Prototype oven recommendations

On Sat, 06 Sep 2014 09:42:10 -0400, krw@attt.bizz Gave us:

>On Fri, 05 Sep 2014 23:04:22 -0700, DecadentLinuxUserNumeroUno
><DLU1@DecadentLinuxUser.org> wrote:
>
>>On Fri, 05 Sep 2014 23:05:12 -0500, Joe Chisolm
>><jchisolm6@earthlink.net> Gave us:
>>
>>>
>>>I have some ESD stackable trays I use.  My sequence is I paste maybe
>>>8 or 10 boards and put them in the trays.  I have a little "box" that
>>>keeps them about 65F to help with the paste life.  I stuff one and
>>>into the oven.  While it bakes I start stuffing another, etc. Or 
>>>depending on the board I might stuff them all.  The stack trays 
>>>really help.
>>>
>>>The trick is to balance your paste life with how long it takes
>>>to stuff and bake.
>>
>>
>>  Paste life includes all time until the reflow run itself.
>>
>>  I am sure lead free has longer open air time than the old stuff.
>
>Perhaps but lead-free has a very tight temperature profile. 



  I'll bet that it a pure PITA.

> The
>difference in temperature between the solder melting and the parts
>wilting is very small and short (time).

  We never had to worry about that at the lower temp older process
either...  unless the belt stopped.

>  The PPoE had a lot of trouble
>with their lead-free solder process in a four-stage oven.  LEDs,
>aluminum capacitors, inductors, and QFNs were the biggest problems.

 So, we can at least say that the new RoHS stuff has cause better,
higher temp polymers, etc to get incorporated.

  One plus on a very short list.  They have massaged a lot of the bugs
out though...  still.

>Balancing the profile between the needs of all of these was like
>balancing a pencil on its point.

  I am sure it is still that way.  Just not the PITA it was in the
initial years.

>  They bought a used multi-stage oven
>(eight or eleven, can't remember) in 2010, when manufacturing
>equipment could be had for pennies on the dollar, and the problems
>went away.

  There is still a lot of liquidous gear out there, but demand is higher
as the economy lifts (a tiny bit) so the prices are higher in that
market currently.

>>  The old stuff made way better solder joint IMO though.
>
>Far harder to get the same quality, that's certain.

  You mean now, right?

>  Harder to
>inspect, too.

  I used to be able to scan an 18 inch square board back in the 0805
days and literally just *see* any bad joints or tombstoned resistors,
etc. in my periphery.

  Now, I can't even count my fingers without a pair of reading glasses.
(exaggerating)

On Fri, 05 Sep 2014 22:13:24 -0700, Don Y wrote:
[snip]
>>
>> I have some ESD stackable trays I use.  My sequence is I paste maybe 8
>> or 10 boards and put them in the trays.
> 
> So your boards tend to be small-ish?
> 
Some PCI size.  Others in the 7x7 range.  It varies.  I do have
some backplane boards that are 14x2 (I think it is).  But those
are TH connectors and such that we do by hand.  I think the
stacking trays are about 9x10x3.

[snip]
>> The trick is to balance your paste life with how long it takes to stuff
>> and bake.
> 
> I am hoping to decouple the two -- by not caring how long the bake cycle
> takes as I'm "done" stuffing before it *started* (so I don't have to
> drop what I am working on to attend to the oven's call).
> 

The clock starts ticking when you apply the paste to the first board
of the batch and does not stop until the last board reflow is complete.
It's not so much an issue of getting the board out of the oven as it is
getting the next board in the oven.  That damn clock keeps ticking.
If you are squirting pads with a dispenser it not too bad to do a panel
or 2, stuff and reflow, do some more, etc.  With a stencil the prep
and clean up time is high so you want to do as many boards as you can
fit in the working window.
   
>> Trying to do something like an oven rack is tricky and would be almost
>> impossible with a double sided board.  You have to be able to control
>> the heat on the "other" side.  Surface tension only goes so far
>> competing with gravity.
> 
> "Oven rack" meaning *toaster* oven rack?  I.e., have you had problems
> doing double-sided in your kit?  (do one side at a time -- heatsinking
> the inactive side?)
> 

The T-962[A] will do double sided no problem because the IR lamps are
on top. A toaster oven with elements on top/bottom would be much
more of a challenge.  I have 3 designs that are ds and never had an
issue with them.  But it's generally just C R and L on the one side,
maybe a discrete semi or LED here and there. The other side has the
bigger parts.  I really push for components on 1 side only.  I only
want 1 pass through the oven.
 
