I am very excited to see this! Looking at the bonding image, it has a date of 2014. Is that a typo, or did the actual bonding not change since that was proposed?
I am very excited to see this! Looking at the bonding image, it has a date of 2014. Is that a typo, or did the actual bonding not change since that was proposed?
Can't wait to play with the P2
That was the leadframe vendor's document. We just used their leadframe.
I took a break from getting the next release together, in order to make a PCB that attaches to the Prop123-A9 board, so that we can run the Prop2 analog test chip. Everything, including parts, is ordered now. By the end of next week, Maybe we can build some of these. I got parts to make 25 units. Only 4 PCBs are coming, but we'll order more once we know it's working.
This test board will only work on the -A9, since it has all 3.3V I/O's on the header connectors. The -A7 board's right-side signals are all 2.5V. We need 3.3V, all around.
This board provides a SPI flash chip, some pluggable pull-up resistors that can be connected to the header pins, four normal I/O's, and TWO full analog I/O pins. It takes 16 I/O's to talk to each pin, plus 9 more for the four fuses in each pin. The other header signals are mainly eaten up by the XTAL/PLL/OSC circuit and TESn and RESn signals. This is going to let us fully use the Prop2 padframe circuitry to act like a real Prop2. It will be neat to run off the real clock circuit. Man, I hope this all works.
Here are some pics of the PCB.
It makes me a little nervous that the test chip's package has the pin-1 indicator and the ID text 90 degrees off from each other.
Chip - is there a simple test that you could do to check the pinout before mounting the chips? e.g. carefully check some power and ground pins with a multimeter? If there's any asymmetry in the pinout then perhaps something simple can be done. I remember once when something similar happened to us with a WLBGA (wafer level BGA) chip, but it was easy to check because you could see the die ;-)
Chip - is there a simple test that you could do to check the pinout before mounting the chips? e.g. carefully check some power and ground pins with a multimeter? If there's any asymmetry in the pinout then perhaps something simple can be done. I remember once when something similar happened to us with a WLBGA (wafer level BGA) chip, but it was easy to check because you could see the die ;-)
Good idea. Yes, there is plenty of asymmetry in the GND pins which can be checked. I'll do that today.
Its a minor point, but since we're testing analog - you have a ground loop caused by that track underneath and two ground entry points along the left hand header. What i'd do is make your top left the single ground entry point, and put a test pin before the ground goes to the 1v8 regulator, for connecting cro's and logic analyzers etc. Connect both ground pins at top left (technically a ground loop but only 0.1"). Disconnect the ground connection at bottom left of the header, or break it with a 0.1" jumper if you really think you might need it for current purposes. Break the ground connection where it leaves the regulator tab, then route a solid ground from the test point at top left to underneath the chip, and connect to the existing ground networks using several vias.
Hope you have a steady hand ... soldering that fine pitch can be tricky, even with a stencil and re-flow oven. You might need all 4 of those PCB boards just to get one good one out of the bunch. ...or at least past a slight learning curve. We made a Compact Flash adapter card with the same pitch I believe of 50 pins at .5mm and even with a microscope and a soldering iron it was a challenge to make sure there weren't any solder bridges.
Understanding Solder surface tension is your friend ... Sometimes counter intuitively using more solder is better than using less solder in a situation like this.
Hope you have a steady hand ... soldering that fine pitch can be tricky, even with a stencil and re-flow oven. You might need all 4 of those PCB boards just to get one good one out of the bunch. ...or at least past a slight learning curve. We made a Compact Flash adapter card with the same pitch I believe of 50 pins at .5mm and even with a microscope and a soldering iron it was a challenge to make sure there weren't any solder bridges.
Understanding Solder surface tension is your friend ... Sometimes counter intuitively using more solder is better than using less solder in a situation like this.
Good to hear from you, Beau!
I don't know how hard this will be, but I was thinking about using lots of liquid flux and a ball of solder on the iron tip. One good pass per side with a little desoldering on the last 2-3 pins will hopefully do the job.
Its a minor point, but since we're testing analog - you have a ground loop caused by that track underneath and two ground entry points along the left hand header. What i'd do is make your top left the single ground entry point, and put a test pin before the ground goes to the 1v8 regulator, for connecting cro's and logic analyzers etc. Connect both ground pins at top left (technically a ground loop but only 0.1"). Disconnect the ground connection at bottom left of the header, or break it with a 0.1" jumper if you really think you might need it for current purposes. Break the ground connection where it leaves the regulator tab, then route a solid ground from the test point at top left to underneath the chip, and connect to the existing ground networks using several vias.
Hope this makes sense.
Good point. When we test the ADC, that will probably make a difference.
"Good to hear from you, Beau!" ... Yes, many things have changed in two years. Never thought I'd have a house on 4 Acres and owned a thoroughbred race horse in our back yard if you had asked me two years ago. :-)
"...I don't know how hard this will be, but I was thinking about using lots of liquid flux..." ...I wouldn't use any more flux than what is in the solder. If you conduct too much heat you can de-laminate a trace from the PCB.
