Rpschultz13
Well-known member
I’ve studied this guys videos, specifically on grounding. Great stuff, very informative:
Getting Started with PCBA at JLCPCB
- Thread starter z2amiller
- Start date
- Featured
z2amiller
Active member
I built a board using through-hole components and wanted to see what the JLCPCB assembly would look like for a purely through-hole board.
For purely through-hole, there's no setup cost and no stencil fee. However, the cost-per-unit for assembly is a lot higher.
For really small runs of boards, I think THT boards would actually be cheaper due to the smaller up-front setup costs.
There's no "Basic vs Extended" juggling because everything is extended but you don't pay a per-extended-part fee
like you do with SMD.
In my example, the boards come out to about $7.50/ea with THT (This is for 5x4-layer boards):
To stack that against a SMT board, here's a board with similar parts count / complexity that is SMT:
You can see the $8 setup and $9 in extended fees, but both the assembly cost per board and the component costs are lower.
The component cost for this one is artificially low since some components come out of my pre-purchased parts library.
But I do notice that the SMT parts are significantly cheaper than the THT equivalents.
So anyway, for runs of 5-10 boards the costs will be pretty similar, but for more than that, the lower per-board cost
and lower component costs will quickly offset the setup costs of SMT.
For purely through-hole, there's no setup cost and no stencil fee. However, the cost-per-unit for assembly is a lot higher.
For really small runs of boards, I think THT boards would actually be cheaper due to the smaller up-front setup costs.
There's no "Basic vs Extended" juggling because everything is extended but you don't pay a per-extended-part fee
like you do with SMD.
In my example, the boards come out to about $7.50/ea with THT (This is for 5x4-layer boards):
To stack that against a SMT board, here's a board with similar parts count / complexity that is SMT:
You can see the $8 setup and $9 in extended fees, but both the assembly cost per board and the component costs are lower.
The component cost for this one is artificially low since some components come out of my pre-purchased parts library.
But I do notice that the SMT parts are significantly cheaper than the THT equivalents.
So anyway, for runs of 5-10 boards the costs will be pretty similar, but for more than that, the lower per-board cost
and lower component costs will quickly offset the setup costs of SMT.
vigilante398
Authorized Vendor
Yeah I have a friend that had a through-hole design he wanted set up for JLC assembly and I was shocked at how much cheaper it was than my usual SMD boards (which admittedly use a large number of weird extended parts).
Rpschultz13
Well-known member
I had NO idea you could get thru-hole assembled like SMD.
vigilante398
Authorized Vendor
Yeah I've been getting mixed assembly (mostly SMD with through-hole film caps) for a while now, but last month was my first time trying a 100% through-hole board. It's funny how much you save without feeder loading fees, setup fees, stencil fees, etc.I had NO idea you could get thru-hole assembled like SMD.
z2amiller
Active member
I had NO idea you could get thru-hole assembled like SMD.
I've been adding some THT parts to all of my designs. Sometimes box film caps for larger values, and always JST connectors for the footswitch board. It adds a bit of cost (since there's a $3.50 THT setup cost even if you just add one component) but it's worth it to me for time savings. I hate soldering wires to things so much.
I also occasionally add DIP sockets rather than SOIC-8 op-amps in case I want to goof around with swapping out different opamp chips, or if I'm building a chorus or delay that uses PT2399 or BBD chips.
really useful @z2amiller thanks!
jessenator
Well-known member
I have a very simple board (literally 1 resistor and 1 diode), but my EDA is not spitting out a proper BOM. Does anyone have an existing BOM CSV I could reference?
The parts I would need are in the basic library, but I want to make sure the columns are right
The parts I would need are in the basic library, but I want to make sure the columns are right
z2amiller
Active member
The bom looks like:
The CPL looks like:
Code:
Comment,Designator,Footprint,LCSC,Quantity
10K,"RV1,RV4",RTRIM3362P,C118956,2
Conn_01x03_Socket,"J1,J2",JST_PH_B3B-PH-K_1x03_P2.00mm_Vertical,C131339,2
100pf,"C19,C21,C22,C23",C_0805_2012Metric,C1790,4
1n5819,D1,D_SOD-323,C191023,1
100R,R1,R_0603_1608Metric,C22775,1
15K,R3,R_0603_1608Metric,C22809,1
3.3K,R6,R_0603_1608Metric,C22978,1
4.7k,R12,R_0603_1608Metric,C23162,1The CPL looks like:
Code:
Designator,Val,Package,Mid X,Mid Y,Rotation,Layer
C15,2.2nF,C_0805_2012Metric,57.45,-76.85,-90.0,top
C16,2.2nF,C_0805_2012Metric,60.7,-76.85,-90.0,top
C17,2.2nF,C_0805_2012Metric,63.95,-76.85,-90.0,top
C18,100uF,CP_Elec_6.3x5.4_Nichicon,60.1,-70.8,0.0,top
C19,100pf,C_0805_2012Metric,85.9,-73.75,90.0,top
C20,10uF,C_0805_2012Metric,96.05,-65.55,90.0,top
C21,100pf,C_0805_2012Metric,89.3,-75.8,90.0,top
C22,100pf,C_0805_2012Metric,92.625,-75.9,90.0,top
C23,100pf,C_0805_2012Metric,89.3,-72.4,-90.0,top
C24,2.2nF,C_0805_2012Metric,74.05,-85.55,0.0,top
C26,2.2nF,C_0805_2012Metric,96.05,-58.575,-90.0,top
C27,2.2nF,C_0805_2012Metric,92.8,-58.625,-90.0,topjessenator
Well-known member
Thanks for copying all of that. Well, it looks like I'm scuppered, and just need to learn a more modern/supported EDA.
