Homebrew PCBs

Today was the first chance I've had to try out the new laminator and
vacuum stuff.  I printed a 5-mil spiral pattern for testing.

Laminator: it comes up to temp pretty quickly (<1min) but I let it
warm up for 30 minutes to make sure the insides of the rollers were
warm too.  At 240F it was running 4 sec on, 27 sec off - I think it
has plenty of range beyond the 320F it was factory-configured for.
This is a GBC 9" with my digital temperature control.

The film stuck after lamination.  Even now I can't scratch it off with
my fingernail.  However, it didn't solve the problems of air bubbles.
I still need to practice getting the film onto the pcb without
bubbles; a better lamination won't help otherwise.  I tried exposing
the bubbles anyway to see if they'd re-laminate after developing, but
the film that was bubbled was also distorted by the heat - apparently
it needs the thermal mass and/or smooth surface of the pcb else it
gets all wrinkly and won't take an exposure properly.

One option is to put paper (or better, something lint-free) between
the film and pcb, and hold it back as the laminator feeds so that the
film and pcb come together right at the rollers.

The vacuum system worked like a charm, though.  Wherever the
lamination was good, the traces were good, with only a 1 mil variation
(due to the printer's precision; I've posted about this before).  I
think I need to cut down my exposure some, though, as I'm getting more
bridging than breaking.  Time to re-calibrate.  I've seen sites say to
use a shorter exposure for finer lines, too.

Microscope photos here, all 200x:

	   http://www.delorie.com/pcb/djspirals/

I made the mistake of re-laminating afterwards with paper over the pcb
to keep it from sticking to the rollers.  It stuck to the paper.  I
was able to peel the paper off without damaging the film too much, but
next time I'm saving the poly film that I peel off ;-)

I tried wet-laminating the second pcb but it appears to also have
bubbles in it.  I might need to do the old squeegie method with the
heat gun, and laminate afterwards just to ensure adhesion.

I did think about using a vacuum to do the lamination.  I came up with
this plan:

top plate
  air gap
flexible vacuum film - mylar, poly, whatever
UV film - stuck to vacuum film (static?)
  air gap
PCB goes here
Lower plate - heating plate, smooth top.

What you do is vacuum both air gaps to as near perfect as you can get,
then release the vacuum in the top gap - the mylar will press the film
against the pcb, with no air bubbles because there's no air.  The
pressure holds the film flat against the PCB.  Now heat up the lower
plate to 240F to laminate it.  Cool, release vacuum.  Perfect
lamination?

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>This is a GBC 9" with my digital temperature control.

I think I have the same model. Would you mind sharing what temps have worked best for you at the default speed? Additionally, do you mind sharing any info on how your made your digital temp control?

Thanks,
Colt

"colt.boyd@..." <colt.boyd@...> writes:
> I think I have the same model. Would you mind sharing what temps
> have worked best for you at the default speed?

I've run exactly one board through it so far.  I ran it at 240F
because that's what the film said to run it at.  I did this to the
laminator because the default 320F melted the film rather than just
stick it to the board.

I might try toner transfer again, now that I can crank the heat up,
but I don't know what temp is best for my printers.  I have an old
laserjet 4p and a new colorjet 2550N.  Hmmm... I have some semigloss
photo paper here to try, too :-)

> Additionally, do you mind sharing any info on how your made your
> digital temp control?

It's just an MCU with an SPI-based thermocouple chip, LCD, a variable
resistor, and a power module:

http://www.delorie.com/electronics/laminator/laminator.pdf

There's six extra I/O connections; I used two for pushbuttons on the
front panel.  One switches F/C, the other cycles through display
modes.  The displays tells me actual/target temp, runtime,
time-on-temperature, and duty cycle.

The internal wiring of the GBC-9 is just right for this.  Remove the
temperature switch (small round metal thing on top of the rollers) and
the wires leading from it.  The two wire nuts those connect to are two
of the connections you need; the other wire nut is the other one -
it's the wire from the power switch, I think.

