So you wanna start an SE/30 Reloaded group buy...

[This Does Not Compute]

You may have seen another thread in which I ran a group buy for a batch of Bolle's SE/30 Reloaded boards.

Let's get something out of the way: I am not going to be doing that again.

This thread is a deep-dive into what that process was like, and in it, perhaps you'll understand why. But if you're ambitious and want to get PCBs manufactured for yourself or others, it will hopefully also give you the information you need to be successful and not repeat the same mistakes I did.

A note: Normally I don't "name and shame" companies when I have a subpar experience, because I prefer to give the benefit of the doubt -- with a sample size of one, it's not statistically significant. This time, though, it's relevant as this project was involved enough for me to need to interact with the company multiple times, and on several levels.

Trial run

I got the idea to build an SE/30 Reloaded board in early 2023. A battery-bombed SE/30 board dropped into my lap, and after I learned about Bolle's project, I decided it would be a fun opportunity to help advance my soldering skills. Problem is, no one was having them made at the time, so unless I wanted to wait an unknown amount of time for a group buy to pop up, I'd need to take on the task myself.

You've probably heard of the various PCB manufacturers based in China. They've invested heavily in integrated ads in YouTube videos, declaring how cheap and quick their services are. PCB's for $5! Who could say no to that?

To get a feel for the process, I wanted to start with a small project. The Bourns Again reproduction chips (https://tinkerdifferent.com/threads...-series-rc-network-t-filters-replacement.1044) made a lot of sense, as their PCBs are simple, small, and use common components. I downloaded the board files, uploaded them to JLCPCB, and basically clicked Next a bunch of times. And indeed, the price did look to be as cheap as they said in the ads: five PCBs were $3.40, plus a $4 engineering fee; the cost of the components and having them soldered to the boards (a process referred to as "assembly") was a further $25. Not bad! I placed the order.

A couple days later I got an email. Because of the design of the board, they decided that an extra fee would apply to route out all the separations between the individual Bourns Again boards in the larger "panel" PCB. That fee was $15, which wasn't too bad, but why didn't their site pick up on that when I uploaded the file? I paid it, and manufacturing began.

Except that wasn't the only fee I had to deal with. The "$5 for 5 PCB" ads are really just for plain PCBs with no components and the simplest of designs, and don't include all of the other things you need to pay for as part of the process -- including the payment itself. JLCPCB doesn't handle credit cards on its own, only payment services like PayPal. That's fine, except they charged me a $1 fee just to pay them. Shipping doesn't get calculated until after production is done, either, and I found that they wanted $9 for economy shipping, which took 2 weeks. Since I was a new customer, they had various coupons available to use that ended up knocking down the total cost by $10, but in the end, what the site initially said would cost just over $30 ended up costing me $49.35.

Do you really want to do this?

The SE/30 Reloaded board is much more complex, and even beyond the technical aspects of having them produced, I had to think about the logistics and costs. Getting five bare PCBs made wouldn't cost much, but there are lots of passive surface-mount (or SMD) components on the bottom, and those would be *very* annoying to populate by hand. Bolle already had published the necessary information for having the bottom side of the board assembled, but that adds not just the raw parts costs but also significantly increases all of the production fees.

And then I got to thinking, if I'm having the bottom side SMD parts assembled, what about the top side? There are lots of chips and a smattering of capacitors that would also need to be installed, and since the majority were still available new (including the infamous UE8), I wondered what it would take to have them added at the factory too. Kai Robinson generously donated his time to help me get appropriate parts identified and put together the necessary files that JLCPCB would need. (Notably, all of the top-side parts were SMD, as JLCPCB can't do wave soldering for through-hole parts. It *can* assemble through-hole parts, but they're all done by hand and charged by the solder joint. We decided that in the interest of cost savings it would be best to leave the though-hole parts for the board buyers to handle, which are simple enough.)

I crunched the numbers and getting five boards made with both sides as populated as possible wouldn't make economic sense. A rough guess put me at well over $100 per board. I only needed one or two boards for myself, but the minimum order quantity was five, so I figured I could sell the extras. But Bolle's licensing on the project was Creative Commons Noncommercial, so I'd have to sell them at cost. And even then, I wasn't sure how many people would want to spend $150 or so on a board they still needed to install additional parts on. Increasing the number of boards would drop the per-unit price, not just through quantity discounts on parts but also by spreading out the one-time costs for production that JLCPCB charges, such as creating the solder paste stencil and setting up tooling.

Some more rough calculations led to 50 boards as being the magic number. From what I was able to tell -- at least, without actually placing the order -- this would result in the boards costing about $65 each, plus a little more for shipping. That was a much more palatable number for sure. There were just two questions I had to ask myself:

1. Do I have the time and ability to manage a group buy? Dozens of people would likely participate, and there's a lot more responsibility involved in managing a project where you take other people's money.
2. Would I be willing to eat the (significant) cost if the project failed spectacularly? I couldn't just tell everyone "oops, sorry that the package was destroyed in shipping" or whatever and not refund their money. With 50 boards, thousands of dollars would be at stake.

I debated for a little while, then decided to go for it. You only live once.

 

[This Does Not Compute]

Parts ordering and the supply chain

There's no real bottlenecks or constraints to getting bare PCBs made. Upload the design, pay the money, then a few days later they're done. PCBs are just fiberglass and copper and solder mask, so there hasn't really been much constraint on the supply chain. The parts to populate those boards, though, have been a giant mess.

JLCPCB works with another company called LCSC as its default parts supplier. LCSC is a component distributor, like Digi-Key or Mouser, and it has a decent selection in its catalog. When you upload the data for a board to a PCB manufacturer, you need to include three things:

1. The board design files -- this is what the board looks like, where the traces go, etc.
2. The board BOM, or Bill of Materials -- the list of parts to be populated on the board, with references to which location each part goes.
3. The CPL, or Component Placement List -- a list of all the component locations on the board that's used by the pick-and-place machines to know how and where to physically put the parts on the PCB.

A neat thing about the JLCPCB/LCSC relationship is that you can include part numbers from the LCSC catalog in the BOM file, and JLCPCB will automatically know what they are and add them to your order. This is what I had done with the Bourns Again boards I had produced; there are only two different parts used, and both were available from LCSC.

Problem is, LCSC doesn't seem to really cater to weird crap like the SE/30 Reloaded boards I was trying to have produced. There are 22 different parts on the bottom side of the board, and 21 of them were available from LCSC. But the top side of the board had 23 parts that could be feasibly obtained and installed, but only six were common-enough for LCSC to carry them.

Thankfully, JLCPCB has another option that can help, called Global Sourced Parts. At the time, you couldn't ship JLCPCB parts yourself directly (this has since changed, though it has its own caveats), but they do allow you to order them from a few different distributors for shipment directly to them. Among those distributors were Digi-Key and Mouser, which generally carried the remaining parts I'd need from the BOM.

Global supply chain shenanigans decided to throw a wrench into the works. Not all the parts were available at any given time, so it turned into a cat-and-mouse game to place orders for what parts were available, then either try to find substitutes that were in stock, or wait for the desired ones. I ended up having to place four separate orders for the global-sourced parts. Each order included parts from different suppliers, who all had different shipping times, so while some parts showed up at JLCPCB within days, others took weeks. I ended up having to devote some time every day just to babysitting the parts-buying process.

Another thing I learned along the way is that JLCPCB will let you try to buy parts they're incapable of assembling. Something clever that Kai had come up with was including sockets on the board for a few of the PLCC chips, like the math coprocessor and SCSI controller. These chips kinda suck to solder, but with sockets, they'd just pop right in. The biggest of the PLCCs on the SE/30 is the "glue" chip, which has 84 pins. We identified an appropriate 84-pin PLCC socket and I tried to include it with one of my global-sourced parts orders, but the next day I got an email from JLCPCB saying they had to cancel it as they aren't equipped to use them for pick-and place.

Eventually, though, I had all the parts sitting in JLCPCB's warehouse and could upload the final files and actually place the order.

Board manufacturing and assembly

I didn't hear anything from JLCPCB for a few days after placing the order. Turns out, there was a holiday in China, and the backlog of orders meant that there was a delay of about a week before things started moving. Manufacturing the PCBs was uneventful for the most part, except for an additional fee to include "edge rails" on the boards -- narrow strips of extra PCB material on two sides that let the boards run through the pick-and-place machines. I wished this had been accounted for when I initially uploaded the file, as JLCPCB's site analyzes the board design and gives you a graphical representation of it even before you place the order. But it was necessary, so I paid it. When PCB assembly started, though, was when the *real* limitations of JLCPCB's site became apparent.

It generally all revolved around the PLCC sockets. Bolle hadn't designed the Reloaded board with them in mind, but confirmed to me that there shouldn't be any major issues. JLCPCB emailed me to say that the sockets I'd ordered had locating pins on the bottom, little plastic nubs that slot into corresponding holes in the PCB so that they get aligned properly. At first, they said they couldn't continue assembly unless I redesigned the boards to include the locating holes. The part numbers for the sockets indicated that they were versions that *didn't* have locating pins -- something Kai had keenly been cognizant of, since the board design lacks them. I asked JLCPCB to confirm the parts, as redesigning the boards wasn't an option due to traces that ran through the area where the holes would need to be.

A few days go by and they email me again, this time from a different representative, saying that they could simply trim the locating pins off the sockets. At that point, I couldn't tell if they were mistaken about the parts I'd ordered, or if the supplier had shipped them the wrong version, but I wanted to keep the project moving. I confirmed that they should proceed, and paid the additional fee they charged to have some poor soul snip the pins on all 350 sockets by hand.

