DSPRelated.com
Forums

[OT] I'm Available for Work

Started by Tim Wescott February 1, 2016
On 2/5/2016 6:04 PM, Paul wrote:
> In article <n92oic$8mj$1@dont-email.me>, gnuarm@gmail.com says... >> >> On 2/5/2016 12:52 PM, Paul wrote: >>> In article <n8vnmb$llf$1@dont-email.me>, gnuarm@gmail.com says... >>>> >>>> On 2/4/2016 5:01 AM, Paul wrote: >>>>> In article <thr2bbdtvrtf39kqkfvrvo89iafhnobm5b@4ax.com>, >>>>> jjlarkin@highlandtechnology.com says... >>>>>> >>> ..... >>>>>> I've noticed lately that people are harassing us, practically begging >>>>>> us, to do analog stuff. But too many are small-scale physics fads, >>>>>> which are unlikely to be enough volume to be profitable. >>>>> >>>>> Get several silly requests one of the silliest that did not want >>>>> redesign but just layout the PCB had >>>>> >>>>> 1/ Bluetooth chip in 6 x 6 mm QFP (had to be QFP) with NO tracks under >>>>> the device (data sheet requirement to avoid interference) >>>>> >>>>> 2/ PCB had to be 8 x 40 mm (remember that 6mm wide QFP) >>>>> >>>>> 3/ several wire points for power >>>>> >>>>> 4/ somewhere for the antenna >>>>> >>>>> 5/ Several other devices that really needed to be aligned to get >>>>> magnetic/gyro/compass axis aligned >>>>> >>>>> 6/ Best of all using a latching Hall effect switch device and its >>>>> Open Drain output to drive the 3V3 rail for all devices. >>>>> >>>>> As NO redeign or change of packages were permitted I said No Quote >>>>> could just see the blame trail game on that one a mile off. >>>>> >>>>> I have seen too many that want to break the lawa of physics >>>> >>>> You didn't explain any of the difficulties to them, you just washed your >>>> hands of it? >>> >>> Repeatedly but they kept saying it would fit and work, so I said NO >>> quote. >>> >>> So as not able to change package or do any other redesign I did not want >>> to be caught in blame trail etc >> >> Sounds like a tough one. I don't typically do just PCB layout, but I do >> like a challenge. Wish I could have had a crack at that. > > OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to > pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly > 3 x 3 mmm underneath the chip must NOT be used must be ground plane. > This is to avoid edges on the Digital signals interfering with the die, > its ground and the bluetooth transmitter. > > Also once you have left at minimum a 0.5 mm gap between copper and PCB > edge you now have a MAXIMUM of 7mm PCB width to fit it in.
Still not sufficient info to analyze the problem. How many of the pins are needed? Which directions? I'm not saying I can do magic. I'm saying I would discuss this with the customer to see what they had in mind. Clearly they either aren't giving you the full picture and they have something in mind or they have no idea what they are doing and it is impossible. I wasn't in any of the conversations so I can't say which. I'm just saying I like tough problems and exploring alternatives. BTW, what is the basis of the 0.5 mm gap around the perimeter? I know there are rules of thumb, but it is important to know the reasons for them so you can tell if they apply or not. -- Rick
On 2016-02-05, Paul <paul@pcserviceselectronics.co.uk> wrote:

> OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to > pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly > 3 x 3 mmm underneath the chip must NOT be used must be ground plane.
"must be ground plane"? That's completely different to "must be empty". put all the traces for the other pins on the back of the board (obscured by the ground plane.) -- \_(&#12484;)_
In article <n93cpa$oml$1@dont-email.me>, gnuarm@gmail.com says...
