Constant current source

Go To Last Post
583 posts / 0 new

Pages

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I was just wondering if it is possible to use pwm and a constant current at the same time.
Leds have specific emmission caracteristics that are only garanteed at the normal rated current.

So I wonder if it is possible to constant current and still dim a led without affecting it's wavelength.

Chris.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you want true color, you need to PWM the current source, not change the current the source regulates at.

The PWM dimming frequency can be way lower, a few hunderd hertz.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Does the wavelength change when a led is dimmed ?

This design uses PWM to produce a DC-current of a pre-defined value.
I did something with an AVR as controller .... should be at the pre-mentioned webpage ..... eh .... http://www.aplomb.nl/TechStuff/NixNies/Switcher.html .... yes, this one

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wavelength is current dependant yes ;)

Ut's necessary to turn on and off the current source quickly ;)

For regular lighting applications I don't think the wavelength shift is a problem; it's more a problem for backlights for LCDs where RGB leds are used and where colour temperature is important.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Two pictures of scope images, with the small I-shape core (upper left)

Upper trace is Fet-gate, 10V/div
Lower trace Fet-source-current, no scale ( I'm measuring over 30 cm of non-inductive wire ;) )

The sat-one is @ 400 mA in the load, the non-sat one with 190 mA in the load

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I put back the toroid, and reduced it's inductance to 26 uH. But it didn't help a lot in increasing the switching frequency. The resistance of the supply-wires and Amp-meter are playing a significant role. Lab supplies deliver 5.05 V, and see what's left ...

Edit: tested with shorter leads, no Amp-meter, dual lab supply @ 6V ---> 14.69V output @ 1 Amp :!:
20 us on, 15 us off

Edit 2: pushing the limits: 1.2 A out @ 14.8 V out ..... using 6V dual supply

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Supertex inc recommend PWM for Led dimming.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Nard

I gone to bed at the time you replied.
Jayjay1974 answered correctly & you passed the me think. :wink:

The graph pictures above is fine & make sense now.
I went to your link & just remembered that I have been there a couple of times & forgot about it.
It's when the thread on Nixie Tube was running on this forum. Nice work.

Cheers
Ken

Hi Nard

I gone to bed at the time you replied.
Jayjay1974 answered correctly & you passed the me think. Wink

The graph pictures above is fine & make sense now.
I went to your link & just remembered that I have been there a couple of times & forgot about it.
It's when the thread on Nixie Tube was running on this forum. Nice work.

Cheers
Ken

Edit 1.
Damm it, I didn't read page 16 of this thread & was responding to page 15.
The pictures from your CRO make sense now.(Thanks)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I gone to bed at the time you replied.
That's what happens, .... being at the other side of the globe :)
Quote:
The pictures from your CRO make sense now.(Thanks)
You're welcome.

A final test tomorrow on 4*NiMH 2000mAh cells .... just to make this story complete. Then the final schematic, and a review. I cleared the desk out already.

When Ossi is back from holydays, I wanna have a look at his auto-transformer design as well. AVR-based of course. Enough old inductors have taken a swim. The latest ones came out better. Getting experienced, I guess.

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi folks,

back again,and really nice brown,
weather at dutch coast was really
good these few days!

Have also seen that this thread is still growing,
very good.

Lets see what happens when I restart my
solder-iron.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It gets hot :P

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If german school-holidays are over,
the durch turn on the sun to full power !

So, perhaps I should turn my push-pull-converter
to full power until it begins to give the same
nice heat as the sun.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey Ossi

You came back too soon: saterday and sunday will be nice and warm summerdays.
But hehe, you've got the tan you wanted :)
I am running the final test now on 4 NiMH's. The problem I expected showed real soon: a "not-so-good" contact in the battery-holder. WACKELCONTACTE :!:
So I juggled the batteries, and that helped. Gotta go back upstairs .... watch the test

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
It gets hot Razz
:lol: ..... smartass :lol:

The test is done.

Outputvoltage was 13.88V @ 488 mA
Batteries came fresh from the charger, 2100 mAh. New one's.

In the plot: the disturbance in the curve was due to Wackelkontakte in the battery-holder.

I posted the datafile as .txt but it is a comma separated datafile

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I wasn't far off with my BPF of 45 minutes 8)

Do them batteries get hot?