>> I understand what you want to do, like baking loaves of bread. Not that
>> easy keeping the temps right.  Take a look at your paste temp profile. 
>> You need to ramp to, say 220C and then make that bump to 240C in 15-30
>> seconds depending on your profile.  Then ideally need to bring that
>> back down in 15-30 seconds.
> 
> As I mentioned elsewhere, external air handler would be an ideal way of
> doing this!  Change air temperature nearly *instantly*! Of course, more
> wasteful of power and you'd need a place to exhaust all that process
> air...
>

It's not the air temp that's important, it's the board temp.  Throw in
some 2oz or bigger planes and the thermal mass starts to add up.  These
little ovens have their limits.  I have some x8 EMI filters that 
are a pain.  I could not get the profile right to push enough heat in
them and not cook the rest of the board.  A big conveyor oven would
have had no problem.  That board is real low volume now so we just put
those on by hand.  I dont think we have shipped one of those boards this
year so it's not worth the time or effort to change the design.
 
[snip]
> 
> (sigh)  Perhaps I should start calling in some favors.  After all, folks
> forget they owe you if you leave them "banked" too long!  ;-) And, it
> would be really annoying to discover folks retiring out of positions
> where they can be of help!
> 
> I guess I will have to deal with the dreaded telephone this weekend...
> 
> Thanks for your comments!
> --don
> 
> P.S.  Any suggestions re: the T-962C I mentioned?  "A bigger version of
> the A?"  Or, does the design not scale well??

I think they all use 4 elements.  As Paul Carpenter said in the other
post, the extra volume makes things worse.  I would not go bigger than a
"A" unless you can find something with more power. I just shoved a L-com
catalog in the tray.  It is 10.5x8 and would probably be OK size for a
board/panel.  Bigger you will have problems, thick planes you will have
problems. I have done 8 layer 1oz boards without problems.

The 962A works for me but if I could justify a larger conveyor oven I
would get it.  A PNP with vision would be more of a help.

-- 
Chisolm
Republic of Texas

On Sat, 06 Sep 2014 07:46:28 -0700, DecadentLinuxUserNumeroUno
<DLU1@DecadentLinuxUser.org> wrote:

>On Sat, 06 Sep 2014 09:42:10 -0400, krw@attt.bizz Gave us:
>
>>On Fri, 05 Sep 2014 23:04:22 -0700, DecadentLinuxUserNumeroUno
>><DLU1@DecadentLinuxUser.org> wrote:
>>
>>>On Fri, 05 Sep 2014 23:05:12 -0500, Joe Chisolm
>>><jchisolm6@earthlink.net> Gave us:
>>>
>>>>
>>>>I have some ESD stackable trays I use.  My sequence is I paste maybe
>>>>8 or 10 boards and put them in the trays.  I have a little "box" that
>>>>keeps them about 65F to help with the paste life.  I stuff one and
>>>>into the oven.  While it bakes I start stuffing another, etc. Or 
>>>>depending on the board I might stuff them all.  The stack trays 
>>>>really help.
>>>>
>>>>The trick is to balance your paste life with how long it takes
>>>>to stuff and bake.
>>>
>>>
>>>  Paste life includes all time until the reflow run itself.
>>>
>>>  I am sure lead free has longer open air time than the old stuff.
>>
>>Perhaps but lead-free has a very tight temperature profile. 
>
>
>
>  I'll bet that it a pure PITA.

It was, in a small company, where I had to worry about the mess left,
yes.  Not so much anymore.  ;-)

>> The
>>difference in temperature between the solder melting and the parts
>>wilting is very small and short (time).
>
>  We never had to worry about that at the lower temp older process
>either...  unless the belt stopped.

Right.  The PPoE's process was golden with leaded solder but went
right out the window, for a year, with the socialist solder.  Along
the way, we were surprised by the parts that caused the biggest grief.
I would have thought BGAs would be a problem but not a big hunk of
metal like a transformer.  Wrong.

>>  The PPoE had a lot of trouble
>>with their lead-free solder process in a four-stage oven.  LEDs,
>>aluminum capacitors, inductors, and QFNs were the biggest problems.
>
> So, we can at least say that the new RoHS stuff has cause better,
>higher temp polymers, etc to get incorporated.

There is some of that but unless you're doing RoHS, most of it is
wasted effort.

>  One plus on a very short list.  They have massaged a lot of the bugs
>out though...  still.
>
>>Balancing the profile between the needs of all of these was like
>>balancing a pencil on its point.
>
>  I am sure it is still that way.  Just not the PITA it was in the
>initial years.

A lot has been learned, sure, and as you note, parts are better.  I
think it's pretty mundane, now.  Everyone knows how to deal with it.
The long-term problems seem to have been overblown.  That isn't to say
that the whole effort wasn't an absolute waste of time and resources,
though. 

>>  They bought a used multi-stage oven
>>(eight or eleven, can't remember) in 2010, when manufacturing
>>equipment could be had for pennies on the dollar, and the problems
>>went away.

I should amend the above to say that the *final* problems went away. A
lot of them had already been solved by that time.

>  There is still a lot of liquidous gear out there, but demand is higher
>as the economy lifts (a tiny bit) so the prices are higher in that
>market currently.
>
>>>  The old stuff made way better solder joint IMO though.
>>
>>Far harder to get the same quality, that's certain.
>
>  You mean now, right?

No, I think the lessons learned and the better tools and processes
have leveled the playing field, at least for those with the resources
to buy the proper tools.  That leaves most of the "garage" outfits
out, though.

>>  Harder to
>>inspect, too.
>
>  I used to be able to scan an 18 inch square board back in the 0805
>days and literally just *see* any bad joints or tombstoned resistors,
>etc. in my periphery.
>
>  Now, I can't even count my fingers without a pair of reading glasses.
>(exaggerating)

You're just getting old.  ;-)  OTOH, at one time I thought 0805s were
small.  Now I almost trip over them (SC70s, 0402s, and .5mm pitch
QFP/QFNs are the norm).  A Mantis or OptiVisor helps.   

Hi Joe,

On 9/6/2014 9:18 AM, Joe Chisolm wrote:
> On Fri, 05 Sep 2014 22:13:24 -0700, Don Y wrote:

>>> The trick is to balance your paste life with how long it takes to stuff
>>> and bake.
>>
>> I am hoping to decouple the two -- by not caring how long the bake cycle
>> takes as I'm "done" stuffing before it *started* (so I don't have to
>> drop what I am working on to attend to the oven's call).
>
> The clock starts ticking when you apply the paste to the first board
> of the batch and does not stop until the last board reflow is complete.

Yes.  I am hoping that *focusing* on stuffing *all* the boards
(instead of stuffing, loading oven, stuffing more, unloading oven,
reloading oven, stuffing more, etc.) and then baking *all* of them
in one batch is a more efficient use of time.

I.e., if I had an "assistant" and staging area so one person could
CONTINUOUSLY stuff boards while the other continuously services
the oven, then there would be no "interruption".

Again, my cookie example:  I can cut 20 dozen cookies from the
(frozen) dough and lay them on 6 or 7 cookie sheets then baking
ALL at once in a lot less time than doing one sheet, feeding it
into oven, prepping a second sheet, removing first from oven,
inserting second sheet, transfering removed cookies onto cooling
rack, then starting to prep the *third* sheet for the oven...

There is a non-negligible "task switching overhead".

> It's not so much an issue of getting the board out of the oven as it is
> getting the next board in the oven.  That damn clock keeps ticking.
> If you are squirting pads with a dispenser it not too bad to do a panel
> or 2, stuff and reflow, do some more, etc.  With a stencil the prep
> and clean up time is high so you want to do as many boards as you can
> fit in the working window.
>
>>> Trying to do something like an oven rack is tricky and would be almost
>>> impossible with a double sided board.  You have to be able to control
>>> the heat on the "other" side.  Surface tension only goes so far
>>> competing with gravity.
>>
>> "Oven rack" meaning *toaster* oven rack?  I.e., have you had problems
>> doing double-sided in your kit?  (do one side at a time -- heatsinking
>> the inactive side?)
>
> The T-962[A] will do double sided no problem because the IR lamps are
> on top. A toaster oven with elements on top/bottom would be much
> more of a challenge.

I think ours allows you to select which elements are on (bake vs. toast,
etc.)

> I have 3 designs that are ds and never had an
> issue with them.  But it's generally just C R and L on the one side,
> maybe a discrete semi or LED here and there. The other side has the
> bigger parts.  I really push for components on 1 side only.  I only
> want 1 pass through the oven.

Understood.  But, there's just not enough real estate on one side.
The 1.25 x 2.5" (by about 1") envelope is pretty much fixed.  So,
I either find more "sides", smaller components, higher integration,
etc.  Keeping in mind that I want others to be able to make these
as well eliminates a lot of options!

>>> I understand what you want to do, like baking loaves of bread. Not that
>>> easy keeping the temps right.  Take a look at your paste temp profile.
>>> You need to ramp to, say 220C and then make that bump to 240C in 15-30
>>> seconds depending on your profile.  Then ideally need to bring that
>>> back down in 15-30 seconds.
>>
>> As I mentioned elsewhere, external air handler would be an ideal way of
>> doing this!  Change air temperature nearly *instantly*! Of course, more
>> wasteful of power and you'd need a place to exhaust all that process
>> air...
>
> It's not the air temp that's important, it's the board temp.  Throw in

Yes.  But, instead of radiant heating, if you expose the boards to
preheated/"precooled" air (e.g., an external air handler delivers a
specific volume of air at a controlled temperature) to the boards,
they will track the temperature of that air.

I don't know how efficiently radiant heat transfers energy into
materials (vs. passing them through a heated "fluid").  But, for
sure, the elements aren't actively *cooling* the materials!  Instead,
the oven relies on evacuating the heated internal air mass.

If, instead, the temperature of the air being pushed into the
enclosure was actively driven to a lower temperature (even ambient),
then you've moved the control loop outside of the oven's interior.

As the oven would *continuously* be exhausting air (even during its
preheat cycle), you can monitor the exhaust air temperature to get
a feel for how much energy is being absorbed (by the oven's interior
and the items placed therein).  (there may be other things worth
monitoring as well?)  You also know the maximum temperature that the
surfaces are exposed to -- that of the incoming air.

The amount of heat you can pump in is then a matter of how much air
you can move, without jostling the components.

Dunno.  I don't design reflow ovens.

> some 2oz or bigger planes and the thermal mass starts to add up.  These
> little ovens have their limits.  I have some x8 EMI filters that
> are a pain.  I could not get the profile right to push enough heat in
> them and not cook the rest of the board.  A big conveyor oven would
> have had no problem.  That board is real low volume now so we just put
> those on by hand.  I dont think we have shipped one of those boards this
> year so it's not worth the time or effort to change the design.

>> (sigh)  Perhaps I should start calling in some favors.  After all, folks
>> forget they owe you if you leave them "banked" too long!  ;-) And, it
>> would be really annoying to discover folks retiring out of positions
>> where they can be of help!
>>
>> I guess I will have to deal with the dreaded telephone this weekend...

Well, first call was met with, "Sorry, we decided it was easiest to
farm out all of our fab -- even the engineering prototypes and first
manufacturing samples".  I am waiting for some number$ to get a better
idea for how much this decision is worth to them...

However, second call was more productive!  I've been offered a small
oven that another friend had previously used in their engineering
department.  Apparently, their Manufacturing guy felt threatened (?)
by Engineering having a reflow oven -- when *his* department had a
*bigger* one!  And, it was relatively easy for him to convince the
powers that be of the needless duplication of efforts involved.  So,
their Engineering staff now schedules prototype runs AS IF regular
manufacturing runs (just incredibly tiny quantities!).

While this seems like a good deal (for me), it means I have to either
let him ship it to me *or* go for a visit (I detest travel!).  SWMBO
has already given me the Evil Eye as she knows if I go for a visit,
I'll end up driving a *truck*/van back!  And it won't *just* have
a reflow oven in it!  (she was thrilled when I stopped my annual
pilgrimages to visit him as it slowed the "rate of accumulation"
here!  :< )

[I'll have to wait until next week for him to get me the details of
what he has to offer...]

Still waiting for third call to be returned.  Perhaps "third time's
the charm"?

>> P.S.  Any suggestions re: the T-962C I mentioned?  "A bigger version of
>> the A?"  Or, does the design not scale well??
>
> I think they all use 4 elements.

So, they are *longer* elements (2500W vs. 1500W)?  I will have to see
if I can find any more detail.  OTOH, the increase in size suggests a
proportional power level would be in the 4KW range.

> As Paul Carpenter said in the other
> post, the extra volume makes things worse.

I have not yet read the review he mentioned.  Still rolling out of
bed...

> I would not go bigger than a
> "A" unless you can find something with more power. I just shoved a L-com
> catalog in the tray.  It is 10.5x8 and would probably be OK size for a
> board/panel.  Bigger you will have problems, thick planes you will have
> problems. I have done 8 layer 1oz boards without problems.

I can stick to 4 and 6 layer boards.  As the boards are small, no
real need for extra thickness.  May have some problem with copper
thickness, though, as many of the devices I am using rely on a
heavy copper plane to pull heat away from the device.  Thermal
reliefs kind of being contraindicated in those cases!  :>

Worst (as far as size) for me are the boards for the network switch.
They tend to be larger (in at least one LONG dimension).  Maybe I
can revisit that design and partially cannabilize an existing switch
(though it would REALLY be a kludge!)

