Efficiencies of Switching Circuits....

[ViReN]

Hello All,

Its time to talk about Efficiencies....

First, I dont know, if this method is correct or not... What i did is...

Hooked up the things (Transistor based circuit.. that was built previously.. with optimized values (for my purpose) ... DMM etc...)... (with things running) ... at what ever (an instance of) the battery condition....
Checked I/P & O/P Voltages (in my case 2.63, 8.79) Volts
Checked I/P & O/P Currents (in my case 150, 35) mA's
(these things changed as the battery looses its *juce*)

So,
Input Power = 2.63 * 150 = 394.5 VA/1000
Output Power = 8.79 * 35 = 307.65 VA/1000

So O/P Power (Devided By) I/P Power = 307.65 / 394.5 = 77.98 something % Approx 78%

Is this correct way ? ....

To add further to this.. I have a query... What kind of efficiencies will Zetex and other circuits give ?

What kind of Efficiency should i expect from various circuits... does any one have any data around ?

I also understand that these efficiencies change with change in I/P Voltage & Current (depending upon the battery condition) ... so it is a chart kind of thing .... It is also likely that Manufacturers quote *UpTo* XX% kind of things, which might be highly missssssleading ... especially when it comes to purposes for running higher loads... at lower i/p voltages... (this much i have understood till now) ...But how good are those XX Numbers ?


Does any body have comparitive efficiencies (typical as well as specific conditions) ... and what kind of efficiencies should i expect.. from a zetex circuits.... Where are the most of losses occuring ? (is the Inductor a Culprit) ... or its the Voltage drop at *diode* that accounts to it ? ... (it was .6 volts that i measured at DMM (i wonder if this reading is correct or taken correctly))

Thanks in Advance,
ViReN

 

[Doug S]

[ QUOTE ]
ViReN said:
What kind of Efficiency should i expect from various

[/ QUOTE ]

Here are a couple of threads that will interest you. BTW, the ArcMania circuit is a Zetex based design:

ArcMania and 1AAA Dorcy data
ARC AAA data

 

[ViReN]

Great Threads Doug /ubbthreads/images/graemlins/smile.gif

Thanks for the information shared...

-ViReN

 

[Steve K]

another quick thought: if you want to use a DMM to measure voltage and current, the DMM will need to measure the true RMS value of the waveform. Most meters are designed with the assumption that voltages are either dc or sinusoidal. If there is much AC on the signal, the displayed value may be off by a significant amount.

Also be aware that efficiency varies with any change in the load or power source. A review of the datasheet for any switching regulator will give an indication of efficiency under various conditions.

have fun!

Steve K.

 

[ViReN]

[ QUOTE ]
ViReN said:
Hello All,

Its time to talk about Efficiencies....

First, I dont know, if this method is correct or not... What i did is...

Hooked up the things (Transistor based circuit.. that was built previously.. with optimized values (for my purpose) ... DMM etc...)... (with things running) ... at what ever (an instance of) the battery condition....
Checked I/P & O/P Voltages (in my case 2.63, 8.79) Volts
Checked I/P & O/P Currents (in my case 150, 35) mA's
(these things changed as the battery looses its *juice*)

So,
Input Power = 2.63 * 150 = 394.5 VA/1000
Output Power = 8.79 * 35 = 307.65 VA/1000

So O/P Power (Devided By) I/P Power = 307.65 / 394.5 = 77.98 something % Approx 78%

Is this correct way ? ....

To add further to this.. I have a query... What kind of efficiencies will Zetex and other circuits give ?

What kind of Efficiency should i expect from various circuits... does any one have any data around ?

I also understand that these efficiencies change with change in I/P Voltage & Current (depending upon the battery condition) ... so it is a chart kind of thing .... It is also likely that Manufacturers quote *UpTo* XX% kind of things, which might be highly missssssleading ... especially when it comes to purposes for running higher loads... at lower i/p voltages... (this much i have understood till now) ...But how good are those XX Numbers ?


Does any body have comparitive efficiencies (typical as well as specific conditions) ... and what kind of efficiencies should i expect.. from a zetex circuits.... Where are the most of losses occuring ? (is the Inductor a Culprit) ... or its the Voltage drop at *diode* that accounts to it ? ... (it was .6 volts that i measured at DMM (i wonder if this reading is correct or taken correctly))

Thanks in Advance,
ViReN

[/ QUOTE ]

FINALLY !!!... I found it...

Checked I/P & O/P Currents (in my case 150, 35) mA's .. its WRONG...... It was a Mistake on My end.... for measuring Currents.. i had not choosed the proper switch position on the meter...