The question remains... is a battery life of slighty more than half an hour acceptable?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, they get warm. But I don't know how it will turn out in real life as I had a fan running to cool the dummyload ..... and it cooled the batteries too. Bit in spite of the cooling .... the batteries get warm ....

IMO 33 minutes is short .... but it's indeed what we could expect. I would never go for 4 NiMH's penlights (AA)

Edit: added picture of setup

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think the next pages of this thread will cover the topics:

"NiMH batteries" and "Wackelkontakte"

both are very interesting and rewarding !

Here a link that gives a kind of "discharge data"

http://www.sanyo-component.com/f...(2100mAh).pdf

I just put 4 old NiMHs into a holder and connected
them to my converter. The input voltage is below 3 volts
and up to now I have not found some where I can get to
4 Volts. I am just charging a few...

Perhaps I built in too many Wackelkontakte......

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

:)

I used new batteries .... so they were not yet on their capacity. They need a few cycles for that. But even then: the discharge current for 4 luxeon leds is high compared to their capacity.
I dismantled the test setup .... time for something new. Elsie f.i.

I also wanna build a converter as you suggested Ossi: the auto transformer type. Correct me if I'm wrong, but I don't see much gain in a push-pull. I'd need 4 fet's.

:?: Who can explain me what the function is for the toroïd in the PC-PS's, in the secondary circuit. Will add a picture ... it's on the other PC. It's part of the output-filter, but what is the smart thing in coupling all outputs magnetically ?

Nard

Attachment(s): 

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A push pull needs two FETs and a center tapped primary winding ;)

A full bridge converter needs four FETs (basically an H-bridge) but is only used for KW range convertors.

For this application, only the simple single inductor boost convertor is the sensible option. All the extra stuff with transformers, FETs only add losses, and with the very moderate power (only 15W) make up for a big part.

What's this autotransformer? I know what it is ;) but don't see how it should work in this app.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Nard

The multi-winding inductor on PC-supplies secondary
has something to do with so called cross-regulation.
(Getting many outputs stable if only one is controlled.)

As far as I remember the idea was quit tricky.
I think there should be an application-note
from UNITRODE (I think belongs to Texas Instruments now)
that describes it. Google for "cross-regulation" or "multi-output" supplies.

I will look also later.
edited: later found that:

http://focus.ti.com/docs/trainin...

http://www-s.ti.com/sc/techlit/s...

http://ece-www.colorado.edu/~mak...

@jayjay

I still like the push-pull. In contrast to the
single inductor boost the peak current in all
components in the push-pull is smaller, that
is not so bad with respect to losses.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It isn't that I dislike the push pull :D It's just that I think that a single inductor is simply the best solution ;) I do like (over)complex circuits though, and I have simulated a few rather complex setups, only to find that the efficiency at these power levels is ghastly :D (Which maybe can also tributed to the fact that simulating transformers isn't easy).

Maybe I'll try a resonant push pull converter, I've been playing with P/P transformers lately anyway.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wow ! That coupled filter for cross-regulation is like magic .... :) Thanks Ossi, for the links. Gotta lot to read .... will make me feel like studying black magic :mrgreen:

As far as the name-giving is concerned: there is apparently consensus about that, but IMHO (yes, this time HUMBLE), I find it confusing. Take an audio-amplifier: we call it a push-pull output with one transistor (fet, valve) in the high end, and one in the low end. For car-audio with a bit more juice: take two push-pulls, and feed one of them with the inverse of the other. That's what I would call a bridge amplifier. Delivers 4 times the power than possible with a single push pull.

Why am I writing this?? Dunno ..... just confused. I will make a note of it (on paper) and hang it over my bedsite :lol:

I think that 80% efficiency for the single inductor is pretty good, considering the low battery-voltage. What I like about the push-pull: it doesn't store energy in the core .... it's a plain transformer. I think it will generate less EMI than the single inductor-one. And more modest currents.

Resonant Push-Pull ..... hmmm, it's getting better all the time ...
Is that the principle where the building of the magnetic field generates the drive for the driver ? And when the core saturates, the build-up drops, thus turning the driver off, field collapses, generates negative voltage ... till all energy is gone and the whole story repeats itself? Neat .... but will be hard to simulate :shock:

Haven't seen Orc for a while .... must be busy.
Ross (Valusoft) is building a converter too. Now this IS a monsterthread, and till the end may take a while.

