Digikey's Physical Connectors Tutorial | Hacker News Reader

Digikey's Physical Connectors Tutorial

digikey.com

124 pointsdragontamer43 comments

I've been on a MIL-DTL-38999 journey over the last few weeks. There's a lot of surplus tooling on Ebay, and they're by far the best connectors I've ever used. Rugged, really high pin density, and impossible to insert wrong. The prices are high for new tools, but there's tons of surplus floating around thanks to government standards. There's also solder-cup versions of a lot of the connectors so that no crimp tools are required. If you want to use the crimped versions, you'll want the M22520/1 basic tool or M22520/2 miniature tool for smaller wire gauges. You'll also need a positioner to hold the crimp in the right position within the tool (these are interchangeable for a ton of different models). You'll probably want the M22520/1-04 turret head. A Chinese company called JReady sells new crimp tools with the positioners for $300. It sounds like a lot, but it's much less than the $800 American made tool and even less than the single purpose crimp hand tools for cheaper Molex and JST connectors. One tool standard, lots of different high quality connectors!

I went down this road a bit too for a project where I wanted to minimise the enclosure complexity, keep everything in and out of the box to nice circular conectors... and yeah it was tempting to get into the MIL-DTL-38999 train, but I needed some high amperage low voltage DC stuff for power (it was a 3d printer project and I needed stuff like 8-14 Amps and 24-48V DC for various heaters, and a lot of these are nice for high density data, but its hard to find one that can do multiple higher power connections. I ended up giving up since I would find so many that got close but in the end the project form factor changed due to cooling issues (larger enclosure for more airflow due to insufficient fan static pressure)

They are definitely cool connectors though if you only need a bunch of data and don't need the higher power density I needed at the time.

The size 12 contacts can do 15-25A. There's e.g. the Amphenol size 25 / 24 connection version, with 12 high current contacts and 12 smaller contacts.

There's not too many that have a variety of pin sizes and that are military qualified, though. And it's not cheap stuff, for sure.

Yeah I probably could have explained a bit more, I did find connections that could do it but it was never quite right in terms of a size pin count or cost trade off for this specific project. The standard absolutely has options for higher current but it comes with a smaller pool of available suppliers making those parts and thus they cost more. I wasn’t prepared to spend $50 Australian on a single pair of connectors for this project, especially since due to wiring limitations I’d have probably had to get two pairs.

They firmly entered my mental toolbox as a go to for a flexible circular connector when I don’t need to join existing physical interfaces like Ethernet or higher speed USB where the signal reliability is potentially worth just using a sealed socket with matching circular screw on connector, but for the sort of thing where I’d have made up a DB9 or DB25, I’m definitely going to be reaching for these again in future (I did get a few of them and used the solder cup versions to avoid needing the crimping tools and the associated extra costs)

The positioners and tools are much cheaper if you buy direct from a DMC distributor. I bought the afm 8 and af8 from patlon in Canada and I got charged the MSRP on the DMC website. AF8 and the universal positioners was like 400usd when I got it and it covers like 90% of use cases. It can't do db pins though. If you buy the 38999 connectors from PEI genesis it also costs a lot less than what you would expect.

We use 38999 series 1 at work for lots of things but mostly for I/O panels on thermal vacuum chambers. Offers up to 61 pins and they are extremely robust which is good for connectors that are going to get a lot of mating cycles. We typically don't use them on the products since they are so large and heavy.

For those of you watching ebay - the M22520 tools are DMC (Daniels Manufacturing Corp) AF8 and AFM8 crimpers.

As someone who just got into using these for an automotive application, there is a great set of lessons on wiring/crimping using these tools from High Power Academy on Youtube or $50 if you pay for their full wiring class (worth it).

This is a good tutorial. Connectors are very tricky purchases; even if you do know exactly what you want, it's easy to buy the wrong version of a part, or forget to buy a necessary part of an assembly, or struggle to find the correctly-mating part that you know you need, or lack the crimping tool that goes with it, and so on. And they are expensive, so errors are costly. The difficulty, and cost, is why a lot of hobbyists (and even professionals) often stick to the same few connectors that they are familiar with. But sometimes you need to branch out; maybe you need something waterproof, or something which handles higher current, etc, and then you can spend hours looking through catalogs.

