approximating a multi-coax feedthrough? / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

approximating a multi-coax feedthrough?

1 2 next

Move Thread

LAN_403

hboy007

Wed May 16 2012, 08:31PM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

some of you may have come across mixed dsub connectors like these:

I have also bought and assembled such connectors with coax inserts and they offer a really nice connector density.
However, I now need them as vacuum feedthroughs for signals on the nanosecond scale.

Could I perhaps use 3 row connectors like this one

and use pins of the central row, each surrounded by six "shield" pins? Of course that's a coarse approximation for a coaxial transmission line segment... but hey... those connectors are available as vacuum feedthroughs.

Any ideas how I should handle the issue? I need at least 4 coax feedthroughs and BNC / SMA sealed feedthroughs seem kinda bulky.

Steve Conner

Wed May 16 2012, 09:42PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I'd start by calculating the characteristic impedance. It should be OK to assume that the six shield pins are a cylindrical shell, and use the formula for coaxial cylinders. You'll also need the dielectric constant of the insulator inside the connector.

If the answer is different from 50 ohms, you'll have a reflection where you connect 50 ohm coax to it. By eyeball I'd guess it is nearer 75 ohms than 50.

The timescale of the reflection, and so the frequency range affected, can be estimated from the dimensions of the coax-to connector joint. The magnitude of it can be estimated by calculating the reflection coefficient from the two impedances. The way the joint is made will have an effect too. Splitting the coax shield into six pigtails, one to each shield pin, might help.

You will also get weird behaviour at still higher frequencies, where the gaps between the shield pins are no longer small compared to the wavelength.

Having quantified the weirdness, the next step is to consider it in terms of the level of signal integrity your application needs, which you didn't mention at all.

hboy007

Wed May 16 2012, 10:58PM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

Thanks for the quick reply. The requirements on signal integrity may vary, depending on whether we'll be using cryogenic analogue switches or drive the device of interest inside a cryostat directly. In the latter case, signal integrity is vital. When controlling the switches, distortions and reflections will create timing problems or invalid states of operation. At least the amplitudes will be well-defined with the MOS switches.

I've estimated the core and inner radii as 1.0mm and 4.0mm respecitvely (relative permittivity of PP is 2.2 .. 2.4). A quick guess with the help of a calculator

yields 56 ohms but I assume it will be more like 60 ohms in the end.

I've also stumbled upon a more flexible numerical calculation tool for arbitrary conductor geometries

(no windows binaries, thanks)

Steve Maurer

Thu May 17 2012, 02:15AM

Registered Member #133 Joined: Fri Feb 10 2006, 10:27PM
Location: Pensacola, Florida
Posts: 47

Note that the termination method used to attach your 50-Ohm coax to the pins is important. The coax dielectric will need to extend all the way down to your feedthrough connector, otherwise your characteristic impedance will increase for the region between the end of the coax dielectric and the feedthrough dielectric (as the dielectric in this region would be air, with a permittivity of 1, resulting in multiple signal reflection points).

Concerning the pin spacing:
1 ns equates to a wavelength of around 300 mm. As long as any apertures are less than 6 mm (1/50th of a wavelength), you should achieve relatively effective shielding. The aperture size will be the longest distance of an opening between the pins. (For a rectangular opening, the longest distance would be the diagonal of the opening.)

Carbon_Rod

Thu May 17 2012, 06:53AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

I don't see how well isolated generic panel mount SMA with a keyed plastic wiring harness is any less convenient. They are really nice to have around when something breaks off in a year....

SSMCX may be helpful if it needs rapid deployment in "complex" situations:

Cheers,

hboy007

Thu May 17 2012, 10:11AM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

SSMCX connectors look great, they may be useful for making the connection to the PCB inside the chamber.

Through this site

I've found some products from Southwest Microwave, page 27 in the pdf:

As Carbon_Rod points out, we can still use panel mount SMA receptacles. What keeps me from just going for this solution is the number of seals that could start leaking.

hboy007

Fri May 18 2012, 08:25PM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

allow me to give you a little update on the results:

using MinGW/Msys

(MinGW Shell)
I was able to cross-compile atlc for windows

1) open the MinGW shell
2) cd '/full/path/to/atlc-4.6.1/'
3) ./configure --host win32
4) make -j 4

the colors are explained here:

adding /full/path/to/atlc-4.6.1/src/ to your PATH environment variable (for a session via
@SET PATH=%PATH%;/full/path/to/atlc-4.6.1 or permanently via regedit at
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Co ntrol\Session
Manager\Environment] ) allows you to use atlc on 24bit bitmaps in any location.

1337372229 1667 FT138508 Pin Arrangement

The impedance turned out to be a little lower than expected... could have been worse though. The electrical field seems to be well contained within the area between the rods.

Carbon_Rod

Sat May 19 2012, 06:49AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

Usually ringing cross-talking lines is something people dread....

Perhaps a clear description of the intended application may yield low cost alternatives to hermetic/cryogenics products.

hboy007

Sat May 19 2012, 10:49AM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

Our setup is built to investigate cryogenic optoelectronic structures.
I need to bring 4-6 fast control signals into a cryostat that has a residual filling of 2mbar of Helium. The connectors are mounted to the hot end. The connector needs to be properly sealed to avoid contamination of the buffer gas. The cold side is at 3.2 .. 4K, small air leaks will coat the optics inside with frozen nitrogen/oxygen.
The connector does not have to withstand cryogenic temperatures, I should have pointed that out in the first place.

The head section of the sample insert isn't that large (8cm in diameter). I can have a wider head milled in our shop but smaller / higher density connectors would be preferable. There are already a bunch of other cables connected to the insert for sensors and actuators.

Carbon_Rod

Sat May 19 2012, 10:22PM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

Indium foils will stick to glass as well as metals, and are easy to cut into vacuum sealing washers.

Additionally, silicone vacuum grease for glass can become a contaminant under some conditions, but in theory one could make a conically-tapered ground glass seal with an old glass CRT lead wire assembly in a counter sunk hole...

However, I would recommend the commercial products to reduce sources of possible error, and simply have leads with regular plugs permanently bonded to the ceramic feed-though on both sides of the vacuum vessel walls.

Best of luck,

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob