JLCPCB 4 Layer boards

[EEVblog]

[isometrik]

Thank you Dave for posting this (and all other) video(s).

I was a bit hesitant to try JLCPCB, for no good/rational reason, but this video convinced me to try them out on my next multilayer design.

Keep up the good work. I am an avid fan of your contents.

[analogo]

Has anybody experience with importing small boards from JLCPCB to Germany or Austria? Problems with the customs?

[janoc]

Has anybody experience with importing small boards from JLCPCB to Germany or Austria? Problems with the customs?

I did it only to France and no problems at all. They declare the package for the customs correctly and unless you are buying something large/expensive the price is generally under the minimum threshold, so I didn't have to pay neither customs nor VAT on import - but I have used regular post. It may be different if you use UPS/DHL/whatever courier service they offer - those will do the paperwork and charge you for it by default, even when it is not required.

[Spark-Doctor]

JLCPCB will not make a false deceleration on the customs paperwork so what you pay is what they declare. There is 0% import duty on PCB's in the European Union but each member state has a different VAT level so you will have to check what yours is. In the UK, it about £45.00 and nearly all my orders come in under this.

 

[Mr. Scram]

What would be the difference between "prototype boards" and production boards, both functionally and price wise?

[Monkeh]

JLCPCB will not make a false deceleration on the customs paperwork so what you pay is what they declare. There is 0% import duty on PCB's in the European Union but each member state has a different VAT level so you will have to check what yours is. In the UK, it about £45.00 and nearly all my orders come in under this.

... try £15

[bodger]

... try £15

Yep definitely £15 it's on the DHL's website, the only thing I wasn't sure of was if it included shipping price too. Turned out it didn't, as I didn't get charged on my $37 order (inc $25 delivery).

It's £39 if the customs declaration is 'gift', but jlcpcb make accurate customs declarations.

[mrpackethead]

I'm using both JLCPCB and AllPCB.     many orders and hundreds and thousands of boards.   

Allpcb is probably a little cheaper for 4 layer pcbs.  The line that does their four layer boards is different from teh two layer line,and its noticablly better.     No issues with mask or drill registrations..         for two layer boards JLCPCB wins.     
both now do electropolished stencils, for $20 or so..  Well worth the money.   

Seriously its at a point now, that its cheaper to desgin 4 layer boards.  The time savings in a four layer board over a two layer, easily make up for it.     Not to mention your boards are smaller, and better designed!

[bodger]

I'm using both JLCPCB and AllPCB.

How does the quality of ALLPCB compare to JLCPCB? I'm thinking about trying them out as they do 10 PCBs for $15 which includes DHL delivery, hence they work out cheaper than JLC for prototypes and fast delivery. Thanks.

[EEVblog]

What would be the difference between "prototype boards" and production boards, both functionally and price wise?

Prototype boards share processes and panel space with other customers.
Production boards get your own panel, your own artwork, more attention to details like reading your manufacture notes on the gerbers, and more care and process control in manufacture, plating, and artwork alignment.
Plus you usually get a sectioned sample so your QA department can test and inspect the quality and via plating etc.

[EEVblog]

Seriously its at a point now, that its cheaper to desgin 4 layer boards.  The time savings in a four layer board over a two layer, easily make up for it.     Not to mention your boards are smaller, and better designed!

Totally. The game has changed.

[BrianHG]

@Dave, in your video, you only examine the alignment registration on the top of the PCB.  You never show the bottom, or, if there are any differences in alignment on each of the 5 PCBs.

Example, an order I made from PCBWay had 3 out of 20 PCB had poor drill alignment on the top only, though they passed continuity.  A second re-order of the same gerbers had the wrong drill size for the mounting holes...

[EEVblog]

@Dave, in your video, you only examine the alignment registration on the top of the PCB.  You never show the bottom, or, if there are any differences in alignment on each of the 5 PCBs.

Yes, sorry, it wasn't a detailed video, a just quick single take 2nd channel unboxing

[EEVblog]

Layer stackup

[NorthGuy]

It looks very good on the video.

But these are big TH components. With thin traces, small vias, and 0.4mm pitch SMTs, it may look very different.

[mrpackethead]

I'm using both JLCPCB and AllPCB.

