Author Topic: Buck or boost?  (Read 6126 times)

0 Members and 1 Guest are viewing this topic.

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Buck or boost?
« on: January 11, 2018, 09:19:58 pm »
Quick project brief.  I bought a small solar system that will run indoors in a window.  I know it won't produce much, but I am only intending on using it to charge battery powered devices.  Phones, eCigs, power banks, etc. I have been charging my USB devices off a repurposed laptop battery (extracted the cell pairs and put them 3S2P config) with a 5V regulator board I got off ebay.  The 3S2P get's charged off the solar load via a proper hobby charger.  Charging USB devices it lasts for ages.

This works, but it has short comings.  I believe it's a linear regulator, it gets hot, it doesn't like to put out more than about 1Amp.

So it got me to thinking, if I want to charge things via USB with more than 1Amp and use QC2.0 or QC3.0 adapter etc.  Or is I want to run anything less than 12V off the 12V Lithium pack I will need a switched step down buck converter.

Now I intend to buy old laptop packs for £8 off ebay and extract the cells, they are low current cells I know.

Is it better/easier/more efficient to use the cells in 1S6P and use boost converters to step the voltage up, or build them in 3S2P and step it down?
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #1 on: January 11, 2018, 09:38:02 pm »
Oh, I forgot one thing.

Cheap and fairly safe.

I have found that switch mode regulators buck or boost are not the cheapest and nor are inductors.

Alternatively I can buy Chinese modules for less than the cost of the IC alone, but I would call them "under supervision" devices and am not wanting to leave such a device running overnight (though I have) or while I'm out of work. and the house is empty.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Buck or boost?
« Reply #2 on: January 11, 2018, 10:11:51 pm »
Buck is almost always more efficient and lower cost than boost. The one caveat I can think of is a failing buck converter can put the full input voltage onto the output while a failing boost converter will typically result in the output dropping down to the input.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #3 on: January 12, 2018, 08:00:27 am »
Buck is almost always more efficient and lower cost than boost. The one caveat I can think of is a failing buck converter can put the full input voltage onto the output while a failing boost converter will typically result in the output dropping down to the input.

Is there anyway to add protection against that?  Something that will, say, cap the voltage and then fail rather than let it exceed tolerance? 

I suppose a zener will cap it, but if the current exceeds it's abilities it will then fail and the full voltage will still get out the output rail.

EDIT:  I was doing a bit of googling, crowbar came up, also Zener clamp.  However I wondered.  If the zener circuit has no resistor to current limit, if the voltage suddenly rises above the zener threshold with no current limiting resistor could a fuse not be used to fail the circuit as current rushes through the zener and blows the fuse?
« Last Edit: January 12, 2018, 08:06:42 am by paulca »
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: Buck or boost?
« Reply #4 on: January 12, 2018, 10:08:52 am »
That's a relatively naive but effective way of doing it. The only problem is that it depends on how much power the zener can sink compared to the fuse. If you think of a Zener's power dissipation as P = I * V, the V is going to be quite high for a zener which means if your solar panel dumps say 2A through a 12V zener, that's going to have to blow 24 watts out of the window until the series fuse blows. There is a possibility the zener will go before the fuse as most fuses are still pretty slow (sometimes seconds). Also the Zener's "knee" is quite soggy i.e. it starts to conduct a little bit, then gets more and then more etc. There's no "snap" moment.

Ergo, that's why you usually use a proper crowbar circuit. This contains a zener, or even better a precision reference and an SCR. When the SCR threshold is hit, the SCR latches down at a very low voltage. SCRs can cope with a lot of current and power. Think clamping a 12A supply in a single TO220 package.

If you look at the following Art of Electronics reference, which happens to be the free chapter: https://artofelectronics.net/wp-content/uploads/2016/02/AoE3_chapter9.pdf pp690 (section 9.13), it will go into detail. The best implementation is the TL431 which is a programmable voltage reference that costs literally bugger all. Add a fat SCR, a couple of resistors and a fuse and you're good.

Also the chapter above contains lots of info on buck/boost converter characteristics and design etc.
« Last Edit: January 12, 2018, 10:10:40 am by bd139 »
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19336
  • Country: gb
  • 0999
Re: Buck or boost?
« Reply #5 on: January 12, 2018, 10:53:42 am »
That's a relatively naive but effective way of doing it. The only problem is that it depends on how much power the zener can sink compared to the fuse. If you think of a Zener's power dissipation as P = I * V, the V is going to be quite high for a zener which means if your solar panel dumps say 2A through a 12V zener, that's going to have to blow 24 watts out of the window until the series fuse blows. There is a possibility the zener will go before the fuse as most fuses are still pretty slow (sometimes seconds). Also the Zener's "knee" is quite soggy i.e. it starts to conduct a little bit, then gets more and then more etc. There's no "snap" moment.

