2024, Chapter 2
Jason’s Law
I’ve joked about “Jason’s Law” recently, which is simply:
Any date I give is wrong.
As in, ask me when something will ship. I’ll make a guess, based on my currently available knowledge…and I’ll be wrong. Sometimes by a week, sometimes a month, sometimes a whole lot more.
For some of you, this is confusing. Why can’t I provide a precise estimate? Don’t I know what’s going on in the entire company at all times? Isn’t there a master timeline or something? Isn’t there, at least, a plan for when we “get things in?”
Oh boy.
I think some of my readers think manufacturing works like this:
- We “order” something, like Vidar 2s.
- They “arrive” (on a container ship, unicorn, or steampunk blimp, not sure here.)
- We ship them to happy customers!
Ah man I really wish. If manufacturing was easy and fun, everyone would do it.
But it doesn’t work that way. There are tons of things that can throw off plans, from missing a $0.08 part to unexpected time off to higher-than-anticipated sales, to glitches in first runs, to weird stuff happening in subsequent runs…and dozens more that we have yet to discover.
So I thought, “Hey, maybe it would be good to do a chapter on Jason’s Law and manufacturing in general, so maybe you can get a better idea of how making things really works.”
“Wait a sec,” some of you are saying. “Didn’t you promise us a chapter on Aegir 2? And now you’re talking missed dates and delays? I’m not getting a good feeling about this!”
And yeah, there you go. Because Aegir 2 will now ship at the end of February.
Yes, I know, we are terrible people. Incompetent. Impossible to work with.
But that’s how it goes.
Refer back to Jason’s Law.
Welcome to the Big Top, AKA How Things Really Get Made
Sometimes it does seem that manufacturing is a 3-ring circus. Heck, today (Tuesday), I worked on five prototypes, including one that is going to production and needed qualified after the in-house-assembled complex and dense SMD PCB was misbehaving, and one that needed some final tweaks on a really cool chassis that had just come in as bare metal. But, at the same time, I had to:
- Look into comments on a new product, to ensure that nothing was wrong with it, and that the second run was the same as the first.
- Source three parts for another product that was past prototype.
- Run APx tests on several misbehaving internal projects.
- Participate in our weekly production meeting, during which we decided to push out Aegir.
- Be part of the weekly digital meeting, where exciting things are happening (and I have to support on the mechanical side.)
- Finalize a BOM (bill of material, a list of everything that goes into a product) for a new part.
- Respond to questions about a new chassis using unfamiliar manufacturing techniques.
So what does my day have to do with manufacturing, and why does it matter with respect to product introduction dates? Perhaps very little, serving only to illustrate the complexity in which we operate.
I mean, you don’t really think that manufacturing is a 3-step process:
- Order stuff.
- Receive boxes.
- Ship them out?
I mean, do you? Because that’s the feeling I get sometime.
Here’s the reality: manufacturing is a complex process involving hundreds of parts for each product, plus the labor to assemble, test, pack, ship, and support them. And literally every part of that process can bring the whole thing to a halt. As in, if you’re missing a $0.08 clip…but it’s a custom clip…and it has a 6-week lead time…and nothing else will work…guess what? You’re 6 weeks boned. Period. (Unless your supplier takes 8 weeks, then you’re boned 8 weeks.)
So what does actual manufacturing look like? OMG, too much. We can start at design, which is an entire process in itself. So let’s start with something simple: let’s start with “doing another run of Vidar 2.”
Note the “another.”
This is a proven product we’ve made before. What can go wrong?
Oh boy. You guys!
Making the "Next Run" of Vidar 2
- Determine if we have all the parts to do it. We have a Bill of Materials of all parts of the product. From this, we can use the ERP system to determine if we have everything we need to do a run of these amps.
- Order any missing parts.Sounds painless? Not if it’s a custom part like a chassis top, bottom, heatsink, heatsink clip, or transformer. The lead times on some of those parts can be 16 weeks. As in, 4 months. As in, 1/3 of a year. If you haven’t been paying attention to stock levels, this can get really ugly really fast.
- Potential delays: See above.
- Choose a PCB assembler.We don’t assemble our own PC boards. We work with PCB assemblers in California, Nevada, Utah, and Texas. Some have different requirements, so it’s important to know which we are working with.
- Potential delay: everyone’s “line time” is taken.
- Order a “kit” of parts.These are typically the more common parts we use, the parts that go on the board. We work with a purchasing company that helps buy and stock many of these parts. Some of them will be as small as an 0603 resistor (those dimensions are in mils, by the way, or thousandths of an inch). Some of them will be big storage capacitors, PCB hardware, pin connectors, power output BJTs, etc. A Vidar 2 has about 300 parts total.