> > On 2/5/2016 6:04 PM, Paul wrote: > > In article <n92oic$8mj$1@dont-email.me>, gnuarm@gmail.com says... > >> > >> On 2/5/2016 12:52 PM, Paul wrote: > >>> In article <n8vnmb$llf$1@dont-email.me>, gnuarm@gmail.com says... > >>>> > >>>> On 2/4/2016 5:01 AM, Paul wrote: > >>>>> In article <thr2bbdtvrtf39kqkfvrvo89iafhnobm5b@4ax.com>, > >>>>> jjlarkin@highlandtechnology.com says... > >>>>>> > >>> ..... > >>>>>> I've noticed lately that people are harassing us, practically begging > >>>>>> us, to do analog stuff. But too many are small-scale physics fads, > >>>>>> which are unlikely to be enough volume to be profitable. > >>>>> > >>>>> Get several silly requests one of the silliest that did not want > >>>>> redesign but just layout the PCB had > >>>>> > >>>>> 1/ Bluetooth chip in 6 x 6 mm QFP (had to be QFP) with NO tracks under > >>>>> the device (data sheet requirement to avoid interference) > >>>>> > >>>>> 2/ PCB had to be 8 x 40 mm (remember that 6mm wide QFP) > >>>>> > >>>>> 3/ several wire points for power > >>>>> > >>>>> 4/ somewhere for the antenna > >>>>> > >>>>> 5/ Several other devices that really needed to be aligned to get > >>>>> magnetic/gyro/compass axis aligned > >>>>> > >>>>> 6/ Best of all using a latching Hall effect switch device and its > >>>>> Open Drain output to drive the 3V3 rail for all devices. > >>>>> > >>>>> As NO redeign or change of packages were permitted I said No Quote > >>>>> could just see the blame trail game on that one a mile off. > >>>>> > >>>>> I have seen too many that want to break the lawa of physics > >>>> > >>>> You didn't explain any of the difficulties to them, you just washed your > >>>> hands of it? > >>> > >>> Repeatedly but they kept saying it would fit and work, so I said NO > >>> quote. > >>> > >>> So as not able to change package or do any other redesign I did not want > >>> to be caught in blame trail etc > >> > >> Sounds like a tough one. I don't typically do just PCB layout, but I do > >> like a challenge. Wish I could have had a crack at that. > > > > OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to > > pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly > > 3 x 3 mmm underneath the chip must NOT be used must be ground plane. > > This is to avoid edges on the Digital signals interfering with the die, > > its ground and the bluetooth transmitter. > > > > Also once you have left at minimum a 0.5 mm gap between copper and PCB > > edge you now have a MAXIMUM of 7mm PCB width to fit it in. > > Still not sufficient info to analyze the problem. How many of the pins > are needed? Which directions?
All pins in use, I had the circuit they wanted laid out. Just get a footptint including pad sizes and sketch it out to see what room you have.
> I'm not saying I can do magic. I'm saying I would discuss this with the > customer to see what they had in mind. Clearly they either aren't > giving you the full picture and they have something in mind or they have > no idea what they are doing and it is impossible. I wasn't in any of > the conversations so I can't say which. > > I'm just saying I like tough problems and exploring alternatives.
You have 47 out of 48 pins in use just sketch it out with pin pitch of 0.4mm see how little space you have
> BTW, what is the basis of the 0.5 mm gap around the perimeter? I know > there are rules of thumb, but it is important to know the reasons for > them so you can tell if they apply or not.
Pretty standard PCB manufacture guideline of avoiding copper at edge of board for when final PCB is cut down to final size to avoid either breaking a track at edge or shorting between layers. Often this is 1 to 2 mm gap of copper to outline. Look at any PCB you have to hand even with a ground plane it does not go to actual edge of PCB. -- Paul Carpenter | paul@pcserviceselectronics.co.uk <http://www.pcserviceselectronics.co.uk/> PC Services <http://www.pcserviceselectronics.co.uk/pi/> Raspberry Pi Add-ons <http://www.pcserviceselectronics.co.uk/fonts/> Timing Diagram Font <http://www.badweb.org.uk/> For those web sites you hate
In article <n93vam$8hr$2@gonzo.alcatraz>, jasen@xnet.co.nz says...
> > On 2016-02-05, Paul <paul@pcserviceselectronics.co.uk> wrote: > > > OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to > > pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly > > 3 x 3 mmm underneath the chip must NOT be used must be ground plane. > > "must be ground plane"? That's completely different to "must be empty". > > put all the traces for the other pins on the back of the board > (obscured by the ground plane.)
Then on double sided board for 0.4mm pitch work out where you fit the vias on the board edge pins and what size they and the tracks must be to avoid shorts, maintain etching gaps/clearances and what process will be required to manufacture it. Remember with pads to solder to and effectively MAX of 7mm available PCB width, then 3mm taken up with central ground plane area how you fit on each board edge side 12 vias to take tracks to other side from a QFP footprint with 47 out of 48 pins used. You will need to maintain some form of track/via/pad minimum clearance. Due to edges with main power and bluetooth pins really not wanting to be on board edges you then have the crystal and other I/O mainly digital down the board edges. Bluetooth really needs to be on non-board edge orientation for shaping and antenna filters with min track length and isolation from other areas so PCB antenna is not in middle of everything else. -- Paul Carpenter | paul@pcserviceselectronics.co.uk <http://www.pcserviceselectronics.co.uk/> PC Services <http://www.pcserviceselectronics.co.uk/pi/> Raspberry Pi Add-ons <http://www.pcserviceselectronics.co.uk/fonts/> Timing Diagram Font <http://www.badweb.org.uk/> For those web sites you hate
On 2/7/2016 6:28 AM, Paul wrote:
> In article <n93cpa$oml$1@dont-email.me>, gnuarm@gmail.com says... >> >> On 2/5/2016 6:04 PM, Paul wrote: >>> In article <n92oic$8mj$1@dont-email.me>, gnuarm@gmail.com says... >>>> >>>> On 2/5/2016 12:52 PM, Paul wrote: >>>>> In article <n8vnmb$llf$1@dont-email.me>, gnuarm@gmail.com says... >>>>>> >>>>>> On 2/4/2016 5:01 AM, Paul wrote: >>>>>>> In article <thr2bbdtvrtf39kqkfvrvo89iafhnobm5b@4ax.com>, >>>>>>> jjlarkin@highlandtechnology.com says... >>>>>>>> >>>>> ..... >>>>>>>> I've noticed lately that people are harassing us, practically begging >>>>>>>> us, to do analog stuff. But too many are small-scale physics fads, >>>>>>>> which are unlikely to be enough volume to be profitable. >>>>>>> >>>>>>> Get several silly requests one of the silliest that did not want >>>>>>> redesign but just layout the PCB had >>>>>>> >>>>>>> 1/ Bluetooth chip in 6 x 6 mm QFP (had to be QFP) with NO tracks under >>>>>>> the device (data sheet requirement to avoid interference) >>>>>>> >>>>>>> 2/ PCB had to be 8 x 40 mm (remember that 6mm wide QFP) >>>>>>> >>>>>>> 3/ several wire points for power >>>>>>> >>>>>>> 4/ somewhere for the antenna >>>>>>> >>>>>>> 5/ Several other devices that really needed to be aligned to get >>>>>>> magnetic/gyro/compass axis aligned >>>>>>> >>>>>>> 6/ Best of all using a latching Hall effect switch device and its >>>>>>> Open Drain output to drive the 3V3 rail for all devices. >>>>>>> >>>>>>> As NO redeign or change of packages were permitted I said No Quote >>>>>>> could just see the blame trail game on that one a mile off. >>>>>>> >>>>>>> I have seen too many that want to break the lawa of physics >>>>>> >>>>>> You didn't explain any of the difficulties to them, you just washed your >>>>>> hands of it? >>>>> >>>>> Repeatedly but they kept saying it would fit and work, so I said NO >>>>> quote. >>>>> >>>>> So as not able to change package or do any other redesign I did not want >>>>> to be caught in blame trail etc >>>> >>>> Sounds like a tough one. I don't typically do just PCB layout, but I do >>>> like a challenge. Wish I could have had a crack at that. >>> >>> OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to >>> pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly >>> 3 x 3 mmm underneath the chip must NOT be used must be ground plane. >>> This is to avoid edges on the Digital signals interfering with the die, >>> its ground and the bluetooth transmitter. >>> >>> Also once you have left at minimum a 0.5 mm gap between copper and PCB >>> edge you now have a MAXIMUM of 7mm PCB