Making best use of buck/boost DC/DC converters

[Sverige]

I’m planning on using generic, cheap LM2596 DC/DC converters to drop my DC bus voltage from around 14V (4S LiFePO4) to the 12V, 9V and 5V my devices will require. All load devices are very low power, around 3W continuous will be drawn from each of the voltages (5, 9 &12V), so well within the rated 3A output current of this particular device which I’ve ordered:

RUNCCI-YUN 8Pcs LM2596S DC-DC Buck Converter Set, 3.2-46V bis 1.25-35V Step-Down Spannungsregler Abwärtswandler, Stromversorgung Step Down Modul, für Arduino: Amazon.de: Gewerbe, Industrie & Wissenschaft

RUNCCI-YUN 8Pcs LM2596S DC-DC Buck Converter Set, 3.2-46V bis 1.25-35V Step-Down Spannungsregler Abwärtswandler, Stromversorgung Step Down Modul, für Arduino: Amazon.de: Gewerbe, Industrie & Wissenschaft

www.amazon.de

Hopefully that listing displays in English. It’s German amazon.

So my questions, to those with experience using these devices:

1) I understand the device efficiency is affected primarily by the size of the voltage drop from input to output, so if I need 5, 9 and 12V all in one location, is it better to daisy-chain the converters, so the output of the 12V converter becomes input of the 9V converter, and likewise the 5V converter is connected at the end of the chain? This way the second and third converters are dropping their input voltage less than if all three devices were paralleled up with 14V input. Ok, the current for all load devices then flows via the 12V converter, but total load connected to all three converters is under 10W, so even with some conversion inefficiencies the first converter is only seeing something like 1A total output current. All converters will be physically in the same place, so no cable runs (more than 10cm or so) to think about.
2) Will cumulative ripple become a concern with three devices all in a chain?
3) Does anyone know if it’s beneficial, or necessary to solder a big electrolytic capacitor across the output side of these converters, or do they have enough smoothing?
4) Has anyone tested the efficiency of these devices to see if they live up to the claimed “up to 92%”? Of course “up to” is fairly meaningless as a spec!
5) Does anyone know if these things have a habit of drifting from their set point output voltage over time? Do I need to keep checking to make sure my 5V DC isn’t creeping upwards at all?



Thanks all.

 

[snoobler]

1) One voltage step is typically more efficient than additional steps, i.e., use one converter for each desired output stepping down once form the source voltage, 14-12, 14-9 and 14-5.

Worth noting that your 4S LFP won't be at 14V for very long. You'll rapidly drop into the low to mid 13s under load.

2-3) don't know.
4) No, but % efficiency is more related to the power being converted. Very low power tends to be very inefficient. Moderate power tends to be efficient (92% "max"), max power tends to be a little less efficient. "Power" is relative to rated power, i.e., 0.5W on a 50W unit would be very inefficient, but 0.5W on 1W unit would be very efficient.
5) I suspect they may drift slightly with temperature. I'd put a small LED voltmeter on each output for fun.

 

[gnubie]

If you are trying to wring every last watt out of the system by avoiding mutliple DC-DC stages look into the design of the converter. Often they use diodes, there is a better way. Look for a board that says it uses synchronous switching, essentially diodes are replaced with FETs giving an improvement in efficiency. They will be built around a different device instead of the LM2596 etc.

I have some sync switched 5V buck regulators, peak efficiency measured is 97%.

You may find that your regulators have 40mV peak to peak of switching noise. Passing up stream noise through the inductor of each successive regulator will tend to knock that noise down quite a bit but ultimately if the planets align and all 3 manage to be switching at the precise same time the noise will add. I don't think it would be tragic, 80mV peak to peak perhaps.

If 40mV is too much for your application some ceramic capacitors will help. There's lots of tutorials about reducing noise in the output of switching power supplies on youtube. Have a wander around the eevblog channel, Dave has done a few noise related clips.

Some of these things have very poor slew rates, ie they are fine with fairly constant loads but when hit with a substantial increase in demand the output voltage can momentarily drop below the point at which circuits misbehave. In those cases an additional low ESR electro on the output will help.

Beware with the LM2596 based boards. There's a lot of relabeling of these devices and of course outright fakes too. Pushing the device to it's alleged ratings, voltage and/or current, will often see boards built around them fail and your load exposed to full input voltage.

 

[Sverige]

1) One voltage step is typically more efficient than additional steps, i.e., use one converter for each desired output stepping down once form the source voltage, 14-12, 14-9 and 14-5.

Worth noting that your 4S LFP won't be at 14V for very long. You'll rapidly drop into the low to mid 13s under load.

