To buy or not to buy, which is the problem? Neoden 4 VS Neoden YY1 VS P20

[Fdr87CC]

It’s been a full month now (even though it feels like years) of investigation for a PNP machine and I am back to the origin of the search (one more time). That is why, with all my frustration desperation, I decided to try and get some help from experienced people: maybe this time I will take my final decision. It may be.
Long story short (no story is ever been made short but I will try) we are in need to set up a little prototyping area in our facility in order to be able to produce prototypes in little time.

The machines I’ve selected (Neoden 4, Neoden YY1 and Mechatronics Systems P20) all have:
-   Bottom camera;
-   Fiducial recognition;
-   Automatic feeders;
-   Comparable feeder capacity.

We don’t care about CPH, but we further need:
-   User friendly software. It would be nice to avoid spending one hour to program it for a 20 minutes run;
-   Reliability. The idea is the following: in the meantime the machine populates the board I am going to take a coffee and, when I came back, the machine shouldn’t be stuck after the 10th step.

We are using passives (1206, 0805, 0603), LQFP100, SOD-123, SOT23, DO-219AB, SOP-8, TO263-7, VSSOP-8, WSON8, SO8 and DPAK.
In this light, which of the mentioned machines would be the most reliable and easy to use? Is it worth spending more money to get a Neoden 4 (or P20) or would the YY1 will be good enough? Any other recommendation worth to mention?

Thanks to your contribution my headache will be out of the picture!

[Jackster]

Personally, I would stay away from Neoden going off this thread https://www.eevblog.com/forum/manufacture/neoden-yy1-pick-and-place-machine-with-under-$3k-price-for-hobbiestlow-vol-usag/msg4292275/#msg4292275

What sort of workload are you looking at putting through a machine? Total boards/panels, components per board, accuracy requirements (size of components) etc?

[Fdr87CC]

Thank you for the support!!

I had a look and I discarded the YY1, but I still see Neoden 4 as an option.

I expect to produce few different boards in a year: between 3 to 5 PCBs per prototype, and no more than 10/12 designs. Worst case scenario would be 60 PCBs, not much (but PCBs may have 200/300 components each, which means a lot of work if I were to populate components by hand). I don’t need to use any panel, just single PCBs. 

Concerning components size:
-   Smallest passive footprint is 0603;
-   LQFP100 with pitch of 0,5mm;
-   Several ICs with 0,4mm and 0,5mm.

[Mangozac]

Thank you for the support!!

I had a look and I discarded the YY1, but I still see Neoden 4 as an option.

I expect to produce few different boards in a year: between 3 to 5 PCBs per prototype, and no more than 10/12 designs. Worst case scenario would be 60 PCBs, not much (but PCBs may have 200/300 components each, which means a lot of work if I were to populate components by hand). I don’t need to use any panel, just single PCBs. 

Based on that I still wouldn't bother. 300 components is a lot to place but you will spend a lot of time programming and mucking around that I still think it would be a break even in terms of time spent. You're probably better off investing in a decent electric assist manual placer and a good method of picking from cut tape.

[mikeselectricstuff]

-   Several ICs with 0,4mm and 0,5mm.

I think any cheap machine is likely to struggle with that - expect to do some manual tweaking between placement and reflow.

-   User friendly software. It would be nice to avoid spending one hour to program it for a 20 minutes run;

Forget that with any Chinese machine, until you write your own software to translate

-   Reliability. The idea is the following: in the meantime the machine populates the board I am going to take a coffee and, when I came back, the machine shouldn’t be stuck after the 10th step.

The biggest issue you are likely to see affecting reliability is feeders. Unfortunately AFAIK no Chinese machine does the sensible thing on a feeder problem, which is to move on to the next part after a few retries and alert all problem parts at the end of the job, so it gets as much as possible done without attention - instead they just stop where they are and cry for help.

[jmelson]

I
We don’t care about CPH, but we further need:
-   User friendly software. It would be nice to avoid spending one hour to program it for a 20 minutes run;
-   Reliability. The idea is the following: in the meantime the machine populates the board I am going to take a coffee and, when I came back, the machine shouldn’t be stuck after the 10th step.

Well, the idea that you can walk away from the machine for 20+ minutes may well be a fantasy.  I had to babysit my old Philips machine every second to try to prevent feeder jams due to the weak pull of the cover tape.  My newer Quad QSA machine with electronic feeders does much better, but I still have tp be ready to deal with feeder issues.  The poor quality of the component tapes seems to be the real problem.  Cover tape can be glued so stronglo it rips the paper tape apart in one spot, and then so weak at another part that the components fall out.  My QSA is a ~  $150K machine when it was new.

Of course, crummy software can be a huge issue.  My old Philips was REALLY easy to program, but it had no vision at all.  My newer Quad QSA has flying vision, and can be real picky about component dimensions, sometimes rejecting a bunch of parts, then you have to tweak the dimensions.  It DOES have reallg GOOD diagnostics for this.  They have "test pick" where it picks up a part and shows the measurement on the screen, so you can tell what the vision system is getting.
Jon

[Mangozac]

The poor quality of the component tapes seems to be the real problem.  Cover tape can be glued so stronglo it rips the paper tape apart in one spot, and then so weak at another part that the components fall out.

This is a really good point too. Tape quality can vary enormously, even from brand name manufacturers and the big distributors.

[SMTech]

The poor quality of the component tapes seems to be the real problem.  Cover tape can be glued so stronglo it rips the paper tape apart in one spot, and then so weak at another part that the components fall out.

This is a really good point too. Tape quality can vary enormously, even from brand name manufacturers and the big distributors.

20 years of doing this suggests this is not a significant issue. 99% of all tape feeder issues can be traced back to either the maintenance or design of the feeder, or the condition of the tape. Short strips from a distributor can be squashed by other things in the parcel during shipping, overly coiled in a bag, coiled backwards (bad news on plastic in particular). These lower volume part can also end up being the kind of thing you end having hanging around much longer than intended, and then you see things like all the glue failing. On our Essemtec the only tape type I see a consistent issue with is the slightly stiffer sticker cover tape you find on plastic tapes typically holding things like LEDs, poly film caps and rectifiers, its distinctly different from other cover tape types as you can stick it back in place after peeling it. The pull off tension appears to be higher so its more sensitive to how slippery the take off spool "clutch" is on the Essemtec feeder. These tapes therefore seem to take a little longer to "bed in" on the feeder and something about them (excess stickyness???) means they are somewhat more prone to not advancing properly. It is clearly not a tape fault or this tape type would not exist, it just falls outside the magic of the very simple and cheap mechanism Essemtec uses to collect cover tape. There are worse mechanisms out there that will fail more often (e.g Mike's Versatronics or the deep pocket feeder on our Essemtec).

[nctnico]

Thank you for the support!!

I had a look and I discarded the YY1, but I still see Neoden 4 as an option.

I expect to produce few different boards in a year: between 3 to 5 PCBs per prototype, and no more than 10/12 designs. Worst case scenario would be 60 PCBs, not much (but PCBs may have 200/300 components each, which means a lot of work if I were to populate components by hand). I don’t need to use any panel, just single PCBs. 

You'll be better off outsourcing. Having a P&P machine only makes sense if you are going to produce a serious amount of boards. Say starting from hundreds a year. There are plenty of places where you can have boards assembled for a low price. For prototyping purposes the LitePlacer looks kind of interesting.

[Mangozac]

99% of all tape feeder issues can be traced back to either the maintenance or design of the feeder, or the condition of the tape. Short strips from a distributor can be squashed by other things in the parcel during shipping, overly coiled in a bag, coiled backwards (bad news on plastic in particular). These lower volume part can also end up being the kind of thing you end having hanging around much longer than intended, and then you see things like all the glue failing.

Regardless of machine/feeder design that's exactly the kind of thing OP is likely to be dealing with. He's doing ultra low volume, so will be dealing with plenty of poor re-reeling, spliced tape and reels that have been sitting around for a while. Which means any intention of being able to just press "go" and produce a dozen boards is just a fantasy.

[Fdr87CC]

Thanks for the suggestions but I think we are deviating from the topic, which is: between Neoden 4, Neoden YY1 and P20 (Mechatronic Systems) which is the best option for prototyping purposes?

It is clear that being low cost machine (it’s a requirement) I will necessarily deal with problems, but that is fine: no one ever bought a Ferrari for the price of a Panda.

And of course we analysed our situation carefully and it makes totally sense (time wise) to have one of those machines, the alternative would be me spending at least one day per board with the -very likely- risk of making mistakes (swaps of resistors and so on).

Given that, I discarded Neoden YY1 thanks to Jackster contribution. Neoden 4 and P20 are still possible options. I was able to find extra info regarding Neoden 4 but I wonder if any of you has experience whit P20.
 

[48X24X48X]

What is your budget in USD?

[Fdr87CC]

Hi 48X24X48X, my budget is around 30K max

[48X24X48X]

And why are you limitting yourself to those machine? There's plenty to choose from if you have USD30K. Read my long blog post on getting a small assembly line (I won't post here), it might give you a better idea. My budget was much smaller than USD30K!

[girts]

Hi 48X24X48X, my budget is around 30K max

N-4 with feeder pack loaded to max + manual stencil printer from technoprint + Eurocircuits reflow = about 16k eur.
Everything is available locally in Netherlands.
Remaining ~13k you can donate to somebody who knows what to do with it.
Or use excess funds to feed some homeless cats in your backyard.

[TonyMontana]

Hey there, how do i access your long blog post on setting up a small assembly line...? what's the posts' name ?

[PCBprototyping]

Thanks for the suggestions but I think we are deviating from the topic, which is: between Neoden 4, Neoden YY1 and P20 (Mechatronic Systems) which is the best option for prototyping purposes?

It is clear that being low cost machine (it’s a requirement) I will necessarily deal with problems, but that is fine: no one ever bought a Ferrari for the price of a Panda.

And of course we analysed our situation carefully and it makes totally sense (time wise) to have one of those machines, the alternative would be me spending at least one day per board with the -very likely- risk of making mistakes (swaps of resistors and so on).

Given that, I discarded Neoden YY1 thanks to Jackster contribution. Neoden 4 and P20 are still possible options. I was able to find extra info regarding Neoden 4 but I wonder if any of you has experience whit P20.

You can check www.tronstolsmt.eu, Tronstol E1 is for prototyping and Tronstol E4 is for low volume production, both have the same recognition, precision and UI. Tronstol's UI is very intuitive and well designed, you can check the programming sequence and UI here: https://www.youtube.com/watch?v=8B8PtCaDMpk&t. It can place 01005 to 35x35 mm components, has laser flying vision and is extremelly precise. E1s drag feeders are very precise if you set them up the right way(center of hole first, then one pull to self adjust the tape, then center of the pocket), they are dead on center for 0201 components. Compared to Neoden 4, it is faster and more precise, but without conveyor and removable feeders(which take a long time to exchange at N4, E1 uses removable feeder banks with automatic calibration). Nozzle exchanger is a plus for E1 also.
P20 has a very intuitive SW for prototyping, but very bad customer support, we used to represent them and had to cancel the distribution due to delayed or missing deliveries and almost no support. It is much more expensive than N4 and E1, adequatelly loaded machine with smart feedes(33 locations, almost half the amount of N4 and E1) will cost you at least 25k. It does not need any component library, you just import the P&P file, set up the fiducials and it will automatically check the component and PCB height and start placing. Recognition of special components is more tricky as you need to set up recognition profiles in IDS imaging SW not in P20s SW. Machine has 1 head, which makes it quite slow.
We found Tronstol the best quality of Chinese machines, which can compare to European standards in terms of precision, build quality and UI.
We have experience with all three manufacturers, so feel free to ask, if you need more info on any of them.

[jmelson]

Well, once you have all your parts in a "part library" so the vision system know the right measurements, then the programming time should not be very high.  I wrote a converter program that reads the P&P file from my CAD system (Protel 99SE) and writes out the file to import to the P&P machine.  I then have to select all the required parts for the build from the machine's part library, assign feeders (the machine can do that, but I prefer to arrange that manually) then run the optimizer that figures out the order to place parts for best efficiency, selects parts from the arranged feeders and then it is ready to go.  I suspect the process flow is about the same on most newer machines.  I usually expect to take at LEAST an hour setting up a new board of low compelxity, and maybe 2+ hours for really complex boards with many unique components.  I generally make 25 - 50 boards at a time.
Jon

[jmelson]

And why are you limitting yourself to those machine? There's plenty to choose from if you have USD30K. Read my long blog post on getting a small assembly line (I won't post here), it might give you a better idea. My budget was much smaller than USD30K!

Yes, for US $30K, you should look at a relatively recent USED machine.  Since you are in the Netherlands, maybe Assembleon machines might be plentiful there.  (I used to use a Philips machine here.)  I am now using a Quad QSA30A (the base machine is a Samsung CP30 with Quad electronic feeders and Quad component alignment cameras.)  Maybe a Samsung CP 40 or CP45 machine.  These high-end production machines are quite remarkable, and generally built like tanks.  Spare parts are still widely available (mostly from the far East) at reasonable cost.
I started running a well-used machine in 2007, and am now on my 2nd machine.
Jon

[PLT]

For info don't buy any P&P from mechatronics , it will answer you very well, send the last good 2019 software, ask you money then he will have permanent problem and find always excuse to not deliver the machine, missing components, worker problem, covid, holidays ....

it's because his company is registered on his name in Germany, the factory is in Poland
he his in personal bankrupt since 2019 when in December he have lose his working team and engineer in Poland
since he try to get new worker but he can't get the raw material since he don't pay them

perhaps its possible to have a serigraphy machine but forget the P&P if you are not a reseller or a German users