What's the 'best' way to get bipolar power from a wall-wart?

[Portabella]

I have 5x little boards available which will turn any dual voltage (AC and DC) into ± 12VDC
Input range: AC 12-16V or DC15-24V

If any of you want to buy such a board, drop be a PN. It's 5$ per piece

[Altitude909]

You might want to check out how Seismic did this thing: https://github.com/seismicindustries/IPS2 .

I use the same vReg to go from +5 to +/-12v and its pretty decent. There are dozens of chips out there for all sorts of different input conditions and loads, you just need to hunt down what you want (which is not trivial, there are a LOT of parts). I HIGHLY recommend purchasing development boards for whatever part youre interested in first, this will give you a properly configured and laid out board to test with to make sure that part you want to use will do what you want it to. The manuals for the dev boards are pretty much the instructions for making your own so highly valuable reading

The Seismic part uses the TI TPS65131 DCDC part, here's the evaluation board for it https://www.ti.com/tool/TPS65131EVM-839 and user guide https://www.ti.com/lit/ug/slvuaw7/slvua ... 131EVM-839

Also worth noting, the complexity of small SMPS devices is like that for a reason, the $5 aliexpress devices will probably get you the voltages they advertise but this will come with 100+ mV noise on the outputs. I have a drawer full of them and none have even come close to something usable noise wise.

[MikeDB]

Also worth noting, the complexity of small SMPS devices is like that for a reason, the $5 aliexpress devices will probably get you the voltages they advertise but this will come with 100+ mV noise on the outputs. I have a drawer full of them and none have even come close to something usable noise wise.

Yeah - it's those things that have given SMPSes a bad name ! Done properly they can have noise lower than a linear supply, but it's so easy to take a short cut to reduce costs .... and kill performance.

[KSS]

Low noise is only one of the classic reasons to employ linear over SMPS. Transient load response is another.

[Thomas73951]

Also worth noting, the complexity of small SMPS devices is like that for a reason, the $5 aliexpress devices will probably get you the voltages they advertise but this will come with 100+ mV noise on the outputs. I have a drawer full of them and none have even come close to something usable noise wise.

That's quite a lot of noise. Do you think some (low pass) filtering would help eliminate the noise and just leave the ±15v DC or is the noise too bad for that?

I haven't done anything with my synthesiser for ages because my current power solution is bad. I'm now wondering whether these are just going to cause more (but different) problems.

[Troubleshooter]

These look very handy to put in pedals. Ordered a pair of 12V and 15 models to try out. Thanks for the tip!

I'm planning a standalone filter box (not part of a modular) that I'd like to power from a wall-wart. I've been looking at various ways to get a negative power rail .

If you're happy to use a 5V wallwart (e.g. some phone chargers, Pi PSU, some CCTV PSU etc) then the TI TPS6513x devices are very good. Can produce +/- 15V 300mA from +5V.

https://www.ti.com/product/TPS65130


Available as a module from China if you prefer.
e.g. https://www.aliexpress.com/item/1005001869133587.html

[MikeDB]

Low noise is only one of the classic reasons to employ linear over SMPS. Transient load response is another.

'Classic reasons' are usually based on hearsay or fantasy, nothing more. As I've said repeatedly, a good linear PSU is always noisier than a good SMPS. Likewise the transient load response in a linear supply is purely down to the large capacitors used to smooth 50/60/100/120Hz. If you need that feature then put the same size capacitors in a SMPS.

Let's be quite clear, there are NO good reasons nowadays to employ a linear over SMPS. Designed properly a SMPS will always outperform a linear supply in ANY application, be it low noise or high power or anything in between. The sooner linear supplies are banned outright the better as they just create so many problems for the grid.

The only problem that needs fixing with SMPSes is people who don't design SMPSes properly, especially those idiots not putting PFC in them to save a few cents !

[Altitude909]

Also worth noting, the complexity of small SMPS devices is like that for a reason, the $5 aliexpress devices will probably get you the voltages they advertise but this will come with 100+ mV noise on the outputs. I have a drawer full of them and none have even come close to something usable noise wise.

That's quite a lot of noise. Do you think some (low pass) filtering would help eliminate the noise and just leave the ±15v DC or is the noise too bad for that?

I haven't done anything with my synthesiser for ages because my current power solution is bad. I'm now wondering whether these are just going to cause more (but different) problems.

