Converting an SMD board to non-SMD - Is it as easy as replacing the components?

[ThermallyFrigid]

Good day!

Today I have a dilemma concerning converting a PCB that was originally an SMD board, into a board that does not use SMD components.

I am wondering if it is as simple redesigning the board so that all the SMD components can be replaced with equal value NON SND components or.........

Does going with non SMD components change the electrical characteristics of the circuit such that the values of components would have to be recalculated?

Thanks

[capt bullshot]

Well, depends ...

For sure, there's a lot of designs that can be converted from THT to SMT or vice versa without affecting the operation at all.

But there's always some special cases, e.g.

- stuff that is RF related and / or has impedance controlled traces
- EMI / EMC might change completely
- thermal design might change, e.g. you'd need a heat sink on a component that is cooled by copper layers on the SMD design

...

[T3sl4co1l]

Eww, moving to THT?..

Well, assuming it's economical or whatever -- and that equivalent components are available at all THT -- yeah, as above.  I would merely strengthen this a bit: EMC is guaranteed to change, that's simply true for overhauling the layout and growing dimensions.

There's certainly plenty of history of moderately wideband designs in THT; the Tek 475 for example boasts 200MHz bandwidth with all THT construction (granted, with the important parts done in hybrids/ICs!).  But that's likely a finely tuned layout.  Repeatable, manufacturable -- definitely -- but maybe not something you can just convert between, the impedances will be different enough to need a lot of work.  But that's also analog, a rare sight these days; digital stuff would be fine.  But then also, how many digital ICs are even available in THT with any kind of bandwidth (i.e. 74LVC or faster)?   (Which, 74LVC itself is, so there's that, but I don't know of any MCUs for example that come in DIP that are much faster than, either 74HC or about halfway between HC and LVC?)

Tim

[strawberry]

flexing board , component reliability is better for TH parts, especially for ceramic capacitors
hermetic Hi rel parts available in TH
I think SMD el.capacitors are BS (hard to rework)

[ThermallyFrigid]

Thanks guys.

It's a fairly simple board with only a LM358 and a TL494, a few capacitors, 6 diodes, a handful of resistors and about half a dozen capacitors
It measured about 2 x 3 in SMD and will measure about 3 x 4 in THT.

The TL494 gives me the impression it's what provides the frequency for the charging process but I'm not sure.

I'm doing it just to make it super easy to check or replace components in the future if they need replacing.
It's a sub board on a $60 LiFePO4 charger that runs a bit hot inside.

[langwadt]

Thanks guys.

It's a fairly simple board with only a LM358 and a TL494, a few capacitors, 6 diodes, a handful of resistors and about half a dozen capacitors
It measured about 2 x 3 in SMD and will measure about 3 x 4 in THT.

The TL494 gives me the impression it's what provides the frequency for the charging process but I'm not sure.

I'm doing it just to make it super easy to replace components in the future if they need replacing.
It's a sub board on a $60 LiFePO4 charger that runs a bit hot inside.

SMD is easier to replace

[ThermallyFrigid]

SMD is easier to replace

I find it more difficult because of the size of the components.
Maybe with more practice I'll find it easier like you do.

I will say I have definitely gotten much better at it.
SMD components scared me early in the process of all this...now not so much.

Do you have any educational videos on YouTube you care to share of your techniques that make it easy?

Thanks

[ThermallyFrigid]

So the THT version is ready for fabrication.
No idea if it will actually work but I'm going to order a few boards, build them and see.

If you see any glaring mistakes please point it out.
Any guesses as to if it will work?
The SMD board is better looking and a good bit smaller.  I'll probably make a few of the SMD boards to have on hand also.
PS
If you happen to have a BtrPower 5A LiFePO4 charger and would like an extra sub board, I'll have a few extras.
Thanks

The original SMD board




The PCB Layout of the THT version



A 3D image of what the final board will look like.....

[Mr.B]

I wonder why you are splitting that track.
There is an alternate route that would perhaps be better.
Pic below.

[ThermallyFrigid]

Thanks !
Excellent.  Just my oversight.
After you've been staring at something for far too long crazy things happen to your vision and your thinking I suppose  haha

I suspect there's more of those if you look too hard   
I'll just be a happy camper if it even works

I use a lot of LiFePO4 battery packs in my Caravan for camping in general so having a few extra chargers on hand while away on the road is vital.
When this one went bad I was 1000 miles away from home and had to spill the beans and buy a new (better) one on Amazon.
To Amazon's credit, it arrived the next day

[mariush]

You know, you don't HAVE TO keep the components in exactly the same locations, unless there's some height restrictions.
You could simplify the layout a bit by shifting some things around.

