Essemtec Fox - thoughts?

[Ice-Tea]

That's just porn.

[SMTech]

I hadn't been paying close attention to how much Essemtec had been tweaking on their machines as we are happy with what we have and would probably be choosing a different kind of machine next time. However the changes are significant, up until the release of the Paraquda sometime around 2010 Essemtecs top end machine was the FLX and below that were Pantera (I think based on the FLX chassis) and then the CLM machines that resembled what they started with (extruded aluminum frame) and they had dabbled at least once with rebadging someone elses machine (Versatec C5). Those who had mid-generation FLX machines that were rated down to 0201 said they did do a little tweaking when placing 0402.
At the same time as the Paraquda they had the Cobra, a machine that vanished quite quickly that was basically a bigger Paraquda with 8 nozzles and carbon fibre employed to counteract the increased weight of the head. I'm merely speculating but think this machine vanished because it was expensive while also lacking features you would find in machines at that price point e.g placement force measurement & outright speed.

Since then they have retired all the lower end machines (I think it goes like this), made the Fox (essentially an improved FLX with the Paraquda software), made a Paraquda gen2 and then retired that, made the Fox gen2 which introduces a 4 nozzle head and linear drives and introduced the Puma which seems to be slightly bigger than a Paraquda was, again adding linear drives. The resulting machines have to be more expensive than their non-linear motor counterparts, however the heads and nozzles look like they are perhaps the same basic design. The IPC speeds in the datasheets are around 50% higher than the old machines, because of that it may even be the case they only offer this generation with HyQ feeders, this is fair enough there are huge compromises to be made when using feeders in banks anyway.

I think it would be fair to say that outside of the mergers that have gone on elsewhere nobody else has made such radical changes to their P&P lineup, it is clearly motivated by something (new machines encroaching on their lower end offerings, profit margin on something built in Switzerland???). The only similarly big change I am aware of is Europlacers new machine the Atom which introduces and a new head/gantry to their platform that previously focused solely on flexibility adding a significant increase in throughput.  Certainly however it makes it very clear these machines are not rivals for TWS, DDM or Autotronik, instead these are rivals for volume production machines with a focus on unconventional setups (they still offer standalone) that those machines are bad at. However I would maintain that for most typical designs beyond ~6kcph is completely pointless unless running a fully automated line, the human will end up rate limiting it.

[Smallsmt]

I saw the FOX on exhibition a really cool machine.
The userinterface looks modern and machine has touch screen control.
As i know they use linear motors and they are really fast.
The feeder system uses feeder groups to place more feeder on a small space.
Machine foot print is small for the high feeder count.
I think it can get moved to a normal office through the doors.
But the price is about 130000 euro in working condition including feeder.
Too high for my usage.

It is a very good solution if the machine is used for production and on small space.

[SMTech]

Don't get too impressed by the attractive outer shell and software UI. The interface is a bog standard interface made bigger to be finger pokey friendly with a popup keyboard. They've done a nicer job than some and the whole package certainly looks the part if you have guests but its not perfect.

[mrpackethead]

I have a line that has an old school DOs inteface.  Its ugly.   but the machine runs relaibly and does lots of work. Do i need a touch screen?  Seriously a PNP line is not about what it looks like.

[mikeselectricstuff]

Mid-high end P&Ps are designed for operation by trained, semi-skilled operators who probably have little knowledge of what they are assembling. This poses different UI requirements to a skilled in-house user assembling their own PCBs.

In terms of flashy UIs, from my system I would say that having a graphical representation of the board can be very useful, making it easy to identify where any problem/missed parts are, omitting parts of a panel etc.

[Kjelt]

A good UI can prevent operator mistakes.
It will pay itself back in no-time.