17 pages .....

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What you describe is ferroresonant AFAIK ;)

A (Quasi) resonant switcher includes a resonance tank (LC tank) which allows to switch during zero current, or zero voltage point which reduces stress. Also, the output is regulated by changing the frequency. It's also voltage and current limited without any electronic means. The magic keyword to google for are ZVS and ZCS.

Another interesting source of information are patents. Difficult to read and to find what you're looking for, but from time to time you can find very nice idea's in them.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hehey!

Quote:
Haven't seen Orc for a while .... must be busy.

Indeed I have been busy. I got a new job this month and have made 11 hour shifts this week, so not much time to spare ;) But I do have access to some incredible stuff now ;) (like CNC, soldering oven, lead-free solder (I'm starting to like it :() etc. Also found a guy to make me an enclosure for free ;) (jeei).

Now that this is sorted:

Thanks Nard for testing the circuit. 33minutes seems... a bit low. I was hoping for something closer to 1h mark. Something like 55 minutes maybe? (92% efficiency)
I have to test that ossis schematic (https://www.avrfreaks.net/modules...). I'll see, if any shops are open tomorrow (yes I STILL haven't gone shopping :() I would really like to get that done. I have so many projects hung up behind that and new ones building up :( But I have no-one else but myself to blame (sadly ;)). Oh well.

Cheers,
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@nard,
just to push (or pull) you to the limits:

There are also so called "half-bridge" converters.
similar to "audio-push-pull" namely:

One controlled switch from midpoint to ground.
One controlled switch from +input to midpoint.
Output filter from midpoint to somewhere.

Now replace the lower active-switch with a diode
and use a inductor as output filter:
You get a down-converter (buck-converter)
(Or, for german-language-lovers Tiefsetzsteller)

Now interchange input and output:
You get a boost converter. (German: Hochsetzsteller)

So, some people may say that your boost circuit
is a half-bridge-circuit.

And if you say a power-converter-halfbridge is
an audio push-pull then you are already building
a push-pull !

Its a confusing world.
or maybe the sun burnt my brain too hard.

To be more on-topic:
I just integrated a current-sense resistor ( 0.2 Ohm)
into my converter and connected it to the ATmega88
ADC ( 1.1Volt reference) and closed the loop. So my
converter starts to get regulated.

Here something about ferroresonant-transformers:

http://www.allaboutcircuits.com/...

I actually own a ferroresonant circuit, built around
1960, but the capacitor is blown.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Rain wrote:

Quote:
I got a new job this month and have made 11 hour shifts this week
And .... happy with it ? Something in the back of my mind whispers ( haha ) you got a software-job. Making money when free from school .... I used to do that too.

Quote:
Thanks Nard for testing the circuit. 33minutes seems... a bit low. I was hoping for something closer to 1h mark. Something like 55 minutes maybe? (92% efficiency)
No way. The 33 minutes is not because of poor efficiency of the converter, it's due to "just 4 NiMH AA-cells". So if you want to stretch the uptime to 1 hour, go to 6 or 7 cells. Almost 80% is very good (I need to say this .... because I worked hard for it)

Ossi, ..... Your push-pull-story is nice. Now it's not just a note at my bedsite .... I need a large Whiteboard :lol:

About the German words: I still use: Soll-wert und Ist-wert. Even in English texts. Yes, the German expressions have a clear soundbite. Btw, I grew up in Nijmegen (Nimwegen), and we could receive the German TV-stations ARD, ZDF and later even WDR. Long before cable came, let alone satelite ;) So in some sense, I grew up bi-langual. Hardly speak foreign languages, and they fade ....

The ferroresonant-transformers: Philps made those. Very rugged. Great for on-shore use, lousy for off-shore. So on a survey-ship we used a special version, also built by Philips: That rectified the ship's power, and used a 50 Hz square inverter that fed ..... a ferroresonant-transformer. Simple, reliable, not very efficient, true, but once installed: problems over. And turning that into a no-break-set was a POC. Lovely times.