LINE goes to the side of the sensor that connects to the motor.
NEUTRAL goes to the other side of the motor.  HEATER goes to the side
of the sensor that goes to the headers.

Warning - if you get this hookup wrong, something in your circuit will
go bang.  Stand back when you first plug it in, in case it does.  Be
paranoid!  Being otherwise could be fatal.

I put the thermocouple under the mounting bracket for the sensor,
which is still in there, just disconnected.  I'd rather put the
thermocouple in contact with the rollers themselves, but I couldn't
find an easy way to do that.

DJ, may I ask a couple of questions about your schematic? (May I ask them even if I reveal all sorts of ignorance on my part? :)

I'm taking VSX-S3 near the upper right corner as a power supply that converts the line voltage to 5V. Is this some sort of pre-packaged module?

Am I right in thinking that you are detecting the 0 voltage crossover in the AC using U4, and this drives an interrupt?

What is the purpose of J1 -- for expansion, reprogramming, or ??

Finally, I think the MAX6675 is the interface to the thermocouple. Does this IC send the temperature of the probe over a serial interface?

Many thanks for taking pity on this newbie!

Andy

> It's just an MCU with an SPI-based thermocouple chip, LCD, a variable
> resistor, and a power module:
> 
> http://www.delorie.com/electronics/laminator/laminator.pdf
>

"awakephd" <a_wake@...> writes:
> DJ, may I ask a couple of questions about your schematic? (May I ask
> them even if I reveal all sorts of ignorance on my part? :)

Sure.

> I'm taking VSX-S3 near the upper right corner as a power supply that
> converts the line voltage to 5V. Is this some sort of pre-packaged
> module?

Yes.  Digikey part number 102-1802-ND.  Ah - typo, it's a VSK, not a
VSX.  Fixed.

> Am I right in thinking that you are detecting the 0 voltage
> crossover in the AC using U4, and this drives an interrupt?

Yes, but not quite.  The capacitor and resistor drop the 120 vac down
to a few milliamps through the opto, but the timing is such that the
interrupt triggers about 45% of the way through the cycle.  I use an
interrupt-driven timer in the R8C to re-interrupt at the end of the
cycle, which is the start of the next one.

> What is the purpose of J1 -- for expansion, reprogramming, or ??

Programming the R8C.  It doubles as a serial console for debugging.
See http://people.redhat.com/~dj/m32c/ for programming info.  That
connector is the "E8a" standard connector for r8c, m16c, and m32c (and
I think r32c) chip families, although each chip uses different signals
off it.  The E8a itself only uses reset and mode, buy my linux-hosted
programmer uses tx and rx as well.

> Finally, I think the MAX6675 is the interface to the
> thermocouple. Does this IC send the temperature of the probe over a
> serial interface?

Yes.  Each 16-bit SPI read returns the actual temperature of the probe
in Celsius (fractions of a degree, that is, plus some status bits).
It's already done the conversion and cold-point compensation.

I added more design files and a board image to my web server:

        http://www.delorie.com/electronics/laminator/

It's not much, but then again, it didn't take much to do that project.

Many thanks. I used to do a fair bit with electronics, many years ago. I'm slowly getting back into it, but things have changed a bit since my 6502 days. :)

I was not familiar with the R8C and related microcontrollers -- I've just been working on getting up to speed with the PIC family. What do you see as the advantages/disadvantages of the M16/M32/R8C vs. the PICs?

I also was not familiar with the VSK modules. They appear to be not extremely cheap, but certainly extremely convenient! Is there something similar I should be aware of that offers higher currents and/or multiple voltages?

Thanks again for your help!!