2-3) don't know.
4) No, but % efficiency is more related to the power being converted. Very low power tends to be very inefficient. Moderate power tends to be efficient (92% "max"), max power tends to be a little less efficient. "Power" is relative to rated power, i.e., 0.5W on a 50W unit would be very inefficient, but 0.5W on 1W unit would be very efficient.
5) I suspect they may drift slightly with temperature. I'd put a small LED voltmeter on each output for fun.

Thanks @snoobler, appreciate the reply. On point 1, I see your point, if all three converters are connected directly to the 14v nominal source (14-12, 14-9, 14-5), then the 5V converter isn’t passing power which has already been converted by two other modules, as would be the case in my proposed 14-12, 12-9 and 9-5 chain. This would be a factor which would tend to increase efficiency, but then on the other side of the coin I have read that the lower the potential difference from I/p to o/p, the higher the conversion efficiency and to that you’ve added that the devices should work more efficiently when they are converting more power (higher current), which the daisy chain configuration does promote.

Heck, the more I puzzle over this the more I think I just need to do it both ways, measure the overall input power in each case and report back!!

If you are trying to wring every last watt out of the system by avoiding mutliple DC-DC stages look into the design of the converter. Often they use diodes, there is a better way. Look for a board that says it uses synchronous switching, essentially diodes are replaced with FETs giving an improvement in efficiency. They will be built around a different device instead of the LM2596 etc.

I have some sync switched 5V buck regulators, peak efficiency measured is 97%.

You may find that your regulators have 40mV peak to peak of switching noise. Passing up stream noise through the inductor of each successive regulator will tend to knock that noise down quite a bit but ultimately if the planets align and all 3 manage to be switching at the precise same time the noise will add. I don't think it would be tragic, 80mV peak to peak perhaps.

If 40mV is too much for your application some ceramic capacitors will help. There's lots of tutorials about reducing noise in the output of switching power supplies on youtube. Have a wander around the eevblog channel, Dave has done a few noise related clips.

Some of these things have very poor slew rates, ie they are fine with fairly constant loads but when hit with a substantial increase in demand the output voltage can momentarily drop below the point at which circuits misbehave. In those cases an additional low ESR electro on the output will help.

Beware with the LM2596 based boards. There's a lot of relabeling of these devices and of course outright fakes too. Pushing the device to it's alleged ratings, voltage and/or current, will often see boards built around them fail and your load exposed to full input voltage.

Thanks @gnubie! I am interested in overall efficiency of course, but then I already bought quite a lot of these converters as they were so cheap(!) and so I’d better make use of them rather than looking for more efficient converters.

My load devices should be somewhat tolerant of a little ripple I suppose. It’s all the typical “black boxes“ you’ll find in a messy pile under the TV in a home. Stuff like wifi router (12V, 3W), home alarm (9V, 4W), HDMI switch (5V, 0.5W), media centre IR control hub (5V, 0.75W), bluetooth audio interface (5V, 0.5-3W, depending on SOC of internal battery).

I don’t think the demand will vary much, so slew rate hopefully not an issue. I’ll look up that eevblog YT channel, sounds interesting.

I definitely wouldn’t trust the “max 3A” rating on these cheap modules, but as you see from above, the power demands of my loads are quite modest, and even in a daisy chain arrangement I can’t see that the first converter would be passing more than an amp or so.

 

[gnubie]

That should all be fine with the typical noise you get from one of these buck regulators and they won't have the sort of current changes that will cause problems either so

 

[rin67630]

I’m planning on using generic, cheap LM2596 DC/DC converters to drop my DC bus voltage from around 14V (4S LiFePO4) to the 12V, 9V and 5V my devices will require. All load devices are very low power, around 3W continuous will be drawn from each of the voltages (5, 9 &12V), so well within the rated 3A output current of this particular device which I’ve ordered:

RUNCCI-YUN 8Pcs LM2596S DC-DC Buck Converter Set, 3.2-46V bis 1.25-35V Step-Down Spannungsregler Abwärtswandler, Stromversorgung Step Down Modul, für Arduino: Amazon.de: Gewerbe, Industrie & Wissenschaft

RUNCCI-YUN 8Pcs LM2596S DC-DC Buck Converter Set, 3.2-46V bis 1.25-35V Step-Down Spannungsregler Abwärtswandler, Stromversorgung Step Down Modul, für Arduino: Amazon.de: Gewerbe, Industrie & Wissenschaft

www.amazon.de

I know these LM2596 / XL4015E based devices. If you really need to supply 3W continuously, they are fine, and you do not need to take any special precautions, if you're really sure to never have a short. The modules have no current limitation and the above-mentioned chips love to produce the magic blue smoke...
If I where you, I would strongly prefer using modules that provide a CV/CC regulation, they are not that more expensive.

