Modifying my cheap Chinese power supplies...

Man, monk-y, we gotta get you a transformer and a diode bridge and a big ol cap.

Then maybe add in a regulator, LC low pass filter, ferrite beads, more caps...

On second thought, mains voltage is dangerous. I'd rather keep ya around. Keep on fiddlin, bruh.
 
Man, monk-y, we gotta get you a transformer and a diode bridge and a big ol cap.

Then maybe add in a regulator, LC low pass filter, ferrite beads, more caps...

On second thought, mains voltage is dangerous. I'd rather keep ya around. Keep on fiddlin, bruh.

I just wanted to make this little thing a bit more versatile. It will only take a few small modifications to do that.
 
I’m going to try and pull some more LM2596 design documents and see if there are any small mods to improve noise rejection as well.
 
So, in the continuing saga of Derek doesn't know shit about power supply components, I give you today's "Dear @vigilante398" questions:

In reviewing the literature on noise and the LM2596 last night, the importance of low ESR electrolytic capacitors was continuously hammered upon. There is wild inconsistency even in the power supplies that I have in my possession.

The 3 larger 10 output supplies represented a cross section of build methods. One has all low ESR electros, one has them on certain parts of the supply and one had none at all.

The 2 smaller supplies I am using on our boards here have no low ESR caps at all. This is an easy enough fix with a little trip top Mouser.

My real question centers around determining the frequency band of the supply noise I am getting and specifying the new ferrite beads to assist in killing some of this noise. Since ohmic impedance/resistance is frequency dependent for these beads, is it as simple as identifying the frequency band the noise "lives" in and use the data sheet to specify the ohmic impedance/resistance?
 
So, in the continuing saga of Derek doesn't know shit about power supply components, I give you today's "Dear @vigilante398" questions:

In reviewing the literature on noise and the LM2596 last night, the importance of low ESR electrolytic capacitors was continuously hammered upon. There is wild inconsistency even in the power supplies that I have in my possession.

The 3 larger 10 output supplies represented a cross section of build methods. One has all low ESR electros, one has them on certain parts of the supply and one had none at all.

The 2 smaller supplies I am using on our boards here have no low ESR caps at all. This is an easy enough fix with a little trip top Mouser.

My real question centers around determining the frequency band of the supply noise I am getting and specifying the new ferrite beads to assist in killing some of this noise. Since ohmic impedance/resistance is frequency dependent for these beads, is it as simple as identifying the frequency band the noise "lives" in and use the data sheet to specify the ohmic impedance/resistance?
So looking at the datasheet you have a fixed switching frequency of 150kHz, which is pretty great. Obviously the higher you are above the audio spectrum (20Hz - 20kHz) the better, and the less likely you are to have audible noise leaking into your signal. But if you're looking to target the switching frequency for noise reduction, 150kHz is where it's going to be at. There are a few notes in the datasheet for reducing ripple (noise), but the summary is "garbage in, garbage out." The cleaner you scrub your input, the cleaner the output you'll get.

As far as low ESR caps, as long as you don't need a huge amount of capacitance, which you usually don't (except in very high current situations), MLCC are hard to beat for the price and I mostly rely on them for power supply filtering. 0805 are my favorite size, and I can get 0805 up to 10uF in voltages high enough to be useful, 1210 gives you even more options. Also for modding existing boards SMD caps are great because if you need more capacitance you can just keep stacking caps on top of each other :p
 
  • Love
Reactions: fig
So looking at the datasheet you have a fixed switching frequency of 150kHz, which is pretty great. Obviously the higher you are above the audio spectrum (20Hz - 20kHz) the better, and the less likely you are to have audible noise leaking into your signal. But if you're looking to target the switching frequency for noise reduction, 150kHz is where it's going to be at. There are a few notes in the datasheet for reducing ripple (noise), but the summary is "garbage in, garbage out." The cleaner you scrub your input, the cleaner the output you'll get.

As far as low ESR caps, as long as you don't need a huge amount of capacitance, which you usually don't (except in very high current situations), MLCC are hard to beat for the price and I mostly rely on them for power supply filtering. 0805 are my favorite size, and I can get 0805 up to 10uF in voltages high enough to be useful, 1210 gives you even more options. Also for modding existing boards SMD caps are great because if you need more capacitance you can just keep stacking caps on top of each other :p

I found 2 sets of Low ESR electros that will work as far as size and value are concerned.

I think I may use these for the beads:


Another thing the data sheets and desgn docs point out is an output LC Filter can further reduce noise.
 
Well I have a mouser cart mostly populated. I just need to figure out the value for the current value for the new Zener.
If I had to guess (which I don't, but will anyway) it's probably 1/2W, which means if you're trying to come from an 18V supply and drop to 12V, you need 6V * output current, so a 1/2W zener will give you 83mA on the 12V line. I would suggest a 1W zener like this guy, that would give you just over 150mA on the 12V tap.