Andy

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:

>
> 
> "awakephd" <a_wake@...> writes:
> > DJ, may I ask a couple of questions about your schematic? (May I ask
> > them even if I reveal all sorts of ignorance on my part? :)
> 
> Sure.
> 
> > I'm taking VSX-S3 near the upper right corner as a power supply that
> > converts the line voltage to 5V. Is this some sort of pre-packaged
> > module?
> 
> Yes.  Digikey part number 102-1802-ND.  Ah - typo, it's a VSK, not a
> VSX.  Fixed.
> 
> > Am I right in thinking that you are detecting the 0 voltage
> > crossover in the AC using U4, and this drives an interrupt?
> 
> Yes, but not quite.  The capacitor and resistor drop the 120 vac down
> to a few milliamps through the opto, but the timing is such that the
> interrupt triggers about 45% of the way through the cycle.  I use an
> interrupt-driven timer in the R8C to re-interrupt at the end of the
> cycle, which is the start of the next one.
> 
> > What is the purpose of J1 -- for expansion, reprogramming, or ??
> 
> Programming the R8C.  It doubles as a serial console for debugging.
> See http://people.redhat.com/~dj/m32c/ for programming info.  That
> connector is the "E8a" standard connector for r8c, m16c, and m32c (and
> I think r32c) chip families, although each chip uses different signals
> off it.  The E8a itself only uses reset and mode, buy my linux-hosted
> programmer uses tx and rx as well.
> 
> > Finally, I think the MAX6675 is the interface to the
> > thermocouple. Does this IC send the temperature of the probe over a
> > serial interface?
> 
> Yes.  Each 16-bit SPI read returns the actual temperature of the probe
> in Celsius (fractions of a degree, that is, plus some status bits).
> It's already done the conversion and cold-point compensation.
>

"awakephd" <a_wake@...> writes:
> I was not familiar with the R8C and related microcontrollers -- I've
> just been working on getting up to speed with the PIC family. What
> do you see as the advantages/disadvantages of the M16/M32/R8C
> vs. the PICs?

A couple of (to me) advantages:

1. Most run on anywhere from 2.7v to 5v (i.e. each chip allows that
   range so you can plug them into pretty much any circuit).  Some run
   as low as 1.8v.  Larger chips support dual power; one half on +5
   and the other on +3.3.

2. Programmable with a simple serial cable (i.e. easy linux support)

3. Internal architecture is readily understandable; very similar to
   the x86

4. large peripheral set, which is mostly standard from the tiniest R8c
   up through the largest r32c (and probably the new RX too)

5. The smallest chips are pretty inexpensive (few dollars) for 16-bit
   20MHz performance.  The largest chips are around $30.  There are no
   8-bit chips.

6. The largest chips have a fully functional external address/data bus.

7. They seem to be very robust against EMI and ESD interference.

> I also was not familiar with the VSK modules. They appear to be not
> extremely cheap, but certainly extremely convenient! Is there
> something similar I should be aware of that offers higher currents
> and/or multiple voltages?

I don't know - I found one that worked and stopped looking ;-)

Another company to keep in mind though is acon - www.aconinc.com They
offer a wide range of dc-dc converters, although they're not cheap
either.

Today I tried 4, 3, 2, and 1 - I didn't expect them to work as I knew
the film wasn't laminated well enough, so I merged the four into one
print (one quadrant per size).  The print for 1/1 was pretty much
solid.  The print for 2/2 had moire patterns all over it.  The 3/3 and
4/4 were printed well enough to expose the mask reasonably.

I reduced my exposure from 5.5 minutes down to 4, but still the traces
were wider than the gaps:  http://www.delorie.com/pcb/djspirals/

I *think* with some tweaking of the print widths and exposure, I
should be able to do 3/3 rules.  As usual, the reason behind obtaining
such results is to make the 5/5 or 6/6 boards that much more reliable.

I'll probably make a test print by hand with specific width/gap sizes
in printer pixels to see what the printer's limits are.  Also,
exposing through a step gauge might help understand what exposure time
is best for such fine traces.