In an earlier post I mentioned that I would do your auto-transformer-push-pull-super-bridged-design as well ..... and as I cleared out the boost-converter, there is again some breathing-room. OTOH, I want Elsie fixed too. And it's gonna be summer .... so it might turn out to: garden, white-beer ...

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Who or what is Elsie? :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Don't you know Elsie ????? Nice girl :!:

(Look in the Mylar-thread .... L C .. )

Nard :)

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's a LCR meter :D

Want that book Art of Electronics?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Resistance is futile…… You will be compiled!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Plons!
Do you have many cutting wires?
http://www.lobzik.pri.ee/modules...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My converter is getting back onto the AVR track now !

Software regulates output-current and displays
input and output voltage via RS232.

In the inductor/transformer design I also
used an "Elsie". Its very valuable !
I will make a comment to the MYLAR thread now.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey

Quote:
And .... happy with it ? Something in the back of my mind whispers ( haha ) you got a software-job. Making money when free from school .... I used to do that too.

Nope, I'm doing nothing with software. I'm on the lowest rank.I'm just soldering ;).Actually I do everything I'm told to do. soldering, testing, programming, machinery etc.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Still doesn't sound too bad, Rain. You're at least in electronics. I will not sum-up all the jobs I did to make some money during holidays. Makes me feel old :)

@AllN: indeed, that's the Elsie I mentioned.

@Zauberer: thanks. Now I know what more I can do with the wire removed from old cores ;)

@Ossi: hehe, that's nice. Curious about the efficiency .... I will rename the Mylar-thread, as it's basically about the LC-meter.

@Jayjay: I saw the book "the Art of Electronics" at ComputerCollectief in A'dam. € 89 IIRC. If you have a better offer, I'd like to hear that :)

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Now that I'm making a list, what I must buy to make some of the designs mentioned here (actually All of them I think) I have come across a small problem.
Ossis design (https://www.avrfreaks.net/modules...) with transformers. It needs 2 non overlapping PWMs right? Ossi, you mentioned, then you are using M88 to control it,right. How are you creating those two signals? I know some of the AVRs have such a mode (non overlapping PWM),but M88 doesn't (at least I can't find it). So how are you doing this? using software PWM? I'm trying to find a suitable tiny to do the job there (2PWM's or 1 with this non overlapping feature) and a few ADCs. I'm currently considering Tiny13.

Any suggestions about that?

Currently in my shopping list:
2x IRLR2703
1x BD140
2x LM393N

Cheers,
Rain

PS I read the previous pages and all of you wanted to get past 17. Now we just need a few more posts and we are there;)

PS2 is there a list of the longest threads in AVRFreaks history?

EDIT:
I found, that Tiny45 has non overlapping PWM mode. I think I'm going to use this. But I'm still curious on how did you, Ossi, solve it?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Use a timer with two PWM outputs, set one to clear on compare match, the other to set on compare match. One with a slightly higher compare value. Fast PWM mode.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Her some info how I do pwm for push-pull:

first:

Initialisation


#define PWMTOP 100
#define DUTYMAX 49
#define DUTYMIN 2

uint8_t DutyCycle ;


// Timer 1 in Mode 8:
// phase and frequency correct PWM
// Top=ICR1
// Output OC1A clear on compare when up-counting
// Output OC1B set   on compare when up-counting
// clock-source: XTAL/1

  TCCR1A=0b10110000 ; 
  TCCR1B=0b00010001	; 
  ICR1=PWMTOP ; 

  DutyCycle=25 ; // initialize to some value

  OCR1A=DutyCycle ; 
  OCR1B=PWMTOP-DutyCycle ; 

In the actual control-loop I do
the following:
(Its only a first shot, but works.)


  if ( Iout<120 ) { DutyCycle++ ; } else { DutyCycle-- ; }
  if ( DutyCycle < DUTYMIN ) { DutyCycle=DUTYMIN ; }
  if ( DutyCycle > DUTYMAX ) { DutyCycle=DUTYMAX ; } 
  OCR1A=DutyCycle ; 
  OCR1B=PWMTOP-DutyCycle ; 

Wit PWMTOP=100 and 16MHz XTAL I get
16MHZ/(2*100)=80kHz PWM frequency.

@jayjay:
Do you think you are able to predict
efficiency of my push-pull design ?
If you need component values, I could
give you what I know about my components.