You should also be careful: the buck converters do not provide any galvanic isolation, if your toys are interconnected, you might get unwanted current loops and, potentially, even shorts.

 

[Sverige]

I know these LM2596 / XL4015E based devices. If you really need to supply 3W continuously, they are fine, and you do not need to take any special precautions, if you're really sure to never have a short. The modules have no current limitation and the above-mentioned chips love to produce the magic blue smoke...
If I where you, I would strongly prefer using modules that provide a CV/CC regulation, they are not that more expensive.

You should also be careful: the buck converters do not provide any galvanic isolation, if your toys are interconnected, you might get unwanted current loops and, potentially, even shorts.

Thanks for your reply.

I am sure that my loads will be low power and their current demands will not exceed even one third of the module‘s rating. One can never exclude the possibility of an accidental cable short, or a device fault which causes a short circuit, and in that case I understand and accept that the module would be fried. At just over €1 each I can live with that outcome, and my DC outlets in the home are all fused so that the main DC bus isn’t taken out by a fault in a device connected to one particular outlet.

What advantage do you think a CC CV module has for powering a device such as a wifi router?

I understand the modules don’t provide isolation from input to output. I don’t see that as a problem in this application unless you could elaborate on why that might be a concern?

 

[rin67630]

What advantage do you think a CC CV module has for powering a device such as a wifi router?

I understand the modules don’t provide isolation from input to output. I don’t see that as a problem in this application unless you could elaborate on what that might be a concern?

As long as you're sure never to have a short, you can go for the CV-only modules.
The buck converter chip will be faster to blow than any fuse.

The thing with the missing insulation is however not to be neglected.
All your toys have been conceived to be powered from galvanic separated power supplies.
If they are not interconnected together through audio/video/cables or non-isolated network lines, it is not a problem.
So charging cell phones, running an alarm system or a WiFi router is safe.

But as soon as you begin to wire audio/video components together with audio/Video/HDMI cables and provide at the same time a common grounded power supply for which they have not been designed, you might run into sheath current/ ground loop problems, that can lead to very weird and unpredictable results.

 

[Sverige]

Thanks @rin67630. The devices I’m connecting to the power modules will not have cabling between them, so no issue with galvanic isolation.

 

[RCinFLA]

The IC's in these buck converter usually have self shut down if the IC gets too hot. Some reactivate when they cool off, some require they cool off and have power removed and reapplied to get them to reactivate.

Put a fuse on the input in case the converter fails.

Avoid a buck-boost converter if you don't need the boost as they are less efficient. Synchonous converters are more efficient. Cheaper non-synchronous converters have a diode which adds voltage drop that must be made up in the down conversion and is a higher percent hit on conversion efficiency the lower the output voltage. For example down conversion to 3.3v with cheaper buck switcher with diode is equivalent to synchronous switcher downconverting to about 4.1vdc to make up for diode voltage loss. That's about 25% loss before switching and coil losses are added in. The one in your picture is cheaper non-synchronous converter as the diode is visible in the picture.

At low load current they typically go into skip mode of operation meaning the output ripple voltage goes up.

Better is totally isolated converter where output side has no common connections to input side., They are more expense.

 

[Sverige]

Thanks @RCinFLA, I note your comments and will maybe look for better converters in future, but for now will try using the ones I have. I don’t see any amazon listings advertising synchronous converters, so maybe another term is used, or one must simply be able to tell by looking at the photo.

 

[gnubie]

There are sync converters on Amazon, the word isn't in their title but it can be in the description. It used to be a fair bet that if you bought a converter off ebay that had the word in the title it actually was one but the rip off merchants soon glommed onto it and now there's heaps that claim to be but aren't.

Get a known name brand board, Drok etc, then take a look at the picture. Check the part number on the thing that has a lot of legs. If it's two of the ones mentioned in this thread it's going to use external diodes. Check the larger components on the board. If you see two or more big packages check their part numbers in Google. If they are FETs and you don't see any biggish parts with D1 or something on the board it's a synchronous design.

Here's a sync one I found on Amazon with the first search. If you take a close look the transistors are 2SD4184 and there's two of them. These are FETs. There's no big diodes visible, so pretty much 100% that this is actually a sync buck module. As a bonus Drok has the same layout board with the same labelled components.