Typically what I do is tack a couple of points with solder (<--it does not need to look pretty) ... Then I move at a steady pace with the iron across the pins while at the same time add a controlled rate of solder to the tip. Don't spend too much time on a pin or the heat will degrade the ability for the solder mask to repel, via surface tension, the solder and you will end up with a bridge. This method takes practice, but generally has good results.
I use a mini-hoof cartridge with my Metcal system for drag-soldering fine-pitch devices. It works very well. The hardest part is positioning the chip and tacking down two opposite corners.
I've soldered the SSD1928 chip in LQFP-128 by hand before. It is very difficult, I'd say, for me anyway.
Like Leon said, the first part is getting it pinned down.
The way I did the rest was to put down a lot of solder to make sure every pin was soldered.
Then, go back with solder wick and suck up all the excess.
But, eventually, I got at stencil reflow oven and that made it much, much easier.
Also, having a microscope or something like it helps a lot.
I would put some flux on the pcb, then use some very fine solder (I like radio shack very fine silver solder) on the traces with a solder pencil. Then lay the chip on top of the pads and use air/oven. I would not risk using a pencil on it if you have not practiced that a lot, prolonged heat can damage easily.
I was suprised how simple soldering is using solder past. Bring paste to the solder pad, heat it up and go. But it should be easy to make some exercise with standard adapter pcb's and cheap chipd of the same pitch. As Beau said: surface tention is your friend. And solderball, that do not merge, are insulated!
But, eventually, I got at stencil reflow oven and that made it much, much easier.
Also, having a microscope or something like it helps a lot.
Sounds like good advice.
Another approach we have used, for 'proof pcb is ok' step, is to pre-tin all the PCB pads and pre-tin the underside of the TQFP.
Then, using a very clean iron, you touch-reflow each pin, or pairs of pins.
Longer PCB PADs can help a little here.
This takes longer, but has good yields.
Connection can be checked visually with a microscope, or using a scalpel blade to try to move the lead sideways.
Once the PCB is proven, then a stencil + reflow would make sense.
I use ExpressPCB quite a bit. They are a little heavy-handed with the tinning. I've found that with a little flux, the 'tin' already on the board will bond these fine pitches very reliably. I use a microscope to line them up, and then apply enough pressure to ensure all the leads will touch once the solder melts.
I think that is harder with these very fine pitch chips...
Hard to get the tip of an iron on just one of these tiny pins.
Maybe under microscope.
Easy to bend pins if you mess up though...
I just remembered that the six pins at the beginning and end of each row are no-connects. That should make this easier, as the solder blob can be left there.
I Just did the GND/VDD continuity checks on the packaged test chips and, indeed, the pin 1 indicator is correct, even though the text is rotated by 90 degrees.
Hope you have a steady hand ... soldering that fine pitch can be tricky, even with a stencil and re-flow oven. You might need all 4 of those PCB boards just to get one good one out of the bunch. ...or at least past a slight learning curve. We made a Compact Flash adapter card with the same pitch I believe of 50 pins at .5mm and even with a microscope and a soldering iron it was a challenge to make sure there weren't any solder bridges.
Understanding Solder surface tension is your friend ... Sometimes counter intuitively using more solder is better than using less solder in a situation like this.
It won't be too hard to solder down. I've done chips that tight individually soldering (for the practice) the pins or drag soldering, (for speed) both work for the initial tack down. For the inevitable solder bridges, flux and solder-wick cleans them up quickly and neatly. More than once, I've just blobbed a chip down bridging most pins then cleaned up with solder wick. The solder joints come out disgustingly perfect and even. A good temperature controlled Iron helps, but it mostly just buys you more time with the solder melted before PCB pads start to delaminate. (And something like the Metcal I use at work will buy you minutes of working time) Finally, the sparkfun turtorials are wonderful and what got me started on surface mout => https://www.sparkfun.com/tutorials/category/2
When I get back home, I'll plug it into the FPGA board, but we did some cursory checks at Parallax.
Here's the 20MHz internal RC oscillator (mode $00):
Here's the 20KHZ internal RC oscillator (mode $01):
These were designed so that fixed capacitors and resistors dominate the RC period. They both have low temperature coefficients. You can see how close they are in frequency to the design goal.
Tonight, I hope to get this test chip intertwined with the the Prop123 FPGA board.
Comments
Can't wait to play with the P2
That was the leadframe vendor's document. We just used their leadframe.
This test board will only work on the -A9, since it has all 3.3V I/O's on the header connectors. The -A7 board's right-side signals are all 2.5V. We need 3.3V, all around.