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
> 
> 2. Programmable with a simple serial cable (i.e. easy linux support)
> 

I'd like to hear more about this one -- are you saying that you don't need anything other than a serial output -- no JDM-type board or anything? I assume there is a pin that must be signalled to put it in programming mode? Does it use a fixed baud rate, or ??

This part sure does sound interesting!!

"awakephd" <a_wake@...> writes:
> I'd like to hear more about this one -- are you saying that you
> don't need anything other than a serial output -- no JDM-type board
> or anything? I assume there is a pin that must be signalled to put
> it in programming mode? Does it use a fixed baud rate, or ??
> 
> This part sure does sound interesting!!

Go to http://people.redhat.com/~dj/m32c/

Click on "Easy R8C/M16C/M32C Flash Programming" and read :-)

The link under it is a sample flash downloader program.

Basically, you need three things to program:

1. To be able to reset the chip (duh)

2. To be able to assert the MODE or CnVss pin (a two-pin jumper is
   fine) during reset.

3. An asynchronous logic-level serial Tx/Rx pair (i.e. serial port)

On most of my boards, I include an FTDI232R chip which lets me control
all those signals (and more) over USB, plus serve as a serial console
when it's running.  However, the standard E8a connector brings all
those signals out anyway, so you could have one usb-based programmer
for all your boards.

I've also done live updates, using one MCU to update another, in my
furnace controller - the gumstix board can reprogram any of the five
R8C chips while the system is running.  (most of the R8Cs have I2C,
they make great "smart peripherals")

Very cool, and very helpful. As I was digging around for suppliers for the FT232R chip, I also came across the Silicon Lab CP2103 chip. One source described this chip as "superior to the FTDI chip" -- but I can't see that there is any significant difference?? Except that the CP2103 is only available in a 28-QFN package, which I'm guessing will be a lot harder for a home-brew novice like myself to solder to the board.

Am I missing something in the comparison between the FT232R and the CP2103??

Andy

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:

>
> 
> "awakephd" <a_wake@...> writes:
> > I'd like to hear more about this one -- are you saying that you
> > don't need anything other than a serial output -- no JDM-type board
> > or anything? I assume there is a pin that must be signalled to put
> > it in programming mode? Does it use a fixed baud rate, or ??
> > 
> > This part sure does sound interesting!!
> 
> Go to http://people.redhat.com/~dj/m32c/
> 
> Click on "Easy R8C/M16C/M32C Flash Programming" and read :-)
> 
> The link under it is a sample flash downloader program.
> 
> Basically, you need three things to program:
> 
> 1. To be able to reset the chip (duh)
> 
> 2. To be able to assert the MODE or CnVss pin (a two-pin jumper is
>    fine) during reset.
> 
> 3. An asynchronous logic-level serial Tx/Rx pair (i.e. serial port)
> 
> On most of my boards, I include an FTDI232R chip which lets me control
> all those signals (and more) over USB, plus serve as a serial console
> when it's running.  However, the standard E8a connector brings all
> those signals out anyway, so you could have one usb-based programmer
> for all your boards.
> 
> I've also done live updates, using one MCU to update another, in my
> furnace controller - the gumstix board can reprogram any of the five
> R8C chips while the system is running.  (most of the R8Cs have I2C,
> they make great "smart peripherals")
>

Hi Andy,

On Sun, Mar 29, 2009 at 8:34 AM, awakephd <a_wake@...> wrote:
> Very cool, and very helpful. As I was digging around for suppliers for the FT232R chip, I also came across the Silicon Lab CP2103 chip. One source described this chip as "superior to the FTDI chip" -- but I can't see that there is any significant difference?? Except that the CP2103 is only available in a 28-QFN package, which I'm guessing will be a lot harder for a home-brew novice like myself to solder to the board.
>
> Am I missing something in the comparison between the FT232R and the CP2103??

They're both pretty similar. Personally, I prefer the FT232R because
they seem to be better supported under linux, especially for non x86
platforms.