[SMTech]

Pretty much anything vaguely recent should at least be intuitive enough to prevent mistakes, some machines will have been built with a mindset that an engineer level person creates the "Recipe" (as Essemtec like to call it) for a product, creates all the components and packages, orientation in tape etc. And a low level operator only has the power to stop, start and load. At that point the machine UI might not offer many opportunities for error to the operator. The steps and tools for the engineer level stuff might well be a separate  package on another machine, you can even have dedicated stations for checking the vision parameters for some machines..

ePlace  gives a graphical representation of the board as configured, that lets you visually check things like orientation using a single piece of software for all tasks

Errors a programmer/engineer might make:

Component orientation, not every manufacturer puts things in the same tape for the same package or the same way round, nor do they make it easy to find out. Visual representation of the component on the board and the tape helps with verifying setup. In Essemtecs case you even get a representation of the feeders on-screen including components in the tapes, multiple pitch and tape types can be stored against a single part number. This means for instance you can have two suppliers of a 2512 resistor one using 4mm and the other 8mm pitch (happens) and the machine will ask prompt you as to which it is when you load the part. It also means for new jobs and new parts you can easily verify what you are loading matches what the machine expects to see. Different CAD packages stick to IPC rules about where 0 degrees to varying levels, Essemtecs software builds up rules to automatically correct these, as a contractor I build up multiple rule files accordingly. 

Bunkum co-ordinates: I don't know how but I have been sent XY files where some components are in completely the wrong place, easy to spot if it is represented on screen.

Incorrect size data: I'll give it its due, the component design section of ePlace really isn't bad and it can have a pretty decent stab at making models for you based on what the camera can see (sometimes comical). Graphical aids here are invaluable. However unless they have snuck placement force measurement into the Fox2, defining component height correctly is critical to both accurate placement and nozzle life. By contrast machines with placement force feedback can compensate for incorrect height and/or board warpage on the fly.

Errors that an operator might make:

Putting in the PCB rotated 180 - this is very easy to do, a visual aid in the GUI can help here but the foolproof solution is not to place fiducials symmetrically, everyone makes stupid mistakes sometimes.
Loading the wrong component in a feeder, this is very easy to do if you don't have checks in place. Essemtec Feeders are "Intelligent" all this really means is each one has an electronic ID and the machine keeps a record of what is loaded on it. However compared to a dumb feeder that has many advantages, meaning you can move it and the machine knows you have done so, it also knows things like the lane type, pitch & pick height. I don't think UI makes much difference here, if you are changing loaded parts a lot the best tools you could add would be a component verification unit & barcoding so you scan parts on and off, both of these are offered by Essemtec but cost extra (a surprisingly rude amount in the case of barcodes).
Loading the wrong file, I don't see the UI helps much here, however unless what you have loaded is a variant it probably shouldn't build. Mitigating tools here would be a job(works order) tracking system and some barcoding, if your building in any volume at all you should at least have works orders and paper tracking. If you don't cross your fingers none of your customers are the type who might audit you.

Dunno that an operator can do much else wrong, mostly they just need a smidge of common sense to respond to errors & prompts sensibly.

[Reckless]

Is it possible to run this all night with no operator present?  How long does it take for solder paste to run out?

By having graphical representation of the pcb does it act do any AOI/SPI functions?

[mrpackethead]

Fully autonoumous PCB Production?     Even on the very high end machines, they have stoppages that require human intervention from time to time.    One operator can look after several lines, but totally human free.. I doubt it.

Fully automated solder pasting happens as well, so this means that the solder paste is stenciled as its required, rather than stackign up boards.

[rx8pilot]

I sure like the compact design of this one. The pricing seems reasonable for what it does.

For me, the big questions are how long does it take to do a changeover. How easy are the feeders to unload/load. Are there any alignment tweaks needed after loading a new part? How easy is it to add new parts to the library and verify them? How gracefull does it handle mis-picks? Does it auto correct pickup locations based on the previous pickup?

For me, placement speed is not interesting. Ease of use, reliability,  and setup speed is king in my world.

[Reckless]

I sure like the compact design of this one. The pricing seems reasonable for what it does.

For me, the big questions are how long does it take to do a changeover. How easy are the feeders to unload/load. Are there any alignment tweaks needed after loading a new part? How easy is it to add new parts to the library and verify them? How gracefull does it handle mis-picks? Does it auto correct pickup locations based on the previous pickup?

For me, placement speed is not interesting. Ease of use, reliability,  and setup speed is king in my world.

From what I saw at the show:
Changeovers very fast
Feeders easy to load/unload
New parts easy to add
Mispicks was a question mark as we saw a number of chips scattered around head due to a wrong nozzle being used
It does auto-correct, it has some AI

Ease of use, setup, reliability are good but im still torn.  The machine was not stable and seemed to move at 100% speed on the show carpet.  Personally I am used to super planted and overbuilt universal genesis machines.  I moved away from Juki as I didn't like the instable nature of their machines at full speed.  I don't like machines rocking but universal makes their machines soo big that the arm has to travel more distance and have to run faster to compensate.  The fox being soo small takes very little production space which is something I do love.

[SMTech]

You can get close to an autonomous factory (machines requesting new reels form storage, robots bringing them to the load station, SPI& AOI rejecting pcbs to a divert magazine etc) but there is no robot to load a feeder with a new reel, or change a feeder or one with enough common sense to sort out those little issues that will crop up from time to time. While these factories exist in the promo vids for SMART factories I don't know who has them, clearly not china, they have heavily staffed SMT lines if YouTube tours are anything to go by. Maybe high end manufacturers like Miele ? This isn't something you'd do with a FOX tho', you'd just buy something faster.

As to the Fox, while admittedly the Paraquda I have is now an older platform it is where the ePlace software comes from and while I'm not on the very latest version, it won't be all that different. It is quite nice to use, there are a few moderately annoying glitches that crop up when moving between screens or selections but nothing that stop the machine carrying out its function unlike some of this which seems unacceptable to me https://smtnet.com/Forums/index.cfm?fuseaction=view_thread&CFApp=1&Thread_ID=21056&mc=6 . Setting up new jobs in indeed super easy, and gets easier & quicker the more you do. As of yet I don't think you can directly import ODB++ which could be a nice feature, nor can you display the assembly layer (or any gerber) as part of the image, something else that could potentially be useful.

I don't know what they would mean by auto-correct, that sounds like what every machine does where it alters the pick position based on the alignment which is nothing special for a machine with optical alignment.
It is compact yes but that does come at a cost, which may or may not affect you, it only handles smaller boards and if you add tray changers to Essemtec machines, the max PCB size shrinks again.
Changeovers: I don't think you could really say these were any quicker on Essemtec than any other modeern machine with intelligent feeders that are easy to load. If you want the fastest that accolade probably goes to Mycronic with their Agilis system - a key selling point of their platform, which is anything but compact. Equally look at Samsung feeders where you can stick the next tape in the feeder while the old one is still in use when working in high volume and it will switch over on its own. However truly rapid changeover is achieved by having enough feeders to load the next lot, on trolleys, while the current job is running or just to have so many permanently on the machine that most of it is already loaded (Europlacer is the king here).
Misspicks: It seems very odd to me that there would be a visible number of picking issues on a demo line at a show, optical alignment does allow you to set all sorts of parameters to accept or reject a part at which point to either return it to its tray or drop it in the reject bin. If your parameters are out or you have the wrong nozzle, sure this can raise the number of them but there's not much excuse on a demo line that was probably placing generic parts on a demo board they run all the time or drawing a twee logo with parts. You probably saw it with HyQ feeders while I run  CLM so I can't comment on those but if they still try and sell the old style deep pocket feeder (which is the only option for anything over 6mm height with CLM feders), they are to be avoided at all costs, utter garbage.
In my experience my highest pick failure is a total failure to pick at all, this typically crops up on one specific lane for no apparent reason and will randomly go awayon another job or a software restart, its not common but enough that its an issue I recognise.

In a Lab with no conveyor it has an unrivalled max feeder count in a tiny footprint, and this is where perhaps the dispensing options make sense (and they are better than anyone elses) however dispensing paste makes no economic sense to most people. As soon as you move to the production environment it has some pretty strong competition on price from machines that will place faster and have fewer restrictions and more features. I like our Paraquda, it is excellent at what it does, the way we use it, but if we suddenly needed a line that handled higher volume, automated end to end, I personally wouldn't even look at Essemtec as I know I can get more for less (and cheaper to run) elsewhere.

[Reckless]

In a Lab with no conveyor it has an unrivalled max feeder count in a tiny footprint, and this is where perhaps the dispensing options make sense (and they are better than anyone elses) however dispensing paste makes no economic sense to most people. As soon as you move to the production environment it has some pretty strong competition on price from machines that will place faster and have fewer restrictions and more features. I like our Paraquda, it is excellent at what it does, the way we use it, but if we suddenly needed a line that handled higher volume, automated end to end, I personally wouldn't even look at Essemtec as I know I can get more for less (and cheaper to run) elsewhere.

This is my dilemma.  I want to run this in production level, I can keep refilling feeders (most of my components have 10k/reel).  I can give up speed IF this machine can be left unattended at night.  The misfeeds have me irked as I dont want issues running long jobs. 

It is a beautiful machine but it reminds me of a high output inline 4 cylinder engine.  Being American I like running huge horsepower v8s.  I am leaning towards a universal with lightning head but really liked the size of this machine.

[SMTech]

I wouldn't want to leave any machine fully unattended and this is particularly true if I NEEDED the output it was working on, could maybe be OK with leave it running something complicated and if production stops for whatever reason "oh well never mind". However if you look at the reported speed for the Fox and you think you already need more than that in a normal working day, I think that means it is the wrong machine. Important also to remember those reported speeds (true for everyone) are hypothetical rubbish. With CLM Feeders I have no PCB that achieves over 6200cph, mostly more like 5300 on the Paraquda, and in theory could gain 20-30% if I had HyQ Feeders. If I were to take the speeds from the brochure the Paraquda has a theoretical speed of 15kcph & an IPC Speed of 10.2K. If you have a fairly typical design you can send it to them and they will give an estimation as to the placement rate you might achieve on that board.

Once you are running an inline machine, the Fox loses some if its advantages, becoming a 120 slot placement machine much like many others, its very narrow front to back, hence the modest max PCB dimensions, but line length can only shrink so much and still fit that many feeders on.

If you look around for consumables like Nozzles you will notice that many of the bigger brands have multiple sources for equivalent parts, there are none for Essemtec and one could certainly consider some of these things expensive/over priced. Get consumables pricing form every option you consider and compare them.

Equally put some proper numbers together for the amount of work you need the machine to do, what feeders you need and how big your panels will be. Get quotes from everyone, make sure they know its a real contest (if you are seriously considering new) and see what happens to those new machine prices.

If you like the Universal lightning head how about the C12pp head on an E series from SiPlace? http://www.e-by-siplace.com/en/e-performance/placement-heads/cp12pp

[rx8pilot]

I can give up speed IF this machine can be left unattended at night.  The misfeeds have me irked as I dont want issues running long jobs. 

I have been looking at new machines for the past few months and I don't believe any of them are at a level that I would count on lights out performance. Perhaps, I would walk away and hope for the best.

In my CNC production experience, I tried full lights-out machining. This is a process that is considerably less fiddly than PCB assembly. It was extremely difficult to manage all the faults well enough to call it reliable. In the end, I called it bonus time - if we were lucky, we would have parts in the morning. If we were unlucky, we would have an alarm, broken tools, or scrapped parts.

I would be pretty happy if I had a P&P line that I could load up and run for a full 8 hour day with only 45 minutes of labor. That person could be working on dozens of other tasks. That would be excellent from a business perspective without the additional challenges related to true lights-out assembly.

[Reckless]

I would be pretty happy if I had a P&P line that I could load up and run for a full 8 hour day with only 45 minutes of labor. That person could be working on dozens of other tasks. That would be excellent from a business perspective without the additional challenges related to true lights-out assembly.

Me too, this is what I am looking for.  I also like saving on floor space but can give that up for less headaches.  I also tried lights out with my Instructables CNC machines (Xcarve and Carvey) and it works with the Carvey as its self contained and very repeatable.  The X-Carve gave us two fires in the beginning due to some component failure on z-axis so I don't feel comfortable running that after hours. 

[Reckless]

I wouldn't want to leave any machine fully unattended and this is particularly true if I NEEDED the output it was working on, could maybe be OK with leave it running something complicated and if production stops for whatever reason "oh well never mind". However if you look at the reported speed for the Fox and you think you already need more than that in a normal working day, I think that means it is the wrong machine. Important also to remember those reported speeds (true for everyone) are hypothetical rubbish. With CLM Feeders I have no PCB that achieves over 6200cph, mostly more like 5300 on the Paraquda, and in theory could gain 20-30% if I had HyQ Feeders. If I were to take the speeds from the brochure the Paraquda has a theoretical speed of 15kcph & an IPC Speed of 10.2K. If you have a fairly typical design you can send it to them and they will give an estimation as to the placement rate you might achieve on that board.

Once you are running an inline machine, the Fox loses some if its advantages, becoming a 120 slot placement machine much like many others, its very narrow front to back, hence the modest max PCB dimensions, but line length can only shrink so much and still fit that many feeders on.

If you look around for consumables like Nozzles you will notice that many of the bigger brands have multiple sources for equivalent parts, there are none for Essemtec and one could certainly consider some of these things expensive/over priced. Get consumables pricing form every option you consider and compare them.

Equally put some proper numbers together for the amount of work you need the machine to do, what feeders you need and how big your panels will be. Get quotes from everyone, make sure they know its a real contest (if you are seriously considering new) and see what happens to those new machine prices.

If you like the Universal lightning head how about the C12pp head on an E series from SiPlace? http://www.e-by-siplace.com/en/e-performance/placement-heads/cp12pp

Thank you soo much for responding and giving your experience with Essemtec machines.  Your CPH numbers are much lower than I was expecting.  My pcbs are small and I don't have many feeders.  My biggest concern is a no headache machine.  I like the form factor of the Essemtec but speed is going to be the issue.  It is soo tiny that it will fit 2-3 units where I could fit one Universal.  I don't know anything about siplace machines but am very keen on learning about other manufacturers that have excellent reputations for being die hard.  It is very confusing picking out a pick and place machine esp USED.  I don't like losing money when taking it off the 'new dealer' lot.  I was half considering the fox new but everything else I prefer used but want something that I can beat to the ground and not have to worry about.  I love industrial machines that pay for themselves esp with no headaches.  One of the reasons I was drawn to universal was to the overbuilt nature of their machines.  Their beams are ridicously thick and heavy but the machine has to move a greater distance.  I compare it to a monster V8 as opposed to a japanese V6.  I may end up buying fuji someday as it seems to be the company everyone respects the most.  I will have to wait it out until I find a deal.  My SMT Tech is a repair man from Panasonic's SMT department. 

[mrpackethead]

As a general rule,  "real world" placement rates seem to be anywhere beween 1/3 and 1/4 of the stated 'maximum' placement rates.   It takes time for boards to load.   For multihead machines they will have them all picking at the same time, and placing at optimal positions for the test, and it will be set up for the shortest parths etc. etc.    Its just a wee bit artifical.   I've seen this across many many differnet machine lines, from different manufacturers.

Runnign a line fully human free, sound great, but i suspect that it woudl cost so much it, would not be worth it.

[rx8pilot]

Runnign a line fully human free, sound great, but i suspect that it woudl cost so much it, would not be worth it.

That is basically what I learned in CNC (an easier process to go lights-out).
It only made sense in if the volume was very high and therefore able to absorb the cost of making it happen. P&P is considerably harder and the costs of achieving a true lights out process would be pretty big.

[Reckless]

As a general rule,  "real world" placement rates seem to be anywhere beween 1/3 and 1/4 of the stated 'maximum' placement rates.   It takes time for boards to load.   For multihead machines they will have them all picking at the same time, and placing at optimal positions for the test, and it will be set up for the shortest parths etc. etc.    Its just a wee bit artifical.   I've seen this across many many differnet machine lines, from different manufacturers.

Runnign a line fully human free, sound great, but i suspect that it woudl cost so much it, would not be worth it.

Do you feel that way about universal instruments?  I just got a gsm flexjet (10k cph) and a dual beam inline7 machine (27k) and they seem very fast.  I am still setting them up but my entire circuit is all tiny chips.

[mrpackethead]

in practical terms;

you would need automated solder paste stenciling or dispencing.   The Fox has solder jet printing, but once you start getting down in size its problematic.   I am hopeful that one day solder paste printing will actually get their and be affordable.. ( MY-500 printer is pretty good, its just $500,000).  An automatic stencil printer is going to hit you up for $50-$60k.

you'll need a board loader, and a board unloader..      A hole nights operation, probalby would deplete some reels, if not the first night, def on the 2nd or 3rd night and then they start getting offset.. Unless you want to restock everyday.   

And theres the sanity check of being able to look at each panel before it goes into the oven....

[mrpackethead]

Do you feel that way about universal instruments?  I just got a gsm flexjet (10k cph) and a dual beam inline7 machine (27k) and they seem very fast.  I am still setting them up but my entire circuit is all tiny chips.

Yes, depending on how big your run is.   IF you start timing from teh time you start the job, till the time you finish, then you'll have a realistic number.. Over time you'll know how your system pans out.     What matters when you have to work out how long a job will take is how long it will take. Not the theortical limits!

In my system, we have a 'theortical' rate of around 24k parts per hour..  Practically we get around 7k.     Having bigger pcbs / or panels will help.. The set up of nozzles, and were feeders are all makes a difference too.
 

[Reckless]

in practical terms;

you would need automated solder paste stenciling or dispencing.   The Fox has solder jet printing, but once you start getting down in size its problematic.   I am hopeful that one day solder paste printing will actually get their and be affordable.. ( MY-500 printer is pretty good, its just $500,000).  An automatic stencil printer is going to hit you up for $50-$60k.

you'll need a board loader, and a board unloader..      A hole nights operation, probalby would deplete some reels, if not the first night, def on the 2nd or 3rd night and then they start getting offset.. Unless you want to restock everyday.   

And theres the sanity check of being able to look at each panel before it goes into the oven....

I bought a vintage dek automatic stencil printer for $500 and it works.  I dont have board loader and unloader yet.  I have 2 essemtec 300fc ovens.  My panels have 50 boards on them and are smaller than 8.5x11.  I can restock everyday or setup multiple reels.  I bought 2 desktop marantz AOI systems.  I only have 2 main pcbs with 30 parts.  I am new to SMT but I want the best pick and place machine.

[Nauris]

I bought a vintage dek automatic stencil printer for $500 and it works.  I dont have board loader and unloader yet.  I have 2 essemtec 300fc ovens.  My panels have 50 boards on them and are smaller than 8.5x11.  I can restock everyday or setup multiple reels.  I bought 2 desktop marantz AOI systems.  I only have 2 main pcbs with 30 parts.  I am new to SMT but I want the best pick and place machine.

Small boards, big panels, you love things overbuild and admire Fuji - you should have bought a CP-6 then!
Beside no worrying about night runs as all your boards would have been done by the first coffee break in the morning.