Another few days, and another email. There was a problem caused by one of the sockets, specifically the one at location UE10. It was too close to two adjoining capacitors for the pick-and-place machines. They asked, can I skip installing that socket? Hell no, those parts were expensive, $2 each, and nonreturnable, so I'd be throwing away $100. I asked if there were any alternatives, and they said they could try to apply an "offset" to the two capacitors -- basically, intentionally install them not perfectly aligned with the centers of their pads -- and see if there would be enough clearance. I asked them to proceed.

A week goes by and I get another email. A different representative was letting me know that the socket at UE10 was interfering with two nearby capacitors. What? I already talked with someone else about this. They helpfully included a photo this time, though, that illustrates the problem:

I explained to them that I'd like them to try an offset on the caps, and if that didn't work, to let me know. This time, they must have gotten the message as they replied a day later with mixed news: The offset would work for one of the caps, but not the other. This was OK with me, as the cap that wouldn't work out (C7) could still be populated by hand when everyone got their boards -- and wasting $20 worth of caps was a much easier hit than the sockets would have been. My mistake for not figuring that out ahead of time, so I decided to eat the cost.

Then...nothing. Over a week passes and I don't hear anything, and the production status never changed on the site. It was a complicated project, so I figured that maybe they were still working on it. I gave them a few more days, then sent an email asking for the status. They replied that they'd never received confirmation of the pick-and-place layout, so production was paused. Apparently I had to acknowledge this in the order details on the site, but never got any notification that I needed to do so. Ugh.

 

[This Does Not Compute]

International shipping shenanigans

I clicked the button like they asked, and about 3 days later, they were marked as finished and headed towards their shipping center. After another day, the package was picked up by FedEx and on its way to the US.

Despite going International Priority, it took 3 days for the box to make it to Anchorage, Alaska where FedEx hands packages off to US Customs. I've ordered plenty of stuff internationally before, and the process is usually just a formality, with clearance sometimes happening even before the package leaves its origin.

This time, though, it was different. An hour after getting off the plane, the package showed a customs clearance delay. I called FedEx to ask what it was about, and they said that the item description wasn't detailed enough. The agent I spoke to was able to update it, then I was told that Customs also needed an "importer registration number." I had no idea what that was, but they sent me an online form to fill out.

Turns out I was dealing with big-league importing. The package was large, heavy, and expensive enough that it was clear I was importing commercial goods, even if I ultimately wasn't making a profit off of them. Uncle Sam keeps track of this stuff, and companies have their own registration numbers they tie shipments to. I'm just some idiot on the Internet with a credit card, so of course I didn't have one. I waded through the multi-page form and found all the options where I could identify as an individual who does infrequent shipments, then sent it back and hoped it would placate them.

It was indeed sufficient to let Customs release the package, but only after two weeks of silence. I called FedEx every few days to ask what was going on and why the box was still in Alaska, but all they could do was send emails to Customs and nudge them to check the paperwork I'd submitted. But 19 days after leaving China, the box with the Reloaded boards was sitting at my doorstep.

I wasn't exactly thrilled with how JLCPCB had packed the boards. There were five bundles of 10 boards each. To make each bundle, they took two PCBs and put them face-to-face with a sheet of foam in between. Then the pair was wrapped once with bubble wrap, and another pair of boards was added to the bundle. At first glance this seemed OK, but as I unwrapped the bundles, I occasionally found a loose capacitor or tiny inductor. To my horror, I found that the padding between boards wasn't thick enough and the jostling from shipping caused them to rub together. Most of the boards were fine, but four of them had either an inductor on the bottom knocked off (strangely, it was always the same one), or worse, one or more of the SMD top-side caps ripped off the board.

Something that was always a little off-putting to me is that, after an order is completed, JLCPCB adds a button marked "Quality Complaint" right next to it. This sort of thing is apparently why, as clicking it opens a support case where you can attach photos and explain what happened.

When I got the box I inspected it carefully and there was no sign Customs ever opened it, so the problem was clearly due to improper packing on JLCPCB's part. Of the four damaged boards, one of them was clearly repairable, but the other three had pads torn off. Here's a lightly-edited transcript of the email chain, spoilered for brevity:

Spoiler: Conversation with JLCPCB customer service

Me:
Four boards arrived with components that had been knocked off. I believe this happened due to improper packing.
Board 1 - C1, both pads damaged.
Board 2 - C12, one pad damaged.
Board 3 - C12, one pad damaged, and L9 missing.
Board 4 - L9 missing.
The L9 components can be reworked but the C1 and C12 components cannot.

