Products > Test Equipment
19" (server) racks and cabinets
(1/1)
matthuszagh:
Disclosure: this is basically a cross-post of this thread on garagejournal: https://www.garagejournal.com/forum/threads/critique-my-plan-for-constructing-server-rack-rails.516681. I'm posting it here too because I'm interested in getting feedback from people who are used to heavy test equipment.

I use 19" racks to hold my test equipment. Most of this is heavy, vintage HP RF gear. I recently moved to a new spot and decided to upgrade the rack rails that provide the vertical weight support. My previous solution for the rack rails worked, but had issues, hence the upgrade. I'd appreciate any feedback and general thoughts on my my new plan.

Old solution:

I use these Navepoint 4-post racks and have been very happy with them: https://navepoint.com/navepoint-42u-800mm-adjustable-depth-4-post-open-frame-rack-with-casters. I want the rails to mount to the internal mounting holes (see attached pictures) so that the front holes are free for rack ear mounting and the back holes are free for other things, like power outlets and small auxiliary equipment. One challenge with the Navepoint racks is that the width between the internal mounting holes is 19.25", whereas rack-mountable HP equipment is 16.75" wide. This means that the rack rails need to be wide enough to overlap the equipment. In my previous solution (image attached) I used Tripp-Lite rails (https://tripplite.eaton.com/support/4POSTRAILSM) that are 3" wide. Unfortunately, since the equipment overlapped the rail toward the edge, they produced noticeable bend and were really not up to holding anything remotely heavy. To improve the weight-bearing capabilities, I added steel and aluminum flat bar between opposite rails (see picture). They're VHB-taped to the rails. This pushed the load to the corners of the rails and significantly improved their load-bearing. For very heavy equipment I also placed another set of rails inverted below the first set. This had a number of limitations / problems. First, the cross supports made inserting equipment difficult, as the back of the equipment would frequently bump up against the cross support. Also, I used these rails because I got a good deal on them, but the retail price is very expensive for what they are and so I can't really expect to get more.

New solution:

I've decided to create custom rack rails out of 2.5x1.5, 3/16" thick angle steel. I can get this at a local metal supplier in 20' segments that I cut there, and it's cheap. I'm aware that 3/16" thick is total overkill, but that's the thickness available in the dimensions I want and it works with HP equipment dimensions (these leave a bit more than 3/16" between the bottom cover and bottom of the front bezel/frame). I then drill mounting holes at the right height and spacing, clean everything up, deburr etc and then bring these to a galvanizer to coat in zinc and prevent rust/corrosion. I drill the holes a bit too large since the gavlanization process will add between 3-10mil thickness (or 6-20mil reduction in hole diameter). I've worked out that a 17/64" drill bit should work well for a M6 screw. It's a bit of work, but it's not so bad. The second picture is one of these rails mounted in the rack before galvanization.

I think this works quite well. I've attached two images showing vertically-adjacent equipment supported by the rails. The HP 8566B is the heaviest single unit I own and I think weighs something like 110lbs. The rails seem to be rock-solid with this on it. The unit below shows that the dimensions work out nicely and there's no wasted space for shelves. Additionally, the rack ear holes line up with the front mounting holes.

An individual on the other forum raised a concern about the twisting action of the weight of the equipment on the rails, given the fact that the weight is placed on the rails away from the corner. So, I decided to calculate the forces and torques on the rails and the resulting forces on the equipment rack and tension on the M6 mounting screws. The last picture shows this. Assuming a 110lb instrument, half the weight will be on each rail, or 55lb. The rail width is 2.5" and the equipment contacts the rail from 1.25" to the end. So, 55lb is distributed over 1.25". We therefore assume the full 55lb weight is 1.875" from the corner of the rail. The fastener will exert the upward counteracting force, to prevent the rail moving vertically. The attachment point is the screw, so the instrument weight will produce a torque on the rail (the r vector is the one from the screw to where the weight is centered on the rail). I calculated this torque to be 103lb-in. The rack will need to exert a force on the vertical part of the rail to prevent the rail from rotating. This force should be somewhat distributed over the side between the screw and corner of the rail, but will probably be mostly concentrated at the corner of the rail. Anyway, if we assume the force only occurs at the corner that would be 118lb. In reality, the force will be somewhat distributed over the vertical edge and so the force at any given point will be less than this 118lb. The screw will need to counter that with a tensile force of equal magnitude and opposite direction to prevent the rail moving horizontally.

118lb seems totally fine for an M6 screw. This also seems ok for the force of the rail against the rack, but admittedly I don't really know (FYI the rack is 16 gauge steel). The holes the rails are mounted to are steel angle themselves, so I really can't imagine this bending/deforming that. And with my heaviest equipment mounted, I see no bending or other issue whatsoever.

If the torque is an issue, I could get L angle with longer legs and have it take up more than 1U vertical space. This would allow placing a screw higher up, which reduces the forces on the rack and screw for the same torque.