Its not trivial to deal the noise, they are VERY fast transients and a typical RC filter wont do much for and the other issue is that many DCDC SMP devices have a capacitive load limit so you cant just pile on big caps hoping to solve it. LC filters are normally used in SMPS designs as a result and properly implementing those isnt trivial.


>I'm now wondering whether these are just going to cause more (but different) problems.

I would say yes, and I wouldnt even bother with them. A good SMPS should cost you about the same as a decent linear one. The power supply is the basis of any electronic device, you dont start with the cheapest shittiest thing you can find, you start with the best possible power and work backwards from there to find the "good enough" point. Having a better PSU than you need is a much better prospect than having a much worse one.

[MikeDB]

A good SMPS should cost you about the same as a decent linear one. The power supply is the basis of any electronic device, you dont start with the cheapest shittiest thing you can find, you start with the best possible power and work backwards from there to find the "good enough" point. Having a better PSU than you need is a much better prospect than having a much worse one.

Great advice for everyone. Why spend 1000s of $/£s on modules and then put a $5 supply on it and wonder why you have noise/unreliability/dead modules ? Buy something decent to begin with when building the case and base everything around that supply.

[KSS]

and was going to make the exact same quote -and near same comment- but was working on reply to MDB first.
--------------------------------------------------------------------------------------------------------
@MikeDB
How about pointing us to a couple good SMPS 'replacements' for the typical PowerOne -and clones/derivs- in the 1A and 3A -/+12-15V ?

And aren't Linear's already fiunctionally banned from many countries? <--Yes, there's exceptions, but the trend and focus is clear.

[MikeDB]

and was going to make the exact same quote -and near same comment- but was working on reply to MDB first.
--------------------------------------------------------------------------------------------------------
@MikeDB
How about pointing us to a couple good SMPS 'replacements' for the typical PowerOne -and clones/derivs- in the 1A and 3A -/+12-15V ?

And aren't Linear's already fiunctionally banned from many countries? <--Yes, there's exceptions, but the trend and focus is clear.

I don't use that sort of power supply so don't usually like to make recommendations. I always use an RS 12V or 48V lots of amps switcher and then DC-DC convert locally on each PCB to the voltages I need, which is what this thread was originally dealing with. But an RS 904-8486 switcher with something like the Behringer CP1A may deliver a reasonable result.

And whilst linears are banned in many countries for electronic uses, someone above was knocking together a halogen bulb AC-AC convertor and something off eBay to rectify/regulate it. There are always people who try to get round regulations.

[EATyourGUITAR]

LM317 is pretty slow in response to big fast transients but other than that, I have no problems with it. If you know the magnitude and the frequency of the transients you expect then it is within spec. Good SMPS design is hard to do which is why people avoid it. I can agree with the benefits of SMPS but if you don't design it for me then I just use linear.

[plushterry]

You might want to check out how Seismic did this thing: https://github.com/seismicindustries/IPS2 .

I use the same vReg to go from +5 to +/-12v and its pretty decent. There are dozens of chips out there for all sorts of different input conditions and loads, you just need to hunt down what you want (which is not trivial, there are a LOT of parts). I HIGHLY recommend purchasing development boards for whatever part youre interested in first, this will give you a properly configured and laid out board to test with to make sure that part you want to use will do what you want it to. The manuals for the dev boards are pretty much the instructions for making your own so highly valuable reading

The Seismic part uses the TI TPS65131 DCDC part, here's the evaluation board for it https://www.ti.com/tool/TPS65131EVM-839 and user guide https://www.ti.com/lit/ug/slvuaw7/slvua ... 131EVM-839

Also worth noting, the complexity of small SMPS devices is like that for a reason, the $5 aliexpress devices will probably get you the voltages they advertise but this will come with 100+ mV noise on the outputs. I have a drawer full of them and none have even come close to something usable noise wise.

Thanks, thats an interesting design. I did consider that TPS65131 part initially, well, it's little brother the TPS65130. I decided against it as I wanted to be able to use up to a 12VDC input and the TI part had a max of 6V iirc. the part I went for in the end (LT3741) allowed upto 16V on the input, gave a better output current at the target voltage and was a couple of quid cheaper.

I hadn't considered using USB C as a power source. now that is a good idea.

[Altitude909]

..
I hadn't considered using USB C as a power source. now that is a good idea.

