp the FB pin in
case the neg supply goes away before the pos supply as
you mentioned earlier. This will kill the output voltage
in case input power goes down while powering something
critical, so the output wont jump up unexpectedly.

Actually, the 3x300uf might be an overkill, probably
one 1000uf would be ok, but it will probably end up
being 1000uf for every amp output, so that's
3 x 1000uf for 3 amp output, etc.
I'll double check this too.

Here's the new version with the simpler neg ps:

Take care,
Al

 

[vicbin]

Hi Al,

Great ! Actually those are not my design, I'm sorta copy it from my other circuit. (the other Adj. PS with so many-many discrete components) /ubbthreads/images/graemlins/grin.gif
Finally after hours of staring those new parts, I finally understand how it works when the neg. supply collaped and how it shutoff the whole thing !

It is so simple and pardon for my "noobeeness", another great thing I've learned today ! /ubbthreads/images/graemlins/bowdown.gif -> Mr.Al

Hi Pablo,

You can save those ICL7660 ICs now /ubbthreads/images/graemlins/grin.gif , trust me, you'll need it for those ICL7107 based DMM for good accuracy and consistent measurements.

Regards,
Vic

 

[PEU]

Hi Al/Vic

yeah! I ordered 2 sets of samples from maxim, and both of them included icl7107/7660 ics /ubbthreads/images/graemlins/smile.gif

They will go to friends there and then forwarded via usmail here.

Also ordered some lm338 from national this way also /ubbthreads/images/graemlins/smile.gif hope they deliver this time (but I'll use the lm350 for the PS)

So, as you can see, I'm in no hurry, and learning a little every day it passes /ubbthreads/images/graemlins/smile.gif

I wait for the single output version

/ubbthreads/images/graemlins/popcorn.gif

Pablo

 

[Burnt_Retinas]

Should the 50uf electro for the -ve supply should be a non-polarised one?. The current polarised one will have reverse charge on it for some time which may lead to premature failure.

Chris

 

[MrAl]

Hello again,

Vic:
I hope you like the way the power is shut down if the neg supply goes
away first. It's interesting that with light load, the thing could
put out power for some time without it.
We're all learning here :-)

Pablo:
So i guess you are making your own volt meters? What are you
going to use for the display?
Were you able to get the 338's as samples or no?

Burnt_Retinas:
Thanks for the input!
Well, when the ac input goes negative at the top (+) of the 50uf cap
diode D4 conducts, and with enough current flowing to the Cin caps
it might drop 1v or so, which means the top (+) of the 50uf cap sees
about -1 volt, and since the second 1N4003 diode drops about 0.8 volts
or so, the difference is -0.2 volts (approx). This means the 50uf
cap will at most, see about -0.2 volts. Since it's rated at 50uf
i think it can take this small neg voltage.
Did i miss something or does this seem ok? I'll double check too.
[LATER]
I've found that there could be 1/4 to 1/2 cycle reverse
voltage across the 50uf cap, if the input power is turned
on during the negative half cycle of the 60Hz input.
Changing this to 500uf reduces the voltage to about
1 volt or so, so i guess we'll just do that.
This change makes the 50uf cap 500uf instead.


Take care,
Al

 

[PEU]

[ QUOTE ]
MrAl said:
Pablo:
So i guess you are making your own volt meters? What are you
going to use for the display?
Were you able to get the 338's as samples or no?



[/ QUOTE ]

I already have the voltmeter, but is nice to receive free stuff and I may end building a second one for current reading /ubbthreads/images/graemlins/smile.gif

Pablo

 

[MrAl]

Hi Pablo,

Oh yes, i thought so, but what are you going to use for
the display of the second one?

Take care,
Al

 

[vicbin]

Hi all,

Just did a little shopping at local shops today, look what I got ! /ubbthreads/images/graemlins/smile.gif

Al,

The sense resistors are ready to serve at 5 Amps even at 1 Ohm ! (by 2 X 2 matrix 40 watts or 0.5 Ohm at 20 Watts !) ...psstt.... 5 cents a pop /ubbthreads/images/graemlins/cool.gif Wondering what that "J" means ?

Have you decided the sense resistor value ?

Also got a new MAX724 (PDF) switcher with 5 Amps capability and the shop keeper said it is better, this will be used for the second phase (5 Amps) after the LM2576 settled.

Regards,
Vic

 

[MrAl]

Hi Vic,

Gee, where do you find those shops? Wish we had a few
around here :-)

Oh good, im glad you got the sense resistors already.
Im thinking of going with 0.5 ohms, because 1 ohm
eats so much power and voltage. I'd like 0.1 ohms,
but i'll have to do a few tests to make sure this will
work ok... havent gotten to that yet, but soon.

The Max724 looks interesting, but they cant seem to decide
on what value for the resistor in the compensation network.
Did you compare the switch voltage drop in the Max chip
to the LM (5 amp) chip?
Also, how much are those Max chips (5 amp)?

Take care,
Al

SOME INDUCTOR SELECTION THEORY

With 100% duty cycle at 50kHz we have a period of 20us,
so 50% duty cycle is 10us.
We then can apply the equation:
dv=L*di/dt

which can be read as:

"The change in voltage (dv) is equal to the inductance (L)
times the change in current (di) with respect to time (dt)".

For our purpose, 'dv' becomes the input - output differential
volts, 'di' becomes the ripple current, and dt is the period (10us).

Since we know the input voltage and the period, we can assume any
output voltage we want in order to determine what the ripple current (di)
will be, and we simply look for a max.

Lets say we have 20vdc input, 10vdc output, and the period is 10us.

Solving the above equation for di, we get this:

di=(dv*dt)/L

Using this equation with our chosen value of 200uH for the
inductor, we end up with the following values:

dv=20-10=10 volts
dt=0.000010 seconds
L=0.000200 henrys

and after inserting them into the equation for di we end up with:

di=(10*0.000010)/0.000200 {amps}

which when we do the math we end up with:
di=0.5 amps

This means at 50% duty cycle we'll end up with about 0.5 amps
ripple current through the inductor.

Hope this helps.

 

[vicbin]

Hi Al,

Btw, I'm really sorry for delay on ferrite core shipment cause didn't have a chance to visit the nearest post office (approx. 1 hour drive) /ubbthreads/images/graemlins/frown.gif . Will do this on next week. Two of this hunkee resistor will go to NJ too , no big deal, just 10 cents ! This will suitable for your 1.5 Amp (LM317) linear on both variation either 1 or 0.5 Ohm at 1.5 Amps ! /ubbthreads/images/graemlins/grin.gif

I'm ready what ever sense resistor value you choosed, 1 Ohm or 0.5 Ohm or even other. Anyway as you said we will have one spare/free op-amp that will be used as multiplier/divider for sensing the drop for measurement using DMM. Just focus on the best result for the current overload loop as you said.

[ QUOTE ]

The Max724 looks interesting, but they cant seem to decide on what value for the resistor in the compensation network.
Did you compare the switch voltage drop in the Max chip to the LM (5 amp) chip? Also, how much are those Max chips (5 amp)?

[/ QUOTE ]

Sorry, can't answer your question about the voltage drop caused I never build 5 Amps switcher before and also this is way beyond me since I don't understand what is that "<font color="red">compensation</font>" used for ? Is this suitable for our need on variable voltage and current ?

Also I found out other brand switcher non National such as this Maxim, Linear or On-Semi has this compensation pin.

Example non National switcher :

- Linear : LT1070 - 5A and 2.5A High Efficiency Switching Regulators Datasheet (PDF)
- On Semi : MC34167: 5A, Step-Up/Down/Inverting Switching Regulator Datasheet (PDF)

Why National "Simple Switcher" family (0.5 to 5 Amps) doesn't have it ? Is this an advantage or disadvantage ? or maybe National has a better design (more "simple") ? Please explain if this compensation network pin for if you got a chance.

Anyway, that MAX724 cost me approx $5.5 a piece ! Not cheap, and if it is not suitable for our project, I can return it and have a refund ! /ubbthreads/images/graemlins/grin.gif

Thanks & regards,
Vic

 

[MrAl]

Hi again Vic,

the LM2678 doesnt required a compensation network, while
the Max724 does. Since they arent clear on what the
value of the resistor is supposed to be, it might
have to change with output voltage and/or current.
I dont know for sure, so i cant say for sure, but it
looks like the National part will work. It's not that
i think the National part is better, just that there is
nothing to adjust so it should work the same as the
3 amp version (almost).

I'll have to check the LT and MC parts you showed, and
see.

It might even be possible that the Max724 can be adjusted
to have a slightly better response than the National part,
but without more information i cant be sure, and i wouldnt
want to take the chance because this sort of adjustment
pretains to the transient response, something most people
cant check once they build the power supply.

For now i think we should stick with what we 'know' (however
little that is /ubbthreads/images/graemlins/smile.gif ) and perhaps later try some new parts
too, which i sure wouldnt mind. Other makers parts would
be interesting too, im sure, and i'll check them out next.

I'd like to get to the most advanced design soon, so im
looking at various ideas. I think the major part of it
isnt really the switcher itself though, that's pretty
straightforward...it's the current limit and 0v adjust
with only one output. I'd like it to work very good
without any problems and nothing to check except a few
simple tests once someone builds one up.

Max724 comp pin:
The pin on the Max724 is for compensating the error
amplifier. Pick the wrong parts and the switcher could
become unstable for some inputs or for some loads.
Pick the right parts and it might have a better response
than the National part. Trouble is, im still in the dark
about it's exact nature, so i'll have to look more into
it. What i did find out so far is that the resistor
has to be chosen carefully, and the data sheet seems to
be kind of vague about what it should be. If it ends up
that it has to be chosen according to output voltage or
output current or even input voltage range (on the
breadboard) then we're out of luck, because ours will vary
from 0 to probably 30v output with current from 10ma to
3.00 amp (or higher) so we probably not have the luxury of
selecting only one value. It could be tested of course.

I made some adjustments in my previous post...we should
start with 50% duty cycle for the inductor selection.

If your inductor takes a bit longer to get here that's
perfectly all right...might be just in time for when i
get the other parts :-)
Please dont make that trip just for that alone, but
rather wait until you have another good reason to
go there too. No reason to rush... the longer we
take to find a good solution, the better it is /ubbthreads/images/graemlins/smile.gif
I thought i would have my parts ordered already too,
but im still looking at a few things. Once im sure,
i'll get what i need. I'd like to make only one
order for everything, and i think i can get that done.


Take care,
Al

 

[PEU]

Hi Vic/Al,

one question related to the inductor, can I use the ones found in PC AT suplies? they are about 1.5in in diameter and the core is about .5in? I have many of these.

If not, I know a local shop that sells toroids, so doesn't matter.


Pablo

 

[vicbin]

Hi Al,

Thanks for the explanation on the inductor, really clear few things out of my mind.

But again, lets forget those switcher, National switcher LM2576 simplicity beats out of those switcher with "compensation" feature. I've read all their datasheet and to my surprise, they're not clearly explained on how to utilise it. Looks like they left their customer in the dark and assuming we build em first and solve it later method. /ubbthreads/images/graemlins/frown.gif


Hi Pablo,

Dunno how to answer that, but I believe it is a switcher's inductor hence you can try it ! /ubbthreads/images/graemlins/grin.gif

Regards,
Vic

PS:
Folks, I will be "offlined" for few days ahead, on the field assignment ! Feeding the family first ! /ubbthreads/images/graemlins/smile.gif

 

[MrAl]

Hello again,

Pablo:
Gee, im not sure about the inductors inside the switcher.
Im sure some of them would be suited, but you'd need a
way to test them--which means you at least need a way to
measure the ripple current though the inductor.
Perhaps build the Ripple Detect Network and measure
ripple voltage, im not sure. Normally you would use a
scope. A scope is a very important tool for these kinds
of power supplies, because there is much concern about
the dynamics which can usually only be observed with
a scope.
Some of the tests points about an inductor would be
1. It's inductance
2. It's change of inductance with dc current or it's
ripple current when used with the chosen switcher ic.

More than likely there is one you can use, but we'd have
to be able to make sure it worked ok.
If you hook it up with the switcher and maybe turn the
input voltage up slowly, you can check the ripple current
to get an idea if it's working or not, or perhaps check
the ripple voltage across the output cap. If the ripple
current is too high (inductance too low) the output
caps might get hot.
When you purchase a ready made inductor, it's got a rating
for highest current so you know if it will work before
you use it.
Sorry this probably doesnt help you that much.

Vic:
Yes, the National switchers look like they will be the
best choice, at least to start with. It takes some of
the guess work out of it. And yes, it's funny that the
data sheets on the comp networks arent that clear with
the other manufacturers. I guess they figure mostly
fixed output applications, which is probably 99% of the
uses of these things.


Take care,
Al

 

[MrAl]

Hello again,

Update:
Just to let you know im still working on this
circuit. Almost done :-)

Take care,
Al

 

[PEU]

Hello Al,

take your time, we got plenty of /ubbthreads/images/graemlins/popcorn.gif /ubbthreads/images/graemlins/smile.gif

Pablo