OTOH, there is nothing *forcing* me to adopt the typical form factor
for such a device.  I *could* rearrange things to better exploit
the capabilities of a smaller oven...  might *look* weird, though...

> The 962A works for me but if I could justify a larger conveyor oven I
> would get it.  A PNP with vision would be more of a help.

I don't want to be in the board fab business.  If there was a fab shop
"next door", I would gladly give them the business:  walk over in the
morning to see what problems they have encountered; wander back home
for lunch; drop by any time they stumble on an "issue" that I need
to examine/address; etc.  Let *them* acquire and maintain the equipment,
train/supervise the staff, etc.

It's like having a tech build your prototype so you know what sorts
of issues he -- and, ultimately, Manufacturing -- will face.  Sending
*off* to a fab house is just not time-efficient at uncovering these.

(Here, any such industry would be at least half an hour drive each
way... perhaps considerably more!  I live in a high-tech wasteland...)

Hi Paul,

On 9/6/2014 12:44 AM, Paul wrote:
> In article<lue55k$g41$1@speranza.aioe.org>, this@is.not.me.com says...
> .....
>> P.S.  Any suggestions re: the T-962C I mentioned?  "A bigger version
>> of the A?"  Or, does the design not scale well??
>
> Have a look at these folks and their description of differences
>
> https://www.estechnical.co.uk/blog/entry/t962c-reflow-oven-troubles
>
> In their view it is underpowered has the same heater and fans as
> T-962A trying to heat a larger volume.

Obviously not the *same* heater -- as the C version draws 2500W
while the A draws 1500.  But, the increased size suggests the C
*should* draw 4000W to have the same *heating* capacity as the A
(assuming the A is not OVERdesigned).

Likewise, a *third* fan would seem to be required to similarly
scale the capabilities of the A.

Without photos of the interior (I will start searching once I
am awake), its hard to know for sure but I suspect they just
put *longer* heating elements in place instead of *more*
elements (the latter giving better coverage).

It really *does* seem like the better solution is an external
air handler approach.  Size *it* for the volume that you want
to control.  And, the maximum temperature that you want to
be able to attain.  The *rate* of temperature change would be
easily accommodated by the face/bypass damper:  25C to 300C (air
temperature) in 0.5 seconds!

As a compromise, Hamilton's "convection oven" approach may
merit further exploration.  Especially if you rearranged the
elements to locate all of them *above* the work (and had
good temperature sensor placement).

(sigh)  I had planned on doing OTHER things today...

In article <luad06$5a7$1@speranza.aioe.org>, this@is.not.me.com says...
> Hi,
> 
> I'm looking for recommendations for a small-ish oven for SMD
> prototype fab.  Small boards (non-palletized).  Probably just a
> couple/few square feet total board area at a time.  No exotic
> processing, fancy controls, etc.  Seldom used -- though I want
> it available "when I need it" (instead of sending off to a fab
> house "a la carte").
> 
> Operating it in a home environment so nothing beyond "220"
> (ideally, repurpose a stove/dryer or other "dedicated" service
> for it).
> 
> (No, I'm not keen on the EZ-Bake/"toaster oven" approach!)
> 
> Thx!
> --don
> 
I bought one of these low cost things, and it has served us well :

http://www.ebay.com/itm/T-962a-/221542342836?pt=LH_DefaultDomain_0
&hash=item3394f3e0b4

http://tinyurl.com/p3zrnxa

On 9/8/2014 11:28 AM, WangoTango wrote:
> In article<luad06$5a7$1@speranza.aioe.org>, this@is.not.me.com says...

>> I'm looking for recommendations for a small-ish oven for SMD
>> prototype fab.  Small boards (non-palletized).  Probably just a
>> couple/few square feet total board area at a time.  No exotic
>> processing, fancy controls, etc.  Seldom used -- though I want
>> it available "when I need it" (instead of sending off to a fab
>> house "a la carte").

> I bought one of these low cost things, and it has served us well :

Yes, Joe (& Wouter) comments agree.

I've been offered a larger oven (more in line with my initial goal).
But, another friend has offered to fab the boards for me -- *if*
doing so will be "uneventful" (i.e., no problems encountered in
their manufacture).

So, I will rethink "dry fitting" each of the designs to verify the
boards are manufacturable.  Then, just kit everything up (probably
in two separate batches so I don't have to invest in *all* of the
silicon before needing it) and let him assemble them for me.  (I'll
do all the testing since it is largely automated).

If I have concerns about a particular design, I may opt to build *a*
prototype, myself.  In which case, a *smaller* oven (than my original
goal) may be a worthwhile thing to have available, here -- even if
it only gets used one or twice!

Thanks, all!
--don