@nard
Probably in a few hours I will give the measurement
data of efficiency together with some calculation
where the losses occur. Takes a little bit
of time to write it up.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks. But what DutyCycle did you use on the first try (without the control loop)?

Thanks
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Efficiency will be between 10% and 95%.

:)

Difficult to predict I'm afraid, a matter of finding the sweetpoint of switching frequency, FET type, transformer design, load current etc.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@bloody-orc

I think I used something like
DutyCycle=40

I just entered this value into my code and measured output:

The PWM cycle time is 12.5 us (means 80kHz).
The on-Time (High-time) for the outputs ist
(with DutyCycle=40) 5us.

A half-cycle lasts for 12.5us/2=6.25us and
the on time is 5 us so that means (on the
secondary side) a duty-cycle of 5/6.25=0.8.

In the test-phase the ATmega was connected to
my STK500 and I changed DutyCycle by changing
code and reprogramming.

@jayjay
If I give you compnent data and so on, how
precise do you beleive can you predict efficiency
(using simulation or whatsoever ?)

Warning:
I ask that, because I think its
very difficult to get good predictions
(efficiency prediction with 1 percent accuracy)
in this area without excellent component models
and much experience.

Edited:
Wow, we made it to page 18, and still beeing
on-topic for most of the time !

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

I ask that, because I think its
very difficult to get good predictions
(efficiency prediction with 1 percent accuracy)
in this area without excellent component models
and much experience.

Exactly ;) So I won't even bother because it will end up anywhere between 70 and 90%, and as such isn't of much use :(

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jayjay wrote:

Quote:
Use a timer with two PWM outputs, set one to clear on compare match, the other to set on compare match. One with a slightly higher compare value. Fast PWM mode.
Are you sure about Fast PWM ? I think it is freq. and phase correct pwm.

I will leave this nice work for you guys .... I need to go back to bed: ill.

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@nard
Hope you recover soon. Best wishes.

Here results I promised:
So, here are some results about my push-pull converter
in its current state.

The converter is now self-contained and all power is
drawn from one source. The AVR (16MHz) controls the output current.

The following measurement is done with a transformer made as
follows:

Core:

toroid core 16mm outer diameter, material N27
Winding: 12+6+6+12 turns with 0.5mm diameter copper

Now measurement results

Output: 13.05 Volts into 25 Ohm load
--> Output power is 6.81 Watts

Input: 4.34 Volts at 1.73 A
--> Input power is 7.51 Watts
--> Efficiency is 6.81/7.51=91 percent
--> power loss is (7.51-6.81)Watts= 700mW

(With 12mm diameter core, efficiency drops to 89 percent.)

Now I will try to provide some information where these losses
may come from. The analysis is guided by the following:

Quote:

The Unknown:

As we know,
There are known knowns.
There are things we know we know.
We also know
There are known unknowns.
That is to say
We know there are some things
We do not know.
But there are also unknown unknowns,
The ones we don't know
We don't know.

—Feb. 12, 2002, Department of Defense news briefing
D.H. Rumsfeld

Things we know:
---------------
Transistor used is IRLR2705 with on-Resistance of 0.04 Ohm
Frequency is 80kHz, its a push-pull with high duty-cycle.
Transformer winding is 12+6+6+12.

Now the unknowns:
-----------------

Thats things we don't know, but can measure
or compute .
(see above) In the analysis I use (for simplicty)
Input-Current Iin=1.8A
Output-Current Is=Ip/4 = 0.45A
Duty-Cycle 100% (so always exactly one transistor an diode conducting).

6 Windings on my transformer have a resistance of about 0.013 Ohm.
(computed from wiredata and length-of-wire measurement).
Primary winding carries approximatly a current of 3/4 Iin and has 6 turns.
--> primary transformer resistive losses approx: 25mW

Secondary winding carries approximatly a current of 1/4 Iin and has 6+12 turns.
--> secondary transformer resistive losses approx: 10mW

Transistor carres approximately a current of 3/4 Iin
--> resistive Transistor losses approx. 75mW

The diode have a drop of around 0.35 Volts and carry current of Iout=Iin/4.
--> diode-on-losses approx. 200mW

