You accept that there's a lower size limit below which capacitors are not useful in this context.
Therefore you accept that there's a lower size limit below which inductors are not useful in this context. The only question is where the limit lies.
In order to be reasonably described as a filter, a structure needs to have series elements above the noise, and by that I mean a reactance significantly above an interconnect resistance routinely encountered.
An inch of 1/2 oz. 10 mil copper track has a resistance of 0.971 ohm (The CircuitCalculator.com Blog PCB Trace Width Calculator). A 150 uH inductor has a reactance of 0.094 ohm @ 100 Hz (2*pi*f*l). It does not meet the above criterion.
Uncounted engineers worldwide set the limit of usefulness roughly where I have set it.
That's why low (and high) voltage power supplies built by engineers look the way they do, all roughly the same. That and the fact that resistors have a frequency response out into the gigahertz. And a few other things none of which have anything do with the choke-fed red herring you trailed earlier, which you have evidently chosen to abandon. I wonder why?
I'm trying to teach some electronics to a novice who has already had his head filled with a load of half-understood nonsense. Everybody else here with an ounce of commonsense has recognized that this guy is trying to run before he can walk, but you're standing on the sidelines shouting 'faster'.
If you cannot see the OP's circuits or mine, then you are to a degree ill-placed to comment, so why do so? If you can see his, then perhaps your efforts might be better spent putting your back into the attempt to explain the practical pitfalls he is likely to encounter, if indeed such sociability is not constitutionally beyond you.
I should know better. There's no point in arguing economics or dimensions with somebody who thinks nothing of building a cubic foot of cubic inch capacitors for a shunt.
Therefore you accept that there's a lower size limit below which inductors are not useful in this context. The only question is where the limit lies.
In order to be reasonably described as a filter, a structure needs to have series elements above the noise, and by that I mean a reactance significantly above an interconnect resistance routinely encountered.
An inch of 1/2 oz. 10 mil copper track has a resistance of 0.971 ohm (The CircuitCalculator.com Blog PCB Trace Width Calculator). A 150 uH inductor has a reactance of 0.094 ohm @ 100 Hz (2*pi*f*l). It does not meet the above criterion.
Uncounted engineers worldwide set the limit of usefulness roughly where I have set it.
That's why low (and high) voltage power supplies built by engineers look the way they do, all roughly the same. That and the fact that resistors have a frequency response out into the gigahertz. And a few other things none of which have anything do with the choke-fed red herring you trailed earlier, which you have evidently chosen to abandon. I wonder why?
I'm trying to teach some electronics to a novice who has already had his head filled with a load of half-understood nonsense. Everybody else here with an ounce of commonsense has recognized that this guy is trying to run before he can walk, but you're standing on the sidelines shouting 'faster'.
If you cannot see the OP's circuits or mine, then you are to a degree ill-placed to comment, so why do so? If you can see his, then perhaps your efforts might be better spent putting your back into the attempt to explain the practical pitfalls he is likely to encounter, if indeed such sociability is not constitutionally beyond you.
I should know better. There's no point in arguing economics or dimensions with somebody who thinks nothing of building a cubic foot of cubic inch capacitors for a shunt.
Not just accept but also stated earlier in this thread - for example that ceramics wouldn't be a valuable addition. However the context in part is set by the inductor sizes, another part of the context is the impedance of the supply.
Indeed, given the caveat I mentioned above. The context here is the choice of capacitor size.
Its not really a limit more of a slippery slope of lower and lower returns on time/money spent on components.
I followed the link and it showed an answer of 0.0971 ohm, 10X lower than you show here. 1/2oz copper is not standard - PCB houses tend to use 1oz. A 10mil copper track is too narrow for power, its a signal track.
All in all - just more chain yanking.
Actual trace resistance is 0.113 Ohms, use Saturn PCB toolkit:
Saturn PCB Design - PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
Actually 1/2oz copper is a standard thickness these days, especially where 10z finish is required for PTH boards, they are etched on 1/2oz the rest is built up during plating. This also limits etch back getting the trace nearer to the desired width. The majority of boards have SMD components on them and using 1/2oz as a starting weight for outer layers is preferable. 1oz is used for inner layers that require that weight as there is obviously no extra weight added during plating. A 10 mil trace would be good for 0.7A, but would have much more inductance than a thicker trace or a copper pour. (And here don't just think planes, on power boards copper pours are used to sculpt areas of copper where maximum current but minimum copper (for EMC reasons) is required.)
Saturn PCB Design - PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
Actually 1/2oz copper is a standard thickness these days, especially where 10z finish is required for PTH boards, they are etched on 1/2oz the rest is built up during plating. This also limits etch back getting the trace nearer to the desired width. The majority of boards have SMD components on them and using 1/2oz as a starting weight for outer layers is preferable. 1oz is used for inner layers that require that weight as there is obviously no extra weight added during plating. A 10 mil trace would be good for 0.7A, but would have much more inductance than a thicker trace or a copper pour. (And here don't just think planes, on power boards copper pours are used to sculpt areas of copper where maximum current but minimum copper (for EMC reasons) is required.)
So marce you mean the bare copper board begins life as 1/2oz but after electroless plating it becomes 1oz? When I said '1oz is standard' I meant the finished item, not the raw material. Assuming then that the conductivity of the deposited copper doesn't vary much (its bound to be a little less dense) from the original on the PCB, the 1oz figure would be the one to use to get the correct resistance.
Since the OP only wants a power supply there would be no need for SMT parts - I'd be inclined to specify 2oz (finished) copper which is what I did on my last batch of PCBs. Given the OP's current specification is for 2A then an absolute minimum trace thickness would be 30mil, however if I were designing I'd use a copper pour which would have at least 200mils effective width, given the physical bulk of the components involved.
Since the OP only wants a power supply there would be no need for SMT parts - I'd be inclined to specify 2oz (finished) copper which is what I did on my last batch of PCBs. Given the OP's current specification is for 2A then an absolute minimum trace thickness would be 30mil, however if I were designing I'd use a copper pour which would have at least 200mils effective width, given the physical bulk of the components involved.
Yes its standard when doing PTH boards, but the finished copper weight will be 1oz or a tiny bit thicker. So the 1oz figure is correct, generally unless you are doing RF or controlled impedance boards having extra thickness of copper is a benefit.
I agree with the 2oz copper or heavier, if you specify 2oz base laminate and have PTH holes you'll get thicker copper due to plating.
The most fun PSU boards I have done had 4oz on one side and some inner layers, with 2oz (finished) on the component side due to SMD components being used...Not to be recommended unless you have deep pockets.
When choosing a trace thickness I use the Saturn toolkit, dial in the maximum Board working temperature and set an over temp of only 10 deg.C.
Etch back compensation can catch people out when they are doing finer SMD designs, luckily the PCB manufacturer will choose a base laminate and process to fulfil your requirements where they can.
I agree with the 2oz copper or heavier, if you specify 2oz base laminate and have PTH holes you'll get thicker copper due to plating.
The most fun PSU boards I have done had 4oz on one side and some inner layers, with 2oz (finished) on the component side due to SMD components being used...Not to be recommended unless you have deep pockets.
When choosing a trace thickness I use the Saturn toolkit, dial in the maximum Board working temperature and set an over temp of only 10 deg.C.
Etch back compensation can catch people out when they are doing finer SMD designs, luckily the PCB manufacturer will choose a base laminate and process to fulfil your requirements where they can.
- Status
- Not open for further replies.