Got a circuit where in hindsight the client decided they need >10 times the juice than initially spec'd. This was truly unforeseen for them and we have to make it work at least for a proto series. Of course, now all those 0805 resistors in the power path would exceed their usual 1/8W spec. Found these at 1/3W:
I plan to run them around 200-250mW. Was expecting those to be really tall to vent off some heat but they are surprisingly flat, only 0.4mm high. Will this become a heat problem on FR-4 for stuff that can run 24/7? Any experience?
<notthisjoerg...@removethispacbell.net> wrote: >Hello All,
>Got a circuit where in hindsight the client decided they need >10 times >the juice than initially spec'd. This was truly unforeseen for them and >we have to make it work at least for a proto series. Of course, now all >those 0805 resistors in the power path would exceed their usual 1/8W >spec. Found these at 1/3W:
>I plan to run them around 200-250mW. Was expecting those to be really >tall to vent off some heat but they are surprisingly flat, only 0.4mm >high. Will this become a heat problem on FR-4 for stuff that can run >24/7? Any experience?
Sounds OK to me. We push regular 0805's harder than that. Zero fails so far.
We experimented with various resistors, 0603 through 1206, with a thermal imager that has plenty of resolution to find the hot-spot temp on all the parts.
Turns out that the actual part thetas are astronomical, and what cools resistors is the copper pads and traces, more than the bodies themselves. An 0603 soldered to big copper pours can dissipate as much as a 1206 ditto; the hot spot temps are the same. That makes sense physically, as theta is determined by the L/W ratio and the substrate thickness, and all the resistors we checked were all 20 mil thick alumina.
A better substrate, BeO or AlN, would help a bit, at great expense.
If there's a ground plane close, that helps a lot.
The easiest thing to do is fire up the board and scope it with a Flir, and see how hot stuff gets. You can borrow mine, or come here and use it.
John Larkin wrote: > On Wed, 23 Jul 2008 09:59:26 -0700, Joerg > <notthisjoerg...@removethispacbell.net> wrote:
>> Hello All,
>> Got a circuit where in hindsight the client decided they need >10 times >> the juice than initially spec'd. This was truly unforeseen for them and >> we have to make it work at least for a proto series. Of course, now all >> those 0805 resistors in the power path would exceed their usual 1/8W >> spec. Found these at 1/3W:
>> I plan to run them around 200-250mW. Was expecting those to be really >> tall to vent off some heat but they are surprisingly flat, only 0.4mm >> high. Will this become a heat problem on FR-4 for stuff that can run >> 24/7? Any experience?
> Sounds OK to me. We push regular 0805's harder than that. Zero fails > so far.
> We experimented with various resistors, 0603 through 1206, with a > thermal imager that has plenty of resolution to find the hot-spot temp > on all the parts.
> Turns out that the actual part thetas are astronomical, and what cools > resistors is the copper pads and traces, more than the bodies > themselves. An 0603 soldered to big copper pours can dissipate as much > as a 1206 ditto; the hot spot temps are the same. That makes sense > physically, as theta is determined by the L/W ratio and the substrate > thickness, and all the resistors we checked were all 20 mil thick > alumina.
> A better substrate, BeO or AlN, would help a bit, at great expense.
> If there's a ground plane close, that helps a lot.
> The easiest thing to do is fire up the board and scope it with a Flir, > and see how hot stuff gets. You can borrow mine, or come here and use > it.
Thanks, John. I think we are ok then up to 250mW. At least the resistors are rated for that. I've got to visit you guys some day and look through the FLIR. And maybe empty a pitcher at Zeitgeist.
Another peculiarity that surprised me was that the TO18 was stated at 83.3C/W to case by ST and the TO92 at the same by ON. Junction to ambient was stated lower (!) for the TO92 even though it's plastic. Strange. I haven't used TO packages in roughly two decades but now I have to, for the proto run. I thought TO18 would have to be better but it carries the collector and then some day it touches something ... bzzzt ... POOF.
>>Got a circuit where in hindsight the client decided they need >10 times >>the juice than initially spec'd. This was truly unforeseen for them and >>we have to make it work at least for a proto series. Of course, now all >>those 0805 resistors in the power path would exceed their usual 1/8W >>spec. Found these at 1/3W:
>>I plan to run them around 200-250mW. Was expecting those to be really >>tall to vent off some heat but they are surprisingly flat, only 0.4mm >>high. Will this become a heat problem on FR-4 for stuff that can run >>24/7? Any experience?
> Sounds OK to me. We push regular 0805's harder than that. Zero fails > so far.
