Anyone any experience with the LumenPnP from Opulo (Stephen Hawes)

[stuartlea]

I'm thinking of pushing the go button on one of these for light duty use....Anyone played with one?

Thanks, Stu

[MR]

Sorry to say but it doesn't seem to be worth it.

Motors -- too small, too low power
Simple gantry design ... just buy a 3d printer and dodge a cheap pnp head on it.
OpenPNP ... not even developed by Opulo...
EMC ... don't even dream about that.
eg.
https://www.m3.tuc.gr/EQUIPMENT/DMU50eco/Sinumerik%20810d/Planning%20Guide%20Edition.pdf

And all those printed parts DIY ... I mean you can print them straight away right now.
Extruded Aluminum is cheap and you can get it everywhere.
There's not even a drag-chain just some dodgy hanging part.

1700$ ok the guy is an enthusiast who should be supported but that's too much.

If I would like to go that way I'd rather buy a chinese machine maybe used, and update their controller to support openpnp if the chinese
software is not sufficient.

A PNP is not difficult to design and build --- it's all about the feeders.
Want 50 feeders? ... setup-time per feeder 3-5 minutes ... 2:30 - ~4:00 setup time for the feeders.
It's quite some work to design reliable feeders.
How about humidity can the machine stand different humidity levels?

- You also need to organize the smd components
- You will have to organize your PCB projects (and revisions)
- possibly stock monitoring / component counter

- you will need a stencil printer
- you will need a reflow oven

certainly it can all be done cheaply but if you want to do it efficiently it will be more expensive.

[mairo]

I wonder if the recent NeoDen move with theirs low cost YY1 is in response to the low cost Index PnP.

[newto]

I've been following the development for a while, and I think it's *almost* there. As MR said, it seems a bit over priced for what you get in the kit, so I hope they can bring the price down with volume.

The thing I'm most interested in is the feeders, his most recent video has a very slim and probably cheap design. If he can get them going reliably for 30-40 bucks that would probably make the design worth it.

[Yellofriend]

Sorry to say but it doesn't seem to be worth it.

I disagree.

He puts lots of thought into his PnP and the cost is reasonable. Bonus: He really uses his own machine - this concept is unknown in China!

I have a Chinese PnP and parts are failing left and right. Mainly due to poor connectors (all no brand from the electronic markets). Then you have dreadful software (mine uses Windows 7).

I believe with the LumenPnP you have at least a user that can give you support.

Of course it's a hobby machine and not for the daily production user.

Disclaimer: I am not related to LumenPnP, I have no LumenPnP, I don't plan to buy a LumenPnP. If I had no PnP I would consider a LumenPnP.

[MR]

Just subscribe to the openpnp mailinglist and you'll get all support you can dream about.

[Pinkus]

Sorry to say but it doesn't seem to be worth it.

Motors -- too small, too low power
Simple gantry design ... just buy a 3d printer and dodge a cheap pnp head on it.
OpenPNP ... not even developed by Opulo...
EMC ... don't even dream about that.
eg.
https://www.m3.tuc.gr/EQUIPMENT/DMU50eco/Sinumerik%20810d/Planning%20Guide%20Edition.pdf

And all those printed parts DIY ... I mean you can print them straight away right now.
Extruded Aluminum is cheap and you can get it everywhere.
There's not even a drag-chain just some dodgy hanging part.

1700$ ok the guy is an enthusiast who should be supported but that's too much.

If I would like to go that way I'd rather buy a chinese machine maybe used, and update their controller to support openpnp if the chinese
software is not sufficient.

A PNP is not difficult to design and build --- it's all about the feeders.
Want 50 feeders? ... setup-time per feeder 3-5 minutes ... 2:30 - ~4:00 setup time for the feeders.
It's quite some work to design reliable feeders.
How about humidity can the machine stand different humidity levels?

- You also need to organize the smd components
- You will have to organize your PCB projects (and revisions)
- possibly stock monitoring / component counter

- you will need a stencil printer
- you will need a reflow oven

certainly it can all be done cheaply but if you want to do it efficiently it will be more expensive.

I should note that this is an open source project from Steven. Yes, you can purchase a pre-assembled Lumen PNP V3 for $1745 but you can also get all files from the Github. And if I say all files I mean all: KiCad files and Gerber files for the PCB, CAD files and STLs for the 3D parts, building plans, firmware files etc. etc.: everything is there to build you own Lumen right now.
On Discord there is a huge community supporting and enhancing the Lumen.
The kit (all parts but 3D parts) with pre-assembled cables etc. were available in mid 2022 for $999-$1145, but it looks like, they only sell fully assembled units right now.

However, if you have a lot of time, you can purchase all parts for approx. $750 (material cost, or use what you have on stock). But then you will need to throw in many hours of your time to build your Lumen. Your time = saved money. Expect a week full time (40 hours) for the first Lumen (plus learning time for OpenPNP).

[Doctorandus_P]

The mechanics of this one look much better:

https://www.microsmt.com.cn/

And there are some other open souce projects such as liteplacer and openPnP

[super7800]

not to necro it too bad, but incase your still looking... i just bought one for work (where price really wasn't a talking point, so why'd i pick this  ). Just need it to aid the interns in assembling demo boards (which we do in house, built to order, for some insane reason)

Haven't gotten it working fully yet, spent a week on it so far. It shipped to me with bricked firmware on the mainboard that simply doesn't work, but after a few days they sent me firmware that does work. The sensor less homing is shit, and I'm going to have to redesign it with physical switches. The supplied cameras are mediocre at best, but do work, most of the time. It is flimsy and i have my doubts about its 0402 claims. For the price, i expected them to have made the OPENPNP setup easier. It is not. I would like to have seen them make a custom fork of it. the build plates are just PCB material. Not even aluminum PCB, regular fiberglass, and flex like crazy.

Now that i'm done hating on it, the good: The feeders are cheap. Like, really cheap. And they are great. My personal machine (SMTMax) uses some proprietary software with the "pin pull" feeders but im considering switching to openPNP just to use these feeders (only offered in 8mm tho) . Once i get it working its likely we will buy one for every passive we use so that they are just saved onto the machine. The feeders are smart and remember all their settings and components. The machine is also "open source" and customizable. All in all expected more for the price. Not better quality, but better "it just works". I need a machine that just worked off-the-shelf, but this ain't it. My (arguably very overpriced) SMTMax machine was assembling test boards the same day i received it.

The backlash is not terrible, and the motors are not underpowered for a belt machine. The new version uses a proper drag chain and is OK, but the weight of it tilts the gantry head backwards!

Another note: The feeders get hot. Real hot. For no reason. They used an inefficient linear regulator!! The inrush current with feeders is insane! many many amps! WTF! Mechanically and software wise the feeders are a work of art. Like legit want to frame one on my wall as an example of 3D printed manufacturing. But electrically not so great imo. Not terrible, but just not as good as they could be, and when compared to the work of art the mechanics are it just pales in comparison.

Ill make another post once i have it in production. all-in-all i would recommend this if you like to tinker, but for actual work without spending weeks getting it perfect, i would say its a no-go. Do i regret starting this saga? no. Would i have pitched a different machine? maybe. I wish they'd offer a full service machine (with PC) fully configured tuned and working. Easily would have paid double for that.

[thm_w]

Another note: The feeders get hot. Real hot. For no reason. They used an inefficient linear regulator!! The inrush current with feeders is insane! many many amps! WTF! Mechanically and software wise the feeders are a work of art. Like legit want to frame one on my wall as an example of 3D printed manufacturing. But electrically not so great imo. Not terrible, but just not as good as they could be, and when compared to the work of art the mechanics are it just pales in comparison.

Ill make another post once i have it in production. all-in-all i would recommend this if you like to tinker, but for actual work without spending weeks getting it perfect, i would say its a no-go. Do i regret starting this saga? no. Would i have pitched a different machine? maybe. I wish they'd offer a full service machine (with PC) fully configured tuned and working. Easily would have paid double for that.

Any idea what the voltage to the feeders is? They use a switching reg to bring it to 10V, then a 1117 linear reg to bring it from 10V to 3.3V, as you say.

Probably should just be 12V and a switching reg to 3v3 if that is possible. Sort of mentioned here:
https://github.com/opulo-inc/feeder/issues/34

[super7800]

the power brick is +24v 5Amp. I assume the feeders are fed 24V. Why they don't just use a dual switching regulator is beyond me.

[johofz]

The sensor less homing is shit, and I'm going to have to redesign it with physical switches.

I thought that homing through fiducial markers was satisfactory. Do you make use of that functionality?

the build plates are just PCB material. Not even aluminum PCB, regular fiberglass, and flex like crazy.

The flexing would be greater with aluminum sheets. Stiffness, rather than tensile strength, is the key factor to consider. When it comes to stiffness, PCBs of sufficient thickness provide adequate rigidity for this purpose while also being lightweight. This is particularly important for shipping costs. While a one-inch aluminum plate would offer superior stiffness, it would be excessive and unnecessarily heavy and expensive.

[thm_w]

the build plates are just PCB material. Not even aluminum PCB, regular fiberglass, and flex like crazy.

The flexing would be greater with aluminum sheets. Stiffness, rather than tensile strength, is the key factor to consider. When it comes to stiffness, PCBs of sufficient thickness provide adequate rigidity for this purpose while also being lightweight. This is particularly important for shipping costs. While a one-inch aluminum plate would offer superior stiffness, it would be excessive and unnecessarily heavy and expensive.

Youngs modulus for aluminum is about 3x that of FR4 (69GPa vs 24GPa).
While the density is only about 50% greater (2.7g vs 1.85g/cm3).

So I don't think what you are saying is correct.

[CrisErving]

OK, if the LumenPNP isn't worth the $2500 ($2000 + $500 for some feeders), what is a good alternative?   I'm looking for something to make mounting a few parts on some boards easier... SSOP's and 0.65mm pitch LCC's, plus 0805 passives (which I could do by hand but they're close enough to the IC's that a little slip would knock them off kilter).

[super7800]

i wouldnt say its not worth it. It is great value for the money. Its just not a great machine by any means, and has a lot of small issues. I've found after assembling (well, having some poor intern assemble) a few thousand simple boards (roughly two terminal blocks, 8 0603 test points, 2 caps, and a part), that it is not a machine that can be left alone. It has to be constantly watched to work right, and makes at least one mistake per panel. It does 0603 and especially 0805 passives great, well as great as can be expected. If all you want to do is have it place passives, this is probably a great machine for you.

Alternatives would be DIY print it + order mechanical parts yourself (i recommend JLC PCBs 3D printing resin service), or find a used machine. My smtmax qm2100 is amazing and beats the pants off of this machine all day, and i bought that used. I can leave that thing alone placing 0402 and 0201 components and be damn sure it won't mess up (software is clunky as hell tho). Honestly, my recommendation would be to find a used machine. Not a big floor standing unit, but something between like the QM2100. I personally wouldn't buy anything that didn't use real linear rails and closed loop steppers or servos, so look for that in a new/used machine.

maybe convert a machine? https://hackaday.com/2020/06/24/how-to-retrofit-a-pick-and-place-machine-for-openpnp-in-detail/

[meshtron]

I recently purchased a (yet to be delivered) TVM802B to be my first PnP.  While I'll be using it "for production," I anticipate low volume and high babysitting effort, that's fine with me and I think in the price range I had available, that was always going to be the case.

I looked at the Lumen (as well as the Pandplacer A1 and the mysterious TornadoSMT V2) and seriously considered it.  I really wanted an OpenPNP machine, but at the end of the day this device - for me - is more a tool than a project.  I felt like the Lumen is probably another 12-18 months away from realistically getting to that point (based on what I can see and what I've read from a number of users).  Mad respect from me to the team for all they've accomplished, but (as an example) there are no feeders yet that are not 8mm, and the lead time was 5 weeks (ish).  Had the unit and feeders been in stock, I might have one in my shop right now.

I think the $2-4k desktop PnP space has a lot of options, very little documentation, and almost anything is going to have good and bad aspects.  For me, ultimately, it came down to the fact that the maximum advertised throughput on the Lumen is "800+cph."  I expect even a fiddly TVM, with passive feeders to (after it's dialed) be at least double that and probably faster.

I hope the Lumen team continues making progress, seems like a great project.