The Input is about 175 mA and output is about 32 mA
So,
Input Power = 2.63 * 175 = 460.25 VA/1000
Output Power = 8.79 * 32 = 281.28 VA/1000

So the Efficiency is 61.11% and NOT 77.98%

I am extreemly sorry for putting up wrong figures... See... I am a newbie... (when it comes to measurements in the circuits)... and i m still learning... Hope you will forgive me for the mistake....:(

Earlier.. i was very happy after seeing 78% of efficiency... later on while playing with the circuits... i got much less... even as worse as 47 % in some configurations... that really lead me to recheck the meter readings... and thats how i found out that it was wrong...

Hope you will understand this... and forgive this newbie for the mistake...


ViReN

 

[unclearty]

Forgive??? What's to forgive? Isn't this what we're all about? I, for one, have been reading ALL your posts with great interest. I find it a great deal of fun to discuss experimentation with other "newbies". Forgive???...I eagerly await watching everyone tinker with their projects and giving us their results. The fact that you continued tinkering with your findings is great. I'm enjoying going home and tinkering with the schematics you've presented.

 

[ViReN]

/ubbthreads/images/graemlins/smile.gif Thanks unclearty ...

I do post each of my experiance here... so that some one who reads it... learns it from my experiance... and wont commit the mistakes that i have done...

I have learnt many things in life (and in CPF) the hard way only /ubbthreads/images/graemlins/wink.gif ... like the blowing off of LED's /ubbthreads/images/graemlins/cool.gif /ubbthreads/images/graemlins/grin.gif with the same switching circuit ...

ViReN

 

[evan9162]

Heh. I say you haven't hit the bleeding edge of experimenting unless you've blown up a few LEDs. I had a bad run and fried two luxeons in a row one day - 12V + DMM probe shorting the circuit = popped luxeon /ubbthreads/images/graemlins/eek.gif

 

[unclearty]

I agree..I had a little medecine bottle on my work bench. Every time I'd cook an led, I'd throw it in the bottle. I just dumped it out last Saturday...it got too full! /ubbthreads/images/graemlins/ohgeez.gif

 

[ViReN]

Oh unclearty, I should also try this thing out... i had just thrown away the blown LED's .... but had maintained a count...

17 Total....
14 mix of Cheap White, Blue, Red, Amber & Green LED's... surprisingly... no UV led is blown yet /ubbthreads/images/graemlins/grin.gif ... and
3... umm... i think... i should keep quiet /ubbthreads/images/graemlins/icon15.gif Nichia's (@$2 a piece + Posting of course)... and I am lucky that I dont even posses any Luxeon's.... Will go on to them ONLY after learning quite a good amount of things... though... there are still chances that Luxeon's would get blowed... coz.. i aint perfect... but atleast... i can have a hope that ... the damage would be less... as compared to having em just yet...

ViReN

 

[evan9162]

Even blowing up a $2 Nichia isn't too bad. But when you pop a luxeon, in the back of your mind a voice goes "well, there goes $10".

 

[ViReN]

Yep... thats true evan....

Thats why.... even after a while now... i havent even thought of getting a luxeon .... not that its too expensive... but i think i am not yet ready for luxeon's yet .... and after all... when you loose luxeon... its for ever... it aint gonna come back... so... why to have it in first case /ubbthreads/images/graemlins/grin.gif ... i will let some more experiance in to myself... so that i will be very very careful about handling Luxeon's .... after all they are the KING of LED's ... and they need to be dealt with respect & caution ... However... after all i too am still a human... and mistakes will occur... and the only thing that i can do is to minimise them /ubbthreads/images/graemlins/smile.gif

I hope... after those Luxeon's... you have not blown any more /ubbthreads/images/graemlins/smile.gif Wish you best of Lux'ing ...

Best Regards,
ViReN

 

[Steve K]

regarding the murder of innocent leds:
when you are building and troubleshooting your circuits, I'd suggest not using the actual led. Instead, use some sort of dummy load. This might be a power zener diode with a zener voltage close to the Vf of the led, or it might just be 2 or 3 1N4001s in series.
Once everything is working just right, then go ahead and install the valuable led!

regarding efficiencies: losses in a switching power supply occur in nearly every component; switch transistors, inductors, caps, etc. Losses can be reduced by using lower resistance (or ESR) parts and by lowering the percent of time spent with the switch transistor operating in the linear region (which usually means a lower switching frequency). All of these measures usually mean that the size of the parts will be increasing!

If the efficiency is low (such as 50%), then you need to go back and see if you wouldn't be as well off with a simple linear regulator or even a simple resistor.
I've been going thru some of these calculations on a home-brew current regulated buck that I'm designing to drive a 3W Luxeon, and if I use a flyback diode instead of a synchronous switch, then the efficiency will be barely better than a linear regulator.

regards,
Steve K.