How does the quality of ALLPCB compare to JLCPCB? I'm thinking about trying them out as they do 10 PCBs for $15 which includes DHL delivery, hence they work out cheaper than JLC for prototypes and fast delivery. Thanks.
[/quote}

On 4 layer boards, my opinion is that Allpcb boards are slightly better than JLCPCB.  2 layer boards its the opposite, the JLCPCB are better.       They have great silk screen on their 4 layer line.  ( its a different printer on the 2 layer ).    Having said that, both companyes have delviered me product that is up to stracch..    Allpcb made a few boo-boos in their early days, but they seem to have settled down nicely. I think they had to figure out who and what they were doing.   They stopped doing Electropolished stencils which was a killer for me, but then started again..         

[electronics man]

Nice, i was not sure about JLC because the price seamed to be too good to be true (£1.56 for 10) but i will use them now that Dave has reviewed them.

[coppice]

Seriously its at a point now, that its cheaper to desgin 4 layer boards.  The time savings in a four layer board over a two layer, easily make up for it.     Not to mention your boards are smaller, and better designed!

For high volumes, the price increase of a 4 layer board over a 2 layer board is small enough that people will often go for a slightly cheaper chip that will demand a 4 layer board, over a slightly more expensive chip that will operate cleanly on a 2 layer board. If the board size is not pre-defined by the case or other factors, its often possible to make a 4 layer board smaller than a 2 layer one, and the price of the 2 boards might be the same.

[asmi]

IMHO the most important advantage of 4 layer boards is that it allows to route hi-speed traces properly. Looking at Dave's photo of cross-section, it looks like they use fairly thin prepreg, which is good for hi-speed traces as 50 Ohm traces will have a reasonable width. And nowadays hi-speed traces are becoming increasingly common and no longer they are confined to "hi-end" boards.
Another advantage of 4 layer boards is that using BGA packages is no longer unfeasible in typical hobbyist lab, as cost of boards has been the bigger detractor. Ability to use BGAs has some important advantages - giving access to ICs in BGA-only packages, or allowing for more compact designs as generally BGA packages are physically smaller and allow more dense routing.
I've also noticed that JLCPCB allows traces down to 3.5/3.5 mil and 0.2/0.45 mm vias as standard process capability, which is a huuuge step up from 6/6 mil traces/0.3/0.55 mm vias process (this is a standard process at the likes of PCBWay or Allpcb).

[ansonbao]

Example, an order I made from PCBWay had 3 out of 20 PCB had poor drill alignment on the top only, though they passed continuity.  A second re-order of the same gerbers had the wrong drill size for the mounting holes...

Sorry for this,please PM your order I will ask our factory investigate the reason and feedback to you, also our after-sale team will arrange refund or reproduce for you.

[EEVblog]

I've also noticed that JLCPCB allows traces down to 3.5/3.5 mil and 0.2/0.45 mm vias as standard process capability, which is a huuuge step up from 6/6 mil traces/0.3/0.55 mm vias process (this is a standard process at the likes of PCBWay or Allpcb).

Really? I didn't notice this, wow 

[xaxaxa]

This is a bootleg plutosdr done using jlcpcb's 4 layer process:


(collab with mrf184)

Unpopulated board:


The BGA on the left is ad9363 (0.8mm pitch), and the vias used are 0.2mm drill diameter with 0.45mm outer diameter (minimum allowed).

Total cost including pcb/pcba prototyping is less than a real plutosdr 

[cgroen]

I have had very good experience too with JLCPCB, I think i'm at about my 50th order or so, a mix of 2 and 4 layer board (from 10 pcs prototype to 3000 pcs at most).
All have been at a very good quality (smallest geometry is for 0.8mm pitch BGA).
I only have one thing that I think could be (much) better! I have tried explaining them that their "ship with already ongoing order" is not working too well. If it is a language barrier or not I don't know, but I have a hard time getting them to understand (or maybe they wont ?) what this is about...

PCBWAY etc all have this feature too and it works perfectly, when adding a new order you can select if you want that shipped with one of your "already in production" orders. You are warned that your complete shipment will be delayed until the order that is finished last, is ready. Which is perfectly fine.

But, JLCPCB.... Well, they have the same feature, and it kinda works.... But, only if you send an order that is accepted immediately for payment.  If it is put on hold for some engineer to look at it, once it clears and you are ready to pay for it, you are no longer getting the option for bundling it together with the "already in production" order.
With the amount of projects/PCBs I order from them, this would have allowed me to save substantially on shipping. I often have projects coming in a row, wanting to send a new design while something is in production with them.
Other than that, all is nice and dandy

[mrpackethead]

Seriously its at a point now, that its cheaper to desgin 4 layer boards.  The time savings in a four layer board over a two layer, easily make up for it.     Not to mention your boards are smaller, and better designed!

