3

I am absolutely new to flex PCBs. I saw some from time to time but never designed one myself.

Now I'd like to get into this topic and the first task would be to create a ~ 10cm board to board interconnect flex PCB with 90 ohm (USB) and 100 ohm (LVDS) controlled impedance traces on it.

I used the KiCAD PCB calculator, asked the PCB manufacturer about the tangent loss (0.01) and dielectric constant (3.3) to calculate trace width and spacing, as I am used to from designing standard PCBs with FR4 material.

KiCAD PCB Calculator

I realized pretty quickly that the trace widths and spacing I'd need to achieve e.g. 45 ohm differential would be way out of spec.

I assume that my assumptions are completely wrong. How do you guys do that? Just ignore the impedance of the flex PCB?

JRE
  • 67,678
  • 8
  • 104
  • 179
pm4812
  • 57
  • 6
  • is the thickness of the substrate really 25 µm? 100 to 200 µm would be closer to what I expect. – Marcus Müller Apr 18 '23 at 13:17
  • 1
    https://jlcpcb.com/help/newsdetail/113-Flex-PCB-Now-Available-at-JLCPCB-From-Special-Offer-of-$25 - yes, the Polymide Layer between the 2 Copper layers is spec'ed as 25 um thick ... – pm4812 Apr 18 '23 at 14:41
  • and if you check the specs from PCBWay: https://www.pcbway.com/pcb_prototype/Stack_up_for_FPC.html - they also spec the thickness of the "polyimide core" with 25um. Seems to be true ... – pm4812 Apr 18 '23 at 14:49
  • 1
    You should not be trying to achieve 45 ohm diff impedance, but rather 45 ohms, or 50 ohms single ended impedance on each trace. – SteveSh Apr 18 '23 at 15:21
  • I'll see if I can find the stackup on the last flex design I was involved with for LVDS. – SteveSh Apr 18 '23 at 15:22
  • @SteveSh that would be great, thank you! I'm running in the fog right now and have no idea, how this can be achieved? Obviously it must work, because others have done it long before I came up with that, but I can't find any suitable documentation / tutorial etc. that could lead me on the right track ... – pm4812 Apr 18 '23 at 15:25
  • 25 u (microns) is way to small to achieve 50 ohm single ended impedance. Bump that up to 150 u and see what you are able to achieve. – SteveSh Apr 18 '23 at 15:31
  • Have you considered using a coplanar waveguide (CPW) structure instead of microstrip? Or, if you have access to a simulator that can deal with arbitrary structures, cutting some lengthwise slots in the ground plane to reduce the trace-to-ground capacitance? – The Photon Apr 18 '23 at 15:50
  • nope, I don't have access to such tools which could simulate arbitrary structures. But with the help pf @Zygis I found some parameters I can play with. My error was to assume that there is only 25um core thicknes availiable, which is at JLCPCB (atm), but e.g. not at PCBway. As I don't plan "heavy bending" atm, I'd go with a thicker core for this project and gain some experience! – pm4812 Apr 18 '23 at 16:29

1 Answers1

2

If you need impedance control, don't use such thin core. You can request them to make what you need.

From the PCB Way page you have linked, at the top: "The following is the usual FPC stackup information. If you need custom FPC stackup, please make note and we will manufacture according to your requirement. Polyimide (PI) is the most commonly used thermal curing insulating material in flexible circuit processing. The thickness range of the material is generally 12.5μm (0.5mil) and 125μm (5mil). Divided into with glue and without glue, the DK with glue is 3.5, and the DK without glue is 3.3."

In the past, I have used 0.075mm core for USB flex.

JRE
  • 67,678
  • 8
  • 104
  • 179
Zygis
  • 58
  • 6
  • thanks for your answer. I just calculated your example with the KiCAD PCB Calculator, using 75 um core, 0.25mm track width and 0.25 mm spacing and the result is Zodd=~19 Ohm (=> Zdiff=~38 Ohm). Does this mean you simply ignored the impedance requirement for USB and it worked anyway, because the frequency is relatively low (12 Mhz)? – pm4812 Apr 18 '23 at 15:23
  • No, do not ignore the impedance requirement. I use Allium for impedance calculations. 0.25/0.25mm seems wrong with this stack-up. 75um core, differential 90R (USB). I get 0.125mm track 0.2mm gap. If your supplier can't make this work, have go with something even thicker then. This was just an example from a real-world design that works at USB 3.0 speeds. – Zygis Apr 18 '23 at 15:32
  • ok, that helped! looks like I can create 90 Ohm and 100 Ohm (both differential) with 100 um core, 0.25mm width and 0.3mm spacing. I guess I start understanding that topic ... – pm4812 Apr 18 '23 at 15:39