Power Supply with Switching Pre-Regulator

[MrAl]

Hi again,

Pablo:
Yes /ubbthreads/images/graemlins/smile.gif i find that im constantly updating the schematic too.

Vic:
I've verified that you can measure the ripple voltage peak
to peak and ripple current peak to peak and divide v/i
to get ohms ESR, at least in theory. How to measure
current, well, that's another story. It's got to be
the ripple current and not the dc current. Maybe no
load, but still need something to catch the peaks if
no scope is available. Maybe modify the ripple
detect network to measure current peaks too.
Then a small resistor in series with the cap, but also
some kind of fast amplifier to amplify the peaks.
I dont think this would be too easy, and how accurate;
would have to be tried, tested and calibrated.
On the scope you can see the pp values and divide.

It's interesting that with high frequency and high value
cap, the cap looks like a short circuit to ac voltages,
but it has a series resistor (ESR) which drops voltage
and so a small but measureable ac voltage appears across
the cap in spite of it's high microfarad value.
Also, if it were really a short circuit an infinite ac
current would flow but because of the series resistance (ESR)
the current is limited to some relatively small value,
but is also measureable.
Both ac voltage and ac current being measureable, it's
then possible to calculate the ESR.


Take care,
Al

 

[NewBie]

Hey Al,

How much does the output voltage sag from no load to full load?

 

[MrAl]

Hi there Newbie,

The spec's for that are about the same as the linear
regulator used. LM317 for example has typical 0.3 percent
load regulation. This works out to about 6mv at 20v out.
For comparison, two external leads of #16 gauge wire each 2
feet long drop about 5mv total at 3 amps out.
I think Kelvin sensing is possible too with this ps.

Take care,
Al

 

[vicbin]

Hi Al,

Sorry about the double caps & inductors question, didn't mean to distract you and I aware how busy you are ! And also I will concentrating on breadboarding the linear part very soon and will update you.

[ QUOTE ]
Pretty simple, and there's no reason why you cant apply this technique once the switcher is operating by simply measuring the peak output voltage with 50% duty cycle and measuring the peak current through the cap. Perhaps we can do the same by measuring the ripple voltage. This might be possible using that 'Ripple Detect Network' i talked about before. If there is I average amps through the cap and there is v volts of ripple, than Rs is probably something like v/I/2. I'll double check this later today. The ripple detect network isnt hard to build either, a few small cheap resistors and caps.

[/ QUOTE ]

Great & thanks ! Will keep this in my high priority to do list ! /ubbthreads/images/graemlins/wink.gif

[ QUOTE ]

It's interesting that with high frequency and high value cap, the cap looks like a short circuit to ac voltages, but it has a series resistor (ESR) which drops voltage and so a small but measureable ac voltage appears across the cap in spite of it's high microfarad value. Also, if it were really a short circuit an infinite ac current would flow but because of the series resistance (ESR) the current is limited to some relatively small value, but is also measureable. Both ac voltage and ac current being measureable, it's then possible to calculate the ESR.

[/ QUOTE ]

Well, for sure I will definately looking into this, again noted into my todo list ! /ubbthreads/images/graemlins/smile.gif

Btw, a bit offtopic but related since you're talking bout high value & high freq cap, I bought some weird caps that I've never seen before and it looks like a firecracker or small dynamite with double fuses to me LOL ..he..he.. /ubbthreads/images/graemlins/grin.gif.

It was my caps supplier that has a "very good" salesmen skill, they "persuaded" me to buy these caps below because they claimed that they're very good at high freq. and very low ESR, really ?

Although I don't believe it 100%, I just bought them anyway because they're dirt cheap, those 4 smaller one 1uF cost me only $0.2 a piece and the big one 10uF cost me "nothing" and as a sample since I purchased quite some electrolite caps for the switcher part from them. /ubbthreads/images/graemlins/grin.gif

Looking at the big one, it is labelled "Audiophiller" brand ? Geezz...is that only suitable for audio circuits ? Since it is non polarised and I didn't see any + & - sign, is it ok to subsitute it for the C2 capasitor in your circuit ? Yeah, don't worry, next time I'll ask again from him the 4.7uF one for free sample ! /ubbthreads/images/graemlins/wink.gif

Size compared with 5mm led.

About Newbie question, (Hi there Newbie ! /ubbthreads/images/graemlins/smile.gif ), is that depends on the load transient response od the switcher ? Linear part should be pretty fast aren't they ?