-- 
Dave Hylands
Shuswap, BC, Canada
http://www.DaveHylands.com/

I've taken to purchasing ready made USB/RS232 adapters and using them
for my purposes. They are so cheap you can hardly mount any IC on a
PCB for the money.
You can easily remove the MAX232 if 5V interface is desired (the
cheapest converters at dealextreme have a slip-on cover instead of
molded plastic, so are easy to take apart).

ST

On Sun, Mar 29, 2009 at 7:34 PM, Dave Hylands <dhylands@...> wrote:

>> Am I missing something in the comparison between the FT232R and the CP2103??
>
> They're both pretty similar. Personally, I prefer the FT232R because
> they seem to be better supported under linux, especially for non x86
> platforms.
>
> --
> Dave Hylands
> Shuswap, BC, Canada
> http://www.DaveHylands.com/
>

"awakephd" <a_wake@...> writes:
> Very cool, and very helpful. As I was digging around for suppliers
> for the FT232R chip, I also came across the Silicon Lab CP2103
> chip. One source described this chip as "superior to the FTDI chip"
> -- but I can't see that there is any significant difference?? Except
> that the CP2103 is only available in a 28-QFN package, which I'm
> guessing will be a lot harder for a home-brew novice like myself to
> solder to the board.

There have been complaints in the past about the quality of FTDI's
drivers, but I suspect they've been fixed by now, and Linux uses its
own drivers for them anyway.  I like the TSSOP package myself,
although I'm using FTDI's QFN on my latest project.

My key need is the ability to directly manipulate the device-side
lines (gpio, dtr, etc) from a Linux program.  I know FTDI can do that,
I don't know if SI can.

> My key need is the ability to directly manipulate the device-side
> lines (gpio, dtr, etc) from a Linux program.  I know FTDI can do that,
> I don't know if SI can.

From the quick read I've done of both data sheets, it appears that the CP2103 only has Windows drivers for the direct control. I am running mostly under Linux, so that gives another plus for the FT232R.

With direct control of the GPIO outputs on either chip, it seems you could set it up to control the reset and mode pins on the R8C/etc. for programming without using jumpers. I suppose you could also use these pins to make a programmer for a PIC, without having to go through the gymnastics that the typical PIC serial programmer has to do, except for needing +12v or so depending on the version of the PIC. Of course, you'd still have a serial output available to act as a terminal, or whatever ...

DJ, I've done a little looking at your website. Do I understand that you have done a port of gcc to output code for the R8C/M16/M32 processors? Is there a minimum program memory size needed to use this rather than assembler?

Thanks so much to all for furthering my education ... now I've just got to find the time to try some of this out!

"awakephd" <a_wake@...> writes:

> With direct control of the GPIO outputs on either chip, it seems you
> could set it up to control the reset and mode pins on the
> R8C/etc. for programming without using jumpers.

That's exactly what I do :-)

> I suppose you could also use these pins to make a programmer for a
> PIC, without having to go through the gymnastics that the typical
> PIC serial programmer has to do, except for needing +12v or so
> depending on the version of the PIC. Of course, you'd still have a
> serial output available to act as a terminal, or whatever ...

I don't know PIC programming, but does it use asynchronous or
synchronous serial?  Does it have any special timing requirements?
The R8C is sufficiently flexible with its timing to allow for
user-level accuracy.

The 13v requirement would take up a lot of board space if you wanted
to embed the programmer, too.

> DJ, I've done a little looking at your website. Do I understand that
> you have done a port of gcc to output code for the R8C/M16/M32
> processors? Is there a minimum program memory size needed to use
> this rather than assembler?

It depends on whether you use the default interrupt vector handling,
which requires at least 1k of flash just for that.  I use gcc/C for
nearly all my R8C programming, the biggest 1B's only have about 1k of
ram and 16k of rom.  checking... one of my test programs uses 2k of
flash and 318 bytes of ram, which fits in the smallest 1B's.

Of course, each C library routine you pull in (malloc, printf, etc)
adds to the needed size.