For example
rotate Q1 and Q2 90 degrees which would cause D3 and D4 to go horizontal, but allows for better traces
UK2 could be horizontal and then you wouldn't have to go on the other side with the trace going to R11
R15 could be turned vertical  and then UK3 and R16 could be shifted to the right so that the green trace from the bottom don't have to move between them
C1 could be flipped so that the green trace doesn't have to wiggle through so many components

C7 and C8 negative to the top to have all electrolytic capacitors ground at the top,
U2 could be rotated 90 degrees to the left


Also, you could probably use some jumper links (0 ohm resistors) to jump over some traces instead of using vias or traces on the bottom.

[T3sl4co1l]

Ah, low density, not a big deal then.

Ground plane could be nice, if more traces can be pushed to one layer (possibly at expense of a few more vias, or longer routes around components?).  Most glaring part I see is, what looks to be GND on the far left pin, hooking around the whole board before the outputs (TL494I and BJTs?).  Probably still not fast enough / enough current flow, to warrant the effort.  Maybe the opamp would get a bit more offset (AC rectification) like this, probably still fine (I mean, it's not a LM358, maybe it would be a concern then).

Jumpers would provide more opportunities for that obviously; I think I wouldn't bother, not for something this simple.  Solve it with vias first, only once that's been exhausted, use some.

Trace can be MUCH thinner, of course, which will greatly aid these things.  But it's simple enough, again, this probably makes little change to the overall layout.

Ed: ohhhh weird, I think you misread R7-R10, give that a double check.  Maybe it's actually like that, but it looks weird.
And if they are actually in series, nothing wrong with reducing them to a single part.

Tim

[ThermallyFrigid]

Ah, low density, not a big deal then.

Ground plane could be nice, if more traces can be pushed to one layer (possibly at expense of a few more vias, or longer routes around components?).  Most glaring part I see is, what looks to be GND on the far left pin, hooking around the whole board before the outputs (TL494I and BJTs?).  Probably still not fast enough / enough current flow, to warrant the effort.  Maybe the opamp would get a bit more offset (AC rectification) like this, probably still fine (I mean, it's not a LM358, maybe it would be a concern then).

Jumpers would provide more opportunities for that obviously; I think I wouldn't bother, not for something this simple.  Solve it with vias first, only once that's been exhausted, use some.

Trace can be MUCH thinner, of course, which will greatly aid these things.  But it's simple enough, again, this probably makes little change to the overall layout.

Ed: ohhhh weird, I think you misread R7-R10, give that a double check.  Maybe it's actually like that, but it looks weird.
And if they are actually in series, nothing wrong with reducing them to a single part.

Tim

Very good eye Tim!
You were spot on.

Does this look better?

[T3sl4co1l]

Still weird: why is R8-R9 connected in series between the bases?

The saturation voltage on the C1/C2 outputs isn't very low anyway, I think, making the direct connection dubious; 2N7000s could be used instead, perhaps (with a somewhat lower value pullup to +5V giving similar risetime), or a resistor divider so that Vbe(off) can be lower.  This would require 3 resistors per channel so I'm not quite sure which way they would've had it originally.

If in doubt, ohm out the original?

Tim

[ThermallyFrigid]

Want to hear something really weird?

At the location of UK3 (unknown 3), I just de soldered that and found 2 resistors piggybacked one on top of the other

I confirmed this was intentional because the other unit is exactly the same.

I've never seen this done.  Why would they solder a 3001 resistor on top of a 182 resistor?So they have a 3K resistor with a 1.8k resistor on top.  They are soldered together?

Why not just use a 4.8k resistor? (or even a 5K resistor) ?

As if that's not enough, the 182 resistor is a 1206 and the 3001 resistor on top is an 0805 I think.

[Venturi962]

Double stacked SMD resistors would be in Parallel, so 3K + 1.8K would be 1.125K.  Maybe they had these values lying around instead of 1K?  Or it was meant to be 1K and they accidentally put a 1.8K.

[T3sl4co1l]

Not just replacing it with an equivalent resistor seems odd, but a parallel combo is about as much effort to rework so I guess that works out.

Stacked chips isn't uncommon, for rework purposes; I've had to do it plenty of times myself.

Tim