Ergo, that's why you usually use a proper crowbar circuit. This contains a zener, or even better a precision reference and an SCR. When the SCR threshold is hit, the SCR latches down at a very low voltage. SCRs can cope with a lot of current and power. Think clamping a 12A supply in a single TO220 package.

If you look at the following Art of Electronics reference, which happens to be the free chapter: https://artofelectronics.net/wp-content/uploads/2016/02/AoE3_chapter9.pdf pp690 (section 9.13), it will go into detail. The best implementation is the TL431 which is a programmable voltage reference that costs literally bugger all. Add a fat SCR, a couple of resistors and a fuse and you're good.

Also the chapter above contains lots of info on buck/boost converter characteristics and design etc.
A crowbar is a good idea, but beware of nuisance triggering, due to small, sharp spikes or noise.

One way round this is to use a zener with higher voltage, than the crowbar and add a capacitor to the crowbar circuit, to slow it down. The zener will absorb short transients, without tripping the crowbar, which will only trip when the overvoltage persists for long enough.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: Buck or boost?
« Reply #6 on: January 12, 2018, 11:07:13 am »
Good point. I have actually had that problem before. Cost me a pass transistor as well  :-DD
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #7 on: January 12, 2018, 11:09:16 am »
So the particulars in this will be:

3S2P Lion 12V -> Buck 5V -> USB* -> Device*  (*=multiple)

For most things anyway.  Other things will most likely be 12V wall wort devices and they usually handle the Lion voltage range fine and won't be bucked.

I haven't yet figured out how to do QC2.0/3.0 but I'll leave that for later.

Cost benefit would depend on what the device is.  If it cooks one of my eCig batteries, I won't be too worried, I have half a dozen of them and they are only £30. 

However my phone is circa £250 to replace.  It has QC and accepts 9V, so maybe it would survive 12V.

Just a thought, but a zener, transistor and latching relay is a bit expensive.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Buck or boost?
« Reply #8 on: January 12, 2018, 05:56:12 pm »
Well assuming the regulator is reasonably well designed, then failure is rare. I don't recall ever seeing any form of overvoltage protection in an off the shelf cigarette lighter to USB adapter.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #9 on: January 12, 2018, 06:20:56 pm »
Well assuming the regulator is reasonably well designed, then failure is rare. I don't recall ever seeing any form of overvoltage protection in an off the shelf cigarette lighter to USB adapter.

I thought about on the way home as I was using just that.  It's been in my car for 5 years.  I had one that failed, but it failed 0V.  Hard to tell if it is just a linear reg or not though.

I have the components.   I will use the reference application schematic in the datasheet for the buck IC and see what I get on a breadboard (low current) if it works I'll move it to a perfboard and load it up.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: Buck or boost?
« Reply #10 on: January 12, 2018, 06:40:37 pm »
I had a national semiconductor TO3 voltage regulator go in/our short once. Was on 5v rail so a ton of TTL got very hot very quickly and expired. I think about 16V hit it all. It only gave up when the transformer secondary went phut.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Buck or boost?
« Reply #11 on: January 12, 2018, 06:56:35 pm »
Linear regulators can also fail shorted, I've had it happen myself. In one case it fried a bunch of stuff, in another it magically fixed an arcade game board that had been dead. No idea what happened with that one but the 5V was up over 10V and the game came up but looked blurry, fixed the power supply and it has worked ever since.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #12 on: January 12, 2018, 07:21:00 pm »
Linear regulators can also fail shorted

And as I discovered the other night, they are a route to ground in reverse.  Added 5V 'instead' of the regulator output, didn't unwire the regulator, disconnected it's power rail....  the smell got to me before I worked out why stuff wasn't working.  I love my current limiting bench PSU, thanks to all on here for telling me it was a good idea to have one for bench stuff.  It only took 0.5A as a result and survived.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #13 on: January 12, 2018, 07:23:01 pm »
Anyway, so that makes tomorrow's project either making an ATTiny85 LED perfboard or a buck converter.  Hmm...   Can I do both in a day.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Buck or boost?
« Reply #14 on: January 12, 2018, 07:45:34 pm »
You can probably do both in an hour if you have all the parts on hand and don't have to correct any mistakes.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #15 on: January 12, 2018, 09:42:10 pm »
You can probably do both in an hour if you have all the parts on hand and don't have to correct any mistakes.

I did the basics of the LED board tonight.  Beer is now adding up though.

Tomorrow morning I have to sort out the solar system.  I'll hopefully get to the buck converter on the breadboard (at least) tomorrow night.

My perfboard skills suck :(
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: Buck or boost?
« Reply #16 on: January 12, 2018, 10:40:46 pm »
Everyone’s perfboard skills suck. Some people are just better at hiding the horrible bits.
 
The following users thanked this post: Brumby

Offline wraper

  • Supporter
  • ****
  • Posts: 16791
  • Country: lv
Re: Buck or boost?
« Reply #17 on: January 12, 2018, 11:15:03 pm »
Now I intend to buy old laptop packs for £8 off ebay and extract the cells, they are low current cells I know.
Don't do that. Savings are questionable at best compared with new good quality batteries and quiet likely you'll get some worn out batteries.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #18 on: January 13, 2018, 12:00:41 am »
Don't do that. Savings are questionable at best compared with new good quality batteries and quiet likely you'll get some worn out batteries.

I'm not buying second hand laptop batteries.  I am testing the cells.  The first set claimed to be 2600mAh, I tested them to be 2600mAh.  Not exhaustive, but indicatively.  Close enough for what I want.

They won't run an eCig though.  They are standard low current batteries.  I'm rating them as 1C charge, opting for 1/2C instead, 1-2C discharge, aiming for 1C+grace and not asking more.  Still that's 2.6 Amp Charge, 5.2 fast charge, 5 amps constant discharge, 10 amp burst.
« Last Edit: January 13, 2018, 12:04:01 am by paulca »
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline viperidae

  • Frequent Contributor
  • **
  • Posts: 306
  • Country: nz
Re: Buck or boost?
« Reply #19 on: January 13, 2018, 07:28:15 am »
It's already been done for you.
Buy a QC 2/3 car charger. It should work fine on 3S
 

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: Buck or boost?
« Reply #20 on: January 13, 2018, 10:25:54 am »
Be aware that some of the XL6009 boost converters sold on Ebay can 'run away' when the input voltage falls below the minimum for regulation, about 3v. The output can rise to as much as 40v, which could do some serious damage to USB equipment. 

The XL4015 buck modules seem quite reliable. Bear in mind these sense current in the -ve so the battery -ve must not be grounded.



I'd go for a 3S or 4S arrangement and buck. Charging can be via similar buck. Bear in mind the need for a balancing/protection board.

To protect the output, put a 5.6v zener across it. If the reg goes out of control this will blow and short the output. You are then relying on the protection board to cut the power. Or, add a polyswitch as well in the battery lead just for good measure.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #21 on: January 13, 2018, 11:10:36 am »
Thanks guys.  I now have a number of options in the post.

* 12V car QC3.0 charger plug (with voltmeter), which takes care of sorting that fuss out for the phone.  And a socket for it to wire an XT60 onto.
* A Chinese 3 port USB buck converter for other  things.  Remains to be seen how stable this is and what it's based on, but it has a transparent acrylic case and I seem to have a fetish or something for those.

I have the parts to build a 3A buck converter, but I'll do that just for fun.  Hopefully smoke free.

Next thing I need to build is a battery protection circuit.  Again this is a rabbit hole as it really 'should' monitor on the balance port and kill the main power rails of the battery if a single cell drops below 2.8V.  However, it's probably cheaper to just buy one from an RC shop.  I'd have to accept the higher RC high current LiPo safety cut off voltage, often as high as 3.3V, more likely 3.0V, but I'll see what turns up.

EDIT: The latter turns out to be trickier to buy than it should be.  The only ones that will actually monitor individual cell voltages AND cut out are expensive.  £25.

I remember looking into this before and the best option is a low cell voltage alarm (£3) and pull the buzzer and connect a relay instead.  However, it will need to be a latching relay or it will oscillate.
« Last Edit: January 13, 2018, 11:40:05 am by paulca »
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #22 on: January 13, 2018, 01:35:18 pm »
So I built the buck converter on a breadboard as per the first application schematic in this:
http://www.ti.com/lit/ds/symlink/lm2576.pdf

I expect my lm2576T-5 buck IC is fake.  It gives me 5.04V open circuit, but with a load pulling 800mA the voltage drops to 4.34V and rises back to around 4.45V.  The IC also gets pretty toasty, which is probably normal.  I don't think it's even worth putting a heatsink on it, not if it sags 0.6V with that small load.

0.6V would be suspiciously like a diode drop, but the only diode is the schotky and it's a 2.something V drop.

Was bought from a UK seller on ebay, but they were about 1/3rd the price of RS/CPC.  Should have known better.
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline paulcaTopic starter

  • Super Contributor
  • ***
  • Posts: 3991
  • Country: gb
Re: Buck or boost?
« Reply #23 on: January 13, 2018, 04:47:48 pm »
I checked the price.  RS - £1.48 each in a bag of 5.  Ebay: £2.99 for 2.  So same price.  They are actually ON Semi.  Doesn't prove they are real though.

Should I really be seeing a 0.4-0.6V drop from the fixed output voltage under load?
"What could possibly go wrong?"
Current Open Projects:  STM32F411RE+ESP32+TFT for home IoT (NoT) projects.  Child's advent xmas countdown toy.  Digital audio routing board.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23017
  • Country: gb
Re: Buck or boost?
« Reply #24 on: January 13, 2018, 05:03:41 pm »
At least the ones on RS are likely to actually work properly.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf