EEVblog

How to reduce your PCB assembly cost & production risk by optimising schematic and BOM.
Dave explains three ways to consolidate your project PCB Bill Of Materials (BOM), to make your products potentially cheaper and easier to manufacture with less supply risk. As well as being an essential good design practice!
A look at a complex FPGA based design with a large number of components, how far can this design be optimised and for what benefit?
github.com/ciaa/Hardware/tree...
The 1 cent voltage regulator: • EEVblog #1147 - 1 Cent...
Forum: www.eevblog.com/forum/blog/ee...
#PCB #Design #Tutorial
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
AliExpress Affiliate: s.click.aliexpress.com/e/c2LRpe8g
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.com/crypto-currency/
T-Shirts: teespring.com/stores/eevblog

[@mozismobile]

You skimmed it, but one thing that can really help cost reduction is the more complex silicon. If you have 3 types of RAM modules in different parts and there's any way to go from 2 of these, 1 of those, and 4 of the other down to 11 of one type that can save you a fukton through both volume discounts on the parts and the wide pick'n'place lanes for big chips. Put lots of different big chips on suddenly "80 lanes" becomes "an 8 lane wide CPU, two different 6 lane RAM chips, then five regulators at 3 lanes each" and almost half the lanes are gone before you start.

[@EEVblog]

Yup! Add in a few large pin count SMD connectors and your are done for! Time for a bigger line.

[@SkyCharger001]

just an idea: It might be even cheaper to use a large capacity RAM chip that can provide dedicated regions to the subsystems that require what otherwise would have been their own RAM-chips. (EG instead of 3 different 6-lane RAM chips, you use one 8 lane RAM-chip ... but you also need to take RAM access contention (and the circuitry for handling it) into account when calculating the amount of lanes saved)

[@urugulu1656]

@@SkyCharger001 if signals dont need to be routed across the whole board for it and if it doesnt require any more address decoding logic and if signal runtimes arent buggering you up sure that could work

[@hannahranga]

I'm curious if for bigger chips you can get custom mixed tapes made to suit. IE 2 RAM A, 3 IC A etc.

[@St0RM33]

@@hannahranga If they are the same package, i'm pretty sure it can be done but manually unless all are from the same manufacturer and you are ordering LOTS of them

[@grumble2009]

My first employer made us reduce BOM line item count, use existing inventory parts whenever possible, forbade certain manufacturers, provide 3 sources for all new components, standardize footprints, and typically trade off parts cost versus insertion cost for manufacturing (resistors are 10 for a penny, but it costs 4 cents to get it onto the PCB). It was a great way to learn about all of the cost and time saving techniques that are glossed over in college. It's all about trade offs, and the sooner you get used to it the better.

[@johnsonlam]

Hi Dave, please do the 1 cents voltage regulator, it's fun to know such a cheap thing works or not.

[@flos251]

Switching regulator : lcsc.com/product-detail/Lack-of-specifications_Diodes-Incorporated_AP3770AK6TR-G1_Diodes-Incorporated-AP3770AK6TR-G1_C260850.html less then 1 cent in 600 qty linear rgulator : lcsc.com/product-detail/Linear-Voltage-Regulators_HX-hengjiaxing-HX6206P282MR_C296236.html ess then 1 cent in 1500 qty this video already exists kzbin.info/www/bejne/ipWxhoJpgriZgNU

[@urugulu1656]

@@flos251 oh thats why i thought wait a sec didnt i see something like that some where already?

[@urugulu1656]

@Arpad Toth without any further information this is just noize

[@lhxperimental]

It definitely does. I have these things running without issue in field.

[@skyem5250]

I bought 78L05 regulators for 0.4 cents each. They work normally.

[@johnsonlam]

Dave is really giving out knowledge that not even teach in institute, great. The project guy should watch this video for several times and learn something.

[@HighestRank]

Why would I need to watch this “several times” when I understood it 💯% -completely the first time? Are ‘project-guys’ supposed to be exceptionally-dense?

[@urugulu1656]

@@HighestRank b/c you often oversee stuff the first time

[@EcoVoltz]

I never post on comments sections but this was so helpful I had to say thanks! Reduced my BOM by 30% on a new project that I was about to send off for manufacture. Crazy good!

[@Amerinidiot1231]

Would love to see Dave go through this design and optimize it and compare his reduced BOM to the original.

[@hypercube33]

And then order a prototype and show off it working!

[@mikeselectricstuff]

..and also on the PSU section, 68nf soft-start cap and 100nf boost cap can also almost certainly be both 100nf

[@craigrotay3732]

Really great one Dave. Really. Seriously. Everyone in our field needs to feel this. Not know it, but feel it.

[@Stoneman06660]

This is even brilliant info for the hobbiest who is build up stuff to learn the trade. Knowing you don't need a 10 pack of every single obscure thing to get the job done is literally gold. Money, but more importantly, so much time saved to get things underway. Smashing stuff.

[@noeliascotti1795]

Hello Dave, I am part of the CIAA Team! Thanks for your review. I will design a new release considering your suggestions. It is important to note that 0201 capacitors have a lower inductance than 0402 caps. Besides, don't forget to consider the temperature coefficient when you replace capacitors (in order to make BOM consolidation). About bypass capacitors... I would not reduce the quantity, if Xilinx says use 100, I will use 100!!!

[@noeliascotti1795]

@Matt herpader It is very risky to not follow the manufacturer recommendations... I prefer to design an expensive board. Anyway I will do the math in order to see how much money we can save with the BOM consolidation.

[@absalomdraconis]

@@noeliascotti1795 : Rather than being risky to just not follow manufacturer recommendations, it risky to not give yourself the _chance_ to follow those recommendations. Once you have the location allocated to a part, you _can_ choose to leave it empty,

[@noeliascotti1795]

@@NerdyNEET jaja I don't want to be rude, so I will only say that the FIRST RELEASE of this board worked fine! And it is open hardware. And it was routed with KiCAD. And we are in Argentina... you will never understand how difficult is to finish a project like this in my country.

[@kevinvermeer9011]

​@@noeliascotti1795 If the first release of this highly involved board worked fine, you deserve the utmost respect regardless of where it was made or with what it was laid out! That's incredibly impressive!

[@aliuzel4211]

Not always good idea to stick with silicon vendors evaluation board specs. They are so relaxed designs even horrible PCB stackups. They are just eva boards. Sometimes more caps do not add additional benefits, rather placement,value and type. I am working on RISC-V SOC board and test card will hopefully be ready in Sept/Oct time.

[@asmi06]

I agree with the overall thrust of the video, but I have to add my 2c regarding the single thing I totally hate - about 0201 decoupling caps. Yes you absolutely do need them, because that is the only size which fits between vias in regular BGA fanout grid without encroaching on via pads (and so risking having solder paste wicking from pads into vias causing solder failures). Been there done that. Going for 0402 requires too many compromises with the fanout (like skipping some balls to make room for the caps), and in large BGA grids a fanout is a PITA even without additional complications. In some cases you will be even forced to use 01005's for the same reasons. On top of that smaller packages have smaller ESL which is often the very reason you use those caps in the first place. I've done quite a bit of FPGA boards, and decoupling caps under BGAs is the only place I ever use 0201s (because manually placing them is a bit time-consuming), and I only have a single value, which I bought a 15k reel of so I'm unlikely to run out of them any time soon. Also while you can get away without some decoupling caps, you have to know your limits because there are cases when they absolutely have to be there. For example, a lot of multi-gigabit serial transceivers require so much current that LDOs are not practical, and designing a 1.0 V ± 3% switcher with

[@EEVblog]

Of course 0201 capacitors have their place, but don't use them (and 0402's for that matter) unless you absolutely need them. I've done countless FPGA's board and I've never had to use 0201's. YMMV.

[@asmi06]

@@EEVblog I agree about 0201's in general - I avoid them as much as I can, but 0402s are my default size - love them. I mostly do multilayer boards nowadays, and space on those boards is expensive, and 0402s allow for tighter placement. Also they are in most cases the cheapest size in terms of price per unit, so I got reels of pretty much all parts I use often.