 
If I had to guess (which I don't, but will anyway) it's probably 1/2W, which means if you're trying to come from an 18V supply and drop to 12V, you need 6V * output current, so a 1/2W zener will give you 83mA on the 12V line. I would suggest a 1W zener like this guy, that would give you just over 150mA on the 12V tap.


My man!

The 300 mA ferrite beads I picked out are a little narrower and taller than what’s installed now but the length is bang on so I’m not concerned.

I’m also entertaining patching in an LC filter at the output like in the data sheets but I’ll do the easy modifications first.
 
Imagine if Derek bought one good power supply instead of the 5-9 chitauri power supplies he has now

If I got a full refund for all 5 I have in my possession, I still couldn’t buy a good power supply!

But yes, I agree. More of a cool experiment and excuse to learn about power supplies.
 
In spite of having my balls busted about this topic (those guys know I’m the sense of humor champion when it comes to good natured ribbing) I am still finding aspects of it that I’m curious about.

Case in point: @vigilante398 pointed out during the discussion on filtering and ferrite beads that the old adage “garbage in, garbage out” applies here.

That got me thinking (Very dangerous. Can you smell the smoke?): part of my plan for the few small mods I plan to do was a new wall-wart that could deliver 2A instead of the 1A these ship with. Part of the reason was that if we tally the rated full load amperage of the power supply, it exceeds the 1A rating of the wall-wart supplying it by 100 mA.

Now granted, there is safety margin inherently built in because most pedals you’d plug into a 9 vDC/100 mA aren’t actually drawing 100 mA. Yet I plan to modify the 100 mA taps with 300 mA ferrite beads to expand the capability of the taps themselves.

So now I’m rambling.

What I’m driving at is, and I think I’m understanding this correctly, that if we want to address the quality of the input power to the power supply itself, the quality of the wall-wart/inout power would be a good place to do so.

So I’m thinking about the following:

1.) Start with a small enclosure like a 1590a

2.) Use a small step down transformer and rectifier to convert 120 vAC on the primary to 24 vDC on the secondary.

3.) Use the same LM2596 regulator setup on a small board with all design document best practices and output LC filter to provide a nicely regulated 18 vDC on the output.

4.) Feed this output to the power supply ensuring at least 2A available current.
 
In spite of having my balls busted about this topic (those guys know I’m the sense of humor champion when it comes to good natured ribbing) I am still finding aspects of it that I’m curious about.

Case in point: @vigilante398 pointed out during the discussion on filtering and ferrite beads that the old adage “garbage in, garbage out” applies here.

That got me thinking (Very dangerous. Can you smell the smoke?): part of my plan for the few small mods I plan to do was a new wall-wart that could deliver 2A instead of the 1A these ship with. Part of the reason was that if we tally the rated full load amperage of the power supply, it exceeds the 1A rating of the wall-wart supplying it by 100 mA.

Now granted, there is safety margin inherently built in because most pedals you’d plug into a 9 vDC/100 mA aren’t actually drawing 100 mA. Yet I plan to modify the 100 mA taps with 300 mA ferrite beads to expand the capability of the taps themselves.

So now I’m rambling.

What I’m driving at is, and I think I’m understanding this correctly, that if we want to address the quality of the input power to the power supply itself, the quality of the wall-wart/inout power would be a good place to do so.

So I’m thinking about the following:

1.) Start with a small enclosure like a 1590a

2.) Use a small step down transformer and rectifier to convert 120 vAC on the primary to 24 vDC on the secondary.

3.) Use the same LM2596 regulator setup on a small board with all design document best practices and output LC filter to provide a nicely regulated 18 vDC on the output.

4.) Feed this output to the power supply ensuring at least 2A available current.
Finding a step down transformer with enough output current that still fits in a small box will be the hardest part there, but you're absolutely on the right track.
 
Finding a step down transformer with enough output current that still fits in a small box will be the hardest part there, but you're absolutely on the right track.

Yes. My Mouser, Jameco, and Digikey searches proved that!

I could also be a simple box that has the advanced filtering and a regulator providing clean 18 vDC to the main supply. Basically plug a 24 vDC/2A wall watt into this box and then I put the input to the supply.

I ended up finding a True Tone CS7 last night for a good price but I’m still pursuing this, especially for my sons little board.
 
So I brought my laptop to the music store today when my Son was at his drum lesson and worked up a quick PCB for the regulated 18vDC patch box.

It will fit in a 1590LB and I’m going to order 5 when I get the Fuzz 2022 board ordered.
 
Back
Top