width to fit it in. >> >> Still not sufficient info to analyze the problem. How many of the pins >> are needed? Which directions? > > All pins in use, I had the circuit they wanted laid out. Just get a > footptint including pad sizes and sketch it out to see what room you > have. > >> I'm not saying I can do magic. I'm saying I would discuss this with the >> customer to see what they had in mind. Clearly they either aren't >> giving you the full picture and they have something in mind or they have >> no idea what they are doing and it is impossible. I wasn't in any of >> the conversations so I can't say which. >> >> I'm just saying I like tough problems and exploring alternatives. > > You have 47 out of 48 pins in use just sketch it out with pin pitch of > 0.4mm see how little space you have > >> BTW, what is the basis of the 0.5 mm gap around the perimeter? I know >> there are rules of thumb, but it is important to know the reasons for >> them so you can tell if they apply or not. > > Pretty standard PCB manufacture guideline of avoiding copper at edge of > board for when final PCB is cut down to final size to avoid either > breaking a track at edge or shorting between layers. Often this is 1 to > 2 mm gap of copper to outline. Look at any PCB you have to hand even > with a ground plane it does not go to actual edge of PCB.
My production boards have traces that are cut by the scoring. Obviously the fab house doesn't have a problem with the board being designed this way. -- Rick
On 2/7/2016 6:39 AM, Paul wrote:
> In article <n93vam$8hr$2@gonzo.alcatraz>, jasen@xnet.co.nz says... >> >> On 2016-02-05, Paul <paul@pcserviceselectronics.co.uk> wrote: >> >>> OK explain to me how you would fit a 48 pin QFP with 6 x 6 mm pin tip to >>> pin tip dimension on DOUBLE sided board 8mm wide. When an area roughly >>> 3 x 3 mmm underneath the chip must NOT be used must be ground plane. >> >> "must be ground plane"? That's completely different to "must be empty". >> >> put all the traces for the other pins on the back of the board >> (obscured by the ground plane.) > > Then on double sided board for 0.4mm pitch work out where you fit the > vias on the board edge pins and what size they and the tracks must be to > avoid shorts, maintain etching gaps/clearances and what process will be > required to manufacture it. > > Remember with pads to solder to and effectively MAX of 7mm available > PCB width, then 3mm taken up with central ground plane area how you > fit on each board edge side 12 vias to take tracks to other side from a > QFP footprint with 47 out of 48 pins used. > > You will need to maintain some form of track/via/pad minimum clearance. > > Due to edges with main power and bluetooth pins really not wanting to be > on board edges you then have the crystal and other I/O mainly digital > down the board edges. > > Bluetooth really needs to be on non-board edge orientation for shaping > and antenna filters with min track length and isolation from other areas > so PCB antenna is not in middle of everything else.
I've dug into this and I don't see the vias being the impossible aspect of this design. I I found one PCB vendor who will make a 12 mil pad (0.3 mm) with a 4 mil hole (0.1 mm) which will fit with 4 mil space along a row of 0.4 mm pads. This fits within the 1 mm chip to edge dimension allowing some extra space for pad extension (can't evaluate that until I know more about the package) and edge clearance. Hole to edge is a concern, but with such small holes it may not be an issue. Via in pad is another option but I'm not sure would offer much advantage over the vias along the edge other than providing more hole to board edge clearance. A second vendor has the capability to make the vias, but their auto quote page didn't allow boards smaller than 12 mm in either direction. They might make it if I ask for a custom quote. This was after spending 10 minutes searching on the web for PCB fab capabilities. I certainly would not bid this as a standard design. I would likely bump the price by a factor of three (or more) to cover the extra work involved in making it all fit. But I'm sure you saw more than what you've describe here. I'm just addressing the issue of the 48QFP routing. -- Rick