Low efficiency on a buck converter

[manick]

I tested a linear regulator and a buck boost converter using a 12V lead from a computer power supply. The output voltage was 5V with the current just over 1A. The efficiency I got on the linear regulator (7805) was 41% which is what I expected. The buck boost converter gave me 39% but the manufacturer claims 73% going from 12V to 3V with a 3A draw.

I uploaded a video on YT. Link below. The buck boost testing of importance starts at 8:00 min into the video. What did I do wrong? Why is the efficiency so low? Apologies if this was asked before. I did a search but could not find this particular problem.

[Buriedcode]

It looks like that converter is actually just a boost followed by a buck  (or a buck followed by a boost) - cascading two converters. 

If they are rated for 3A, and don't have thier input/output ratio at the most efficient, say stepping down 12V to 36V then 36V stepped down to 5V (silly i know) then it could be that each converter has ~64% efficiency.  Which gives over all 0.64^2 = ~40%

I haven't seen cascading buck/boost converters often, but when I have it was mostly for the large input/output ranges, not for efficiency.

[manick]

Buriedcode that seems weird but the maths presented does explain the low efficiency. I can't think of a simple test to do other than put it under load again and see if both converters are getting warm. If so I will assume both are in operation. During the recorded test I did not check the boost side at all.

Let me test that theory and report back.

[IanMacdonald]

The LM2596 has quite a high forward drop, hence it is not all that efficient in low voltage applications. IIRC it's a BJT darlington output stage. The XLSemi devices such as the XL4005 have lower forward drop due to using a MOSFET power device. Beware of fakes though. Seen a couple of fakes which could lose regulation under some conditions, and potentially damage the load.

[manick]

Buriedcode when testing I did not pay attention to which converter I checked. Seems the boost converter is getting warm and not the buck. Weird. The capacitor nest to is is getting warm too. That is at the original 1.2A

I increased the resistance so I am drawing 0.74A at 5V so 3.7W. The supply side is taking 0.6A at 12V so  7.2W. At that setting the efficiency is 51%. Better at lower current it seems.

Dropping the voltage to 3V I am getting 0.42A output so 1.26W, 0.23A @12V input so 2.76W. Efficiency of 46% Not too great either.

I am starting to suspect the boost converter or even the capacitor adjacent to it is faulty.

[manick]

I ended the video with that possibility. Chinese could do anything.

It seems the boost converter is the one getting hot. I even tested it boosting a 5V to 8V and it got hot still. Seems something in that section is defective. It I cannot figure it out I may try the brute force approach of removing that regulator probably without proper checks on how it affects the rest of the circuit.

[ez24]

Reading along

[Buriedcode]

I have quite a few similar modules, but just bucks or boosts, and a couple have a pseudo-current regulation using a PCB trace low side resistor, and a cheap (LM358) opamp.  So I assume those multiturn trimmers on there are for setting output voltage and current limit/regulation. 

Nevermind, found this:  https://www.electronicspoint.com/attachments/lm2596lm2577_p04-jpg.17900/

So it appears it is meant for (lead acid) battery charging from a wide input voltage.  My point is, its not simply a buck, or a boost, or even two separate modules, but what someone considers a reasonable "module" for charging lead acid batteries.  I wouldn't think the current regulation is accurate, but often it doesn't' have to be.  These sorts of modules do work, but not much thought has gone into them before they are mass produced.

I'm not going to bash 'chinese stuff' because well, china produces a vast range of quality from the laughable to the highest quality.  Often we see stuff at the lower end.  But most of the single bucks/boost modules I have work just fine - albeit with a fair ammount of ripple on the output, but considering they cost <$2  US, I can't complain.

[manick]

Buriedcode yes I agree the Chinese do produce top of the line stuff too but at the lower end like this module I am never sure with them. Guess its part of the risk buying cheap stuff.

Thanks for the pic. I got most of that figured out. What is a bit confusing is the boost converter is overheating even when its boosting the voltage. There aren't that many passives so I am tending towards a defective boost converter. I will attempt to desolder that and see.

[ez24]

But most of the single bucks/boost modules I have work just fine - albeit with a fair amount of ripple on the output, but considering they cost <$2  US, I can't complain.

I have gone through every (several of each) buck on Aliexpress.  The only failure was ripple on the most expensive buck.  Overall I find Chinese bucks to be very good.

I also do not run them over 50% advertised rating.

[james_s]

It's worth noting that many (most?) of those LM2596 and similar converters are not genuine parts at all but something else that has been printed with that part number. They may or may not come close to the efficiency of the real deal. The cheap buck modules do seem to work well though within reasonable limits.

[manick]

ez24 its just odd that the buck works but the boost overheats. I hope to get a chance soon to remove the boost regulator and see if it still works. I will also order another one. Hopefully I will have better luck.

[manick]

james_s the print isn't extremely clear or clean so maybe this is not an original part. The device is just over $3 so anything is possible at that price point.

[james_s]

I can almost guarantee it's not a genuine part, I don't remember if it was that one or a similar chip but someone tested one and found it wasn't even running anywhere near the same frequency as the part it claimed to be.

I'm not too surprised the boost part is running hotter, in my experience a boost converter is harder to get right.

[manick]

Update
I desoldered the boost package alone. The device works fine in buck mode. Test results as follows:
Output side 1.09A, 4.66V so power 5.08W
Input side 0.60A, 12V so power 7.20W
Efficiency - 71% which matches the manufacturer's value of 73%.
The weird part now is the voltage does not remain stable under load. Under no load its 5V but under 1A load it dropped to 4.66V

Conclusion the boost package was defective. I think james_s is right. This is probably not a genuine NS part. If it is maybe the buck boost design I have is not very good.

I will do an update video so there is this post and something on YT out there for reference if someone else has this problem.
Thanks for all the suggestions.