[@ANTALIFE]

Super useful video!!! I used to work for a PCBA manufacturer so have first hand experience with what happens when BOM's are all over the place. I recall a number of customers never bothering with consolidating their BOM's (even when we reminded them about it), so it was common to see 5 variants of a 10N capacitor (X5R/X7R/0603/0805/1206...) on the same board... And when this happens 5-10 times on a single board assembly cost really start to add up as you have the machine operators spending more time kitting up, more nozzle operations, the job taking up a larger chunk of the production line...

[@akhurash]

Interesting video. I’m going through this at work now. Another topic is cost reduction. We were using a lot of 0.1% resistor for pull-up resistors... That was a waste.

[@EEVblog]

Yes, that would need a separate video as I explained. But BOM consolidation is essentially a cost reduction step in it's own right.

[@bertblankenstein3738]

0.1% seems like a really tight tolerance for pullups. I bet if you went to 1% or maybe 5% tolerance you'd save a lot.

[@IACooper]

Even where 0.1% resistors are justified, watch the ppm values: a 20ppm 0.1% resistor is a lot cheaper than a 10ppm 0.1% resistor, and if you don't need the tighter temperature spec, don't waste the money...

[@enginerdy]

I usually specify these in the schematic with the loosest value required (5%), then you can make a "sub allowed" note in the BOM.

[@akhurash]

Bert Blankenstein I agree. Whoever worked on the schematic before didn’t care about cost it seems.

[@grooeygroo]

Jones, you absolute legend. Well done taking an easily-overlooked subject and infusing it with your enthusiasm. I learned heaps. Thanks!

[@passenger000]

what a coincidence to have Dave go through BOM list as I am about to do my very first PCB project soon :) thanks!

[@flymypg]

I've done package optimization in layout, especially getting to single layer when I needed to run traces under components. Going to a larger package can save the day!

[@KerbalLauncher]

The best way to do BOM consolidation is to hand assemble one of the boards ;).

[@BerndFelsche]

With 0201's ??

[@absurdengineering]

@@BerndFelsche It can be done, but so can be lots of ridiculous things :)

[@vintlabs]

Great video Dave - and many thanks for pointing out the interactive HTML BOM in Kicad! I didn't know about that and it's going to be massively useful!

[@stephenbell9257]

Regarding the "Muntzing" of the supply bypass caps I think that this shouldn't be done on a new design until a first prototype has been pre-qualified for EMC issues. It is better to initially fully bypass the chips, in accordance with the manufacturer's recommendations, rather than failing the initial EMC test and having to do another time-consuming PCB re-spin to add extra bypass caps. Once the prototype has been EMC tested and, if the design has a comfortable margin, then the Muntzing process can be done as part of a value engineering process. This generally doesn't require a PCB re-spin as this can be accomplished quickly and simply by modifying the BOM.

[@urugulu1656]

yeah just leave them footprints unpopulated...

[@enginerdy]

At what point do you put it back through EMC to validate your changes?

[@HighestRank]

Those caps weren’t MUNTZED, they were substituted for a value more common to the rest of the board, a local ‘jellybeanization’ if you will, on the project only. Whether that ruins their intended-functionality as far as the purpose of bypass caps in that area remains to be calculated, but if so, then the end result is almost always worse than having muntzed them.

[@absalomdraconis]

@@enginerdy : You put it back through EMC after whoever is Muntzing it goes back a step. Note that this _does_ require the Muntzer to have some functional EMC probing capacity- not perfect capacity, mind you, just something to let them work out relative change. They also need to "calibrate" (compare) their equipment to the initial EMC spec, to counterbalance any inadequEcies of their own equipment.

[@kalhana1]

I agree. Also, PI (power integrity) simulation can help with cutting down extra caps.

[@davidvanhorn498]

Back in the early 80's I was designing the first terminals for Verifone, and Karl Chang (VP eng and my boss) told me that the ideal design has as few different parts as possible. Engineering is the ART of compromise!

[@riccardoz2953]

Thanks Dave, i love a tons how u explain stuff, your passion and your knoledge... plus how u speech it's amazing.. thanks again | Best regards from italy !

[@MadRC]

Hay,Hope your staying safe and well, enjoying the content as always :). Thank you for all you bring to the community

[@adammujaj]

Sometimes it's useful to use weird values... I recall a few years ago there was a capacitor shortage, and 'normal' values were in short supply with long lead times. However if you changed your 1uF for a 1.1uF suddenly stock was easier to get!

[@marklewus5468]

I got my first job as an EE in 1980. In those pre-Internet days even small manufacturers had a dedicated purchasing agent. It was not a big company, and our typical production runs were in the low hundreds. Shortly after I designed my first board the purchasing agent sat me down and gave me the same lecture that was in your terrific video. He told me that buying a unique one cent resistor might cost the company $50 for his hour or more on the phone. On a production run of 100 it turns a one cent resistor into a 50 cent resistor. I never forgot the lesson and to this day even on hobby projects, every time I place a resistor I’m thinking, “can I use a 10K 0603 here?” Thanks for the reminder!

[@MrZnarffy]

1 cent voltage reg??? PLS PLS PLS :D

[@flos251]

this video already exists kzbin.info/www/bejne/ipWxhoJpgriZgNU

[@pdp11]

This kind of stuff is what makes this channel so great!

[@brianhginc.2140]

One comment about too many of 1 component value mentioned @26:15 , If you simplified/consolidated down your BOM and there are free reel slots on the pick and place machine, instead of a worker always changing reels, they would just program and place 2 or more reels of the same value in the P&P machine.

[@EEVblog]

Yeah, forgot to mention that! Less downtime on the machine.

[@jakrol]

Great video, Dave! I'm about to finish schematic capture of my first large component count board... has analog, digital and medium power RF (50W PEP) on it... this video came at a great time for me!

[@MadsonOnTheWeb]

This is one of the best videos on your channel.

[@sepphuber7434]

I like BOM items #28 and #29: Two different 1K resistors with 1% and 5% tolerance =)

[@absalomdraconis]

I can see it in high-spec analogue (shoot, go to .1%!), but if you aren't dealing with delicate filters it's hard to imagine the sense in it.

[@MarineMickk]

Great video Dave! I was lucky enough to work at a company that had in-house assembly capability. I had a front row seat to see all these DFM implications. Sometimes a very small change in design makes a huge difference at assembly. Keep 'em coming Dave!

[@EEVblog]

That would be neat!

[@bengrant4724]

1c voltage regulator yes please!

[@369gems]

Well done. Very interesting. Thanks Dave.

[@pmaurin]

This was and awesome video Dave! Thanks for sharing your knowledge.

[@electronic7979]

It was a nice and useful video. I watched with pleasure. thank you.

[@ronniepirtlejr2606]

I remember watching one of your videos Dave, with the bypass capacitors, you basically need them unless you really want to sit in probe & spend a lot of time figuring out what it is you really need for the end production instead of just playing it safe.

[@TheDefpom]

When I'm designing something (not like I do a lot) I consider the standard resistor values and duplicated values at the design stage, too tedious for me to go back and do it later.

[@decee1157]

Yeah. E12 or E24 all the way

[@AJB2K3]

Jellybean as many components as possibly?

[@HighestRank]

The design stage often includes hacking together a prototype and woes be to us if we hadn’t shopped for pricing quantities while purchasing its samples.

[@00Skyfox]

Thumbs up for the 1 cent voltage regulator testing!

[@diggleboy]

Great advice video, Dave, on BOM consolidation. I've had to do this in the past where I tested multiple parts in the lab for their specifications, cost and supply chain availability. Most of the parts came from China, but there were times I had to consolidate parts from the parent company's parts list to save on costs because of bulk volume purchase discounts. BOM parts consolidation is an essential process to sustainable manufacturing and multiple supply chains and vendors should be sourced and considered, just in case of a supply chain shortage or disruption like what we're experiencing now with the pandemic. Spend the time to calculate the consolidated part and have it tested, then approved by the design engineer, prior to substitution. This saves time, money and headaches.

[@desaoaraujo]

Great stuff! After I started to worry about production cost on my designs, all these aspects really came to mind. Even selecting the least amount of drill sizes. Reduce the number of every different parameters to as low possible!

[@ThisMicrophoneSoundsCheap]

Thanks for being online during the premiere. Please keep it up and consider using scheduled premieres on fairly short notice, like 1-3 hours or so, so we can watch it live in full, rather than joining in the middle then rewatching. It's a widely hated feature because it tends to be shamelessly abused to get around youtube not showing notifications to old subscribers, by youtubers that have a million subscribers but only 1k views per video. You being online in the chat during the premiere makes all the difference, as it adds value. Even if you wouldn't have been online, in the specific case of educational videos, being able to ask questions live to other technical viewers (and see the replay) could add value to your content.

[@EEVblog]

Yep, will announce before hand next time.

[@PepeSanches4]

I would love to see more about optimization on pcb design stages! Love the channel!

[@malgailany]

I learned something new. Thank you!

[@jr0th]

Thanks Dave, nice video as usual. Some quick infos as I'm used to get PCBs manufactured: - some assemblers are charging a minimum count of each part, so having 1 special value resistor may get you charged for 25 or 50, - some are counting parts per PCB, so you may really lower costs by panelizing the boards yourself, - they are charging for reel placement (around $3 per reel if not a common part), - number of special parts may be limited to 10 or 20. My personal rules and tips: - never get a resistor or capacitor that's not a basic part, - consider you have 1k, 10k, 100k resistors on board, 100nF, 10uF capacitors, and pick every R and C from these, - sometimes it's better to replace huge capas with THT variants, and assemble these in-house (or buy an automated soldering machine) - always double-check the footprint with the manufacturer datasheet (or even create the part yourself), I have tons of PCBs that were faulty because the footprint has errors (mostly from KiCad libs)

[@BerndFelsche]

Had similar with Eagle. Always check the footprint.

[@ArielSasson]

thanks for enlighting us!

[@tedmich]

EE angered by inelegant solutions, video to follow! Bravo Dave!

[@apepost]

You're confusing the imperial component sizes from the BOM with the metric sizes in the data for the Yamaha Z:TA-R YSM40R. The imperial equivalents for the size limits of the YSM40R are 008004 for the ultra-high-speed head, 01005 for the multi head, and 0201 for the flexible head.

[@HighestRank]

Since when didn’t imperial sizes use units?

[@chriscoleman2128]

Watched all the way through 👍🏻

[@RobinHilton22367]

Yes please. Love to see the sub 10 cent parts series of products.

[@rodrigomaero]

don't know why but I feel proud to see the CIAA here, even though I haven't contributed to it in any way other than see it in college (being worked on by teachers) before the release

[@Factory400]

This topic is endless........glad you touched it Dave.

[@EEVblog]

It could be endless, yes!

[@galileo_rs]

Loved the P&P machine sound effects ;)

[@abisaiamatalo2769]

Great video. Why would someone give a thumbs down to this video? They need their heads checked. Keep up the good work. Following all the way from Kenya. Asante Sana ( Thanks a lot).

[@Narwaro]

As a physicist that only does this from time to time when needed, I really appreciate this because nearly everything I have ever done is really (mostly high-end) low-volume specialised boards because there is no off the shelf hardware that you can use. So its a really cool learning experience for me thanks ;D

[@MrDazzlerdarren]

Confuser? Too much watching of the Mr. AvE there! :-D

[@EEVblog]

Nope, had no idea he called it that, I've called it that for decades.

[@urugulu1656]

@@EEVblog well in todays day and age the REAL confusor is a sliderule...

[@kevinvermeer9011]

@@EEVblog AvE has mentioned your channel in the past, looks like Dazla has reversed the causality.

[@manteksarai8322]

Trying to learning from this good channel.. ❤

[@grerlab2777]

Always interesting with educational videos ... Thank you...

[@McTroyd]

I'm one of the (apparently) few who went through all 5 of your Padauk tutorial videos, and liked it. It was super informative, and so is this video. I'd happily watch a video about a 1 cent regulator... or any other insights you might have in this sort of value engineering. Thanks!

[@DerinTheErkan]

At uni we had a simple flashing LED board (just the usual 556 + 4017 business) which we gave out as a kit while teaching kids about electronics. In my experience, kids are an even bigger target for BOM consolidation than PNP machines. I took the design from 5-6 resistors and 3-4 caps to pretty much two values of each. Beyond making them simple to assemble, I'm sure the people putting the parts into the kits appreciated it too

[@urugulu1656]

also kids may not really give a shit on different values until they understand why they matter by messing them up.

[@DerinTheErkan]

@@urugulu1656 true, the original values are mostly a case of following the confuser as Dave puts it

[@urugulu1656]

@@DerinTheErkan also with the simple circuitry in such a thing whats the worst thing that can happen by changing the values ? its counting at .7hz instead of 1hz or what ever... as if in a bit of kit thats sole purpose is to educate children and get them hooked on electronics that would even matter you could probably even get it down to one value for each. actually what you need the 556 for. you dont need more than one 555 for the one oscillator it takes...

[@erikmjelde4428]

great video with good information

[@radiotek4023]

Thanks Dave good advice i didn’t consider the pick and place machine. Yes please on the 1c reg

[@alexiekola]

Yes please more vids on consolidation need more input!

[@Gejuch2233]

It's fun to watch, when I'm a operator of SMT assembly line :D BTW. I'm working on Fuji NXT machines(3 bases with two modules for 45 slots for 8mm feeders) with 225 slots with module for trays(around 40 slots). I Have only one request for people making they own PCB for SMT assembly, please place the fiducials in places that doesn't have near viases or similar thing. It will complicate the process a lot, thanks :D I'm from Poland if someone is curious, and sorry for language mistakes, I've never was good at this :/

[@arunkumar-dg4sp]

Good explanation

[@bertblankenstein3738]

My bom consolidation consists of sifting through the resistors and caps I have and plugging them in to the breadboard for my ultra low volume production hobby projects.

[@almostanengineer]

I think I’d love to see that 1c regulator, and as for the bom consolidation, that could be a pretty good live stream.

[@mpvlad89]

I’m sad I can only like this video only once! Please do the regulator video too!

[@petersage5157]

Oh yes please! I would love to see a more in-depth BOM consolidation analysis of a board like this. Your "Holy Grail" of one resistor and one capacitor value intrigues me. One project I have on the back burner is a SMD clone of a solid state guitar amplifier that many choob fanbois love for its choob-like tone. (It's the Fender Bronco amp, and I've zeroed in on the asymmetric differential amplifier at the input as responsible for simulating the distortion of a single-ended common cathode triode gain stage.) Lots of opportunities for series-parallel resistors and capacitors to consolidate the BOM down to just a few values. One of my "Holy Grail"s of choob amp design is replacing electrolytic reservoir capacitors with blocks of polyputthekettleon in parallel, since wet electrolytics tend to dry out. Surely parallel MLCCs can be used in lower voltage reservoirs, since the microphonic effect is voltage dependent.

[@Gloworm17]

I hope KZbin doesn't think you are talking about a certain explosive device. Its captions sure did. Let's hope the demonitizing gods don't strike!

[@EEVblog]

Not demonetised, yet...

[@bigjd2k]

Christopher Bassett The BOM explosion! 🤣

[@MrGoatflakes]

@@EEVblog you might not have been demonetised by KZbin yet, but were you demonised? 🤔👹

[@IanScottJohnston]

Generally, I get a working prototype with whatever the datasheets dictate (unless it's obvious) and go back later and consolidate the BOM. On large boards where you might have a good number of low tolerance components as well as lots of high tolerance ones for instance........it can SOMETIMES be more cost effective overall to upgrade the high tolerance ones to low tolerance even though the circuit doesn't need it just in order to get the next supplier price break and thus an overall BOM cost reduction. When it's your own money........you enevitably end up scrutinizing & trying everything!....LOL!

[@enercotec350]

Great Video!!!

[@hardwareful]

14:30 BOM consolidation / small capacitor values: load capacitors for crystals can be very important to get right for stability and exact frequency - so there might be caps in your design that need to be that exact value BUT! many xtal manufacturers (e.g. Abracon) offer variants of their parts with different nominal load capacitances. There usually is a window in which the xtal can be selected to use a more common capacitor already present on your board. ST has excellent resources on oscillator design, precision and reliability, informing the issue from the controller-side design.

[@lolaa2200]

At the other end of the scale is the little hobbyist single unit project. You sometimes can't buy from big name catalog because in some country they just don't want to deal with individuals, so you are stuck with whatever part your local small dealer have in stock. For this i often make a pre-draft of my schematics then set up the BOM and try to purchase the parts. Then only when i have sourced all parts (99% of time not exactly the one i intended at first) then i redesign the whole project with whatever i have been able to source. And yes small local dealer don't have 13.3K resistor, some of them don't even know that exist at all. In this situation it's actually more convenient/cheaper to use a trimpot wherever you need a very specific resistor value cause if they don't have the exact valut you can just take the next bigger value and trim it down. Recently i was designing an audio efect unit and for the filter part i needed to have relatively precise values of potentiometer, but was unable to find all the value needed in the same package form (there was only 3 !) because for small local dealer a 10k pot is a 10k pot, the physical dimension doesn't matter (sic) So at the end the BOM is almost the first step of the project, not the last as the theoretical industrial flow tend to tell us.

[@gato712]

Very interesting.

[@Kayram1]

Awesome talk. Please do 1 cent voltage regulator video

[@IgnacioMoya]

People from the CIAA Project should be watching this video

[@bukitoo8302]

I've already told them. They are releasing this new design for education. github.com/ciaa/Hardware/tree/master/PCB/EDU-FPGA

[@RyanVasquez6089]

damn, this was the PEP talk i needed

[@tinygriffy]

17:00 maybe the design house had a bunch of 11 and 13 pF in stock and didn't know what to do with them :D

[@urugulu1656]

that would be valid but why would you have them in the first place? up until now i didnt even know you can get those values in caps. and i dont think they are really neccessary...

[@ZomB1986]

If you look at farnell and sort resistors by price, you find the cheapest ones are all oddball values

[@Arachnoid_of_the_underverse]

Very interesting video Dave ,maybe you could attempt an optimisation on one of these projects and guestimate the cost saving you could have made with your consolidation process?

[@jethrobo3581]

Fantastic video! The TPS65400 is a superb quad power management unit (95% Eff.). Please let us see the 1 cent VR in action!

[@MrGoatflakes]

We definitely need to know about 1c voltage regulators :P

[@tallfred500]

Depopulating parts as a result of board testing can be a good way to reduce BoM costs on subsequent builds, but leave the pads if you have space.

[@ingmarm8858]

Nice video. I've generally noticed in my industry that the number of nutjob component values is directly related to how little of the designers own money is tied to the product. If there is almost no accountability there is no limit :-)

[@mariaselectronics3888]

LOL even though my project was a small batch of 15... I was thinking about this just before ordering because of the $ aspect. I actually changed a few parts to reduce the amount non basic parts for my JLCPCB order. Thanks for the great video! :)

[@mikemike7001]

This is the kind of practical knowledge - among all the other stuff - that makes this a great channel. I'd love to see you take it to the next level and delve into a data sheet or two. For example, assuming the four compensation networks for the TPS65400 don't really need all those different component values, what values would you actually use? And why?

[@mrwonk]

A 1 cent voltage regulator! Yes please!!!

[@ovalteen4404]

Alternate titles: Is Your BOM a Bomb, Or the Bomb? The PCB Band: You Dropped the BOM On Me How I Learned to Stop Worrying and Love the BOM I'll show myself out.

[@UweGrensing]

We SAP people have a story about a manufacturing consultant on the phone at an airport who was heard saying: "Now press F8 and check if the system explodes the BOM!" Don't know what happened to him....

[@douggale5962]

Just make sure you say the long form "Bill of Materials" when you are talking about the project on your phone at the airport.

[@volodymyrzakolodyazhny7740]

:))

[@dimitridekkers4553]

I am an SMD operator. We are ussing Mycronic machines like the MY100 and MY300. These are great and easy to operate PP machines.

[@anno5936]

Ah, GREAT... one day off and i'm watching a video about exactly that ^^ BTW, even in the smallest facillities i've ever seen there's been 2 modules attached to each other (in case of the NXT Fujis 45 parts/each).

[@rustyosgood5667]

I am an Engineering Manager at a Tier 1 CM (in the R&D department). You may be surprised at how "immature" some big customers are. Smaller customers are not expected to know much of this DFM stuff but we frequently get LARGER customers who have no idea. Our rule of "thumb" is to engage with your supplier as early as possible in the product life cycle (design / cocktail napkin?).

[@TheSadButMadLad]

2nd pass consolidation on a BOM is like the practise in programming where you optimise after you've got it working. Build a prototype board with confuser calculated values, make sure it works, then see what you can optimise out. Same in programming, get it working first, then optimise. In both cases you then know if your optimisation worked and you aren't fighting original bugs and issues in the original design.

[@HenryLoenwind]

As a software engineer I disagree to a certain degree. I've run into too many designs where optimising meant to throw the whole thing away and start from scratch---usually on software that already was productive and strict data compatibility had to be maintained. A major PITA, just because some inexperienced (or lazy, or underpaid) coder thought "it is working" was all the thought they had to invest. One of the qualities an experienced software designer/coder brings to the table is that they build their designs in a way that they can easily be optimised if needed and that they chose semantics that lend to being optimised.

[@TheSadButMadLad]

@@HenryLoenwind If you're optimising that much that a redesign is necessary, then you're not optimising - you're upgrading. Either that, or the original design was invalid from the start. Optimising is a little step by little step process. And after each step, a full regression test is done to make sure you've not introduced bugs. Agree about the qualities of an experienced software engineer, as I have practised the art for 35 years.

[@adrian10anm10]

ANOTHER VIDEO LIKE THIS PLEASE !! SAME TOPICS !!

[@martylawson1638]

Noticed all the termination resistors on the DDR lines. AFIK most FPGA's have internal resistors for termination of common IO impedance. So all of these resistors are probably redundant. I've also had good luck getting rid of series termination resistors by picking the right IO drive strength. For instance, on the Spartan 6 chips 3ma drive strength series terminates a 120ohm twisted pair almost perfectly. (very useful for a CYA connector or built-in connection to a logic analyzer.) Another good off-brand use of FPGA IO pins is as high speed analog comparators. Because they support so many different differential logic formats it's just cheaper to put a high speed rail to rail comparator on each pin instead of 4-5 different differential input stages.

[@glaubhafieber]

Just put some pride in it. First people ask “why does it take so long? That’s expensive“ and later they see that my expensive work saved them $$$

[@movax20h]

There is one more thing about this schematic and PSU. The TPS65400 do have default sequencing of 5ms by default from the factory. This can be adjusted programatically via EEPROM programming of the chip, or via external capacitors to SS3 / PG3, etc. Check section 8.6.2.15. So the C85 (68nF), C84 (68nF), C97 (68nF), C98 (68nF), can all be just probably. 1) Default will be probably all good what you need. 2) You can change it and burn into EEPROM to do other timing if needed. 3) It looks if you don't program it this way, the capacitors are required still tho (a sad thing), but they don't need to be really all 68nF. Any reasonable value will be all fine in most situations. Anything 30-100nF will probably work, as sequencing will still work, but exact timeing doesn't need to be so precise. It is important to annotate schematics and group parts to you and other that some values doesn't matter too much, and can be safely changed. I don't blame the designer for using the values recommended in the data sheet. Sometimes your volume is really small, cost doesn't matter much, and you don't want to spend few more weeks on optimizing your BOM, which do require a bit of mental burden to do, where there are other more important things to focus on.

[@UncleWalter1]

Heh. A very timely upload. I've been assembling some boards recently. Not my design. So many values could be optimised out.

[@hamishdavidson3368]

How are you solar panels fairing at home. Just watched a replay when you installed them. Did you keep any data with reference solar radiation and weather data. Have the cells maintained their rates outputs? Maybe you could do a follow up video on them.

[@AdamKeim]

Just a note - There are a couple items in the board BOM that are marked DNP, so this would bring the total reels down to 160 or lower. Still not taking into account width though, there are some pretty wide parts on there.

[@allmycircuits8850]

Lots of devices designed by company I work in had the same "military" connectors for power, communications, telemetry etc. so it could (and actially WAS lots of times) be connected wrongly. When I started to design my device, I used different connectors so it's fool-proof. But then I was heavily critisized for this solution. Found out there was a rule: order 10% more components of each type "just in case". So having 10 connectors of the same type requires to buy just 1 extra, while my "different connector at each place" requires doubling of quantity (10% extra for ONE connector is still 1, rounded UP). So yeah, BOM consolidation at its finest.

[@theantipope4354]

Nice video Dave. Another vote here for a 1¢ VREG video.