Is there a connector/crimp tool set that's recommended for hobbyists? I make just a couple devices per year in a year where I'm actively pursuing the hobby. Mostly audio, some home automation and on the fence if I should dive a bit deeper into either synthesizers or robotics. Up to now I've only done screw terminals and soldered plugs. Spending even a couple hundred bucks on a single crimp tool seems like such a hurdle, especially as basically every tutorial or open source design I see seems to be using different connectors.

edit: nvm, I just never did proper research. It seems like you can get a high quality crimping tool like the Engineer Pad-11 for just ~50 bucks and it does a bunch of common connectors like JST-SHD. I don't know why I thought I had to spend hundreds on single purpose tools.

Though physical connectors in electronics are a boring subject, I do think its one of the "important" subjects that's commonly overlooked in electronics tutorials or books.

The fact is that there are many companies with many combinations of connectors and wires. Much as how a mechanical engineer has to inevitably learn the difference from a Philips head screw vs a Torx screw, electrical engineers will need to learn the difference between D-Sub, 0.1" pin headers, and other such connectors, as well as the pros/cons of each.

Im my experience: If there's one thing that makes sense to stock at your own little lab - to avoid waiting 1-3 days for getting that little thing that blocks that project you want to complete - it's a wide assortment of connectors. The rest can often be improvised in many ways using various microcontroller boards and voltage level converters etc that you'll probabably already have.

The next most useful thing: A great assortment of fasteners!

I think the thing to do here is to buy like M2x60 through M10x60 (A4 stainless, insex screws) and then use that nut-retapping trick to get that custom screw length that you need.

"Do you mean cutting the screw to length with a nut on it, which then fixes the thread as you unfasten it?"

Yes. Looks like I've been too chatty for today, HN's rate limiting brutally tells me to shut the fuck up.

(The especially evil thing about this: You don't know until you press submit if your comment will be accepted or not. You could be writing the greatest comment of the week, but you have no idea if it will be automatically refused or not. You can only write and hope for the best. Fuck everything about that.)

Google is really not helpful in finding out what the nut-retapping trick is :>

Do you mean cutting the screw to length with a nut on it, which then fixes the thread as you unfasten it?

This works well. I would suggest acquiring a triangle file and a blade file. You can fix cut thread ends easy and even some cross threaded sections.

You don't always have nuts to go on threads items. E.g. when a rod is threading into a block.

I might be a bit weird, but I find fitting proper connectors onto hardware super satisfying.

I suppose I see a good connector as a good polish to an interface, similar to what might be found in software, for example. I get the same type of satisfaction that comes from polishing the interface for some API or other product I’ve been working on for a while.

It just makes it feel more like a finished product. The final touch. The step moving away from jerryrig wires and oscilloscope probes to rugged connectors.

Call me weird but I enjoy the LEMO connectors. I've used them for over a decade, and they just feel nice to plug in and remove.

Not weird at all! I've gotten lost in old LEMO catalogs many times.

I've seen racks of CAMAC modules that all use LEMO. I rarely see them elsewhere. I've heard they are very popular in Europe.

So 3d maker pro is selling affordable 3d scanners with LEMO connectors on them.

I've also seen them in military type gear.

I will often spend hours researching different kinds of electrical connectors instead of....any other thing that I'm actually supposed to be doing. If I am feeling extra depressed I'll go order myself some Anderson PowerPole housings or maybe a kit of aviation connectors. Or read http://www.mattmillman.com/info/crimpconnectors/dupont-and-d... yet again.

And then there's the madness that is the physical USB-C plug which is sometimes also Thunderbolt over the same connector: https://www.intel.com/content/www/us/en/architecture-and-tec... along with sometimes DisplayPort and/or HDMI: https://en.wikipedia.org/wiki/USB-C#Alternate_Mode_partner_s...

I will presumably forever be mystified why I can only get "fast charging" on my Pixel 4a with USB-C-to-USB-C into the Google power brick, and never USB-A-to-USB-C into any power brick labeled as "fast charge"

> USB-C-to-USB-C into the Google power brick

This is USB Power Delivery (USB-PD for short). It is exclusive to the USB Type-C connector, and will work with any Type-C, PD-compatible charging brick. Bricks with various wattages are available (e.g. for charging PD-compatible laptops as well as phones).

> USB-A-to-USB-C into any power brick labeled as "fast charge"

This is legacy fast charging via any of a myriad of proprietary FC standards (Qualcomm QC, Samsung AFC, Apple 2.4A, etc., etc.) Since your Pixel does not support any proprietary fast charging standards, it will not work.