- Potential delays: any one of the parts on the PCB can hold it up. Most will have “alternates,” from other suppliers, but some won’t. No Microchip microprocessors? Too bad, sit and wait.
- Get the kit and PCBs to the PCB assembler.Go back and forth with any questions. There will typically be questions even on recurring runs.
- Potential delays: PCBs delayed due to exotic materials or specs, missed questions mean your run slips.
- Wait for the first articles and qualify them.Typically a PCB assembler will do “first articles,” as in, a couple of boards done to the current BOM with the parts supplied, so you can check it and make sure it’s OK. Sometimes it goes smoothly, sometimes there are surprises.
- Potential delays: errors on first articles, “alternate” parts not truly alternate, parts variability causing problems—hell, dealing with variance is a big part of engineering. I should write a chapter on that.
- Wait for production boards. Production will take some time after first articles are approved, especially if it is a large run.
- Potential delays: PCB supplier has their own people out on holiday/sick leave/family time.
- Prep the line. Concurrently with the production boards, or later if labor is tight, prep work can begin on chassis. This sounds easy, but Vidar 2s have a ton of fiddly thermal strips, clips, binding posts, transformer mounts, front panel boards, AC inlet, etc that need to be done right.
- Potential delays: miscounted inventory means you’re missing a critical part, like a thermal strip or binding posts; now you’re back 3-4 weeks. Argh. Or just nobody to build it at the time.
- Perform prelim QC. Vidar 2s need to be programmed, “Flirred up,” as in, we look at every board with a thermal camera, bias checked…and only then is the board ready to drop in and assemble.
- Potential delays: something wrong with the board, no manpower, etc.
- Build a batch. Again, sounds easy, right? No. It’s much more than dropping the board in. You need to ensure the thermal clips are attached correctly, you need to set the bias, you need to check the front panel operation…lots of things can go wrong.
- Potential delays: as above, lack of labor.
- Burn them in. Vidars, like every one of our products that has an AC power cord, are burnt in for a day or more. As in, they go on racks, they get turned on and left on, to catch any early failures. They don’t get on the racks themselves. No do they jump off the racks and dance into final QC like Mickey’s magic brooms.
- Potential delays: labor, or an unexpected number of failures on burn-in.
- Do final QC. Now we can finally send the racks of Vidar 2s to instrumented and listening test. Literally 100% of everything we do is checked on instruments and every product is listened to. Yes, down to Magni. This is one of the most important parts of the process, and our test and QC staff are our most experienced.
- Potential delays: staff time again; new product ramp-up; broken or misbehaving test equipment, bad cables…oh boy, lots of strange stuff happens here.
- Clean and pack.Products don’t clean and pack themselves either. Heck, Elon’s robots can barely fold shirts. Call me when they can handle a 25-lb amp.
- Potential delays: shortage of boxes and inserts, missing manuals (yes, no kidding), staff shortage.
- Ship it. When you place your order, do you think it is whisked off human-free by an Amazon-esque robotic sled? Nope. Ask Art, or his crew, what’s really involved in this.
- Potential delays: believe me, we try to do this as fast as possible, but again, box and insert shortage, human shortage, lots of stuff can derail this.
- Swap. Any early field failures are swapped out. Not many of these, but they happen. And it is part of the process. You can’t just make things and ship them into the ether unsupported. This is part of manufacturing.
- Support. People have questions and concerns. We do our best to answer them. And, no kidding, this is one of the hardest things to get right. One of our longest-standing and most experienced persons in this department went on long-term leave recently, and we’re still getting back on our feet. But we’re finally making some real progress.
- Service. And, if there are problems with a product, we need to provide support, both inside and outside of warranty. It’s not unusual for us to repair products made 12-13 years ago. Not many, but you’ll see a reallllllllly old Asgard or Valhalla once a month or so. Still repairable. Still will probably be fine for many years.
Oh yeah and repeat this for the 20+ other products we make.
Okay, so maybe those last 3 things aren’t exactly parts of the manufacturing process, but it’s definitely part of being a manufacturer. If you’re thinking of getting into the business of making things, not only do you have to think about what you’re making now, you also need to consider what you have made.
And, when you get right down to it, even all those steps above really oversimplify manufacturing. Because manufacturing is really multifaceted, involving:
- Many different kinds of parts. Is it a steel stamping? A machined aluminum piece? An injection-molded plastic part? A standard resistor? A semi-custom potentiometer? A precision-spec’d inductor? Some will be fully custom, requiring mechanical engineering, CAD, and detailed drawings with materials and finishing specs, and some will be fully off the shelf. Knowing which to choose to best optimize price, design, and quality is a hugely important skill-set.
