LT1303-5 DC to DC Boost Converter efficiency issue

[jsquaredz]

I just bought a Linear Technologies LT1303-5 boost converter. I plan to power a 12mA load with it at 5V off two AA batteries. After lashing up the typical circuit I am not getting the efficiency shown in the data sheet.  I did a video here LT1303-5-two-cell-to-5v-boost-converter-supply   Please take a look and see if you can find some issue.  Thanks in advance. 

PS Be gentle.  I never did a video on electronics.  I am usually behind the camera :-)

[jsquaredz]

Thanks for the feedback.  I used the 10A scale on the meter, so the voltage drop across the shunt was relatively low. Based on the data sheet I should have been about 85% efficient when running at 17mA output current at 5V with a 3V input.  That is 33mA input at 3V. I measured about 70mA on the input. Thats an efficiency of  under 40%.

The 22uH inductor I am using is the higher end of the range with the 5V fixed output model.  It is also the size they show in the example circuit. 

Take a look at the video please.  I am sure you will see what is wrong.  I have some waveforms on the scope that someone may be able to interpret. 

If you can guys can give me a few things tests you want me to try I will do it and let you know the results.  Thx.

[georges80]

Take a look at the video please.  I am sure you will see what is wrong.  I have some waveforms on the scope that someone may be able to interpret. 

Looking at the video the first thing I noticed by a quick pan through it is you have a switcher on a plug proto board - eek!. I work with switchers every day and I'd never try to proto one on anything but a PCB with due consideration to layout, ground plane(s), high current paths, sensitive small signal control paths etc.

cheers,
george.

[jsquaredz]

Fair enough.  I didn't consider that. The switching frequency for this part is about 150k so I didn't take any precautions.  I will try to put something together on a veraboard. Can't really make up a quick custom pcb, so hopefully I can keep it tight and do better than the last time. 

Any issues with me socketing the LT1303? 

Take a look at the video please.  I am sure you will see what is wrong.  I have some waveforms on the scope that someone may be able to interpret. 

Looking at the video the first thing I noticed by a quick pan through it is you have a switcher on a plug proto board - eek!. I work with switchers every day and I'd never try to proto one on anything but a PCB with due consideration to layout, ground plane(s), high current paths, sensitive small signal control paths etc.

cheers,
george.

[Skimask]

Went thru the same basic scenario a few months ago.  Until you get that on a decent PCB with a lot of ground plane area, you'll keep fighting the same battle...and losing.

[jsquaredz]

Thanks.  I suppose its a trap for young players like me...

Went thru the same basic scenario a few months ago.  Until you get that on a decent PCB with a lot of ground plane area, you'll keep fighting the same battle...and losing.

[mariush]

Solid ground needed.. very short loop with the inductor and the diode... the diode leads are too long and the diode has poor performance... yes, it's listed in datasheet as OK, but just OK... for the cases where someone needs to use leaded components. Doesn't mean the peak efficiency mentioned in datasheet is done with such a diode.

Capacitors have to be low esr ... Nichicon HE is lowish esr (about 0.2 ohm is not very low esr these days ) but for this application, tantalum, polymer or ceramic are more suitable. 

[Skimask]

Thanks.  I suppose its a trap for young players like me...

Yep, trapped me for a good week or two trying to figure it out.
https://www.eevblog.com/forum/beginners/still-trying-(and-failing)-with-the-lm2588-5v-gt-12v/

Trying to generate +12v/-12v using a LM2588 and the reference circuit in the datasheet.
1st try used a solderless breadboard, failed horribly.
2ndtry used a solder 5-pad-per-run board, semi-sorta worked, but still consider it a fail.
(See page 3 of the thread for pic's)
Switched over to a chunk of single sided copper-clad PCB with a few holes drilled in it.  Worked like a champ.  The calculated efficiency (watts in vs watts out) isn't quite what I hoped, but it's close enough for me. (Page 4 of the thread)

[Niklas]

Contact resistance in the board can cause some of the dissipation. Think of every connection point as a small resistor.

The multimeters are probably assuming constant current with no ripple in the DC A-range. The oscilloscope measurements at around 22 minutes shows ripple in the input and output voltages. The inductor is charged with magnetic flux using current from the input and is then discharged to the output, creating a non-DC current waveform. This will cause the multimeter to show an incorrect reading that must be compensated for the actual signal's Crest factor. Also, the multimeter's accuracy can vary with the Crest factor.

Check the wave forms in Fig 3 and 4:
http://en.wikipedia.org/wiki/Boost_converter

http://en.wikipedia.org/wiki/Crest_factor

[jsquaredz]

I did a followup after implementing some of the suggestions here. 

http://wp.me/p4y7Xy-f

[jsquaredz]

Thanks for the info.  I appreciate the feedback.  Please look at my followup.  I eliminated the breadboard and tightened up the circuitry.  Let me know what you think of the current iteration.

Contact resistance in the board can cause some of the dissipation. Think of every connection point as a small resistor.

The multimeters are probably assuming constant current with no ripple in the DC A-range. The oscilloscope measurements at around 22 minutes shows ripple in the input and output voltages. The inductor is charged with magnetic flux using current from the input and is then discharged to the output, creating a non-DC current waveform. This will cause the multimeter to show an incorrect reading that must be compensated for the actual signal's Crest factor. Also, the multimeter's accuracy can vary with the Crest factor.

Check the wave forms in Fig 3 and 4:
http://en.wikipedia.org/wiki/Boost_converter

http://en.wikipedia.org/wiki/Crest_factor

[Niklas]

That looks a lot better.

I would suggest that you take a closer look on the inductor. This one looks quite close to the one you have on the test board:
http://www.tdk.de/productsearch/e532_tsl0709.pdf

Could it be that you are using a filtering choke instead of a storage choke? "For power line" makes me suspicious. The difference is in the ferrite material and its magnetic losses. For a filtering choke you would like to have higher losses for the AC current components and low for the DC components. A storage choke, like those used for voltage conversion, should have as low loss as possible. Loss = less efficiency.

[poorchava]

Looks like this might be the case.

[jsquaredz]

Well thats interesting.  It certainly didn't say choke on digikey and admittedly the data sheets sucks.  Power "line"  does sound suspect. 

I will order up some Sumida RCH110's. 

That looks a lot better.

I would suggest that you take a closer look on the inductor. This one looks quite close to the one you have on the test board:
http://www.tdk.de/productsearch/e532_tsl0709.pdf

Could it be that you are using a filtering choke instead of a storage choke? "For power line" makes me suspicious. The difference is in the ferrite material and its magnetic losses. For a filtering choke you would like to have higher losses for the AC current components and low for the DC components. A storage choke, like those used for voltage conversion, should have as low loss as possible. Loss = less efficiency.