Small batch PCBs with blind vias?

Hi, all,

Anyone have a favourite board house that can do blind vias on a 4-layer board? I've been using Gold Phoenix, but they only do blinds on 6 layers and up.

(I want to sprinkle a bunch of vias on the back of a test board, so that I can add extra decoupling between the analogue power pour and ground--those 46 GHz transistors again.)

Also, for some reason Eagle's default blind via ratio (thickness divided by drill diameter) is set to 0.5, i.e. a blind via in a 20-mil layer has a minimum drill of 40 mils, which is ridiculous. What ratios do you folks usually use?

Thanks

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

Never done it myself; PCBs are confusing enough already.

At 46 GHz speeds, the planes themselves are the best caps; keep the dielectrics thin.

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
 Click to see the full signature
Reply to
John Larkin

Sure thing. It's just that there are a few different pours, and they aren't that big, so I want to be able to sprinkle some nice 0402 bypasses on the back of the board if needed.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

why do you need blind vias, running out of space?

-Lasse

Reply to
Lasse Langwadt Christensen

A 0402 package with a dead short in there would be around 0.2nH, which as 46GHz is around 50ohms Z. I'd consider much smaller packages there. Like this:

formatting link

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

I want to have the option to put a bunch of 0402 decaps on the back of the board, to stitch the power pours to the ground plane. Parallelling up all those via stubs reduces the overall inductance.

It's a test board for a 10 MHz laser noise canceller, which sounds easy until you say that the cancellation target is 70 dB up to 10 MHz, i.e. I care about phase shifts of 0.1 degree at 10 MHz. Matching like that is not happening without active tweaks.

There are three photodiodes, and a total of 6 AC tweaks (j omega and omega**2, per PD) plus five DC tweaks: one offset voltage, two offset currents, and r_ee' of the differential pair (actually a differential quad with Darlington drive).

This board is to test the AC tweaks, which use 2N7002s to adjust a series resistance in each PD for the first order (j omega) tweak, and varactors to adjust the frequency compensation of a bootstrap loop wrapped around a fast closed-loop current mirror to get the second-order (omega**2) tweak. (The capacitance of the 2N7002s is on the high side, but since it's only a few ohms' worth of resistance, the time constant is no problem.)

The real board will use an MCU for the job, but the test board just uses pots, so that I can verify that the basic scheme works and that I can keep all the zillion-gigahertz transistors from misbehaving.

We're having a try at selling a few instruments as a family. My son Simon is doing the firmware and my younger daughter Magdalen is doing the layouts and graphics. (They're both drinking from a firehose at the moment--partly from having me around, but also from learning a whole lot of new stuff from scratch on their own.) Mags is also working at learning circuit design, but it's at a pretty basic level ATM.

Having other folks around who also have a fire in their bellies is a wonderful thing. Being able to admire one's own children for real stuff is even better.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

why can't you use normal vias?

I don't have any kids but I can imagine :)

-Lasse

Reply to
Lasse Langwadt Christensen

We're trying to keep the layout constant, and add some fall-back positions. If everything works OK, the real board will use the exact layout of the protos, but if not, we can replicate the design with the blind vias and back-side caps. Once there's revenue coming in, a board spin is not such a big deal.

Using normal vias takes up space on the component side, meaning that the layout can't be as tight as it otherwise would be.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

None that you know of anyway :)

Jamie

Reply to
Maynard A. Philbrook Jr.

vias can be really small if you use a thin pcb, but sometimes you gotta do what you gotta do

btw. I think the reason for the big ratio on blind vias is that if they are made by peck drilling and then plating there is a risk that air gets trapped in the hole and the plating fails

if the 4 layer board is made by effectively stacking two finished two layer boards there should be no difference in rules, but it adds extra steps to plate the through vias

-Lasse

Reply to
Lasse Langwadt Christensen

I'll ask someone Monday. (Weekend has started.) It was a Rogers prototype, but I am not sure of all the features they offered. It was 4L.

Don't know about 20 mil. I used a 13 mil blind "drill" for a 10 mil Rogers 4350 core recently. (ViaSystems did the board.) Obviously, it does get bigger the farther you drill.

Reply to
Simon S Aysdie

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.