For high volumes, the price increase of a 4 layer board over a 2 layer board is small enough that people will often go for a slightly cheaper chip that will demand a 4 layer board, over a slightly more expensive chip that will operate cleanly on a 2 layer board. If the board size is not pre-defined by the case or other factors, its often possible to make a 4 layer board smaller than a 2 layer one, and the price of the 2 boards might be the same.

In deed. :-)

[mrpackethead]

IMHO the most important advantage of 4 layer boards is that it allows to route hi-speed traces properly. Looking at Dave's photo of cross-section, it looks like they use fairly thin prepreg, which is good for hi-speed traces as 50 Ohm traces will have a reasonable width. And nowadays hi-speed traces are becoming increasingly common and no longer they are confined to "hi-end" boards.
Another advantage of 4 layer boards is that using BGA packages is no longer unfeasible in typical hobbyist lab, as cost of boards has been the bigger detractor. Ability to use BGAs has some important advantages - giving access to ICs in BGA-only packages, or allowing for more compact designs as generally BGA packages are physically smaller and allow more dense routing.
I've also noticed that JLCPCB allows traces down to 3.5/3.5 mil and 0.2/0.45 mm vias as standard process capability, which is a huuuge step up from 6/6 mil traces/0.3/0.55 mm vias process (this is a standard process at the likes of PCBWay or Allpcb).

Allpcb are in the process of standardising this as well. They said they have to compete.  Its a real race to the bottom.

[EEVblog]

FYI, I just found two active sockpuppet accounts for PCBway, now banned.
It's like playing whack-a-mole, the PCB companies just looooove this forum.

[mrpackethead]

Heres a good question.   ( have some answers, but i think the question is good ).

Why would we typically put the signal  layers on teh outside of a 4 layer pcb, and a gnd and pwr in the middle.

[cgroen]

Heres a good question.   ( have some answers, but i think the question is good ).

Why would we typically put the signal  layers on teh outside of a 4 layer pcb, and a gnd and pwr in the middle.

1) to (easier) fix "fuckups"
2) It looks nice with the artwork

[EEVblog]

Heres a good question.   ( have some answers, but i think the question is good ).
Why would we typically put the signal  layers on teh outside of a 4 layer pcb, and a gnd and pwr in the middle.

Because the components have to connect to the signal traces, so you'd have the drop them all down to the inner layers, and that's silly given that a good lot of (majority?) of SMD layout traces can be done on the one layer, no need for vias.
Also, easier to inspect, trace, rework, hack, and do things like current field probing.

[matseng]

Is it just me being a bit old and stupid or do  they really not have any DRU and CAM files to download for the most common PCB packages?

Sure - I could make them myself, but why risk getting one of dozen of parameters wrong and fsck up things when they just could supply an official and tested DRU.  Actually even the manufacturers get things wrong - I just discovered that the German company Aisler had put 0.05mm (2 mil) as the minimum annular ring. ;-)

They are cheap, that's true, but I'd rather pay $10 and have 100% confidence that my design is ok than pay $2 and possibly have to spend a few days with a Chinese customer support trying to convey what they feel is non-compliant or problematic with my design.

[Bud]

Heres a good question.   ( have some answers, but i think the question is good ).
Why would we typically put the signal  layers on teh outside of a 4 layer pcb, and a gnd and pwr in the middle.

Because the components have to connect to the signal traces, so you'd have the drop them all down to the inner layers, and that's silly given that a good lot of (majority?) of SMD layout traces can be done on the one layer, no need for vias.
Also, easier to inspect, trace, rework, hack, and do things like current field probing.

Also this arrangement provides better supply decoupling because the ground and power planes are closer together which increases distributed capacitance between them.

[asmi]

Also this arrangement provides better supply decoupling because the ground and power planes are closer together which increases distributed capacitance between them.

It is negligible for most 4 layer stackups as internal layers are just too far away from each other. Power and ground layers need to be 0.3 mm (or less) apart to have any usable capacitance.

[Mr. Scram]

FYI, I just found two active sockpuppet accounts for PCBway, now banned.
It's like playing whack-a-mole, the PCB companies just looooove this forum.

Too bad, their overt main account was looking useful.

[EEVblog]

Also this arrangement provides better supply decoupling because the ground and power planes are closer together which increases distributed capacitance between them.