I tried following a few tutorials using a few scripts, but it's just not spitting out what I need. I imagine if I send it, it'll either get returned or it'll populate 1 of 48 mini PCBs. Blurg. At the very least, JLC at least confirmed that the routing I had should work and it will all count as panelized.
z2amiller
Active member
Good news is, you _should_ be able to see this state of affairs in the UI. After you upload the BOM and CPL files,
it shows you a rendering of the board. And someone at JLC does check it, even it you don't have the option checked,
when they're setting it up they'll email you if they spot anything that looks really wrong.
z2amiller
Active member
Good News Everyone!
As of about a week ago someone committed a patch to the Bouni KiCad tools to fix the in-app search
functionality, and the components are searchable again. You don't need to update the plugin itself, but
you do need to re-download the parts library ("Download", in the upper right side of the tool).
Unfortunately there's still no filter for 'Promotional Extended' - I guess I need to file a feature request!
As of about a week ago someone committed a patch to the Bouni KiCad tools to fix the in-app search
functionality, and the components are searchable again. You don't need to update the plugin itself, but
you do need to re-download the parts library ("Download", in the upper right side of the tool).
Unfortunately there's still no filter for 'Promotional Extended' - I guess I need to file a feature request!
thunderaxe
Active member
i'm curious if there are any good SMD choices for 1uF caps in the signal path. should i just go with electrolytic?
vigilante398
Authorized Vendor
Does it have to be SMD? JLC also does through-hole assembly, they have plenty of options for through-hole 1uF film caps. That's what I do for caps in the signal path.
thunderaxe
Active member
it doesn't have to be SMD, but i know that there are tons of pedal companies doing everything SMD, quite a few of them are getting their PCBs made through JLC, and using SMD 1uF caps in the signal path, so i'm wondering what they're doing. i can't find any good 1uF SMD options on JLC other than electro
z2amiller
Active member
Nothing that I've seen that is SMD and super small for the audio path.
JLC will wave solder in a film capacitor for you, though.
(C125386 is a the box film capacitor I use, 4.5mm wide).
This is usually what I do for values between ~220nf -> 1uF.
The first through-hole component you add costs something like
$3.50 for setup, but after that it's like 2-3 cents per solder joint.
So you can actually still do box film caps everywhere if you want.
Honestly for me when I'm doing the first pass of a design,
lately I've just used the 'Basic' MLCC ceramics (heresy!).
I've thought about creating an overlapping footprint with holes for the
through-hole film but also with a SMD footprint for a ceramic I
can have PCBA'd so I can choose later and not spend as much
money if I screw up a design. (Also I'm not 100% convinced
that ceramics are bad in the audio path but I'm half deaf anyhow)
JLC will wave solder in a film capacitor for you, though.
(C125386 is a the box film capacitor I use, 4.5mm wide).
This is usually what I do for values between ~220nf -> 1uF.
The first through-hole component you add costs something like
$3.50 for setup, but after that it's like 2-3 cents per solder joint.
So you can actually still do box film caps everywhere if you want.
Honestly for me when I'm doing the first pass of a design,
lately I've just used the 'Basic' MLCC ceramics (heresy!).
I've thought about creating an overlapping footprint with holes for the
through-hole film but also with a SMD footprint for a ceramic I
can have PCBA'd so I can choose later and not spend as much
money if I screw up a design. (Also I'm not 100% convinced
that ceramics are bad in the audio path but I'm half deaf anyhow)
That's actually a great idea.
thunderaxe
Active member
the conventional wisdom i've seen repeated in multiple places is that C0G/NP0s are fine for the signal path but X7Rs and worse are too unstable in their capacitance to use for the signal path. and C0G/NP0 pretty much only exists in the pF range.
X7Rs are totally fine for a 100nF power filter cap though because that doesn't need to be a perfect value.
You've heard or even used plenty of circuits with X7R caps in the signal path. One way to minimise the "downsides" is to over spec the voltage of X7Rs and Electros.