DIY Triax text box

[dr.diesel]

In my attempt to be as cool as robrenz (yeah I know I'll never get even close) I'd like to build a similar enclosure.

These male to male Pomona triax cables.

These female bulkhead connectors.

The tip of the bulkhead connectors would terminate to the same Tyco connectors used by robrenz, the guard connection left unconnected.

And some all aluminum enclosure, haven't decided on the exact model yet.  A couple questions, in terms of low noise performance:

1 - Does the size of the enclosure have any effect?  I'm not thinking coffee table size, but perhaps a cigar box size?
2 - Exterior conductivity?  Even though this will be sitting on a proper antistatic mat, would a painted or otherwise non conductive exteriour offer better low noise performance?
3 - Probably my main question, with the bulkhead connectors above, this setup would ground the outer shield on all 4 connections together at the AL enclosure, instead of the somewhat normal just force low shield.

Aside from the triax price robbery, what do you guys think?   

[German_EE]

What do I think? I think that some folk take things too far. Elsewhere on the board people are making fun of speaker cables that are hundreds of dollars a metre yet here we are considering test equipment cables at $119 each just to add an extra zero or two on the meter. Does it really matter that you measure 1.234567V instead of 1.235V?

[dr.diesel]

What do I think? I think that some folk take things too far. Elsewhere on the board people are making fun of speaker cables that are hundreds of dollars a metre yet here we are considering test equipment cables at $119 each just to add an extra zero or two on the meter. Does it really matter that you measure 1.234567V instead of 1.235V?

This isn't audiophoolery, with instruments such as the K2450, a proper test box is required to make accurate measurements down that low.  The world way down low, is not only important and practical, it's also very interesting and a fun place to play and experiment.

[robrenz]

What do I think? I think that some folk take things too far. Elsewhere on the board people are making fun of speaker cables that are hundreds of dollars a metre yet here we are considering test equipment cables at $119 each just to add an extra zero or two on the meter. Does it really matter that you measure 1.234567V instead of 1.235V?

Triax cables and guarding for low currents is NOT audiofoolery. When talking about .00000000000001 resolution amp values you better be using them.  Good used top name cables can be had for much less on fleabay.

[Richard Crowley]

In a previous job, I made several shield boxes around 1m cube for low-current (and low-capacitance) measurements for the process development of the Pentium-class microprocessor products.  We used the HP equivalent of the SMU, and ultimately the leads ended up in microscopic probe micromanipulators and we probed the test-die during the process development.  We used the aluminum enclosure both to shield the measurement environment from electromagnetic noise, but also to block light because many of the un-protected devices are light-sensitive.  This is the machine we used to hold the wafers on a vacuum-chuck and probe them under the microscope...



And for automated probing, I invented probably the first guarded probe-card...
https://www.google.com/patents/US4983907

[dr.diesel]

Very cool Richard!  Dave/Chris, sounds like a great Amp Hour episode to me   

[dr.diesel]

3 - Probably my main question, with the bulkhead connectors above, this setup would ground the outer shield on all 4 connections together at the AL enclosure, instead of the somewhat normal just force low shield.

No thoughts on the 4 parallel shield grounds?  Personally I don't see it being an issue, but often case you only see the one, perhaps no specific reason.

[robrenz]

Suggest checking out the Keithley Low level measurement handbook on the topic.

[dr.diesel]

I have a copy, read a bunch of it, dang pdf with no available hardcover to purchase, 

[Yago]

Very cool Richard!  Dave/Chris, sounds like a great Amp Hour episode to me   

Indeed!
*doffs cap to Richard*

[dr.diesel]

Well, after a cross-eyed marathon, skimming the KLLMHB:

 - Numerous references to Force Low tied to enclosure/box, nothing specifically about the other 3 outer shields.
 - If you look at the Keithley model 8010 test fixture, it shows the outer shields grounded together on the back, at least for the 2657A instrument.  This would mimic my idea/implementation, so I guess it might answer my question.

I can always isolate the other shields, test and experiment, which is 1/2 of the fun or more.

Got an engine to build first anyhow, I'll ponder and read further while I get this project out of the way.

[ivaylo]

You don't seem to be squeamish about spending money on this so you have options. I tried everything including building my own cables and boxes - https://www.eevblog.com/forum/projects/diy-triaxial-test-leads-and-shield-box-(with-diode-leakage-test)/msg474986/#msg474986
Doable but I called it mixed success in terms of fun/price ratio for the project...

[dr.diesel]

Thanks ivaylo, your rectangular enclosure looks to be very similar to what I plan to do.  It also looks like you've got all of the outer shields grounded/tied together at the enclosure, which was my biggest question.  I believe I found support for this in the K8010 documentation, but has it been working well for you?

I've has triax in my eBay saved search list now for a month, so far nothing cheaper than just buying Pomona/digikey, 

[ManateeMafia]

I am also interested in adding one for my Keithley 6514. I would like to experiment with measuring capacitor leakage. I have a HP 16058A test fixture which already has the adapters and wires for different components including axial parts.

The leads that Robrenz and ivaylo made look like they would be compatible. The box already has several connectors, I am not sure how much shielding would be need to be added but I would like to know other people's thoughts or experiences with this fixture.

[ivaylo]

The difference between my leads and those by robrenz is that I don't take the guard shield all the way out as a separate connection on the fixture side. It is still connected on the instrument side, goes all the way to the fixture, but then I leave it floating there and only have the ground and inner wires making contact with anything on the fixture. I read the fixture recommendations by Keythley and all and they recommend a bunch of different things, from having just a grounded enclosure to having that plus an inner guarded box (connected to one of the guard connections?). Anyways, in all this I feel comfortable with the ground connections, having them connect together at the fixture, etc, and less comfortable with dealing with the guard stuff (I think I understand the concept but don't feel like doing anything practical about it except connect the middle guard shield of the cable to the instrument).

[Richard Crowley]

The difference between my leads and those by robrenz is that I don't take the guard shield all the way out as a separate connection on the fixture side. It is still connected on the instrument side, goes all the way to the fixture, but then I leave it floating there and only have the ground and inner wires making contact with anything on the fixture.

That may or may not be adequate depending on the physical configuration of your experiment. It may be the case that leaving the sensitive measurement node hanging out in the breeze, unguarded, would be OK.

For some of the longer distances (several inches up to a foot) INSIDE my measurement boxes, I used ordinary BNCs and coax where the shell of the BNC and the braid around the coax was connected to the GUARD node.  The GROUND node was automatically connected to the measurement box via the triax bulkhead connectors. The return side of the DUT circuit was another guarded measurement node, not simply the circuit ground/shield/screen.