There's not a huge lot in it for the JLC 4 layer board in question, as the layers are near the surface with a thin pre-preg anyway.
I'll be doing a video along these lines tomorrow actually, I hope the measurement turns out interesting enough, we'll see.

[EEVblog]

Also this arrangement provides better supply decoupling because the ground and power planes are closer together which increases distributed capacitance between them.

It is negligible for most 4 layer stackups as internal layers are just too far away from each other. Power and ground layers need to be 0.3 mm (or less) apart to have any usable capacitance.

I don't think that's the case. As per previous post, I'm doing a measurement on this tomorrow, and I expect to get a usable plane response plot on the JLC board. We'll see...

[EEVblog]

FYI, I just found two active sockpuppet accounts for PCBway, now banned.
It's like playing whack-a-mole, the PCB companies just looooove this forum.

Too bad, their overt main account was looking useful.

He's now begging to be let back on, maybe I'll post a poll in the supporters section 

[coppice]

Heres a good question.   ( have some answers, but i think the question is good ).

Why would we typically put the signal  layers on teh outside of a 4 layer pcb, and a gnd and pwr in the middle.

1) to (easier) fix "fuckups"
2) It looks nice with the artwork

For a lot of military and high reliability applications everything is put on an inner layer, except for the actual component pads. It greatly reduces reliance on the integrity of a conformal coating. It makes working with the boards in development a major PITA, though.

[Mr. Scram]

For a lot of military and high reliability applications everything is put on an inner layer, except for the actual component pads. It greatly reduces reliance on the integrity of a conformal coating. It makes working with the boards in development a major PITA, though.

Are the outer layers the power planes in those cases?

[asmi]

I don't think that's the case. As per previous post, I'm doing a measurement on this tomorrow, and I expect to get a usable plane response plot on the JLC board. We'll see...

I use this PDF presentation for layer stackup planning. Attached is a graph of plane capacitance vs distance between planes from this file (page 58). As you can see, at 0.3 mm (12 mil) capacitance is about 75 pF/sq in, so for typical 4x4 in (10x10 cm) boards you will get 1.2 nF of capacitance, which is not enough for any appreciable decoupling. If you have a distance of 0.13 mm (~5 mil) - this is a distance between copper of a typical 0.2 mm 1Oz/1Oz core - you will get 2.8 nF for the same size of a board, which is much better. Although of course this is just a generalization for typical hi-speed digital boards (like CPU/FPGA boards), and it's best to do a simulation to determine what's best for any specific project. There are cases when you might face the opposite problem when you have too much of a capacitance.

[Bud]

Also this arrangement provides better supply decoupling because the ground and power planes are closer together which increases distributed capacitance between them.

There's not a huge lot in it for the JLC 4 layer board in question, as the layers are near the surface with a thin pre-preg anyway.
I'll be doing a video along these lines tomorrow actually, I hope the measurement turns out interesting enough, we'll see.

It may not be a lot userul for a arduino circuit but i expect it to be useful for RF because of low inductance, where lump capacitors become questionable. People mostly think of decoupling caps as 0.1uF or something, this is misconception when it comes to RF, where a 10pF may be doing better job. You should do a test using a VNA to 500MHz or something, to get full taste of difference between lumped and distributed capacitance, if you have a VNA. An arduino project scope of measurements will not give that sense.

[coppice]

For a lot of military and high reliability applications everything is put on an inner layer, except for the actual component pads. It greatly reduces reliance on the integrity of a conformal coating. It makes working with the boards in development a major PITA, though.

Are the outer layers the power planes in those cases?

What part of everything except the actual component pads was unclear? If you put the power planes on the surface you haven't reduced the exposure of important areas to the environment at all. You need 2 more layers than you would for a more conventional board, with traces on the 2 surfaces. Sometimes you need even more, if the increased number of vias clogs up signal routing too much.

[Mr. Scram]

What part of everything except the actual component pads was unclear? If you put the power planes on the surface you haven't reduced the exposure of important areas to the environment at all. You need 2 more layers than you would for a more conventional board, with traces on the 2 surfaces. Sometimes you need even more, if the increased number of vias clogs up signal routing too much.

I'm wholly unfamiliar with these kinds of boards, that's why I  asked. I thought maybe it not being traces but planes reduced susceptibility to moisture and mechanical damage, which I guess it does to a minor degree. I'm quite fascinated by high reliability designs and always like to learn more about them.