Them:
Thanks for reaching out and so sorry to hear about that.
Could you share us some images of the parcel you recieved? If the parcel arrived well? And could you share us some images how the boards were packed?
For the three boards with pad damaged, may I know if there is any reworking helpful?
After checking, if it is some improper operation on our side, we will also apply some compensation accordingly. Would it be OK?

Me:
Hello, the cardboard box doesn't show any damage. The problem is that the boards were packed face-to-face with only one layer of foam separating them. There were 10 boards to a bundle, with one layer of bubble wrap between each pair of boards. It was the boards rubbing against each other that caused the damage. I've attached two photos showing how they were packaged. There was no additional padding inside the box, just the five bundles as shown in the first photo. I don't think I can rework the boards with damaged pads, the pads were torn off the PCB completely with very little trace remaining to solder to. These will need to be replaced.

Them:
Thanks for your information
1-May I know if there is a lot of spacing in the outer box to pack these boards? Could you remind of that?
2-Actually the packing method you shown can fit our standard. For the parts removed, they seems to be larger. Would you consider adding another foam layer would be helpful? If yes, for future orders, you can contact us after you place the order and inform us to add one more foam layer between the boards.
3-For the defective boards, we are so sorry. There are three of them that cannot be reworked, is it right? Would it be OK that we apply a $15 coupon for you as compensation? For the one with L9 missing, kindly help to rework so it can be useful.
Sincerely sorry for the trouble caused.

Me:
There was no additional padding in the box, just the 5 bundles as shown in the photo. There was some extra space inside, yes. Correct, I cannot rework 3 of the boards because there are pads that were torn off. A $15 coupon is not acceptable as these boards cost me $60 each. A $200 coupon would be acceptable so I can have replacement boards made.

Them:
Sorry that I did not notice that you used some own stock in this order.
Since the compensation value has been beyond the compensation standard, please allow me some to check and report further.
We will update you asap.

Them (the next day):
We consulted with the factory.
For the boards with pads removed, could you please try to scraped off the solder ink and solder with red glue to fix the components?
We sincerely hope you can have a try again.
For this case, it happened in transit. When the parcel is on the way, we can hardly control. There may be some violent collision and affect the boards. We are sorry for that. Would it be Ok that we apply a $50 coupon for you as compensation?

So I have a few problems with this:
1. They barely glanced over the info I provided, I had to repeat myself a few times.
2. They semi-admitted that only one layer of foam between the boards wasn't sufficient, and that I could contact them for future orders to request additional padding. How was I supposed to know that was even an option to begin with? JLCPCB offers zero information explaining how they package orders.
3. Their first offer of compensation is...$15? Are you for real?
4. They're really insistent on getting me to just repair the boards. Which I *might* be able to do, but should I really have to? If your new MacBook showed up with a big dent in it because they packaged it poorly, would Apple tell you to fix it yourself?
5. Ultimately, the best they could do was a $50 coupon for my troubles. Ugh.

A decent part of my experience was simply due to ignorance on my part, and I'll admit that I learned a hell of a lot about PCB manufacturing in the process. But I also think that companies like JLCPCB (and probably PCBWay, though I have no experience with them) are really better off handling smaller, simpler projects -- like the Bourns Again filter chips -- or unpopulated PCBs. The timezone and communication differences made me feel very separated from what was going on, and I often worried that the boards would arrive with some sort of massive problem because they just wanted to get them out the door. That level of discomfort would be OK for a cheap personal project. This one, though, involved not just a lot of money, but a lot of *other people's* money. So let's get to that next.

 

[This Does Not Compute]

Money can't buy happiness, but it can buy PCBs

My initial estimates placed the per-board cost, for a run of 50, at around $65 each. Do the math and that's $3,250. I try to run my YouTube channel as a business; that is, I want it to be self-sustaining. So far that's worked out, but $3k was a stretch in terms of budget. Thus I'd need to collect payments up front.

The challenge is that there was no way to calculate the true *final* costs at the beginning. To get to my $65 estimate, I:

1. Uploaded the files to JLCPCB, selected the necessary settings (quantity, finish, component assembly, etc) and took note of what it listed as a price;
2. Went through the BOM of parts that LCSC didn't stock and got prices from Digi-Key; and
3. Added a small amount more as JLCPCB charges fees for every type of component it has to load on the pick-and-place machines.

To give myself a little wiggle room, I asked for $66 per board as a deposit from everyone interested. Thankfully, everyone who had signed up was serious about the project and paid promptly. I included a note that once the boards had been produced, I'd send an additional invoice to everyone with their actual shipping costs, plus any "true-up" amount (the difference between their deposit and the final, actual production cost).

In total, 47 boards had been spoken for. I wanted two for myself, and kept one spare as a contingency (in case a board proved defective, got lost in shipping, etc). I counted those three boards as ones I'd pay for myself, so I was already $198 into the project personally.

The first things I paid for were the gobally-sourced parts. As mentioned, this ended up getting split across 4 orders to accommodate parts availability. Something nice is that JLCPCB's interface lets you search by part number, then it will show the prices and availability across all the suppliers, so you can buy from whoever is cheapest. All my initial part cost estimates came from Digi-Key, but ultimately they ended up coming from six different distributors. This allowed me to optimize for price, and I was happy to see that there didn't seem to be any shipping costs involved.

Later I realized why: JLCPCB tacks a profit margin onto parts ordered this way. Ultimately, the amount of money I saved by buying from the cheapest supplier ended up being negated by JLCPCB's margin, and then some; my estimate was $70 short (which, in the grand scheme of things, is not bad!). Thus, when estimating parts costs, search for them in the Global Sourced Parts catalog to get accurate prices, instead of Digi-Key or Mouser's own sites (even if the parts are ultimately coming from them). The total for all of the parts that I had to order this way? $2,676.72.

After those all came into stock and I placed the board production order, here's how all the remaining costs and fees broke down:

PCB manufacturing: $385.22
PCB assembly: $198.06
In-stock JLCPCB (non-globally-sourced) parts: $114.82
Edge rail fee: $19.90
PLCC socket alignment pin removal fee (I still don't know if they needed to actually do this): $11.50

Then I got hit with a number I wasn't expecting to be so high, shipping the boards to me. The cheapest option was an eye-watering $210.24.

But I wasn't done buying stuff yet! My gut said that the boards wouldn't come in static bags (my gut proved to be right!), so I bought a pack for $27.33. A box of shipping labels was $10, and enough bubble wrap to package up all the orders was $32. The only thing I didn't have to buy were shipping boxes, as the USPS gives away free Priority Mail boxes (which was the way I planned to ship the orders out anyway).

So, the aggregate numbers:

Upfront money collected: $3,315
Total cost for the boards and shipping them to me: $3.658.12
Final per-board cost: $73.16

So my per-board cost estimate was off by $7.16, which I don't think is too bad. At this point, I prepared the final invoices for everyone. This consisted of:

1. $7 per board true-up for the JLCPCB costs,
2. $1.38 per board for packing material costs, and
3. the actual shipping costs from me to the recipient's address with no markup (for most people this was less than $10)

These numbers were generally rounded in the buyer's favor, and I ate the difference. I don't know how much this totaled up to, nor do I really want to.

Since JLCPCB couldn't assemble that capacitor at position C7, I bought appropriate tantalum caps from Digi-Key so I could include them with the boards. I'd leave soldering them in place up to the buyer, but I figured the least I could do would be to include the part. And as a further thanks/apology for how long the project took, and also for not being able to include a socket for the glue chip like originally planned, I included the Bourns Again boards I had previously made. Both of these were on my dime.

I had also massively lucked out with the four Reloaded boards that arrived damaged. One of them was able to be easily reworked, so it was good to go out to one of the buyers. I had earmarked three from the batch of 50 for myself at the beginning, and so the remaining three damaged boards stayed with me. As of this writing, I haven't had a chance to see if I can rework them. If I can't, well, I guess I have a $50 coupon to have some more made :-(

General advice

If you've gotten this far and are undeterred in your desire to run a group buy for SE/30 Reloaded boards with the top-side parts populated, here's some general advice.

1. Get a reasonable commitment from prospective buyers *before* deciding on how many boards to be made. Remember, Bolle's design is CC-Noncommercial, so you're not allowed to make a profit off of it. This was perhaps the most difficult part of the project. If it was your own product that you could sell for profit, you'd set a target margin and if your costs ended up higher than expected, you'd just eat into that and keep the retail price fixed. It's impossible to know exactly how much the final cost for a run of these boards would be until you've actually placed all the orders, so the best you can do is estimate like I did. The cost scales with quantity, so you need to know that number first.

2. Plan to have more boards made than are spoken for. Some people who show interest may back out at the last minute, but it's more likely (in my experience) that the final tally of paid-for boards will be higher than the initial estimate. You'll also want to account for extra boards in case they get damaged in shipping or lost on the way to a buyer. Don't forget to also include whatever quantity of boards you want for yourself in the total. Which brings us to...

3. Expect to lose money on this endeavor. Hopefully it's not a lot, but there are many moving parts to this project and even though it's all being done at-cost, it's not fair to buyers if you screw up with the math somewhere along the way. I paid out of pocket for the boards that I wanted for myself, and I also ate any last-minute costs that came up. That said, I was very fortunate to have buyers who were reasonable and thankful, so a modest ask for more funds later (the true-up and shipping costs) wasn't a problem.

4. Transparency is key. You're just some person on the Internet that strangers are trusting their money to. In my case, all of those strangers (and a few friends) even sent me the funds via PayPal as friends/family transfers, so while there were no fees, they also didn't get any of the buyer protections. It's very important that you communicate with the group clearly and frequently. I sent an email with every major step in the process, and more frequent ones as the actual board production was happening. As soon as the box of finished boards showed up, I pulled one out and sent a photo of it to everyone.

5. Get your ducks in a row. 95% of my time on this project was spent dealing with spreadsheets and paperwork. Most of that was managing the parts BOM and navigating the supply chain situation, with the rest of the time going to ordering parts, checking order status, emailing with JLCPCB, etc. Actually placing the PCB order took maybe 15 minutes, but getting to that point took dozens of hours of work. Having everything ready, and keeping on top of things, made the final "place order" button much easier to click.

6. Most important -- be willing to learn and have an appetite for problem-solving. While I got help along the way, which I can't understate how much I appreciate, I knew I couldn't get others to do everything for me. I had to figure out how BOM and CPL files are formatted and what I needed to do to edit them, along with navigating JLCPCB's (sometimes buggy) site. In the end, I learned a ridiculous amount, even though in reality I just scratched the surface. It's very fulfilling to see the boards go out to fellow Mac nerds, with the knowledge that they were the first SE/30 Reloaded boards in the world with so many parts assembled at the factory. The struggles were absolutely worth it to me, but they may not be for you. Hopefully this gives you better insight into what producing and selling PCBs is really like. I had no idea how deep the rabbit hole went until I jumped down it.

So you actually wanna do it

You read everything above, right? Read everything above if you haven't already. There's no TL;DR.

You'll need two sets of files:

1. The design files, you can download the Zip archive of them from Bolle's GitHub page: https://github.com/TheRealBolle/SE30
2. The BOM and CPL files (both CSVs)

For the latter, you have two choices. If you only care about the bottom side of the board being populated, just grab the BOM and CPL files from Bolle's GitHub. Throw them all at JLCPCB and it'll mostly walk you through it. There might be a couple parts that aren't in stock at JLCPCB, so you'll need to order them through the Global-Sourced Parts portal.

If you want to go for it with some top-side parts, though, I've attached my own BOM and CPL files you can download. Note that you *will* need to edit the BOM depending on whether you want to include PLCC sockets. Here's a spreadsheet that lists all of the part details, which should help you when ordering the globally-sourced parts and getting the final BOM file sorted:

SE/30 Reloaded BOM Worksheet

docs.google.com

Ignore the "Top-Side Special Parts" tab; that's for the parts that JLCPCB *can't* assemble (either because the parts aren't obtainable, or because it would cost too much). The spreadsheet contains more info than should be in the BOM file, so don't just export it as-is. Here's how to read the BOM:

--Comment: This column is a brief description of the part, useful for the JLCPCB folks to make sure the right parts get installed in the right places
--Designator: Lists all of the places on the board that part needs to be installed
--Footprint: Describes the physical dimensions of the part for the pick-and-place process
--LCSC Part #: A JLCPCB-specific column, if you're using parts from JLCPCB's own parts catalog (not global-sourced parts), put the part number here

Any parts you *don't* plan to have JLCPCB assemble should be removed from the BOM file before you upload it.

If you decide to go for it with the PLCC sockets, you'll want to decide how to handle the UE10 situation. To make things easy, you could just remove C5 and C7 from the BOM and install them yourself. Alternatively, you could leave C5 on there and ask them to apply an offset to it so it'll fit. For C7, you might be able to find alternative caps that would fit, or just leave it off the BOM. I chose the latter.

I'm happy to answer questions here as I can, but regretfully I can't take an active role in anyone's own group buy due to time constraints.

(I should have turned all of this into a video, shouldn't I?)

 

[Kai Robinson]

I think a video on this would be fascinating, also, the script is already written!

 

[MacOfAllTrades]

Thanks for reporting. Naming and shaming, imho, is well within reason given your experience and numerous (bad) interactions with them.

Thanks for all that work - and I'm saying that not even having participated in the buy! I'll give my working se/30 board an extra kiss tonight

 

[Patrick]

thanks so much for this writeup. those of us in this group buy had a small taste of what you were going through. But for the most part you kept the complexity of the logistics hidden from us.

I appreciate knowing how hard a project like this can be. But also, that it CAN be done. thats great !

 

[trag]

(I should have turned all of this into a video, shouldn't I?)

I vastly prefer the written version. Videos are a frighteningly inefficient way to consume information in most cases.

That said, what a great summary. This is a terrific narrative. Concise, but with all the important points covered.

Also, thank you again for running the group buy.

 

[MacKilRoy]

I think a video on this would be fascinating, also, the script is already written!

Did anyone else visualize and hear the "TDNC" intro music and Colin's voice throughout the whole saga, or was that just me?

Seriously, though, I agree that a video would be great.

 

[RetroViator]

What an amazing summary of something that has clearly taken a lot of your time. In my opinion, your final cost was reasonable given what is pre-populated on the board. I also appreciate you posting this here, where it can be found and helpful to others in the future. Videos are great for storytelling, but it’s nice to have a more permanent reference.

 

[cy384]

With all due respect to Bolle's great technical work on this project, this is a perfect demonstration of the problems with trying to produce hardware without any margin (i.e. non-commercial licenses). I've seen the rule of thumb to sell at a price of absolutely no less than 2.5x your parts/assembly cost. You did a huge amount of unpaid work and also ate extra costs to make it happen. If a piece of hardware can only exist through heroic effort, it's simply not going to be regularly available to people who want to have one, and anyone running group buys gets burnt out.

Especially with a large board that's using uncommon parts, through hole parts, hasn't gone through a DFM process, etc., honestly I'm very impressed with how well you did. I hope the community finds your writeup very educational.

 

[This Does Not Compute]

You did a huge amount of unpaid work and also ate extra costs to make it happen.

This is true, but I would have had to do almost as much work had I just produced five boards for myself, and my total out-of-pocket expenditure would have been *much* higher. The economies of scale really came into play with this one.

 

[JDW]

@Bolle
In light of all that Colin went through, can the "non-commercial" terms you established be modified to make group buys much easier? If not, could you, perhaps, start selling mostly populated boards?

---

What we do at my workplace (we make mostly electronic security systems, mind you, and mostly for 12V vehicles) is deal with our contracted factories in Taiwan, which we have dealt with for decades. We know everyone there and everyone there knows us. Indeed, I am the primary point of contact. Trust has long been established, and we do everything for the best interest of each other. Everything is done in English, and I know how to avoid communication issues. Trust and good communication is the first important step to success with any mass produced item.

Getting a brand new design made at our partner factories isn't as hard as you think. We basically just provide our factories with our schematic designs. And no, those designs don't even need to be made in a professional layout app. You could draft it in illustrator and send it off, and they will input the design into their preferred software, and generate the PCB too, which is no small task when you want something done right. They will discussed how many layers it should be, and sometimes that can vary during the design process when you wish to mitigate noise (e.g., by moving from 2-layer to 4-layer).

Our factory will also source the parts for us, and even sell us small quantity samples to test (usually costing several hundred dollars), which is an important step before you mass produce anything. You ALWAYS want to make a fully working set of samples before you mass produce anything. But in the case of our designs, that's easy. Not so with the SE/30 Reloaded boards because they will not come 100% populated.

Anyway, our factories will of course manufacture the PCBs in bulk, and then they will perform testing IN TAIWAN (they make their own test jigs to achieve that) before they ship to us in Japan. We then perform 100% QC checks before we ship to our customers, because in Japan, customers demand absolute perfection. Why? Because the selling cost is high.

But we have never asked our factory partners to build anything as big and complex as the SE/30 motherboard. And again, because the board wouldn't be fully populated, testing that at the factory isn't an easy task at all.

But the even bigger problem is that everything we buy is in quantities of 500pcs or higher. (That's called a MOQ, for those of you who don't know.) So that high MOQ, in combination with Bolle's current "non-commercial" terms is what I could never consider offering a group buy myself. And asking my company to do it would not work either is they won't do it for "no profit," and overseas licensing and royalty terms can get messy.

Clearly, there is demand for the SE/30 Reloaded board, but only when it is mostly populated. Of course! Me too. I don't want a bare board. This isn't an SE Reloaded board after all. Thru-hole is easy. SMD is hard. Very hard!

I am a complete idiot because when I stopped getting email updates about replies to Colin's group buy thread, I then got too busy to remember to come back to that thread on my own. A couple months went by, and then I heard people were getting their boards! Ack! I missed it. A total blunder on my part.

Moving forward, there will probably be more and more demand for these mostly populated SE/30 boards. Even if they were $100 each, people would likely pay it. But would a 500pc manufacturing run be feasible? Probably not. Because who is going to eat the cost of 500pcs (which all must be paid to the factory at the time of shipping), and then the recipient of the said boards would need to stock 500pcs in hopes of selling them all off over a 2-5 year span of time? (50-100 might sell right away, but I doubt all 500pcs would.)

Some individual probably could go about it like Colin did and use one of the mentioned Chinese firms. But as you could read, it's an enormous time commitment, and there is financial risk as well as the risk of trouble with product defects or assembly blunders.

So there is the dilemma. It's a fantastic reloaded board, and Bolle did an amazing job. But it's now all about how to get that into the hands of people who want it. As Colin so eloquently describes in his opening posts in this thread though, it's no small task.

With all the many things Colin must juggle, I stand in amazement at how well it was able to pull it off, despite the many problems. Bravo, Colin!

 

[TT/30]

Has bolle clarified what the license means somewhere? I was thinking about this topic of CC-BY-NC-SA licensed boards since I just had a simple board made with this license. If someone does a bunch of value added work that is clearly documented in this post, can they not charge for that effort and labor? To extend that logic further, if you build a board or buy one made on a non-profit basis and later decide to sell your machine, does that mean you have to pull the board out and scrap it? I wouldn't think that's the intention of making a board CC-BY-NC-SA, but maybe it should be clarified.

 

[JDW]

Has bolle clarified what the license means somewhere?

I sent @Bolle a private message on Discord yesterday, linking this thread for him to review.

 

[This Does Not Compute]

Has bolle clarified what the license means somewhere? I was thinking about this topic of CC-BY-NC-SA licensed boards since I just had a simple board made with this license. If someone does a bunch of value added work that is clearly documented in this post, can they not charge for that effort and labor? To extend that logic further, if you build a board or buy one made on a non-profit basis and later decide to sell your machine, does that mean you have to pull the board out and scrap it? I wouldn't think that's the intention of making a board CC-BY-NC-SA, but maybe it should be clarified.

Here's my interpretation, based both on the nature of the project and also my gut: The board was designed and released at no cost to the retro community for the benefit of the retro community. Sometimes people release stuff NC so they can protect their own ability to sell the item (commonly seen in the 3D printing community), but Bolle isn't selling these. That's why I felt OK doing a group buy with the provision that it was at-cost -- Bolle didn't make any money, but neither did I, and the "profit" that I personally received was just the opportunity to actually get a board for myself at a far lower cost. Does my work adding the components to the board somehow transmogrify it sufficiently so that I could sell it for a profit? I don't think so, since it wasn't a complete board and the buyers still had a lot of work to do to make them usable (plus the fact that I had help from Kai and Bolle himself). The "spirit" of the group buy Reloaded boards was still the same as what's represented on Bolle's Github page: an incomplete PCB that you can transplant parts onto.

Do I think it's OK if someone sells a complete SE/30 with one of these boards inside it? Yes. Since the buyer is getting more than just Bolle's design, they're getting a whole, working system that the seller has put a lot of time and money. There could also be an augment for being able to sell just a complete Reloaded board for the same reasons, it's become "greater than the sum of its parts". The limitation I'd place is that such sales should be individual ones only, e.g. you built the SE/30, played with it for a while, then got bored and want to sell it. Building a batch of boards and reselling them for the intent of profit would cross the line.

 

[TT/30]

Yeah, it's more of a general question of CC-BY-NC-SA boards and what that means for a maker/builder. There are some grey areas with the value added from turning the PCB into a PCBA (partial or not ) and derivative works. I wouldn't discount the effort of assembly as much since it's a significant amount of work to do it well and every unit requires assembly. I don't think anyone in the community wants to go against the goodwill of bolle's efforts. However, I would not be surprised if someone eventually creates another version of an SE/30 compatible board and makes it a commercial effort, which has happened with other 68k projects where there is enough demand. But I'd hope that energy went into doing something different, like making a more modern version like what Analogue did with the SNES, etc.

 

[Bolle]

To be honest... I went for that license because everyone else seems to use it as well. I don't have the time and I don't want to deal with any of that licensing stuff at all. There's already someone on eBay in Germany commercially selling unpopulated boards (they do have a lot of other boards for sale as well that they just pulled from github for various other stuff)
I personally don't care enough to go through the effort to have eBay take that listing down - I couldn't find an easy way to do it after looking for one for 5 minutes and kind of lost interest.
If one is taken down another one will pop up eventually... idiots gonna idiot all the time and I have better ways to spent my time than going after them.

There's also a few people that offer replacement services I think which if you ask me is totally ok as well. There's actual work involved with that so I'm totally fine with it.

For group buys like this happening inside a closed community I totally don't have anything against the people organizing it making back at least the costs for a few boards to keep on their own if not more considered all the work that's involved. I think I clarified that as well in PM when talking to @This Does Not Compute about the group buy.

 

[TT/30]

Thanks for clarifying @Bolle!

Just to add another thought, I have participated in a non-profit Kickstarter where they plan on donating extra proceeds to a charity relevant to the hobby, so maybe that's another option for someone trying to make the venture non-profit. Or issue partial refunds, but that is probably more of a hassle for the organizer.

 

[croissantking]

@This Does Not Compute

I too, would be interested in a video! You clearly put so much care into the whole process from start to finish, but were let down in certain key aspects by JLC. The poor packing and their 'hand-washing' is especially disappointing. I hope you can get your three boards working and while not pretty I'm totally confident you can patch up the ripped pads. C1 and C12 have very short traces leading to vias, so you can just add a piece of repair wire on each side, feeding it through the via and soldering in place. Connect to the caps, tacking down with a dab of hot glue or epoxy resin to give it mechanical stability.