> We experimented with various resistors, 0603 through 1206, with a > thermal imager that has plenty of resolution to find the hot-spot temp > on all the parts.
> Turns out that the actual part thetas are astronomical, and what cools > resistors is the copper pads and traces, more than the bodies > themselves. An 0603 soldered to big copper pours can dissipate as much > as a 1206 ditto; the hot spot temps are the same. That makes sense > physically, as theta is determined by the L/W ratio and the substrate > thickness, and all the resistors we checked were all 20 mil thick > alumina.
> A better substrate, BeO or AlN, would help a bit, at great expense.
> If there's a ground plane close, that helps a lot.
> The easiest thing to do is fire up the board and scope it with a Flir, > and see how hot stuff gets. You can borrow mine, or come here and use > it.
Thermal vias are your friend. Vias in pads are now standard procedure for most PCB houses. Put a thermal via in each pad and one under the center of the part. Tie TVs to heat spreader on another plane. JL, tell us more of your FLIR, type, cost and performance. Cheers, Harry
>"John Larkin" <jjlar...@highNOTlandTHIStechnologyPART.com> wrote in message >news:n1qe84pvvg5bfomsm6uvu2e8a42mf35ouk@4ax.com... >> On Wed, 23 Jul 2008 09:59:26 -0700, Joerg >> <notthisjoerg...@removethispacbell.net> wrote:
>>>Hello All,
>>>Got a circuit where in hindsight the client decided they need >10 times >>>the juice than initially spec'd. This was truly unforeseen for them and >>>we have to make it work at least for a proto series. Of course, now all >>>those 0805 resistors in the power path would exceed their usual 1/8W >>>spec. Found these at 1/3W:
>>>I plan to run them around 200-250mW. Was expecting those to be really >>>tall to vent off some heat but they are surprisingly flat, only 0.4mm >>>high. Will this become a heat problem on FR-4 for stuff that can run >>>24/7? Any experience?
>> Sounds OK to me. We push regular 0805's harder than that. Zero fails >> so far.
>> We experimented with various resistors, 0603 through 1206, with a >> thermal imager that has plenty of resolution to find the hot-spot temp >> on all the parts.
>> Turns out that the actual part thetas are astronomical, and what cools >> resistors is the copper pads and traces, more than the bodies >> themselves. An 0603 soldered to big copper pours can dissipate as much >> as a 1206 ditto; the hot spot temps are the same. That makes sense >> physically, as theta is determined by the L/W ratio and the substrate >> thickness, and all the resistors we checked were all 20 mil thick >> alumina.
>> A better substrate, BeO or AlN, would help a bit, at great expense.
>> If there's a ground plane close, that helps a lot.
>> The easiest thing to do is fire up the board and scope it with a Flir, >> and see how hot stuff gets. You can borrow mine, or come here and use >> it.
> Thermal vias are your friend. Vias in pads are now standard procedure for >most PCB houses. Put a thermal via in each pad and one under the center of >the part. Tie TVs to heat spreader on another plane. > JL, tell us more of your FLIR, type, cost and performance. > Cheers, > Harry
It's a model E45 with the wide-angle (germanium!) lens. You can basically touch a part and still focus, so you can easily resolve the hot spot temp on an 0603 or a SOT-23. It's amazing... image a board and learn all sorts of stuff.
The PC interface, sadly, is insane. It's USB, but it's a network device, so you have to use their very klunky PC software. They should have made it work like all the other cameras in the world.
Cost about $10K, and worth it. I think some new technologies are coming on board, so prices will drop.
In that second pic I posted, we thought that the FPGA (under the blue heat sink) was getting too hot as clock frequency increased, so we were looking into clock gating and tedious stuff like that. A few minutes with the Flir showed that the dacs (the white-hot blobs) are in fact heating the FPGA.
On a sunny day (Wed, 23 Jul 2008 11:52:06 -0700) it happened John Larkin <jjlar...@highNOTlandTHIStechnologyPART.com> wrote in <urue84dk5os9eiihvr5d990o7lvhl7q...@4ax.com>:
>Cost about $10K, and worth it. I think some new technologies are >coming on board, so prices will drop.
>In that second pic I posted, we thought that the FPGA (under the blue >heat sink) was getting too hot as clock frequency increased, so we >were looking into clock gating and tedious stuff like that. A few >minutes with the Flir showed that the dacs (the white-hot blobs) are >in fact heating the FPGA.
>John
Xcuse me, but would not a finger test show that too? That teh DACS were much hotter then the FPGA? I am not against modern tech, but 10k is a lot of money for something you can also do with a temp probe or your finger. ;-)
<pNaonStpealm...@yahoo.com> wrote: >On a sunny day (Wed, 23 Jul 2008 11:52:06 -0700) it happened John Larkin ><jjlar...@highNOTlandTHIStechnologyPART.com> wrote in ><urue84dk5os9eiihvr5d990o7lvhl7q...@4ax.com>:
>>Cost about $10K, and worth it. I think some new technologies are >>coming on board, so prices will drop.
>>In that second pic I posted, we thought that the FPGA (under the blue >>heat sink) was getting too hot as clock frequency increased, so we >>were looking into clock gating and tedious stuff like that. A few >>minutes with the Flir showed that the dacs (the white-hot blobs) are >>in fact heating the FPGA.
>>John >Xcuse me, but would not a finger test show that too? That teh DACS were much >hotter then the FPGA? >I am not against modern tech, but 10k is a lot of money for something you >can also do with a temp probe or your finger. >;-)
Consider a series of images, of the whole board and of regions, taken at different clock rates. That tells us, very quickly, where the power is going, and sometimes it's unexpected. It's very hard to quantify, or even feel, the temperature of an 0603 resistor or a SOT-23 opamp, and one board can have hundreds. If you stick a thermocouple to a small part, you suck all the heat away. And the simple IR thermometers have rotten spatial resolution.
One of our VME boards can sell for more than the Flir cost. It's been a great tool, sort of an instant thermal design review.
It's also cool for troubleshooting. Suppose a power rail is collapsed. You can often image the board and find the part that's hogging the power. Or add another power supply, force a few amps, and image the traces.
And with the tax considerations, the net price is about half.
>>On a sunny day (Wed, 23 Jul 2008 11:52:06 -0700) it happened John Larkin >><jjlar...@highNOTlandTHIStechnologyPART.com> wrote in >><urue84dk5os9eiihvr5d990o7lvhl7q...@4ax.com>:
>>>Cost about $10K, and worth it. I think some new technologies are >>>coming on board, so prices will drop.
>>>In that second pic I posted, we thought that the FPGA (under the blue >>>heat sink) was getting too hot as clock frequency increased, so we >>>were looking into clock gating and tedious stuff like that. A few >>>minutes with the Flir showed that the dacs (the white-hot blobs) are >>>in fact heating the FPGA.
>>>John >>Xcuse me, but would not a finger test show that too? That teh DACS were much >>hotter then the FPGA? >>I am not against modern tech, but 10k is a lot of money for something you >>can also do with a temp probe or your finger. >>;-)
>Consider a series of images, of the whole board and of regions, taken >at different clock rates. That tells us, very quickly, where the power >is going, and sometimes it's unexpected. It's very hard to quantify, >or even feel, the temperature of an 0603 resistor or a SOT-23 opamp, >and one board can have hundreds. If you stick a thermocouple to a >small part, you suck all the heat away. And the simple IR thermometers >have rotten spatial resolution.
>One of our VME boards can sell for more than the Flir cost. It's been >a great tool, sort of an instant thermal design review.
>It's also cool for troubleshooting. Suppose a power rail is collapsed. >You can often image the board and find the part that's hogging the >power. Or add another power supply, force a few amps, and image the >traces.
>And with the tax considerations, the net price is about half.
>John
Some time ago I posted a little auto-zeroing microvolt meter that I used at GenRad to trace drops on PCB traces, to locate shorts.
...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 |
> It's also cool for troubleshooting. Suppose a power rail is collapsed. > You can often image the board and find the part that's hogging the > power. Or add another power supply, force a few amps, and image the > traces.
I just donated it to a client but in the days when that still happened a lot I used a >10A power supply for that. Crank it up ... *WHADDABAM* ... aha, t'was that high-faluting PWM driver again. Only with eye protection, of course. Sometimes all it takes is a D-cell.
>> It's also cool for troubleshooting. Suppose a power rail is collapsed. >> You can often image the board and find the part that's hogging the >> power. Or add another power supply, force a few amps, and image the >> traces.
>I just donated it to a client but in the days when that still happened a >lot I used a >10A power supply for that. Crank it up ... *WHADDABAM* ... >aha, t'was that high-faluting PWM driver again. Only with eye >protection, of course. Sometimes all it takes is a D-cell.
>[...]
I used to "cure" pin-hole shorts in alumina thick-film hybrids with a 5V/100A supply... rarely did anything more than just go "click" ;-)