USB-PD is not exclusive to the USB-C connector. USB-PD rev. 1 was finalized in 2012, whereas USB-C 1.0 was finalized in 2014.

This is technically correct but AFAIK USB-PD 1.0 saw essentially zero adoption.

Power Delivery ("PD") requires chips inside the cable's connectors to work. The chips provide information on the cable's capabilities. No chip, then standard voltage, standard current only.

Thunderbolt is 40 Gbit/s over two serial lanes. That's 20 billion bits per second over a single link. I'm not an expert in this field (i.e., the field of high-throughput cables) at all, but that doesn't sound very realistic to me for regular cabling. But I guess the use of copper, gold contacts(?), differential signaling, shortness of the cable (or alternatively active components), and an immense amount of shielding make it possible. These cables are quite expensive.

The cable ID is only required for the higher end of possible currents. You can still get 60W (20V, 3A) through a non e-marked cable with USB-PD, just not 100W (20V, 5A).

This is such a rabbit hole. The automotive/milspec side of things is very deep. I ended up building a $1000 wiring harness for my $500 riding mower. Madness.

One that's not mentioned in the article, but is a beautiful assembly of engineering is all of the https://www.neutrik.com/ connectors - speakON, powerCON, etherCON, opticalCON etc.

Amazing solid bits of hardware with ingress protection, strain relief, and good usability for environments with a lot of equipment movement.

One of the best connector resources I've found: https://www.mattmillman.com/info/crimpconnectors/

This is very good, thanks!

I had to read this twice.. why on earth would one want just a single power pin?

> designed to have four signal pins and one power pin

turns out it is a typo. Datasheet shows one signal and four power pins.

> Figure 6: Assmann’s AWP 10-7240-T ten position, free hanging socket.

I giggled at this.

Why are crimping tools always so dang expensive?

Low volume, and everyone who isn't a big company just buys clones.

If you are a company or care about reliability, you have to get the OEM version but hopeful you amortize it over thousands of crimps.

My guess is that they need to be cut using EDM (slow) and need expensive CMM machines to make sure the resulting die shapes match the tolerance. I know you're probably referring to OEM hand crimping tools that cost on the order of 500€/$, but even decent quality non-OEM tools (e.g. Engineer PAD-11) aren't exactly cheap. Though, I hear IWISS ones are usable if you're on a budget.

The generics are pretty cheap, but I guess it's a relative matter. I got a PA-20 for $35 on Amazon and it works with all the Molex micro-fit I've thrown at it. Each crimp takes much longer than if I had the $500 purpose-built tool from Molex itself, but I do so few it doesn't matter.

The higher end (Daniels / DMC) are very precise tools that you should calibrate with a go/no go gauge. If you handle one you can see why they are fairly expensive - everything is precise, heavy, and without slop and must remain so over thousands of crimps.

One thing that startled me was how low the mating cycle ratings of some connectors are, especially in the fancier end of the spectrum. For example u.fl coax connectors which are fairly common in computing are rated to only 30 cycles, and same is for the df40 connector in Raspberry Pi CM4. Of course those ratings are probably fairly conservative, but still it has made me bit more conscious when continuously plugging/unplugging stuff.

For a recent project of mine I tried a few different connector types but eventually ended up just using RJ-45 (ie Ethernet) connectors. Relatively cheap pre-made cables of almost any length. Twisted pairs also makes it useful for balanced signals. Main downside is the relatively low current rating but that was not a too big of a concern for this application.

I'm doing the same. L-com makes a solid 24 AWG, double shielded, LSZH Cat5E cable that is just about perfect.

https://www.l-com.com/ethernet-cat5e-double-shielded-etherne...

Anyone have a good suggestion for a connector that allows stacking of pcbs (so they are parallel to eachother)?

I commonly see the .1" pin headers (single or dual row) and sockets used for this.

https://www.te.com/usa-en/products/connectors/pcb-connectors...

Yes, this is what I use now, SMD versions of the .1" pin headers (SMD allows to stack indefinitely), but:

- The distances you can get are limited to a small range (roughly the height of the male header pins).

- The 1-row versions are not shrouded, meaning that it's easy to mis-plug the pcbs (put pin 1 into the slot for pin 2 accidentally).

A Sullins connector doesn't have this problem, but of course doesn't allow parallel mounting of pcbs but at a 90 degree angle.

I had a laugh at the idea of ONE ARTICLE about CONNECTORS but this was surprisingly thorough

Reply on news.ycombinator.com