 

[ViReN]

Steve K,

Great Idea... for emulating the LED's ...

Can you throw some part numbers (for better transistors) / make ?

I have taken care of a couple of things.. just to check in the difference... and that's series resistance of an Inductor.. & reducing the capacatance value... (for higher frequency operation) (it sure does make a difference) ...

For most of the time i am using Boost Regulators (as i have less of voltage at my end) ... ...

ViReN

-ViReN

 

[NewBie]

Hey ViReN

Every converter has a peak efficiency point, and it is often not at the lightest load. The trick is to learn how to consider various aspects of the devices. For example, looking beyod the Rds(on) of a MOSFET, but to consider other characteristics of the device, such as total gate charge (and watch at what voltage it is specified for), recovery time and the amount of charge on the body diode, and trading off the various aspects of the devices, such as these, and the selection of the other components for the frequency you want to run at. The dead time between the two switches can also be a loss factor.

Here is an example of a part that is the most efficient at about 100W, which I utilized for the Thor mod I did.
http://www.linear.com/pdf/1680fa.pdf

Another issue is that not all inductors are created equal. As the switching frequency changes, different types of ferrite/powdered cores have different points at which they work better. Also, many inductors such as the often seen DO1608 and others, at their rated max current, can drop to 70% of their inductance or beyond, some even drop down to 25% of their inductance at their rating. In this case you'll get higher peak currents, and larger swing over the B-H curve (yes, there are losses in the ferrite too), which causes additional loss. As you get into higher and higher frequencies, using multiple strands of fine wire can reduce the losses due to the eddy currents in the copper wire in applications where you have large swings in the inductor field. The the stronger the field the more this becomes an issue. Some inductors will even be wound with "foil" turns, orienting the foil in an axis which reduces the eddy currents induced into the turns. In certain applications this can reduce the losses by 60%, even at frequencies as low as 80,000 cycles per second (80KHz). As you get up into higher frequencies, the skin effect starts making its effect known and again, many fine strands of wire bundled together can reduce the losses here.

Here is a snippet of a few of then many inductors I tested awhile back:

If you look carefully, you'll note in one test some very low ESR capacitors were added, and an actual change in efficiency can be seen. So this leads into capacitors, where the ESR of the capacitors also causes additional loss. Now if you are switching at say 300KHz, if you dig deeper into the datasheets, you'll notice that different capacitors have higher and lower losses at 300KHz, but most folks only look at the numbers which are specified at 60Hz, and miss part of the problem, as capacitors work differently at various frequencies.

Further more, resistances in the wire/switch leading to the converter, resistances in the board traces cause additional loss. In the first case, resistance in the switch/wire/flashlight body cause the input voltage to drop, causing the converter to draw additional current in order to boost to the output requirement, as it needs X amount of power in to make X amount out. As the current (I) in goes up, the switching devices in the converter dissipate more power (I^2*R). Plus you have larger swings in the inductor which cause it's losses to go up, both from resistance and magnetic losses. And for the board layout, many grasshoppers will use narrow traces (like 2mm), which are also resistive, causing more losses. It's usually best to try to make the power loops as planes instead, keeping these paths as short as possible. If doing more than one layer, consider using muliple vias for the power loops.

Here is a real simple primer of a few things to consider, yes, it is for the buck, but the same principles apply.
http://www.planetanalog.com/features/OEG20021211S0027

Most importantly, have fun!

 

[ViReN]

Great Links "NewBie" ... (Jarhead) /ubbthreads/images/graemlins/smile.gif ....

I got to know a load of things from your reply.... Now i will try to get more efficiency out of the Transistor based circuit....

What do you guys think ? using MosFET's instead of transistors for switching can increase the Efficiency ?

Thanks & Regards,
ViReN

 

[NewBie]

They sure can.

The trick is doing it right, which is why I provided the basic MOSFET losses link for you. It is not only about driving the MOSFET properly, but also selecting the right MOSFET.

Here is an example of a decent low voltage MOSFET.

http://www.fairchildsemi.com/ds/FD/FDC6000NZ.pdf

Note the 0.028 ohms at only 2.5V on the gate, it will handle 7.3A continous, and 20A pulse, a total gate charge of only 8 nC, body diode recovery time is 16 nanoseconds, and only 4.3 nC charge in the body diode.

But you'd definitely not want to use this in the application that I posted to the link before, there are better choices of MOSFETs to utilize for that.

 

[Cones]

Deleted.

Sorry wrong thread /ubbthreads/images/graemlins/blush.gif