Kelvin sensing ? Is that the same as "remote sensing" ? That is great !!!! /ubbthreads/images/graemlins/clap.gif

Protection from "hard" short circuit at the output, CC loop mechanism also play the role of protecting the "overall" circuit right ?


Hi Pablo,

My mistake, I thought plain copper clad board. /ubbthreads/images/graemlins/wink.gif
But still expensive though, I can do for you for 3 LSs T series, and that's for 3 boards !! /ubbthreads/images/graemlins/grin.gif

What do you mean by "metalized" thru hole board ? Is that vias ?

Regards,
Vic

 

[MrAl]

Hi Vic,

No that's ok, i would rather be reminded.
I'd like very much to hear about your linear once it's running.
Audiophiler? Well, if you find one 47uf i guess you can use it for C2.
Are they mylar caps there?
Yes, remote sensing would have two leads extra leads (sense leads) going
out to the load itself, but i think we shouldnt complicate this circuit
any more. It would require breaking a few connections and attaching
two leads for remote sensing.
For short circuit, CC loop plays a role yes, but so does the current limit
of the linear and switcher probably too.

Take care,
Al

 

[koala]

I'll probably report back when dealer stocks his LF412 dual op amps. For now /ubbthreads/images/graemlins/popcorn.gif.

 

[MrAl]

Hi (Vince right?),

Oh that's all you need? I think Digikey stocks them.

Take care,
Al

 

[MrAl]

Hi again,

Here's the latest drawing, with the protection diode.

Oh, BTW, the load regulation is 0.3 percent which is 0.006
volts at 2 volts output (full temp range), not at 20 volts
output.
At 20 deg C the load regulation is typically 0.1 percent,
which is 2mv at 2v output.

Take care,
Al

 

[vicbin]

Hi Al,

I don't know what kind or type of that caps, the seller was saying it is MKT type ? Is that Mylar ? If it is mylar, is it true they're low ESR & high freq cap ?

I need to find a polypropylene cap for the 4.5 dmm, since it needs a very low leak cap for accuracy and was goggling around and found out plastic caps poly-xxxxxx types are so confusing to identify. Any idea ? /ubbthreads/images/graemlins/icon3.gif

Please explain the underlying mechanism and the reason of the value of these resistors : R4=100, R3=2k or 2.5K, R17=100.

Btw as you pointed out for 30V out, the R3 (Volt pot) is 2.5K ... this value also hard to find here ! Did you find a cheap 100Ohms pot yet ? /ubbthreads/images/graemlins/crazy.gif

Are those R24=100, C10=0.01uF sorta compensation for the voltage loop too ?

Load regulation about 2mv for 2 Volts, I can live with that ! /ubbthreads/images/graemlins/grin.gif

Regards,
Vic

 

[MrAl]

Hi Vic,

MKT i think is polyester...better than electrolytic, but not as good
as low esr type. What would you like to use these for?

You need to find a polypropylene cap? Are you sure they didnt want
Polycarbonate or Polystyrene? Do you know what the dmm uses this for?

Probably the only way to make sure you get the right kind of cap
is to find one in a catalog and check the manufacturer's specs.
They'll have to state what kind it is and what it can (or cant) do.

The reason for R4 is because this sets the current for the resistor
network that sets the LM317 up for a certain output voltage while
conducting the minimum current required for all the regulator types
(LM317/x/x). It's a little high for the LM317 (12ma instead of 6ma)
but it's not a big deal and allows using the same network for all
the regulator types. The value of this resistor is chosen to satisfy
the minimum current requirement for all of the regulator types.

R3=2k or 2.5k is simply the adjustment pot for voltage adjustment (coarse).
When the resistance is increased, the output voltage goes up provided
current regulation doesnt kick in. The value is set to provide full
output with max resistance.

R17 has the same value as R4 and keeps at least some resistance between
pin 7 of IC5 and the ADJ pin of IC2 in case both pots R3 and R16 are
adjusted to minimum resistance, so that if the collector/emitter of Q3
conducts heavily (current limit) it can pull down the ADJ pin without
sinking excess current from pin 7 of IC5. The value is somewhat arbitrary,
but 100 ohms conveniently allows exactly -2x the voltage of the reference diode
used in the current limit adjustment circuit to set the output voltage to
0.000v with min resistance of both R3 and R16 (and -2x is easy to obtain with
one op amp section).

R24 and C10 slow down the regulator slightly which damps an otherwise
overshot response for some values of load cap. I added 10uf on
the output to help swamp the low problematic value also instead of 1uf.

I tested my small 'configureable' power supply i built a short while ago
(LM317 regulator) and i found at 1.4 volts the output changes about 15mv
from no load to 1.2 amps load. This sounds a little higher than i wanted,
but i think it's because i didnt pay close attention to wiring the circuit,
using #24 gauge wire too /ubbthreads/images/graemlins/smile.gif
Interesting, the output INCREASES by 15mv!

BTW, i got your shipment today! Thanks very much.
You're very generous Vic!
More in a PM to you in a minute...


Take care,
Al

 

[PEU]

Hi Al, please let us know when the project reaches FINAL status, I have all the samples arriving on monday, and I still have time to order some difficult parts from digikey, such as multiturn pots etc.

/ubbthreads/images/graemlins/popcorn.gif


Pablo

 

[vicbin]

Hi Pablo,

I will try this coming weekend to breadboard the circuit on the linear part and post the result here. Wish me luck ! /ubbthreads/images/graemlins/smile.gif

Update:
How much do you pay for that multiturn pot ? Is that included the low 100 Ohm ? ...sigh...I still can't find it here ! /ubbthreads/images/graemlins/crazy.gif

Vic

 

[MrAl]

Hello again,

Pablo:
We're really close now, and the pot values shouldnt
have to change anyway. Keep in mind however that this
circuit should be built up and tested at least once
to verify everything works completely as expected.
Normally before many circuits would be built and sold
at least one breadboard is built and tested completely.

Vic:
I cant wait to hear how you make out with the linear part.
Guess what else? Out of all the parts i have, i had
only three electrolytic caps of any significant uf values,
but i was lucky enough to find an assortment of them
at a local RS for about $4. There were a few 1000uf and
two 3300uf of at least 25v, so ill be able to test the
switcher part very soon, complete with your toroid core.
Too bad the esr is probably higher than the uf value /ubbthreads/images/graemlins/smile.gif
but perhaps they will work out good enough anyway.
Im wondering if we can use one of those smaller
toroids for the post switcher filter...it might
be possible. I noticed even 22uH and 100uf knocks the
ripple down quite a bit (simulation). The decrease has
a lot to do with the esr of that cap though.

Oh BTW, if you use 2k for the pot and 500 ohm for the
fine pot you'll get just what you need: 2.5k total !
No need for 2.5k for you then.


Notes on the adjustment pot's "turnage":

I currently use a high quality 10 turn Bournes pot
of 1k value on my simple LM317 ps, and that gives me about
1.23 volts to 7 volts adjustment range, but even with
the ten turn pot it's difficult to adjust right to the
millivolt...for example say i want to set it at 1.400
volts, it's kinda hard to adjust...might be 1.399 or 1.401
...although it is possible... but it would be nicer if a little
better than that. Because we'll be using 2k to adjust
for 0-20 volts, that will be a little harder to adjust
(right to the millivolt) even with a 10 turn pot.
If we want better, perhaps a 2k coarse pot, a 500 ohm
fine pot, and a 25 ohm super fine pot? It's up to you,
the third pot would be wired in series with the other
two.


Take care for now,
Al

 

[PEU]

is this capacitor enough for Cin ? /ubbthreads/images/graemlins/crackup.gif

/ubbthreads/images/graemlins/popcorn.gif

Pablo

 

[MrAl]

Hello again,

Pablo:
WOW! Where can i get a couple of those nice caps?
I need a good filter cap for my AAA flashlight /ubbthreads/images/graemlins/smile.gif

Vic:
Some less than good news...the diode you sent CSD04060
isnt a good idea for the simple switcher because it
has such a high forward voltage drop and probably will
keep the switch transistor (internal to the LM2576)
on during times when it's supposed to be off.
This probably would result in overheating of the LM2586.
With my app (1.5A) i might be able to get away with it,
but at 3 amps output i really wouldnt use this diode.
1N5822 would be better, or a little higher current rated
Schottky. This diode looks sorta like three 1N5822 diodes
in series, so the voltage drop is triple.