- Various critical partners. You’ll be working with people making chassis. People making boards. People making cable harnesses. People making transformers. People making brackets. People buying and kitting parts. People painting things. And lots more. Having a working knowledge of what your partners do, and finding the best way of working with them can be make or break.
- Multiple manufacturing disciplines. You may choose to make some or all in-house. What makes the most sense? What can you support?
- A critical mesh of employees and contractors. Everyone works a bit different, and everyone has their strengths and weaknesses. Creating a team that is excited about what they are doing—whether it is building a Magni or coming up with an entirely new idea—is the ultimate goal. Or at least that’s what we’re going for.
- Making the best of multiple surprises. In manufacturing, the only guarantee is that every day is going to be a surprise. Some for the better, many for the worse. Parts you expected would all be the same, aren’t. But then again, maybe someone wanders in with an insanely great idea…and it works.
Aaaaaaaand production timelines you expected would be dead-easy to hit…aren’t.
Yeah. Welcome to Jason’s Law.
The Impact on Aegir 2
"Yeah yeah yeah, we know making things is hard," someone says. "So why will Aegir 2 be so late? I thought you said you had boards and chassis a week ago!"
Sigh. It really comes down to faster-than-expected sales of other products and lower-than-expected levels of staff.
As in, we have a choice to build Tyrs, Yggdrasils, Vidars, and Bifrost 2s in order to keep up with higher-than-expected orders…or we can move Aegir 2 up in front of it and endure complaints about out-of-stock status on a bunch of other products. After a discussion on Tuesday, we decided to build the other products, and move Aegir 2.
“Well, hire more people,” someone says. “Get it done!”
Yeah. A couple of problems with that:
- We have skilled employees doing complex work. You don’t pick up how to put together a Tyr or Yggy in a day. So “just adding some people” is something with a much longer timeframe. We probably wouldn’t have anyone building Tyrs or Yggys for months after hire.
- Whoever we hire deserves to have a steady job. As in we have to be sure they have plenty of work. Hiring now, as we are going into slower seasons (spring is slower than winter, summer is slowest) could be a problem. It wouldn’t be fair to say, “Sorry, no job now, go home.”
“But why can’t you plan for correct staffing levels?” someone asks. “I don’t see how that’s so hard?”
Simple: because schiit happens.
Sometimes employees have unexpected family issues (both good and bad). Sometimes their vacations don’t line up perfectly. Sometimes they have health problems.
And, in every case, you have to roll with it. When someone isn’t a perfect automaton, you can’t simply chuck them in a ditch at the side of the road and go looking for replacements. No, not even in Texas.
Aside: we’ve had our share of gotcha questions about “how are our employees in Texas? Are they OK?” You know, implying that we’re all a bunch of uncaring hicks down here and we’re hiring cheap and firing fast. In reality, our Texas and California employees are paid very well and enjoy the same benefits, including benefits not mandated by law in Texas. So let’s put these questions to bed.
What happened in staffing was several things, including seasonal crap (plain old flu and COVID), some extra family time off, and surprise health things that have nothing to do with winter illness and are none of your business. All of which meant we had fewer people to make things.
So, yeah: more demand and less capability to make things. You do the math.
Here’s where that leaves us with Aegir 2:
- Current estimate (not by me, by Elvis, head of large/complex unit production) is that it will be available by February 28.
- Closeout Aegirs are getting fairly thin, so it’s possible they will be gone by Aegir 2 launch. Sorry about that, I know some people wanted to hear what others thought about the comparison.
To help you decide if Aegir 2 is for you, here’s a brief outline of the differences:
- More power: 25/50W 8/4 ohms, 100W mono.
- Less heat: revised Continuity runs cooler, revised standby consumes only 1-2 watts.
- Less invasive protection: 2X more output devices, so less overcurrent and thermal shutoff.
- New tech—Halo™: mixed-mode feedback output stage, low damping factor of 10 may be controversial. Nonswitchable. Always Halo. Unless in mono, then no Halo.
- Toroid transformer. Not a big deal, but some people like them.
- More money: $899, $200 more than outgoing Aegir.
Is it a better amp? To me, absolutely. But I don’t know what you’ll hear. I think it breaks down to: if you want to save some money, Aegir is a heckuva deal right now.
So yeah. Apologies on the delay. But sometimes that's how things work out. I hope this short chapter helps you understand where some of these delays may come from.
Only problem is the pi2aes is somewhat limited in sample rate and format support, so this will be an upgrade, and at least in theory should match the ability of modern USB to buffer.