This board provides a SPI flash chip, some pluggable pull-up resistors that can be connected to the header pins, four normal I/O's, and TWO full analog I/O pins. It takes 16 I/O's to talk to each pin, plus 9 more for the four fuses in each pin. The other header signals are mainly eaten up by the XTAL/PLL/OSC circuit and TESn and RESn signals. This is going to let us fully use the Prop2 padframe circuitry to act like a real Prop2. It will be neat to run off the real clock circuit. Man, I hope this all works.
Here are some pics of the PCB.
It makes me a little nervous that the test chip's package has the pin-1 indicator and the ID text 90 degrees off from each other.
1.8V regulator looks pretty massive... I've been using the 1117 versions lately. I think it's smaller...
Fingers crossed for success
Good idea. Yes, there is plenty of asymmetry in the GND pins which can be checked. I'll do that today.
Haha. Don't we all.
Isn't one of the corner chamfers larger than the other? Hard to tell from the pictures.
Sandy
Its a minor point, but since we're testing analog - you have a ground loop caused by that track underneath and two ground entry points along the left hand header. What i'd do is make your top left the single ground entry point, and put a test pin before the ground goes to the 1v8 regulator, for connecting cro's and logic analyzers etc. Connect both ground pins at top left (technically a ground loop but only 0.1"). Disconnect the ground connection at bottom left of the header, or break it with a 0.1" jumper if you really think you might need it for current purposes. Break the ground connection where it leaves the regulator tab, then route a solid ground from the test point at top left to underneath the chip, and connect to the existing ground networks using several vias.
Hope this makes sense.
Hope you have a steady hand ... soldering that fine pitch can be tricky, even with a stencil and re-flow oven. You might need all 4 of those PCB boards just to get one good one out of the bunch. ...or at least past a slight learning curve. We made a Compact Flash adapter card with the same pitch I believe of 50 pins at .5mm and even with a microscope and a soldering iron it was a challenge to make sure there weren't any solder bridges.
Understanding Solder surface tension is your friend ... Sometimes counter intuitively using more solder is better than using less solder in a situation like this.
Good to hear from you, Beau!
I don't know how hard this will be, but I was thinking about using lots of liquid flux and a ball of solder on the iron tip. One good pass per side with a little desoldering on the last 2-3 pins will hopefully do the job.
Yes. I used their free program. I started realizing it was just as easy to design the board as it was to explain to somebody else what I needed.
Good point. When we test the ADC, that will probably make a difference.
"...I don't know how hard this will be, but I was thinking about using lots of liquid flux..." ...I wouldn't use any more flux than what is in the solder. If you conduct too much heat you can de-laminate a trace from the PCB.
Typically what I do is tack a couple of points with solder (<--it does not need to look pretty) ... Then I move at a steady pace with the iron across the pins while at the same time add a controlled rate of solder to the tip. Don't spend too much time on a pin or the heat will degrade the ability for the solder mask to repel, via surface tension, the solder and you will end up with a bridge. This method takes practice, but generally has good results.
Like Leon said, the first part is getting it pinned down.
The way I did the rest was to put down a lot of solder to make sure every pin was soldered.
Then, go back with solder wick and suck up all the excess.
But, eventually, I got at stencil reflow oven and that made it much, much easier.
Also, having a microscope or something like it helps a lot.
Sounds like good advice.
Another approach we have used, for 'proof pcb is ok' step, is to pre-tin all the PCB pads and pre-tin the underside of the TQFP.
Then, using a very clean iron, you touch-reflow each pin, or pairs of pins.
Longer PCB PADs can help a little here.
This takes longer, but has good yields.
Connection can be checked visually with a microscope, or using a scalpel blade to try to move the lead sideways.
Once the PCB is proven, then a stencil + reflow would make sense.
Hard to get the tip of an iron on just one of these tiny pins.
Maybe under microscope.
Easy to bend pins if you mess up though...
It won't be too hard to solder down. I've done chips that tight individually soldering (for the practice) the pins or drag soldering, (for speed) both work for the initial tack down. For the inevitable solder bridges, flux and solder-wick cleans them up quickly and neatly. More than once, I've just blobbed a chip down bridging most pins then cleaned up with solder wick. The solder joints come out disgustingly perfect and even. A good temperature controlled Iron helps, but it mostly just buys you more time with the solder melted before PCB pads start to delaminate. (And something like the Metcal I use at work will buy you minutes of working time) Finally, the sparkfun turtorials are wonderful and what got me started on surface mout => https://www.sparkfun.com/tutorials/category/2
LOL.........I laugh for the weirdest reasons, sometimes.
When I get back home, I'll plug it into the FPGA board, but we did some cursory checks at Parallax.
Here's the 20MHz internal RC oscillator (mode $00):
Here's the 20KHZ internal RC oscillator (mode $01):
These were designed so that fixed capacitors and resistors dominate the RC period. They both have low temperature coefficients. You can see how close they are in frequency to the design goal.
Tonight, I hope to get this test chip intertwined with the the Prop123 FPGA board.