Keithley 2600 Front Breakout

[ch_scr]

The Keithley 2600 Series SMU are nice instruments, but are only available with back connections. If you, like me, want to use these precision industrial tools, in a more frivolous way as general bench instruments, the back connections make little sense. I immediately wanted a more traditional front breakout like on a dmm or a Keithley 2400. Since my unit came without feet or tilting bail anyway, but with the rack mounting kit & the 4x "10-32x3/8PPHSEM" screws, the idea to mount the breakout boxes in these places off to each side seemed very natural. I fashioned the mating part after the rackmount kit and designed the box to fit 5 "Hirschmann PK 10 A" terminal posts without their plastic back parts. My stackup is: terminal post, printed plastic panel, big washer, nut, cable lug, knurled washer, small washer, nut.  As you can see in the picture, the style mounted on the unit has the connection mirrored, and no feet. After looking at another unit, I realized feet would be beneficial for ventilation, so 3d models with and without feet are attached. Also variants for mirrored and non-mirrored layout are provided to tend to you liking. The backpanel comes either with just a hole, or a DB9. I will try a high density DB9 with a row of guard pins for now (pinout see below), but am very interested in your opinion on affordable options that minimize impact on low current performance. Intended use would be a curve tracer desk fixture that uses the same cables. For cabel I used inner cores of a multicore stage audio wire. The shield on the Hi & Hi Sens is connected to guard, while the shield on the Lo & Lo Sens is connected to the green "filtered ground" on the back and the green terminal post in the front. While I am sure this cable & connection is not optimal, it is my best effort at a first shot.

[Hydron]

Interesting timing on your post - I've just made the same thing for my Keithley 236!

Mine is just a prototype at the moment, it's using the correct triax connectors but some temporary junk banana sockets (not even matching!) plus 3D printed front and back panels (on an extruded two-piece case).
The final version will have some shrouded banana jacks (which should give quite a long surface leakage path) and black PCBs as front and back panel - is just waiting on me having enough designs finished to be worth placing a PCB order from China and to order the jacks.

The design also has a Earth/LO shorting switch (to match the removable link on the back of the 236) and a triax jack (replicating the main HI/Guard/Earth triax connection) for use with an external shielded boxes for noise sensitive measurements.
I should be able to play some clever tricks with exposed tracks on the front-panel PCB to guard most of the sensitive potential leakage paths, but even with the rough 3d printed + scrap banana jacks prototype I'm not really seeing much leakage (possibly a few 10s of fA, and maybe not even that!).

For mounting, the 236 only has one screw hole at the front (some annoying imperial thread, I found a matching screw in my junk screw pile but no idea what size/thread it is!), so I've made some "ears" on the back panel to fit into some case features and keep the enclosure level.

I'll release the design/gerbers and some more pics once I'm done and have tested it's final version - could be a while though before I place my PCB order.

Cable wise, I'm using 3 triaxial cables, with the jacks on the back of the case salvaged from some ebay 1553 bus couplers, and the plugs coming from US ebay. Overall not cheap unfortunately!
If anyone wants some of the jacks let me know and I can dig up the seller (and possibly even re-send, as I think it's UK only), they need salvaging from the couplers (which are potted, but with an easy to remove compound) but worked out at only about 3 GBP each, which actually makes that part cheap.

[tautech]

Suggest you post or link this project here:
https://www.eevblog.com/forum/testgear/replacement-knobs-feet-and-fittings-for-test-equipment/

[jjoonathan]

Cool! As it happens, I had the same instrument and the same problem    I went with a slightly different solution that tucks the ports underneath. It's a less traditional arrangement, but it preserves the front profile a bit better. It certainly won't accept one of those four-post low-thermal-shorts, but it does accept the probes I regularly use: kelvin clips, regular clips, and banana->BNC adapters. I opted against bringing out a ground port in order to make room for a hypothetical future PCIe slot in the center that I could use for test coupons and really take advantage of that $2 for 10 JLPCB deal. I haven't gotten around to adding it yet, but I have felt its absence so I probably should at some point.

Originally I planned to 3D print both the feet and the bar, but the prints were so weak that major deflection happened every time I plugged in a banana->BNC adapter. I tried bracing the prints (two sections, to fit on my printer) but between the dimensional inaccuracy and the glue required to make it work, it looked like trash. So I printed the feet and milled the bar out of aluminum. I forgot to account for the taper in the Keithley front panel edge when roughing my stock, but even a mm of aluminum is plenty to hold a screw in so it still worked. I'm very happy with how it turned out!

I put the prints on thingiverse, but less than a year later and they're 404 and thingiverse claims to never have heard of me. My email history says otherwise, so I'm not going crazy. Is there a new, hip place to upload print files that might leave my account up for more than a year?

[jjoonathan]

I didn't really care about low current measurements (I have a proper electrometer for those) so I didn't use triax/guarding and I opted for suspicious and cheap aliexpress banana ports. I did use coax and ground the aluminum bar, though. This combination creates a sort of worst case so I thought I'd take some leakage measurements for the low-current-nuts out there. I get a remarkably consistent:

±202V / ±50pA  = 4TΩ

[ch_scr]

The bar below certainly looks nice as well!
Having talked to some people on here, up came the idea of using a high-density Sub-D connectors and using the same trick that is done on the back of the K2600 units without triaxial connectors. Namely, putting guard potential pins around sensitive pins. I have attached a proposed pinout that should do just that. I have been told SUB-D has decent thermal coupling between pins and they are available with gold-plated copper pins so it should check all the boxes (leakage, offset voltage, low current)?
Connector pair would be "163A16649X" (M) and "164A17059X" (F) from CONEC at about ~10-20$ for a pair
Only downside is they are just specified at Class 3 - 50 plug cycles

[Hydron]

You can also get "sub-miniature" connectors with co-axial pin inserts - the most well known is probably the 13W3 used by Sun microsystems monitors (13 is total number of pins, W3 being the number of co-ax inserts). You might find other options and pre-made cables by trawling ebay, I suspect they were used in other equipment too (I've seen them come up). This is one of the routes I was looking at myself before finding a source for cheap-ish triax jacks.

[ch_scr]

I've modified the original post to put up a V2 of the parts and to show how they look. Feet comply now to the original so stacking & modifying a single channel unit should work fine. Additional backplate that takes a DB9 connector added. As a bonus a replacement feet model is also included

[TheSteve]

My very budget solution I threw together really quick with parts I had on hand in 2017 before I decided to sell the 2602.