BTW, im getting ready to build up the switcher now so i can
run some tests on the big toroid you sent.
I've also looked at a few numbers and it looks like we
might even get away with a lower value inductor, maybe
around 100uH. I'll post more as soon as it's built,
and i've already got the little pc board drilled out
to fit the LM2576 with it's odd pin spacing (non 0.1 inch).
Im planning to try a ready made (and easily obtainable)
inductor also that i already have on hand for the lighter
current output apps.

I was also giving some thought to winding a special
inductor. Heck, if we have a nice size toroid like
this to work with we can easily wind a secondary onto
the inductor and use it as a transformer to obtain
all sorts of interesting (and isolated) power supply
outputs:
Negative of almost any voltage level
Positive double with half output current
Positive half with twice output current
Multiple windings for several outputs

Since they can all would be isolated, this would be
very useful for supplies that really have to be isolated
(like the panel meter dc supply).
And so few parts makes this look pretty attractive...
Two caps,
two 1/4 watt resistors,
one low cost Schottky,
one inductor (with or without secondary windings)
one LM2576 TO220-5 package.
All fits on one small pc board.

Sound interesting?


Take care,
Al

 

[PEU]

My vote goes to build the PS and then start with this multiple windings toroid.

Look, I finally received all the samples/stuff I purchased in the previous 1 1/2 month !!!

/ubbthreads/images/graemlins/popcorn.gif


Pablo

 

[MrAl]

Hi Pablo,

WOW, all them parts /ubbthreads/images/graemlins/smile.gif

Can i ask what you're planning for all those SCR's and
power transistors?
I see you're all ready to go to the LM1084 LDO too.
Plenty of LM2576's too i see
How many power supplies do you plan on building?

I cant guess what that is because i cant quite see it.

Also, yes the multiple windings thing is just an idea
for the future. I agree let's get the main board going
first.

Take care,
Al

 

[PEU]

Hi Al,

I only plan to build one PS, these power transistors & SCRs were fillers /ubbthreads/images/graemlins/smile.gif for the 33063/34063 samples I ordered from Onsemi, also not detailed (and free too) in the picture are 25x npn smd transistors and 25x 2A smd shottky diodes.

If you have some ideas on how they may be used, I'm all ears /ubbthreads/images/graemlins/smile.gif

Regarding the toroid I have, here is a picture:

Outside diameter is 27mm and inside diameter is 14.3mm. It seems to be standard for PC AT Power suplies (I checked 3 and all of them have this one)

Pablo

 

[MrAl]

Hi there Pablo,

Fillers? What do you mean?
What are the 33063/34063 items?

Well, what are the part numbers of the npn smd transistors?
The 2A schottkys part numbers? Maybe for approx 2A switcher supplies...
such as running an LS or other LEDs on a 12v automobile electrical
system...ideal.
Any other kinds of regulator circuits you need?

Regarding the switcher for the PSw/SPR (this thread)...
Im going to have to use a lower current diode (1A) to
start if the big diode Vic sent keeps the switch transistor
'on' (transistor in the LM2576 package), at least until i get
my other parts (very soon now) for this thing.
I'd like to be able to test up to the full 3 amps at least however,
so one diode or the other will have to work. The 1N5822 diodes are
really cheap, luckily /ubbthreads/images/graemlins/smile.gif Just a matter of getting some.
Maybe i could check the NTE site too, as there is a shop near here
that carries some of their parts. Heck, even 1$ for one diode would
be ok since i only really need one for now.

Is that the toroid you plan to use with this ps? If so, what is the
height of the toroid? You already said 27mm OD, 14mm ID, right?
Do you have any idea what kind of material it is made out of?

Take care,
Al

ADDED A FEW MINUTES LATER:
Called the shop, they have the diodes but they are
$1.85 each !! He he, geeze. I guess they're in a
high rent district /ubbthreads/images/graemlins/smile.gif I only need one for now
anyway though...

 

[PEU]

Hi Al,

The 33063/33064 are MC33063 and MC33064 regulators (smd), the schotkys are smd 40v 2A (onsemi ss24t3), the npns are smd 30v 2a switching (onsemi mmbt489lt1), when I said filler I mean, they charge you $11 for 10 sample items, so I "filled" the order with these. All the smd regs are for building lambda ill-pill regulators, the npn&schotkys are for unregulated step-ups.

Back on the main track: The height of the toroid is 11.3mm, I don't know the specs for the core, but they are pretty common in PC-AT power supplies.


Pablo