NeoDen YY1 Pick And Place Machine With Under $3K Price for Hobbiest/Low vol Usag

[Styno]

Unfortunately, I received word from Neoden that they will not be updating the software to utilize two fiducials as opposed to the single fiducial that it currently uses.  In its current state, it uses the routed edge of the board as an alignment point, which, as you probably know, is not a very accurate means for alignment0.  The reasoning for not upgrading the software according to the person I spoke with is that the machine is designed as a low-cost machine and therefore they will not be making changes. I assume this will mean other issues with a machine will go unresolved.   

This smells a lot like intentionally crippling the machine to protect the more capable (and profitable) machines. Proper alignment of the board using 2 fiducials should not cost more to implement than using one fiducial and one board edge. A typical bad marketing overruling best technical solution example.

I would not buy a machine that is so obviously intentionally crippled like this.

[Styno]

Some things I've noticed:

Thick components/plastic tape is a bit annoying, but I should have read the manual better in the first place. Solved the issue by loosening the screw in the feeder connected to the metal tension strip and swapping peeler friction "damping ring" for the "1.5N" insert. You can test if you have the right tension by using the manual test on the peeler.

The needle for advancing tape is quite aggressive, if it's even slightly off center you'll have parts flying everywhere. I also adjusted it a bit higher (it has slots) because even when aligned it was knocking parts around.

Not so much a response to 'newto' but just a trigger to get started.

I partly operate a low-end but professional 25 year old p&p machine for about 3 years now and if I learned anything over that period is that feeders, especially the feeder design, are easily the most critical aspect for a productive and pleasant operational experience.

Getting the X/Y/Z movement and accuracy required is relatively easy, go slow enough and even I can make that work. 0402 and 0.5mm pitch QFN's is just easy with vision these days. Every software has it's quirks but has to be really bad not to be useful, and easy to improve over time. But if the design of the feeder is wrong you'll end up regretting buying the machine (if you're honest), because you end up continuously tinkering with settings, fixing all sorts of things all day like mis-picks, issues with tape advancing, cover tape peeling and bouncing components even if the feeder worked well just before. Really, feeder issues due to inherent design flaws can (and certainly will!) significantly reduce productivity and your mental health. Pay a bit more and get a machine with a good feeder design. I know because you will hear me swearing every time over the push-feeder design of the Dima Optimat.

By good design I mean that it at least has an individual cogwheel for advancing the tape for each feeder (electric of pneumatic activated) and individual cover tape spool tensioning that is well regulated. Drag- or pushfeeders and feederbanks are an inherently bad design choice in my opinion.

[newto]

I'm under no illusions that this is a particularly good or robust machine. I think it fits very well with our workflow, but would be miserable for the vast majority of production. Right now we have only 1 board that actually needs dedicated assembly (a few daughter boards that are simple enough to be done by hand in a few minutes, but it was worth bundling with the order for the main boards). I'm pretty sure the shop didn't even set their machines up, and just did them by hand (which is why it cost us so much per board). I expect the first few boards to be a huge pain with a lot of manual re-work, but if we get to less than 10 minutes of manual work doing setup, machine babysitting, re-work, and the hand placed components, that machine pays for itself in 6 months. And even if were to manage only 1-2 boards a day we beat the turnaround of the shop.

If you had more than 2-3 designs (or had to swap components in the feeders often), had to build  more than a couple dozen boards a day, or your boards were using mostly 0201 or 0402 parts, I would stay well away from this machine

For us, we only do 50-100 boards a year of one design, and it's all 1206 (although I plan to redesign it to 0603 and add a second product with as many shared components as possible in the future). It's going to sit idle in the corner for most of the year, pre-loaded with components, and then whenever we do a new batch of boards we'll double check the feeder settings and placement, and then do 15-20 boards.

[bugrobotics]

I also inquired about a machine and asked specifically about the software bugs and update process.  I was told that to update the software you would need to buy a new control board and touchscreen. 

[mikeselectricstuff]

I partly operate a low-end but professional 25 year old p&p machine for about 3 years now and if I learned anything over that period is that feeders, especially the feeder design, are easily the most critical aspect for a productive and pleasant operational experience.

This.
1000% this.

[newto]

I also inquired about a machine and asked specifically about the software bugs and update process.  I was told that to update the software you would need to buy a new control board and touchscreen. 

Ooof, I hope that's not true. Building a machine without software updates in this day and age is ridiculous

[Fire Doger]

I also inquired about a machine and asked specifically about the software bugs and update process.  I was told that to update the software you would need to buy a new control board and touchscreen. 

I got the same response. That's a deal breaker for me. The only reason to buy it is to convert it to openpnp machine....

I'm pretty sure the shop didn't even set their machines up, and just did them by hand (which is why it cost us so much per board).

This doesn't make sense. Every shop will do what is cheaper for them. Making a program for a machine, teaching the parts, the feeders, etc is time consuming for the first batch.
If it takes me 1 day to make a program you gonna pay 1 day of engineer work plus machine time plus company profit.
If someone do it by hand in a couple of hours you will pay for the couple of hours of a simple worker (simple workers are not paid as high as engineer afaik) plus company profit.

JLC has cheap assembly foe low qty because they cut down the tooling time and preparation.
You use their parts so teaching is already done and parts are already mounted on the machines.

[drgerber]

Hi,

we bought a Neoden4 when they were still pretty new and now also jumped on a YY1 as a backup machine.
In-house we are mainly doing prototyping and small batch assembly for our embedded processor systems. Mostly 0402 passive components - sometimes 0201.

YY1 arrived quickly and setup was pretty easy. Some learnings from the ND4 helped for sure and kept the frustration level low.
Machine needed to be recalibrated for reliable 0402 placement but the on-screen guide made that easy.
Feeders are FDM 3D printed. They are more reliable than they are looking but tweaking the friction for the peelers is not that much fun.

So after adjusting everything the first few boards were pretty good for what it is. Have the feeling that the feeder settings could still use some tweaking but picking error is low enough for now. Also quickly drafted a single 8mm feeder and printed it with our SLA printer. Came out good and might use it once the next board with 0201 parts comes - so far only 0402 in the YY1 here.

Was a little surprised reading that they say the software is not updatable. Nevertheless it looks like they did a good job. Vision works well and I don't see a problem having only one fiducial for that type of machine. I think once you know the weaknesses you can find your way around them.

Don't see a reason you would really need a software update. For a low-budget machine it does its job pretty well. Of course it can't compete with larger and more expensive machines - no software update could fix that anyway.
If you are looking for shorter feeder setup time and higher accuracy you have to invest quite a bit more. For prototyping the YY1 is the right choice and I'm sure we will have some fun with it.

[RadioActivePilot]

I recently received my YY1.  I was looking for a low cost option that would get the job done. I thought about openpnp but I didn't have the time to learn, program, build if necessary etc.  For the price, it will pay for itself in a short time. I build about 200 boards per year.  Mostly simple boards, but 1 has it's share of "fun".

Build process for the YY1 took about an hour.  I took pictures along the way.  Raise the frame height, remove packaging, add panels, and insert the SD card. Nothing too exciting. 


I had two boards I needed to get done.  One had 8 components, the other 9.  Instead of loading data from the cad file, I manually programmed the machine.

I did experience a random "jam the head into the side" event.  Required a reboot.  The reboot button received a lot of use.  One thing that is an issue for me is swapping nozzles. Haven't been successful.  It gives me a Z axis error every time after swap.  Reboot required.  I'll look into this more. For now I used a passive component, small nozzle and a large one for ICs.  No swaps. 

Initial thought was, man... This is going to be a pain, as nothing was placing properly.  Simple 1206 caps were significantly off.

And the SOIC parts, like basic parts, we're off by half a pin. Eventually, I found the password for parameters, and calibrated the head/cameras.  That, in conjunction with slowing down placement, and I was a happy camper.  Tuning the cameras was not hard but took some work to dial it in. 

Anyway, back to the first board... All this was by using the vibration feeder and the flexible tape feeder, no reel/peel feeders yet.   I actually treated the bulk feeder trays as a flexible feeder. I just set the component in the lower right hand corner of each square, set the location of the first one and the pitch and it picks them up as if it were a flexible feeder even though it was the bulk trays. It's a very flexible setup.

I was able to place 20 boards in no time at all.  Vibration feeder is nice.  Would be nicer if there was room for more tubes.  But whatever.

The flexible feeder strips I used were great.  Set up the pick for the first location, pitch, number of parts, and it cycles through the strip, and back to the first position.  After failing to pick, it pauses and gives the option for you to skip that particular component.  I wish there were more of these strip feeders, of which you can place anywhere, in any orientation.  In fact, I am 3D printing some now.  More in that later.

Bulk parts... Yeah, great in concept, but getting the machine to recognize how many are in a tray and picking one somewhat near the center doesn't work very well.  I did think about reflections and making the base darker or more matte as someone mentioned but I have not tried that yet. Most of the time it will pick up a component but it will definitely place maybe one out of 10 or two out of 10 parts in the wrong orientation. The flexible strip feeders don't have this issue because you set the orientation of the component in which it is picked up. At least I think that's how it works and it will continue to work, as I had no orientation issues picking up from the flexible feeder. Also the camera field of view is just such that it is possible for it to pick up a component from the next bulk feeding position over. So, it is critical that you don't put components near the edge of the bulk feeder box that the camera can see Other than that, I let it place some bulk parts and I just picked them up and corrected them upon inspection if it positioned it 90° from normal.

I eventually added a tape and reel in position one. It is easy to set the location of the needle and the part pick location. I set the needle as far back as possible, as to not cause vibration and component to pop out of the tape.  Would be nice to have this motor driven, not a solenoid so you could control the speed.  Brutal if it is off!  This was a 3 mm square QFN. It placed it in the exact position needed, and also on all the boards in my panel (5x2) I did not have any issues with the peeler mechanism.  I plan on loading a good majority of the feeders so we will see how that works with different reel material.

Running a panel of boards worked for the most part. There was an issue with the third board over, for some reason it thought it was off in space somewhere. It couldn't find the fiducial. Eventually I did a reboot and started the job over and it found it. You can select the particular board you want in a panel and the component that you start with by selecting the component. So, if it placed half the board you can start where it left off after a reboot.

Because I had such luck treating the bulk feeder positions as a flexible feeder, in my 3D print model, I modeled several arrays of cutouts, where I can put my large ICs.  Actually, I have a real TQFN IC tray that I am going to mount in my 3D print. The TQFN tray is actually fairly sophisticated. So it's easier just to make a place for a section of the TQFN tray (it holds a lot of parts, so I'll cut it to fit).  Also included in my 3D print will be an additional 10, 8mm strip feeder locations. These will be used on boards that may get only one placement. In my particular setup, I can fit 25 components in each strip before refilling. So I can do 25 boards and then refill those feeders.  I also created indentations for 0603, 0805, 1206 packages. Instead of using the buggy bulk component option, I can just load resistors and capacitors into these indentations and treat it like flexible feeder.

Before ordering the machine, I asked about firmware upgrades. He did mention that it would be a board swap and they would walk me through it should that ever need to take place. I wasn't told that I would have to buy a new control board, but who knows.

Also, someone mentioned losing components just to feed the tape peeler. For $7 you can order a mouse reel from Mouser... It's cut tape but on a reel with a leader. You have to watch the pricing though.  Sometimes it's cost of beneficial just to order a reel of 5,000 resistors for 20 bucks instead of 1,000 resistors for 10 bucks plus a $7 adder for the reel fee.  Or if your component count is low just use the flexible feeders and feed cut tape in there. Again, build your own cut tape flexible feeders, 3D printed. When my print is done, I will run some tests to see how well it printed and how well the machine picks up the parts. I might make the CAD models available. 

Overall, I'm satisfied so far.  My board jobs have been piling up and am happy to knock them out.

Next step is to import CAD data and run a job instead of manually programming.

As I work with the machine, I will be happy to post about progress. I have even taken a couple videos I might find someplace to post them. If anybody has any questions or wants me to try anything let me know!

[ali_asadzadeh]

Thanks for sharing, It would be helpfull if you post some videos on youtube about programing the device also about the issues that you have and if you could have solve them or not.

[mikeselectricstuff]

Before ordering the machine, I asked about firmware upgrades. He did mention that it would be a board swap and they would walk me through it should that ever need to take place. I wasn't told that I would have to buy a new control board, but who knows.

That is just ridiculous, and indicates that either they are clueless or have just bought in the board/firmware from someone else & have no control over it.
There is simply no excuse for such a poor design decision these days.


 

[MakeIt]

@RadioActivePilot

Strange, you can place a QFN 3x3mm with only 1 fiducial. Was it 0.4mm pitch.
Can you share a video of the QFN being placed?

[RadioActivePilot]

@RadioActivePilot

Strange, you can place a QFN 3x3mm with only 1 fiducial. Was it 0.4mm pitch.
Can you share a video of the QFN being placed?

I didn't do a close up video, but will make a special video today or tomorrow.  I won't place the whole board, just that part, sans paste.  I'll even swap out boards, as a trial for two unique boards, and place another.

It was MAX14588. 0.5.

What's the best way to share video?  Add a YouTube channel? 

[Jackster]

I am so glad I did not go with Neoden for my PnP machine.
Going off this topic and how a few of their users have reported issues and have had 0 help, really goes to show how crappy NeoDen the company is.

Replace the control board for basic software upgrades? Nah this is a joke of a machine...

Buyer beware.

[liviux]

Well, they do have a "part" as the first fiducial and the second point as a fiducial sounds weird indeed. Many said they imported the first board fiducial as a part and made that first part in the list. But still why did they choose to do it this way? Plus most of the time you need 3 fiducials diagonal and horizontal to be extra careful you are aligned well with each board you place in the machine. Waiting for it myself as I think it might be better than the Boarditto on which I had loads of issues with 0402(most of the sizes on my boards) But after days of tinkering and babysitting I could place around 290 total components on both sides in a roughly 38x38mm board so quite a dense layout. But loads and loads of manual corrections. In the end, did around 40 boards with it and I genuinely hate it still losing 2 hours easily for 4 boards on one side. Hope that yy1 is better for prototyping stuff. I could go for higher quality even from Neoden but for now, space is a luxury I don't have...

[newto]

I'm guessing what they've done is made the entire software as flashed onto an IC that doesn't have the option of doing firmware updates from removable memory (or they didn't set it up properly to allow it). If this wasn't a machine paid for by my work, I'd open it up and have a look for a UART port somewhere.

I actually whipped up a set of parametric bulk and cut tape feeders with openscad, so that I could make some with bigger boxes for some of our bigger lose components (not super recommended, the upward camera can't see them even in big IC mode, so I have to rely on putting the parts in the exact corner of the box). The strip feeders also allow for wider tapes and variable lengths. Also for variable sizes of neodymium magnets. I did screw up the pins because I forgot that the bottom sets were so close together, and the pair on the right were so close to the metal board holder.

Unfortunately I technically designed it on company time, so I'll need to get permission from my boss to post them online, and I'm sure someone else could come up with a much better design.

I'm still reasonably happy with the purchase (wasn't my money), but definitely disappointed  in the lack of software updates, found another minor issue, when adjusting the strip feeder type (can't remember the name) the move button is extremely slow compared to when in other modes, so if you want to move it from below the board to above, it takes forever. The workaround I found is you can interrupt the auto move to the default location by pressing an arrow buttonf, so at least you don't have to go all the way back across the board, but it's still super slow.

I wonder how much of a market an openpnp drop in board would have...

[Smartbeedesigns]

Got my neoden yy1 in yesterday.

One of the metal brackets that holds one of the stepper motors was bent in shipping. Neoden is sending a replacement bracket thankfully.

I was able to rig it up so I could still move the gantry around to test some stuff in the mean time.

One issue I've come across is that if I set the strip feeder to advance in any increment other than 4mm, it still will only incriminate the tape 4mm. So for instance I've got reels of buttons and connectors that are on 8mm and 16mm spacing and the tape feeder only increments them 4mm at a time which is a problem. I don't know if this is only happening during the pick test and would otherwise work as intended during an actual job run. But since I've got the bent bracket issue I can't test it that fully yet. I sent neoden a video of the issue, so hopefully I hear back from them soon.

[newto]

One issue I've come across is that if I set the strip feeder to advance in any increment other than 4mm, it still will only incriminate the tape 4mm. So for instance I've got reels of buttons and connectors that are on 8mm and 16mm spacing and the tape feeder only increments them 4mm at a time which is a problem. I don't know if this is only happening during the pick test and would otherwise works as intended during an actual job run. But since I've got the bent bracket issue I can test it that fully yet. I sent neoden a video of the issue, so hopefully I hear back from them soon.

It's only for the test mode, seems to work properly for actual picking. Why they don't use the spacing you enter literally on the same page, I have no idea

[Smartbeedesigns]

One issue I've come across is that if I set the strip feeder to advance in any increment other than 4mm, it still will only incriminate the tape 4mm. So for instance I've got reels of buttons and connectors that are on 8mm and 16mm spacing and the tape feeder only increments them 4mm at a time which is a problem. I don't know if this is only happening during the pick test and would otherwise works as intended during an actual job run. But since I've got the bent bracket issue I can test it that fully yet. I sent neoden a video of the issue, so hopefully I hear back from them soon.

It's only for the test mode, seems to work properly for actual picking. Why they don't use the spacing you enter literally on the same page, I have no idea

That's good to know! I was pulling my hair out thinking something was wrong. But ya, that's a strange way to go about it. "Hey do a pick test to make sure things are correct.....oh btw we won't be using the pitch you entered. Youre just going to have to Yolo it and do a job run...hope it's right!" Lol

[jmelson]

I partly operate a low-end but professional 25 year old p&p machine for about 3 years now and if I learned anything over that period is that feeders, especially the feeder design, are easily the most critical aspect for a productive and pleasant operational experience.

Yup!  I ran a 20+year old Philips/Yamaha P&P for 13 years.  It was VERY old-school, with mechanical jaws to center parts on the nozzles.  That was not accurate for the work I was now doing.  Now, I have a 20-year old Quad/Samsung machine with flying vision and auto nozzle changer.  The Philips/Yamaha feeders were advanced by a plunger on the head, and springs peeled the cover tape and advanced the component tape when the plunger retracted.  This ended up not giving suffcient pull on the cover tape.  A CONSTANT issue!  I had to watch the machine like a hawk and pull on the tails of the cover tape to keep them from failing to advance.
The Quad machine has electronic feeders that have DC motors to advance the tape and a rubber roller to pull the cover tape.  The rollers get soft and have to be replaced every decade or so, but they pull the cover tape WAY better than the Yamaha feeders.  I have VASTLY fewer feeder issues on this machine as well as much more accurate part placement.

So, yes, the feeders are KEY!!!  I am super skeptical about those pin feeders, yes simple, and works OK for just a couple boards, but I do small production runs of 25 -50 boards at a time.  I'm getting at least 3000 - 4000 PPH on this machine.
Jon

[jmelson]

Running a panel of boards worked for the most part. There was an issue with the third board over, for some reason it thought it was off in space somewhere. It couldn't find the fiducial. Eventually I did a reboot and started the job over and it found it.

I do boards with HASL, and sometimes the fiducials have specular or frosty finishes, and it confuses the image recognition.  On boards where this becomes an issue, I go over the fiducials with an ink eraser to make it all matte.  then, the recognition works 100%.  Best to decide in advance and do this before applying solder paste.
Jon

[mikeselectricstuff]

Running a panel of boards worked for the most part. There was an issue with the third board over, for some reason it thought it was off in space somewhere. It couldn't find the fiducial. Eventually I did a reboot and started the job over and it found it.