Its not really being leveraged here, its just "dumb" USB C meaning that its just the connector and it doesnt have any of the cool high voltage stuff. USB C is pretty fricken complicated to implement properly if you want to use any of the voltage switching features

[plushterry]

Its not really being leveraged here, its just "dumb" USB C meaning that its just the connector and it doesnt have any of the cool high voltage stuff. USB C is pretty fricken complicated to implement properly if you want to use any of the voltage switching features

hadn't even thought that far, I was just thinking about using USB C as a power adapter

[MikeDB]

hadn't even thought that far, I was just thinking about using USB C as a power adapter

The only decent USB C adapter I've found is the Raspberry Pi 4 one which gives a proper 5V 3A without sagging or excessive noise. There may be others that are good, but I haven't seen one yet :-)

[bawbag]

Has anyone explored using the LT3581?

https://www.analog.com/media/en/technic ... 3581fb.pdf

Page 19 of the datasheet has an interesting configuration using two of the ICs in a master/slave configuration to provide +-12v.

[Portabella]

Has anyone explored using the LT3581?

https://www.analog.com/media/en/technic ... 3581fb.pdf

Page 19 of the datasheet has an interesting configuration using two of the ICs in a master/slave configuration to provide +-12v.

see first post on this page

[bawbag]

Has anyone explored using the LT3581?

https://www.analog.com/media/en/technic ... 3581fb.pdf

Page 19 of the datasheet has an interesting configuration using two of the ICs in a master/slave configuration to provide +-12v.

see first post on this page

I don't understand - I see that you're selling some boards, but how does that relate to the LT3581?

[Portabella]

Has anyone explored using the LT3581?

https://www.analog.com/media/en/technic ... 3581fb.pdf

Page 19 of the datasheet has an interesting configuration using two of the ICs in a master/slave configuration to provide +-12v.

see first post on this page

I don't understand - I see that you're selling some boards, but how does that relate to the LT3581?

It features them dude

[bawbag]

Has anyone explored using the LT3581?

https://www.analog.com/media/en/technic ... 3581fb.pdf

Page 19 of the datasheet has an interesting configuration using two of the ICs in a master/slave configuration to provide +-12v.

see first post on this page

I don't understand - I see that you're selling some boards, but how does that relate to the LT3581?

It features them dude

I see a couple of TO-220 package regulators (LM series?) but not the DFN / MSOP package of the LT3581. Are they mounted on the underside of the board?

[plushterry]

... TI TPS6513x devices are very good ...

... Available as a module from China if you prefer.
e.g. https://www.aliexpress.com/item/1005001869133587.html

so the couple of these I ordered turned up yesterday. thought I'd give em a test to see what they were capable of. I got the 15V version. There's no specs quoted for their design for 15V, but it says +400mA and -350mA for the 12V version.


in all the shots, yellow trace is the negative output, the pink trace is the positive.


this is with no load attached, with 5V input, 8mA current draw.

this is with a small load (44mA) on each side, 5V input 332mA current draw

for this one I added a 220uF cap across each rail (extra to whatever they already have on the board) and repeated the 44mA load, 5V input 332mA draw. (didn't seem to make any noticeable difference)

with double the load, 88mA per side. there was a 638mA draw for this one

so not great really. hopefully the one I've come up with will not be quite as bad. will find out when they arrive

[plushterry]

Without the extra 220uf caps across the rails, the negative rail would not even start up with the 88mA load and the positive rail was a complete mess.

[MikeDB]

Not good. Sorry for suggesting that board but it looked like it should be reasonable as they usually follow the datasheet.

Is that 240nS per division ? If so the spikes are over 20MHz, and the cycle is at 5MHz, which isn't where this device operates normally.

On the output 220uF isn't going to be that effective at those frequencies. Try a few 100nFs.

Can you put the scope on the input lead and see what it is doing to the voltage source ? I suspect that might be doing something silly. Next thing would be to put a 1 Ohm resistor in the positive supply lead and monitor the current waveform being drawn.

Another thing to try is to lift the negative input resistor (R7 in the data sheet) so as to disable the negative supply and see if the positive improves, as it is possible the negative supply is going into power-save mode. This would indicate the grounding is possibly poor and the negative output is affecting the positive feedback, or vice versa, or both.

[plushterry]

no worries, I knew it was going to be a gamble!


thanks for the suggestions, I will have a play with it tomorrow, should go to bed now really instead of getting stuck into something else.