FYI I also calculated that even if every rack unit were filled with my heaviest instrument (something like 75lb across 4U), I'd still be below the weight spec of this rack.

I know that these rails won't necessarily work for other equipment. That's ok; I can use my Tripp-Lite rails and cross bars etc for those if I need to.

Other thoughts:

Many proper, heavy duty server cabinets (e.g., APC AR3100) seem to have the internal mounting holes spaced at a closer width. This would significantly ease the rail requirements since they could be less wide and the torque on them would be less. But, these are expensive, larger, much heavier and I'd have a hell of a time getting them into the upstairs bedroom I've converted into a lab. These do pop up on places like craigslist, and I'm in the bay area, but I'm still in a rental and even if I manage to get these things in here I eventually need to move them out too. I also don't have much extra space and don't like the fact that these are deeper than the racks I have.

If I had heavy duty cabinets like the one mentioned above, I could keep an eye out for commercial rails on places like craigslist. I need quite a few of these, and the retail prices are pretty expensive, so that's not really an option. I like the fact that if I need a new rack I can buy one and the price is decent. And I can make my own rails and add them as I need. Having to always keep an eye out for used options and having to live with what I get doesn't sound enjoyable to me.

There's a whole lot of stuff I didn't discuss, but I didn't want to turn this into a novel. Anyway, please let me know thoughts and I'm happy to post more detail on whatever particular subtopic as needed. Oh and if anyone wants to replicate something like this, I have a whole bunch of notes on the process and just general goals/thoughts/etc and would be happy to share those.
tv84:
My main concern would also be the post bending, as stated in the other thread. With all that setup, the posts are the weakest point.

Have you thought in some vertical spacers next to the posts (or rack ears but without putting screws on the equipment...) to make sure the equipments don't derail horizontally and somewhat protect the inward bending of the  posts (at least in the front posts)?

Also, remember for your calculations that the equipments may not have the weight uniformly distributed (front-back).

If all else fails, just give it a try for some months and see how the stack behaves. We're are not talking about instantaneous changes so you could be on alert and see how it goes on a month by month basis. Unless you live in a earthquake zone...

PS: Don't forget to put some fixed cross beams at some heigths or screw equipments with rack ears. Those are fundamental to keep the rack integrity, given such weigths.
Shock:
Mistakes in this video - top loaded, equipment not fastened, rack not secured (no anti tip, no floor and wall fastening), likely a cheap rack without any structural support.

matthuszagh:

--- Quote from: tv84 on May 26, 2023, 08:17:45 am ---My main concern would also be the post bending, as stated in the other thread. With all that setup, the posts are the weakest point.

Have you thought in some vertical spacers next to the posts (or rack ears but without putting screws on the equipment...) to make sure the equipments don't derail horizontally and somewhat protect the inward bending of the  posts (at least in the front posts)?

Also, remember for your calculations that the equipments may not have the weight uniformly distributed (front-back).

If all else fails, just give it a try for some months and see how the stack behaves. We're are not talking about instantaneous changes so you could be on alert and see how it goes on a month by month basis. Unless you live in a earthquake zone...

PS: Don't forget to put some fixed cross beams at some heigths or screw equipments with rack ears. Those are fundamental to keep the rack integrity, given such weigths.

--- End quote ---

Thanks, this is helpful. Almost every instrument will be rack-ear fastened too (the picture is not representative), so that should help with cross support. The rack also has very solid side-to-side and front-to-back cross beams at the top, so that should help.

Unfortunately I do live in an earthquake zone (bay area california)...

In terms of the screw tension/shear calculations, yeah I can calculate pessimistic bounds too (heavier weight, non-uniform distribution, pushed to one side, etc.), but the stresses on the screw are so far below the ratings that's hard to imagine it making a big difference. The M6 screw, based on calcs should handle tension and shear of more than 1k lbs. I'm so far below that. For instance, if the full 110lb weight was fully located at the furthest edge of one rail, and all tension occurred on one screw (not realistic) the tension would be about 314lb. I think the screw should handle >2k lb, so that's not even close. Also, the shear, which I've estimated at 60% of the tension strength, only cares about the weight, and I certainly won't be putting 1k+ lb on one rail.

I haven't done any calcs for the strength of the posts re bending, but I imagine, even without lots of addition cross support that they should hold up pretty well. It's effectively 16 gauge steel L angle, and the width of the perpendicular side is 2in, which has to be very difficult to bend. I suppose the other leg is much shorter and is mostly holes so maybe that could give way and then the whole thing would buckle or something.
tv84:
Just envision this scenario: if one of the upper equipments skips the rail, it will be supported by a lower one. Putting all the weight in that one (without being noticed by you) can exacerbate the bending or contribute to its own derailing and trigger a domino effect... This is a bad (not worst) case scenario but can happen specially in Bay Area if the spacers or the screws are not in place.
Navigation
Message Index
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod