FYI, I had to shoot half this video again because I didn't plug the wireless mic cable all the way into the camera and it made intermittent contact half way through. Given that I don't have any sort of script it came out very different the 2nd time around. Not sure which one would have been better. But those who follow me on Twitter already know this.
Oh, gawd. 30 years ago I had an instrument/sensor that normally ran cool, but under specific conditions would run hot as hell. And we couldn't add a heatsink or vias to ground. So we stacked a Peltier cooler between it and the case, and simply dumped power into it when things got hot. Yes, the power supplies were oversized...
That's making me wonder what the thermal conductivity of the standard aluminium nitride ceramic SMD resistors is. Could a high value resistor be used as a cheap heat conductor?
I was thinking about doing an estimate based on something like the power derating curves, but that will only give the (overall) component to ambient value. Also, using it in a different setup the mode of heat transfer could get different and ruïn things ....I'm not sure about the last bit, but I think that you should just try it! . Don't forget to check that the case of the SMD resistor. Also, make sure that the resistor's enclosure doesn't create shorts...
Aluminum nitride thermal conductivity 320W/mK. Aluminum oxide 30W/mK. So about 10 times worse. That said, resistors cost at least ~20x less than these, so you could pepper your board with 0603's.
@@jaro6985 320W/m.K is for a single crystal which is expensive, but for an electrically insulating polycrystalline ceramic it is anywhere from 17-285 W/m.K (ref: www.scielo.br/scielo.php?script=sci_arttext&pid=S0366-69132004000300012 see table 1 for physical properties of AlN, SiC, BeO and Al2O3 - resistivity etc.)
I too would love to see a deep dive on thermal design! Maybe if enough of a new wave of join up to support on Patreon or similar Dave will do a proper video deep dive? :-)
We ran into a situation where these were a great option when designing a cubesat. There are very sensitive magnetometers on board to help determine the satellite's orientation, and current loops throughout the bus would cause pretty significant interference. Because of this, the aluminum structure could not be electrically grounded. Our RF amp was pumping heat into the ground plane and we needed to dump it into the structure while being electrically isolated. SMD thermal jumpers came to the rescue! Instead of trying to mount a heatsink directly to the IC, we could expose a region on the edge of the RF board which was thermally coupled to the structure. SMD thermal jumpers allowed heat to travel from the ground plane to this exposed edge.
I could see these being super useful for drone motor drivers. Super tiny, but so much power. Also for current sense resistors, given the temperature coefficient being directly related to current reading error.
They usualy use aluminium heatsinks for a reason, most of their PCBs is usualy covered with MOSFETs. The board itself couldn't take that much thermal load as it's pretty much saturated and the MOSFETs connect to the ground and power traces anyway.
No great help there, for drone ESCs you certainly want direct thermal compound on top and bottom. in plus you do not have enough space for redistribute with vias.
The board experiment would actually be an interesting video. In particular compare their thermal performance to standard resistors. Sure, I know that they are isolated, but a 10M SMD resistor is for many applications (particularly higher power ones) effectively open.
The ceramic base wouldn't be a great conductor of heat I'd imagine. Although you can heat up both pads of a resistor from the one end, so it's possible.
you can also get like 40 smd resistors for the price of one, so even if it's like 5-10:1 it might be worth to just have 10-20 100Mohm resistors around your part.
@@ProjektMacu on smd To220 and sot428 sized componants. sot428 is around 6*6 mm, so you have 18mm to lay your resistors on all 3 sides, which I'd reckon could fit 10 resistors though it depends on what size you use.
@@EEVblog SiC could be be good enough at ~150W/m.K and should be cheaper to produce since it does not require nitrogen gas at 100 and 500 atmospheres of pressure to produce. EDIT: Although the dielectric loss is 100x higher, maybe not. But still might be good enough for SMPS's. (ref: www.scielo.br/pdf/ce/v50n315/a1150315.pdf )
There is nothing wrong with beryllium oxide as long as you don't breath the dust or eat it. If you pulverize and inhale the resulting dust from your electronics, beryllium oxide is not for you! While you're at it, quit making and breathing asbestos dust. This world is nuts. "toxic" > Reeeeee! Reeeee!
yeah, I wonder how existing designs might be improved, because some pcb designers don't care (probably don't have funds and time to care) about thermals on some components.
"THEMAL LOL" Fun fact: There are rune inscriptions in Norway where the vikings added the typical "R↓" corrections above the misspelled word. Not many of course, but they do exist!
Honestly, I presume any reverse engineer worth their salt is actually going to measure the resistance (and capacitance for capacitors, even if they have to pull it out of circuit). Which would give the game away pretty quick.
Bonus points for putting GND on the outside layer, not only do you get thermal dissipation (especially with the thermal vias or GND stitching on the board edges) but now you've doubled it as a faraday cage. Win-win for thermals and FCC Part 15 compliance.
Very nice to present theories of practical problems that never discussed before like thermal design consideration in power electronics... Much better than commercial reviews ...
I seriously once routed in a little hole underneath a few times, then went on a thermal-pad straight to the case. Obviously that only works if you have the space underneath the component but it worked very well.
Though this looks like a game changer passive thermal solution, i am predicting an active thermal solution in similar smt form. They would use the peltier effect to both remove heat and act as a sensor/(power source for backup purposes). Their application area could be based on high performace, compact high current FET driving application.
Seems like these could be useful for on-board temperature measurement. Connect your temperature sensor (RTD, or whatever) to the pin you're trying to sense with one of these and you might have a pretty good measurement.
Thanks for bringing these components to our attention! I'm putting them in my next board design, as the are a perfect solution to heat sinking a bank of isolated SMD power transistors.
another reason to use these could also be wanting to reduce capacitance/size on some node, in that case you might want to go for the standard aspect ratio (eg 0402, 0603) ones. will certainly consider these in my next design
ah, should have watched to the end, you already mention that, still wanted to emphasize that these might be very useful even in situations where you could easily satisfy your thermal requirement by simply using a larger pad and aren't pressed for space for electrical performance / emc considerations
as i work on very low volume and expensive products i am not all that bothered about cost, having to go through an additional prototype cycle or even spending an extra week on a design is always a lot more expensive than having another couple of dollars on the bom over the entire lifespan of the product. i usually try to do everything i can think of to make a design work and pass emc tests on the first try. smd feedthru capacitors are also great btw :)
That's a fascinating product. We do a lot of designs for explosive atmosphere where it can be difficult to meet de-rating requirements due to all the applied safety factors. I am going to bring these up with my team.
8:39 - Also the copper on your top and bottom layers is usually thicker, so that helps with heat transfer. I've used the layer flip technique on some designs where heat or current carrying capacity was an issue.
Thanks Dave for this! Did not know about this and would be handy for a one off stepper motor controller I'm making. And yes, please do a video on actually reviewing these under various test conditions.
In addition to allowing you to tie live pins to the existing ground/power planes for thermal reasons, these jumpers would also allow you to group the thermal pads of multiple isolated components together to share a common heatsink for cost-saving (rather than a small dedicated heatsink for each), without needing any other isolation layers. Think stacked transistor layouts, multiple equivalent channels of a device, battery balancing circuits, etc...
I'm slogging through this stuff working on RF amplifiers. Sometimes, the package you want puts a transistor collector at RF potential (eek!). These are really cool... I'd love it if you would do an experimental video showing a real board and taking measurements of different arrangements!
Hi Dave interesting devices. For a Gamechanging we need a defined insulation. Vishay clames >1.5KV ac withstand voltage👍, but as you know, we need formal insulation approvals and ratings like CTI, dti, PD, working voltage and so on. Otherwise it may be not usable.
I used to work in the lab of an electronics company and the amount of thermal testing that was done was insane! On some low voltage but high power boards, six layer PCBs were used to soak away extra heat. Unfortunately it did make removing copper foil wound transformers a bugger to remove!
When you do a test, could you compare to a variety of regular ceramic capacitors of similar capacities and voltage rating? They could also have some thermal conductivity as well.
If you have many layers you can also play with planes to transfer heat through the FR4 substrate. I did this once where every second layer was connected to the heat source and every other to the GND-plane. The component was almost impossible to de solder because of this! :D
Interesting, but I think you could often do just as well with simple pcb layout techniques. FR-4 has a thermal conductivity of about 0.25 W/mK. So if you have a 10mm x 10mm land on the surface of the pcb, and the ground plane is just below it with 0.1mm of FR-4 in between, the thermal resistance from the land to the ground plane will be about 4 K/W. And a 0.1mm layer of FR-4 should be good for about 3kV (maybe 1.5kV with aging) isolation. So you can get pretty good thermal conduction without any thermal vias if you have enough room on the surface for a large land to spread the heat.
Ok we are currently doing a redesign of an older board that has a thermal sensor to monitor some of the hotter components on the board. Just in case something burns out and we need to shut it off. The current "solution" was to put the SMD thermistor near the hot spot and call it a day. It does work surprisingly well actually. While i was doing the redesign i remembered this video and put 2 thermal jumpers from the potentially hot components to the SMD thermistor. One jumper for each pad. I've yet to receive the new boards, but i believe they should work way better. Thank you so much for this Dave, my boss was practically jumping up and down when i told him what i had done. He was very pleased.
You know I watched this video intially as something 'oh cool, how informative', but as it reached minute 15 I was astonished at the results. This is definitely going to be a game changer. Thanks for the video again Dave!
"Dave what happened to your PSU project?" "David2 left and it's been in stasis ever since." Maybe release the project and let the community finish it, then run some group buys or straight put on your store/other distributors.
the problem with the PSU project is, that for it to become a product, it needs one vision. If you let the community at it, you will get conflicting visions and go nowhere. What would work is realising the hardware and a "driver packet" type firmware. That would also branch out, but quickly, a few branches would establish them self as "good for a specific purpose".
@@Arek_R. Who would 1st have the knowledge to do (correctly) what Dave's wanted 2nd Actually having and wanting to spare free time on an open source project thats not that "open" 3rd Do all of that (basically working completely for someone else) for free Those would be the major problems
@@Arek_R. Dave is the kind of guy, that keeps things in mind. Creating a symbol is not mindless busy work - it is preconceptuallizing how you plan on using the IC, reading the pin description, checking the specification against surrounding parts, taking notes for the layout phase... and then double checking if the b-variant of the part would not fit better. It might have a bigger footprint, but that simplifies the fan out the inductor. I'm tasked with mentoring a student currently. It is fun, but preparing the work for someone else is more work then when I would to that specific task. It is the unified vision and a gut feel of which optional parts are worth in context of other constrains that pop up during the design.
And here I am pissed off at work for having to reflow a board short on MSOP 10s with a beefy thermal plane. I gotta say Dave if it wasn't for me growing up with your videos I wouldn't have a job. Thanks man.
I think the best way to use this part would be to put it between the SMD land and a mounting screw that attaches the board to chassis. Probably could conduct a lot more heat than the ground plane.
Afro and all others are here also. I had ordered some for my last project. But they interfered with my design and would short some parts. At the end I used some vias for the LDO and for safety I thermalglued an old school alu fin block on top. Never heard any problems from the customer.
I was thinking it’s cool and all to have another way to get heat out on large package, but for some reason my mind was blown when you showed the picture where they wick heat out of a resistor from its pad. Thank you for sharing this!
Super. It is not that often that a new thing like this comes out, very interesting. Now, on the calculator thing, I wonder why you did not change the calc on the last scene...
Modeled thermal systems for years (not electronics, but mechanical and fluid systems). Videos like this make me appreciate just how complex electronic designs are. Who would have thought so much thermal design in that itty-bitty iphone. :) Oh, and kudos for the 'big fat wide jumpers' comment. Anyone that studies thermal conductivity of materials understands the need for that. Great video, interesting and shows me just how much I don't know (which seems to be more and more every day lol )
Using the short fat ones may not be viable sometimes because you still have to maintain the gap between conductors. The voltage breakdown is still a thing.
Tbh, although this is nice and dandy, nothing beats a quantitative thermal analysis with Finite-Element Method software. These are expensive as Dave suggests, but with minimal effort and some math you can do a nice DIY piece of software to thermally model your design with incredible accuracy.
@@EEVblog that's the thing, there are a lot of ready-to-use open-source libraries and programs. All you need to do is adapt it and tailor to the needs of your application. I may have been slightly misleading with the "minimal effort" wording -- it will take some effort, but nothing too taxing.
How will this affect repair? Is it going to be tougher to reach the heat you need to replace these SMD parts? Have to heat the entire ground plain to even star moving the solder.
These thermal jumpers seem good, however i do think you can do the same inside the board. If 4 or more layers are available there is an easy fix, without using a much extra space and components. I use this all the time for thermal coupling of the plane with the power device. Just interleave the planes. If the thermal source plane is on top, do thermal vias, use the space in between to couple the GND plane. Make the top "source" plane overlap the GND plane on layer 2. Connect the 3. layer with vias to the source plane on top. This make a a good capacitor as well as provide thermal coupling. The capacitance is not that big and in most use cases not relevant.
Nice video 👍👍👍 Could you make a video about Quartz crystal oscillators, how they can be used to transmit a simple AM audio signal and explain how it exactly works.
Now include the thermal jumper inside the transistor package and you've got even better performance. Wait, surely transistors with some ceramic insulation on the tab have always been a thing?
True, but isolated tab in SMD is pretty rare, though in through hole it is a common thing. I have not seen any SMD devices with an isolated heat slug in the package, even CPU and other devices with integrated heatsink attach areas tell you that plug is an internal connection to some voltage, typically substrate and thus ground. To put this in the package means the package size grows quite a bit, and the price grows a lot as well from the extra assembly steps and lower yield. Probably there is a market for making these in a package that matches SMD thermal pads, with a solderable top and bottom that has the ability to allow you to solder this to the board, put the power device on top and solder it as well, with 3 vias to pads to take the connections through. Make any Dpak device isolated tab, and have easy thermal transfer to the board. going to be an expensive part though, I can see it being easily $1 in volume just from the amount of Aluminium nitride alone, plus the platings and certifications.
@@SeanBZA I guess it's all a bit specialized, but having the isolation right in the transistor, or in between it and the borad, just seems so much less like a hack. Plus you dont need extra space around it on the board and there is no need to place any special copper area for each transistor that needs isolation. Only issue with a separate insulator piece would be assembly, but it's probably not too hard to place both parts and reflow them at the same time.
Dave could you make a video covering SDI video and how it works? Many of the videos I have found just go over what SDI is without really covering the details such as how the information is transmitted.
Hi Dave! Finally a tech-video again! Thanks! As there does not seem to be any temperature gradient on the rightmost rectangle-shaped thermal pad in the thermal image I suspect there might be filled and capped vias that are not visible. /Tomas
@@EEVblog Yes, thermal-stuff is actually interesting and one of the most important topic since this is often the limiting factor when it comes to power conversion designs. As I am sure of that you are very aware of!
It might be tricky to manually solder these jumpers. You might need to preheat entire board first to 150 deg C , because your soldering iron or heating will not be able to melt the solder on such pad that has good connection to big heat sink like ground plane. And you don't want thermal reliefes on these pads either, because then this jumps will operate worse later.
If you moved to this stack up GND Signal Signal VCC And you had some way to get a thermally good but electrically bad connection to GND/Vcc which both look the same to high frequency then you wouldn't need thermal vias or thermal jumpers. Actually, a thermally conductive but dielectric solder mask would do the trick. Essentially the solder mask would be one big thermal jumper. This stack-up is good for RF emissions too. You can't rework such a board and SMD parts would need a via for every non VCC/GND pin but such is life.
Interesting part! One note though, I think if you do the math, you'll find negligible benefit to not soldermasking a board for thermal reasons, the soldermask is very thin and has a very large surface, and to get any kind of temperature difference across that you'd have to be pumping a LOT of heat through it, so much that if you're that close to the bleeding edge of staying in spec for your junction temps, your product will fail in the real world , you know, some dust builds up somewhere and reduces thermal transfer, the device is in the sun, etc.. I'm just saying there'll be bigger fish to fry
This might actually be relevant to my interests. I do software/firmware, but I try to at least follow the other aspects of design. I assumed the ground plane would typically be used as a heat sink, but it's good to know about options like this for cases where parts have to be electrically isolated but could benefit from being thermally connected.
If you didn't care about cost you could make them out of (synthetic) diamond. It's got 7 times better thermal conductivity than beryllium oxide. The cost for the diamond itself wouldn't be that much either, but shaping them would be more time consuming.
Why are the data signals usually on the outside pcb-layer (of a 4 layer board), anyway? Wouldn't they be better protected from outside interference if they were surrounded by a flood-filled ground+power plane?
I am little concerned that the heat will also pass through the tiny solder joint on the side of the jumper and it is of the same size as the jumper cross-section itself. Lead solder is not particularly good in transferring heat (even worse than solder paste sometimes, hence soldering coolers to CPU is roughly as effective), this is why heat pads are made very wide on the bottom of some components. I wonder if they would be able to increase heat transfer by making the joint wider but well... it seems they've investigated multiple designs with a thermal camera. hm...
I've added this comment because Dave repeatedly called the ground plane a heatsink, and it really isn't unless you remember the path to ambient. It's fair to add a warning that the ground plane of your PCB is not a magic heat remover -- it is a very large sink of heat and transfers the heat to a very large area which changes the transfer ratio to the next stage (i.e. the silpad, etc.), but if the PCB is in an insulated case, the heat will still build up (thus a large delta temp); and if the ground plane is internal it will be insulated from any thermal conductances on the outside. I know Dave knows this -- this is a notice to readers that the key of the thermal jumpers is to transfer heat to the next thermal transfer stage *without* the electrical conductance. This allows a connection to metal heatsinks, cases, other planes and ground planes which require electrical isolation; but, in all cases the steady state temperature caused by the transfer of heat will still build up without a path to ambient.
It would be nice to see a thermal analysis done on these but I don't think it would be worth all of the hassle. Unless, of course, a good comparison was made against vias.
FYI, I had to shoot half this video again because I didn't plug the wireless mic cable all the way into the camera and it made intermittent contact half way through. Given that I don't have any sort of script it came out very different the 2nd time around. Not sure which one would have been better. But those who follow me on Twitter already know this.
Whats the deal with all of those calculators? :)
p.s. yes, it would be wonderful if you could make some short demonstration with a custom PCB and thermal camera in future.
I’m so endlessly thankful you’re so dedicated to sharing all of this with us
@@pixlfactory I was also wondering ... 4 "modern" Calculators and one slide rule ... and in the DM42 the batterie is flat ... needs a new CR2032
If you were paying attention to the calculators, you weren’t paying attention to the talk!
Love the calculator swapping :-) subtle.
>uploaded 25 minutes ago
>comment is 17 hours ago
…Patreon I guess?
@@sageinit Patreon, subscribestar, or forum supporter.
After the slide rule I was waiting for an abacus. Wait, **that** calculator might be actually older XD
Calculator trolling..
I noticed that also :)
Oh, gawd. 30 years ago I had an instrument/sensor that normally ran cool, but under specific conditions would run hot as hell. And we couldn't add a heatsink or vias to ground. So we stacked a Peltier cooler between it and the case, and simply dumped power into it when things got hot. Yes, the power supplies were oversized...
Are Peltier electrical insulators? I have SMD resistors that get quite hot, and I could put one of them on top the resistors
That's making me wonder what the thermal conductivity of the standard aluminium nitride ceramic SMD resistors is. Could a high value resistor be used as a cheap heat conductor?
or a low value that you just burn out during burn in :)
we all know if we give you that SMD Thermal Jumper you sure use them to electrify some sausages
I was thinking about doing an estimate based on something like the power derating curves, but that will only give the (overall) component to ambient value. Also, using it in a different setup the mode of heat transfer could get different and ruïn things ....I'm not sure about the last bit, but I think that you should just try it! . Don't forget to check that the case of the SMD resistor. Also, make sure that the resistor's enclosure doesn't create shorts...
Aluminum nitride thermal conductivity 320W/mK. Aluminum oxide 30W/mK. So about 10 times worse. That said, resistors cost at least ~20x less than these, so you could pepper your board with 0603's.
@@jaro6985 320W/m.K is for a single crystal which is expensive, but for an electrically insulating polycrystalline ceramic it is anywhere from 17-285 W/m.K (ref: www.scielo.br/scielo.php?script=sci_arttext&pid=S0366-69132004000300012 see table 1 for physical properties of AlN, SiC, BeO and Al2O3 - resistivity etc.)
Classic EEVBlog tutorial ❤️
Long time no see, glad to see you’re doing fine. Still use your videos as reference today!
Love your videos!
long time no see🎉
We've missed you, Afro! Hope all is well!
Hey man, hope you're doing good and love your videos. When are you going to post again ?
+1 would be keen on seeing independent testing as well as deeper dive on thermal design!
I too would love to see a deep dive on thermal design! Maybe if enough of a new wave of join up to support on Patreon or similar Dave will do a proper video deep dive? :-)
Seconded
Yes, and maybe do the Big Clive thing and throw in some low value caps and high value resistors in the mix for completeness
We ran into a situation where these were a great option when designing a cubesat. There are very sensitive magnetometers on board to help determine the satellite's orientation, and current loops throughout the bus would cause pretty significant interference. Because of this, the aluminum structure could not be electrically grounded. Our RF amp was pumping heat into the ground plane and we needed to dump it into the structure while being electrically isolated. SMD thermal jumpers came to the rescue! Instead of trying to mount a heatsink directly to the IC, we could expose a region on the edge of the RF board which was thermally coupled to the structure. SMD thermal jumpers allowed heat to travel from the ground plane to this exposed edge.
I could see these being super useful for drone motor drivers. Super tiny, but so much power. Also for current sense resistors, given the temperature coefficient being directly related to current reading error.
They usualy use aluminium heatsinks for a reason, most of their PCBs is usualy covered with MOSFETs.
The board itself couldn't take that much thermal load as it's pretty much saturated and the MOSFETs connect to the ground and power traces anyway.
No great help there, for drone ESCs you certainly want direct thermal compound on top and bottom. in plus you do not have enough space for redistribute with vias.
The board experiment would actually be an interesting video. In particular compare their thermal performance to standard resistors. Sure, I know that they are isolated, but a 10M SMD resistor is for many applications (particularly higher power ones) effectively open.
The ceramic base wouldn't be a great conductor of heat I'd imagine. Although you can heat up both pads of a resistor from the one end, so it's possible.
you can also get like 40 smd resistors for the price of one, so even if it's like 5-10:1 it might be worth to just have 10-20 100Mohm resistors around your part.
@@satibel Has to be a big part then, otherwise there is no space to get all those resistors around it.
@@satibel yeah, but who has space for 10 resistors?
@@ProjektMacu on smd To220 and sot428 sized componants.
sot428 is around 6*6 mm, so you have 18mm to lay your resistors on all 3 sides, which I'd reckon could fit 10 resistors though it depends on what size you use.
My first thought was literally: "Just tell me they're not made with beryllium oxide!" :D
Aluminum nitride yay!!!
Keep watching, there is a beryllium oxide version!
@@EEVblog SiC could be be good enough at ~150W/m.K and should be cheaper to produce since it does not require nitrogen gas at 100 and 500 atmospheres of pressure to produce.
EDIT: Although the dielectric loss is 100x higher, maybe not. But still might be good enough for SMPS's. (ref: www.scielo.br/pdf/ce/v50n315/a1150315.pdf )
There is nothing wrong with beryllium oxide as long as you don't breath the dust or eat it. If you pulverize and inhale the resulting dust from your electronics, beryllium oxide is not for you! While you're at it, quit making and breathing asbestos dust.
This world is nuts. "toxic" > Reeeeee! Reeeee!
This is fantastic - almost all my design work is power electronics where thermal considerations remain dominant throughout the design process.
yeah, I wonder how existing designs might be improved, because some pcb designers don't care (probably don't have funds and time to care) about thermals on some components.
"THEMAL LOL"
Fun fact: There are rune inscriptions in Norway where the vikings added the typical "R↓" corrections above the misspelled word. Not many of course, but they do exist!
"Tholfir Kolbeinsson carved these runes high up" is one of my favourites.
As an ME, I love seeing the crossover of EE stuff. That's basically exactly how we are taught steady-state heat transfer. Awesome as always Dave!
Q: What's the difference between an EE and a ME?
A: EEs work with currents less than 10mA .
Thermal jumpers confusing reverse engineers since day one!
Just print some random numbers on them :) Or maybe make thermal jumpers with multiple pins.
Honestly, I presume any reverse engineer worth their salt is actually going to measure the resistance (and capacitance for capacitors, even if they have to pull it out of circuit). Which would give the game away pretty quick.
And make them black if they get a lot of pins. 🤣
What would these read on an LCR tester?
Oh come on dont be petty engineers. Just use sensible design practices and don't make it a complete PITA to troubleshoot for repair.
11:10 "...there's no resistance in them..." Actually, they're extremely large resistances. ;-)
Well, yeah :->
I was thinking the same HAHAHAH
He meant no conductances in them
Bonus points for putting GND on the outside layer, not only do you get thermal dissipation (especially with the thermal vias or GND stitching on the board edges) but now you've doubled it as a faraday cage.
Win-win for thermals and FCC Part 15 compliance.
Very nice to present theories of practical problems that never discussed before like thermal design consideration in power electronics... Much better than commercial reviews ...
I seriously once routed in a little hole underneath a few times, then went on a thermal-pad straight to the case.
Obviously that only works if you have the space underneath the component but it worked very well.
I'm now already redesigning a board I'm working on! :D
10:42 that's what i tell my wife but she won't listen!
love how the datasheet has the thermal imagery, mayaswell be a sales pamphlet take my money
Though this looks like a game changer passive thermal solution, i am predicting an active thermal solution in similar smt form. They would use the peltier effect to both remove heat and act as a sensor/(power source for backup purposes). Their application area could be based on high performace, compact high current FET driving application.
Seems like these could be useful for on-board temperature measurement. Connect your temperature sensor (RTD, or whatever) to the pin you're trying to sense with one of these and you might have a pretty good measurement.
Thanks for bringing these components to our attention! I'm putting them in my next board design, as the are a perfect solution to heat sinking a bank of isolated SMD power transistors.
another reason to use these could also be wanting to reduce capacitance/size on some node, in that case you might want to go for the standard aspect ratio (eg 0402, 0603) ones. will certainly consider these in my next design
ah, should have watched to the end, you already mention that, still wanted to emphasize that these might be very useful even in situations where you could easily satisfy your thermal requirement by simply using a larger pad and aren't pressed for space for electrical performance / emc considerations
as i work on very low volume and expensive products i am not all that bothered about cost, having to go through an additional prototype cycle or even spending an extra week on a design is always a lot more expensive than having another couple of dollars on the bom over the entire lifespan of the product. i usually try to do everything i can think of to make a design work and pass emc tests on the first try. smd feedthru capacitors are also great btw :)
Heh, the calculator keeps changing. Nice touch.
Beautiful, this made me want to add the 0612 and 1225 THJP models to my PCB library, can't wait to use them, Thanks Dave !
Thank you for pointing this part out. I never knew this was a thing. I could have probably used those in some designs.
That's a fascinating product. We do a lot of designs for explosive atmosphere where it can be difficult to meet de-rating requirements due to all the applied safety factors. I am going to bring these up with my team.
Awesome tutorial! I am glad that you are talking about thermal design. It is not getting enough attention.
8:39 - Also the copper on your top and bottom layers is usually thicker, so that helps with heat transfer. I've used the layer flip technique on some designs where heat or current carrying capacity was an issue.
Thanks Dave for this! Did not know about this and would be handy for a one off stepper motor controller I'm making. And yes, please do a video on actually reviewing these under various test conditions.
The previous video is actually a pre-requisite for this one. Tried to watch this one straight and omg I had no idea of what's the drawings was about.
In addition to allowing you to tie live pins to the existing ground/power planes for thermal reasons, these jumpers would also allow you to group the thermal pads of multiple isolated components together to share a common heatsink for cost-saving (rather than a small dedicated heatsink for each), without needing any other isolation layers. Think stacked transistor layouts, multiple equivalent channels of a device, battery balancing circuits, etc...
I'm slogging through this stuff working on RF amplifiers. Sometimes, the package you want puts a transistor collector at RF potential (eek!). These are really cool... I'd love it if you would do an experimental video showing a real board and taking measurements of different arrangements!
Great collection of calculators. Love the slide rule
Thanks for sharing this product. I’ll keep it in mind next time I have significant thermals to deal with.
Hi Dave interesting devices. For a Gamechanging we need a defined insulation. Vishay clames >1.5KV ac withstand voltage👍, but as you know, we need formal insulation approvals and ratings like CTI, dti, PD, working voltage and so on. Otherwise it may be not usable.
I used to work in the lab of an electronics company and the amount of thermal testing that was done was insane! On some low voltage but high power boards, six layer PCBs were used to soak away extra heat. Unfortunately it did make removing copper foil wound transformers a bugger to remove!
Very interesting part. Not necessarily for the kind of products I usually work with but still a great new design choice.
Yep, not everyone, but when you need it it's killer.
When you do a test, could you compare to a variety of regular ceramic capacitors of similar capacities and voltage rating? They could also have some thermal conductivity as well.
If you have many layers you can also play with planes to transfer heat through the FR4 substrate. I did this once where every second layer was connected to the heat source and every other to the GND-plane. The component was almost impossible to de solder because of this! :D
These are super cool and I'd love to see you get them and make an hands on video.
Thank you very much. Very informative. Otherwise I would not have noticed these parts! Niche new option!
Quite interesting. Will try it myself, that is for sure. Very useful video. Thanks Dave!
Interesting, but I think you could often do just as well with simple pcb layout techniques.
FR-4 has a thermal conductivity of about 0.25 W/mK. So if you have a 10mm x 10mm land on the surface of the pcb, and the ground plane is just below it with 0.1mm of FR-4 in between, the thermal resistance from the land to the ground plane will be about 4 K/W. And a 0.1mm layer of FR-4 should be good for about 3kV (maybe 1.5kV with aging) isolation. So you can get pretty good thermal conduction without any thermal vias if you have enough room on the surface for a large land to spread the heat.
Ok we are currently doing a redesign of an older board that has a thermal sensor to monitor some of the hotter components on the board. Just in case something burns out and we need to shut it off. The current "solution" was to put the SMD thermistor near the hot spot and call it a day. It does work surprisingly well actually. While i was doing the redesign i remembered this video and put 2 thermal jumpers from the potentially hot components to the SMD thermistor. One jumper for each pad. I've yet to receive the new boards, but i believe they should work way better.
Thank you so much for this Dave, my boss was practically jumping up and down when i told him what i had done. He was very pleased.
Fighting with the design of a boost led driver managing the heat, if first review gets too hot, I will give a go to these guys....,good video,thanks!
You know I watched this video intially as something 'oh cool, how informative', but as it reached minute 15 I was astonished at the results. This is definitely going to be a game changer. Thanks for the video again Dave!
Thx man, was looking for this part, ya demystified it all for me!
a test setup and results would be very useful....hoping for next video with test results and the performance of those jumpers
That Is great! Thanks! I'll probably use them in my next design. :)
00:00 Casio fx-50F,
01:17 SwissMicros DM42,
04:27 casio fx-740p,
05:41 FABER-CASTELL TR2,
10:08 Faber-Castell TR3,
how much did your collection worth?????
i think i have all them wright.. tell me what i won?
The last two are actually the same device.
@@EEVblog awesome
"Dave what happened to your PSU project?"
"David2 left and it's been in stasis ever since."
Maybe release the project and let the community finish it, then run some group buys or straight put on your store/other distributors.
the problem with the PSU project is, that for it to become a product, it needs one vision.
If you let the community at it, you will get conflicting visions and go nowhere.
What would work is realising the hardware and a "driver packet" type firmware. That would also branch out, but quickly, a few branches would establish them self as "good for a specific purpose".
@@sarowie Let Dave manage any volunteers and have them do exactly what he is looking for?
@@Arek_R. Who would
1st have the knowledge to do (correctly) what Dave's wanted
2nd Actually having and wanting to spare free time on an open source project thats not that "open"
3rd Do all of that (basically working completely for someone else) for free
Those would be the major problems
@@Arek_R. Dave is the kind of guy, that keeps things in mind. Creating a symbol is not mindless busy work - it is preconceptuallizing how you plan on using the IC, reading the pin description, checking the specification against surrounding parts, taking notes for the layout phase... and then double checking if the b-variant of the part would not fit better. It might have a bigger footprint, but that simplifies the fan out the inductor.
I'm tasked with mentoring a student currently. It is fun, but preparing the work for someone else is more work then when I would to that specific task. It is the unified vision and a gut feel of which optional parts are worth in context of other constrains that pop up during the design.
And here I am pissed off at work for having to reflow a board short on MSOP 10s with a beefy thermal plane.
I gotta say Dave if it wasn't for me growing up with your videos I wouldn't have a job.
Thanks man.
+1 to making some test and contrasting to thermal vias and ground plane on inner/top layer.
This is the kind of video that made eevblog great!
I think the best way to use this part would be to put it between the SMD land and a mounting screw that attaches the board to chassis. Probably could conduct a lot more heat than the ground plane.
Possible. It would radiate better externally of course.
Afro and all others are here also.
I had ordered some for my last project. But they interfered with my design and would short some parts. At the end I used some vias for the LDO and for safety I thermalglued an old school alu fin block on top. Never heard any problems from the customer.
I haven't seen Dave so excited since the days when he used to play baseball with DMM's !!
@EEVblog This video is sooo nice...i count the calculators in video :D Thanks for this video and for new technics in heatsinks :)
I was thinking it’s cool and all to have another way to get heat out on large package, but for some reason my mind was blown when you showed the picture where they wick heat out of a resistor from its pad. Thank you for sharing this!
Super. It is not that often that a new thing like this comes out, very interesting.
Now, on the calculator thing, I wonder why you did not change the calc on the last scene...
Would be great to have another video with peroper experimentation! Thanks for the content :)
I'm wondering what schematic symbol to use for this part :|
Very interesting! I guess this could be used for thermal management of sensitive analog circuitry as well.
Modeled thermal systems for years (not electronics, but mechanical and fluid systems). Videos like this make me appreciate just how complex electronic designs are. Who would have thought so much thermal design in that itty-bitty iphone. :) Oh, and kudos for the 'big fat wide jumpers' comment. Anyone that studies thermal conductivity of materials understands the need for that. Great video, interesting and shows me just how much I don't know (which seems to be more and more every day lol )
Using the short fat ones may not be viable sometimes because you still have to maintain the gap between conductors. The voltage breakdown is still a thing.
You're the man , only one ad before the video
Tbh, although this is nice and dandy, nothing beats a quantitative thermal analysis with Finite-Element Method software. These are expensive as Dave suggests, but with minimal effort and some math you can do a nice DIY piece of software to thermally model your design with incredible accuracy.
I'd hardly think writing your own custom FEA stuff is "minimal effort" for most people.
@@EEVblog that's the thing, there are a lot of ready-to-use open-source libraries and programs. All you need to do is adapt it and tailor to the needs of your application. I may have been slightly misleading with the "minimal effort" wording -- it will take some effort, but nothing too taxing.
Loved the calculator parade, Dave.
I'd be interested to see real-world thermal tests, I understand that's a lot of extra trouble though.
I need some of those SSD heatsinks for a 3D printer board upgrade.
these are going to be great for brushless motor drivers
I'm really surprised these didn't previously exist, hindsight is 20/20 though. They sound like they would be very useful in the cell phone market.
How will this affect repair? Is it going to be tougher to reach the heat you need to replace these SMD parts? Have to heat the entire ground plain to even star moving the solder.
These thermal jumpers seem good, however i do think you can do the same inside the board.
If 4 or more layers are available there is an easy fix, without using a much extra space and components.
I use this all the time for thermal coupling of the plane with the power device.
Just interleave the planes. If the thermal source plane is on top, do thermal vias, use the space in between to couple the GND plane.
Make the top "source" plane overlap the GND plane on layer 2. Connect the 3. layer with vias to the source plane on top.
This make a a good capacitor as well as provide thermal coupling. The capacitance is not that big and in most use cases not relevant.
Nice video 👍👍👍
Could you make a video about
Quartz crystal oscillators, how they can be used to transmit a simple AM audio signal and explain how it exactly works.
Cheers mate! Great rundown.
Now include the thermal jumper inside the transistor package and you've got even better performance.
Wait, surely transistors with some ceramic insulation on the tab have always been a thing?
True, but isolated tab in SMD is pretty rare, though in through hole it is a common thing. I have not seen any SMD devices with an isolated heat slug in the package, even CPU and other devices with integrated heatsink attach areas tell you that plug is an internal connection to some voltage, typically substrate and thus ground. To put this in the package means the package size grows quite a bit, and the price grows a lot as well from the extra assembly steps and lower yield.
Probably there is a market for making these in a package that matches SMD thermal pads, with a solderable top and bottom that has the ability to allow you to solder this to the board, put the power device on top and solder it as well, with 3 vias to pads to take the connections through. Make any Dpak device isolated tab, and have easy thermal transfer to the board. going to be an expensive part though, I can see it being easily $1 in volume just from the amount of Aluminium nitride alone, plus the platings and certifications.
@@SeanBZA I guess it's all a bit specialized, but having the isolation right in the transistor, or in between it and the borad, just seems so much less like a hack. Plus you dont need extra space around it on the board and there is no need to place any special copper area for each transistor that needs isolation.
Only issue with a separate insulator piece would be assembly, but it's probably not too hard to place both parts and reflow them at the same time.
Dave could you make a video covering SDI video and how it works? Many of the videos I have found just go over what SDI is without really covering the details such as how the information is transmitted.
Hi Dave! Finally a tech-video again! Thanks! As there does not seem to be any temperature gradient on the rightmost rectangle-shaped thermal pad in the thermal image I suspect there might be filled and capped vias that are not visible. /Tomas
Yeah I thought that was a bit interesting.
@@EEVblog Yes, thermal-stuff is actually interesting and one of the most important topic since this is often the limiting factor when it comes to power conversion designs. As I am sure of that you are very aware of!
I have to keep repeating this - distracted by the changing of the calculators. Squirrel! LOL
It might be tricky to manually solder these jumpers. You might need to preheat entire board first to 150 deg C , because your soldering iron or heating will not be able to melt the solder on such pad that has good connection to big heat sink like ground plane. And you don't want thermal reliefes on these pads either, because then this jumps will operate worse later.
in Russia we put more wood in stove for more thermal.
If you moved to this stack up
GND
Signal
Signal
VCC
And you had some way to get a thermally good but electrically bad connection to GND/Vcc which both look the same to high frequency then you wouldn't need thermal vias or thermal jumpers. Actually, a thermally conductive but dielectric solder mask would do the trick. Essentially the solder mask would be one big thermal jumper. This stack-up is good for RF emissions too. You can't rework such a board and SMD parts would need a via for every non VCC/GND pin but such is life.
Interesting part!
One note though, I think if you do the math, you'll find negligible benefit to not soldermasking a board for thermal reasons, the soldermask is very thin and has a very large surface, and to get any kind of temperature difference across that you'd have to be pumping a LOT of heat through it, so much that if you're that close to the bleeding edge of staying in spec for your junction temps, your product will fail in the real world , you know, some dust builds up somewhere and reduces thermal transfer, the device is in the sun, etc.. I'm just saying there'll be bigger fish to fry
This might actually be relevant to my interests. I do software/firmware, but I try to at least follow the other aspects of design. I assumed the ground plane would typically be used as a heat sink, but it's good to know about options like this for cases where parts have to be electrically isolated but could benefit from being thermally connected.
Lovely, how your leprechaun (or troll or hobgoblin or whatever it is) assistant changes his calculators during the video.
That board looks very sharp!
If you didn't care about cost you could make them out of (synthetic) diamond. It's got 7 times better thermal conductivity than beryllium oxide. The cost for the diamond itself wouldn't be that much either, but shaping them would be more time consuming.
Why are the data signals usually on the outside pcb-layer (of a 4 layer board), anyway? Wouldn't they be better protected from outside interference if they were surrounded by a flood-filled ground+power plane?
I have still to make a video on that. PCB already made and assembled, and a very specialised kit of kit for testing has become available to me.
I am little concerned that the heat will also pass through the tiny solder joint on the side of the jumper and it is of the same size as the jumper cross-section itself. Lead solder is not particularly good in transferring heat (even worse than solder paste sometimes, hence soldering coolers to CPU is roughly as effective), this is why heat pads are made very wide on the bottom of some components. I wonder if they would be able to increase heat transfer by making the joint wider but well... it seems they've investigated multiple designs with a thermal camera. hm...
Great video, loved the Strictly Ballroom reference.
Heck, found one of these at work on the prototyping assembly lane, almost forgot to look into it... great sum up !
I've added this comment because Dave repeatedly called the ground plane a heatsink, and it really isn't unless you remember the path to ambient.
It's fair to add a warning that the ground plane of your PCB is not a magic heat remover -- it is a very large sink of heat and transfers the heat to a very large area which changes the transfer ratio to the next stage (i.e. the silpad, etc.), but if the PCB is in an insulated case, the heat will still build up (thus a large delta temp); and if the ground plane is internal it will be insulated from any thermal conductances on the outside.
I know Dave knows this -- this is a notice to readers that the key of the thermal jumpers is to transfer heat to the next thermal transfer stage *without* the electrical conductance. This allows a connection to metal heatsinks, cases, other planes and ground planes which require electrical isolation; but, in all cases the steady state temperature caused by the transfer of heat will still build up without a path to ambient.
It would be nice to see a thermal analysis done on these but I don't think it would be worth all of the hassle. Unless, of course, a good comparison was made against vias.
Yeah, if done it would need to kill a lot of birds with one board.
Would be very cool to see you having a play with these please do design a PCB to try em out :D
Yeah, i would really like to see this comparison.
These kinds of technical ceramics have been around since the 1960s, so I am a bit surprised to only see them being utilized in this way now.