Amazon.com: UCTRONICS DC 6V 9V 12V 24V to DC 5V 5A Buck Converter Module, 9-36V Step Down to USB 5V Transformer Dual Output Voltage Regulator Board [2 Pack] : Electronics

Buy UCTRONICS DC 6V 9V 12V 24V to DC 5V 5A Buck Converter Module, 9-36V Step Down to USB 5V Transformer Dual Output Voltage Regulator Board [2 Pack]: Power Converters - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

www.amazon.com

 

[rin67630]

To power my 5V instruments from a 12V battery, I use these synchronous buck modules:
Fine buck converters


They have an own consumption below 1 milliamp, even below 100 microamps, if you remove the useless red LED.
They are perfect to feed a Raspberry Pi / Arduino / ESP-modules, and charge the usual USB stuff incl. power banks from a 12V battery.
They are very well protected, i never managed to burn any of them.
They cost under 1$ shipped, bought by 10 units.

The only drawback is, that they are not adjustable.

 

[sunshine]

Stuff like wifi router (12V, 3W),

I've found these to be very tolerant of voltage variation, 10v - 16v.
The wifi cuts out below 10v and the other parts somewhere below that but will need a manual restart upon restoration of power.
The Sagem and TPLinks keep any changed settings however NetGear ones revert back to the the ISP's settings if you have picked up any last year models to extend the wifi.

 

[RCinFLA]

Yes, most don't state it. You have to look to see if a diode is visible on module picture indicating they are likely not synchronous. Sometimes, but rarely, there are reverse polarity protection diodes. Best if you can read the IC part number from picture and look up spec.

Modules are sold under name 'mini-360' for less then $0.50 each. About size of postage stamp. They use MP23070N chip which only goes to 23vdc maximum input. Spec attached.

There are others that take higher input voltage if needed.

 

[Hedges]

When switchers are cascaded in series, a common practice is to have them operating at significantly different switching frequencies.
You probably don't have control over that, might possibly be able to select.
Best to have each output come from a single conversion.

Some modules tolerate brownout on their input, but some die. Make sure your battery voltage range and BMS or whatever don't violate their input specs.

 

[Sverige]

Thanks guys, good info

What about this one - would it be synchronous do you think?

Droking 5v Regler, 5 stücke Mini Spannungsreduzierer DC 4,5-24 V 12 V 24 V Abwärts bis 5 V Buck Converter Board 3A Volt Step-down Transformator Netzteil Modul: Amazon.de: Baumarkt

Droking 5v Regler, 5 stücke Mini Spannungsreduzierer DC 4,5-24 V 12 V 24 V Abwärts bis 5 V Buck Converter Board 3A Volt Step-down Transformator Netzteil Modul: Amazon.de: Baumarkt

www.amazon.de

 

[rin67630]

Yes, most don't state it. You have to look to see if a diode is visible on module picture indicating they are likely not synchronous. Sometimes, but rarely, there are reverse polarity protection diodes. Best if you can read the IC part number from picture and look up spec.

Modules are sold under name 'mini-360' for less then $0.50 each. About size of postage stamp. They use MP23070N chip which only goes to 23vdc maximum input. Spec attached.

There are others that take higher input voltage if needed.

View attachment 26051

I know these modules as well. Although they seem to provide the right technique, all the ten modules I have received had an extremely high own consumption and got hot at idle. The microscopic trimmer is very difficult to adjust and one fall apart as I touched it.
I suppose I got a batch of rejected devices, but I'm not been willing to renew the experiment.

 

[rin67630]

What about this one - would it be synchronous do you think?

Droking 5v Regler, 5 stücke Mini Spannungsreduzierer DC 4,5-24 V 12 V 24 V Abwärts bis 5 V Buck Converter Board 3A Volt Step-down Transformator Netzteil Modul: Amazon.de: Baumarkt

Droking 5v Regler, 5 stücke Mini Spannungsreduzierer DC 4,5-24 V 12 V 24 V Abwärts bis 5 V Buck Converter Board 3A Volt Step-down Transformator Netzteil Modul: Amazon.de: Baumarkt

www.amazon.de

They look very interesting, I like the idea of the soldering bridges.

I have found them here as well:
https://www.aliexpress.com/item/32729546380.html
I have ordered three pieces to test.

I especially like that they have provided an enable input. With other modules, I used to use these kind of buck converters as a "power digital outputs" for my microcontrollers, quasi as relays to directly control e.g. power LED lights or DC fans without needing to design an extra circuit board.

 

[rin67630]

I've found these to be very tolerant of voltage variation, 10v - 16v.

You're right. All the current devices with a 12 V input I have used, had already (a) buck converter(s) inside and could be directly fed from the battery.