I do boards with HASL, and sometimes the fiducials have specular or frosty finishes, and it confuses the image recognition.  On boards where this becomes an issue, I go over the fiducials with an ink eraser to make it all matte.  then, the recognition works 100%.  Best to decide in advance and do this before applying solder paste.
Jon

Is there an option to manually do the fids,i.e. looking at the camera view and adjusting by eye ?
As well as for vision issues, This can be handy of you forget to put fids on a board & have to use another hole or pad

[Styno]

The Philips/Yamaha feeders were advanced by a plunger on the head, and springs peeled the cover tape and advanced the component tape when the plunger retracted.  This ended up not giving suffcient pull on the cover tape.  A CONSTANT issue!  I had to watch the machine like a hawk and pull on the tails of the cover tape to keep them from failing to advance.

This is a perfect description of handling the Dima mechanical feeders as well. I even modded some of the feeders to add much more spring pull on the cover tape.

The Quad machine has electronic feeders that have DC motors to advance the tape and a rubber roller to pull the cover tape.  The rollers get soft and have to be replaced every decade or so, but they pull the cover tape WAY better than the Yamaha feeders.  I have VASTLY fewer feeder issues on this machine as well as much more accurate part placement.

Electric feeders sound like pure joy to me, especially if they are 'smart' and talk to the machine so it automatically knows which component is where and how much there is left on the reel. I want to buy a second hand Dima MP-200 that has this feature. It also has a high feeder count (on four sides), compact size, is quite cheap for a serious pick & place and has respectable cph. But management doesn't want a new adventure with an old machine and mfg that left the business.

I do boards with HASL, and sometimes the fiducials have specular or frosty finishes, and it confuses the image recognition.  On boards where this becomes an issue, I go over the fiducials with an ink eraser to make it all matte.  then, the recognition works 100%.  Best to decide in advance and do this before applying solder paste.

I also find that HASL, especially when left exposed, will develop oxidation or another way of 'frosting' and the fiducials won't reflect well in IR so the crappy vision cannot accurately recognise the fiducial. I use a block with fine sandpaper to make the panel fiducials shiny again and try a gum next time, it sounds less aggressive than sandpaper if it works:)

Is there an option to manually do the fids,i.e. looking at the camera view and adjusting by eye ?
As well as for vision issues, This can be handy of you forget to put fids on a board & have to use another hole or pad

On the Dima, you can disable fiducials entirely but can also teach another board feature (pad, hole or via) as a fiducial as long as it's recognisable enough ('uniqueness factor' in Dima lingo).

[mikeselectricstuff]

]I also find that HASL, especially when left exposed, will develop oxidation or another way of 'frosting' and the fiducials won't reflect well in IR so the crappy vision cannot accurately recognise the fiducial. I use a block with fine sandpaper to make the panel fiducials shiny again and try a gum next time, it sounds less aggressive than sandpaper if it works:)

Maybe try desolder braid ?

[jmelson]

Running a panel of boards worked for the most part. There was an issue with the third board over, for some reason it thought it was off in space somewhere. It couldn't find the fiducial. Eventually I did a reboot and started the job over and it found it.

I do boards with HASL, and sometimes the fiducials have specular or frosty finishes, and it confuses the image recognition.  On boards where this becomes an issue, I go over the fiducials with an ink eraser to make it all matte.  then, the recognition works 100%.  Best to decide in advance and do this before applying solder paste.
Jon

Is there an option to manually do the fids,i.e. looking at the camera view and adjusting by eye ?
As well as for vision issues, This can be handy of you forget to put fids on a board & have to use another hole or pad

My Quad machine does not have this option during running, but it does have a very good diagnostic mode that allows you to test the quality of the fiducial recognition while setting up a run.  If fiducial recognition fails during a run, it rejects the board and I can clean the fiducial mark and send it through again.  It also has the ability to use a through hole as the fiducial.  You just set the fiducial to "black" instead of "white".
Jon