The smoothing inductor has a DC-Resistance of about 0.1 Ohm (measured)
and carries current of Iout=Iin/4.
--> resistive inductor losses approx. 25mW

Controlling AVR controller draws 18mA (measured) from Uin.
--> controller loses approx 75mW

Current measuring shunt resistor has 0.25Ohm
and carries current of Iout=Iin/4.
--> sense resistor losses approx. 75mW

So uo to now we could identify ( 25+10+75+200+25+75+75 ) mW
= 485 mW of losses that we know quite good.
But lets go on:

But there are also unknown unknowns,
------------------------------------
The ones we don't know
----------------------

Switching losses in Transistors and diodes.
AC losses in transformer and inductor.
AC losses in input electrolytic capacitor.
Losses due to magnetizing current in Transformer.

But the above calculation shows, that all these unknown losses will
probably not add up zo more than 700mW (Known Loss)-480mW (Identified Loss)
=220 mW unidentified loss.

That all I can say about it....
Comments welcome......

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You have basically loses in every component :D

But FET and transformer are the biggest ones. Also capacitors with a lot of ESR will be a loss of power. Best is to use ceramic X5R high capacitance caps.

But the efficiency you've got now is quite excellent, 91%. Squeezing out more efficiency will be progressively harder :(

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for all the testing. I'm going to test that last circuit with LED's in the coming week (hopefully). But on the toroidal core. you said, that 16mm outer diameter is good, but what if it is bigger? will it be better or worse? (smaller is worse as you said). Also, does the height of the toroid pay very big role? And maybe you could tell me the color of the toroid (they seem to be color coded by material).

Thanks again

Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Its more the cross-section of the
toroid thats important. The smaller one
(12 mm outer diameter) has a cross-section
of about 3x5mm^2=15mm^2 (and I used some more
windings in order to compensate for that) and
the larger one (16mm outer diameter) has
a cross section of about 5x6.5mm^2=32.5mm^2.

My loss-estimate shows, that there are probably
not too many losses in the transformer at this
size. So making it bigger will not of too much gain.

My core is black, but that means nothing. I
have a lot of cores with the same color, and very
different materials or so.

I think color kan only be used if you already know
the manufacturer. In my case its a core I ordered
some time ao, and therefore I know its material
(N27). Perhaps I may find a datasheet.

But you should take all the efficiency-issues
no too serious. Its just a sport. Increasing
the efficiency from 80 Percent to 90 percent
can be a hard job, and it does not mean anything
for the operating-time of your product.

I think other aspects (easy to get components,
robust design and so on) may be more important
for you.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks
I was asking because, if I can't find that exact size core,then I was wondering if I should take a bit bigger or smaller one ;)

I was thinking of using IRLR2703 FETs and self made inductors; maybe for the final design I'll buy the 68uH one. The diodes are the one I have mentioned before (from PC PSU). Sadly I don't remember the name of those and I don't have them near me. (maybe in the evening). For driving I'm going to use tiny45 with non overlapping PWM mode (easier and "safer" than using 2 channels). It also has a nice ADC I can use for current measurement (But first I have to figure out how to do that, as I have never done it before). I imagine it can't be that hard: just measure voltage drop across a resistor and calculate the current using Ohms law; right?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

bloody-orc wrote:

Thanks Nard for testing the circuit. 33minutes seems... a bit low. I was hoping for something closer to 1h mark. Something like 55 minutes maybe? (92% efficiency)

Cheers,
Rain

Hi Rain and the other team members,

Can I have a reality check please? If Nard's 80% efficiency produced 33 minutes, how do you get 92% efficiency being 55 minutes? I would have thought that 100% would be 33 x 1/.8 = 41.25 minutes. Have I missed something?

Regards,

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey Ross,
Well I have no idea actually;) I don't know where did I get those numbers, it was late and I probably missed some keys on my keyboard.

But something on a brighter note: I found a new shop in my town. 68uH choke (smd) 1.2A 0.2Ohm and just 0.7€!. also M16 for 3.8€. Amazing! ;)
I know this 0.2Ohms in that choke is less than ideal, but I can use that right?

Cheers,
Rain

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Battery capacity depends on the load current, with a more efficiency supply the current drawn from the battery is less, and as such more capacity. Less is more :D

How pronounced this effect is, I don't know.

Pages

Topic locked