EEVblog
ฝัง
- เผยแพร่เมื่อ 3 พ.ค. 2018
- Can a product work without ANY bypass capacitors?
Dave has some fun finding out if "Muntzing" the Gigatron TTL computer works, i.e. removing all the bypass capacitors.
Lots of interesting talk on bypassing, grounding, probing, signal integrity, ground loops, loop area, noise and crosstalk.
TLDR; Don't remove all your bypass caps, they are important!
Forum: www.eevblog.com/forum/blog/eev...
Links:
www.electronicdesign.com/board...
web.mst.edu/~jfan/slides/Archa...
ewh.ieee.org/r3/enc/emcs/arch...
www.ti.com/lit/ml/slyp173/slyp...
www.xilinx.com/support/docume...
www.xilinx.com/support/docume...
Bypass Capacitor Tutorial video: • EEVblog #859 - Bypass ...
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
Stuff I recommend:
kit.com/EEVblog/
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.com/crypto-currency/
T-Shirts: teespring.com/stores/eevblog
💗 Likecoin - Coins for Likes: likecoin.pro/@eevblog/dil9/hcq3 - วิทยาศาสตร์และเทคโนโลยี
Nothing is better than a GND that's a few volts above GND.
@@Antagon666 What?
It reminds me of one time watching a guy say he didnt need any resistance in the drain of a FET amplifier. Yup he was right it worked perfectly until the devices started to age and certain loads were presented. Then the oscillations and instability started and because the smart engineer had not even bothered to put a shorting jumper in place of the resistor it meant a complete board replacement. Cost savings about $800.Cost to replace boards about 100K. Damage to customer relationships immeasurable. Its like putting oil in you car. You can save money by running with just 2 litres and the car appears to run just fine.
After both you and The 8-Bit Guy were a bit "Now what?" after assembling the Gigatron, it's fantastic that you came up with this fun and interesting way to get so much more out of the thing. Thanks.
Sometimes products that otherwise would have warranted just one video just turn out to be useful for other stuff and remind me of other video ideas. This video actually lead me to yet another idea I can do with the Gigatron, as it just happens to be a good practical example of something I've wanted to talk about.
This is exactly what the Gigatron is useful for !
The thing that made me appreciate bypass capacitors was when I made a Arduino-powered PIC programmer and I was getting all sorts of noise and random chip resets, or it would just drop out of programming mode in the middle of programming, causing it to never finish successfully. Then I realised I'd forgotten the bypass cap, added it in and all my problems went away.
Yep, they are important, and can cause all sorts of weird and wonderful problems.
Even *adding* a bypass cap can cause problems sometimes, not just lack of.
And who knows, more signal integrity issues may arise if the final product is used in an electrically-noisy environment, or with all peripheral devices plugged into the PCB at the same time... (Example, you can see here, Dave didn't connect the Audio port up.)
...unless you want random glitches every now and then.
As an Electrical Engineer,
I was often heard saying;
You can never be too thin, too rich or
have too much power-supply bypass capacitors.
Yes, digital logic is very forgiving of power supply ripple.
Repeat the test on an Analog system;
for instance an Analog-to-Digital Converter with some Op Amp inputs.
Going to get big loss of fidelity in A/D conversion.
Analog systems (RF like WiFi, Bluetooth; Audio, capacitive touch, Sensors)
will work poorly or not at all without small capacitors supplying quick, clean energy at a moments notice.
Power-Supply bypass capacitors are like studs in the wall that hold up the house.
You can continue to remove studs(or capacitors) till the roof fails and then place that last one back.
But come a storm and the house cannot withstand the new force and will fail.
So to, a minimum capacitor design will not be robust against the full range of field environments.
When the design must work 100x100 (100% of the time for 100% of units);
the minuscule price of the power-supply bypass capacitors is well worth the cost.
I recently hooked ATmega328p (not Arduino, just bare AVR) with internal RC 8 MHz and 64 ms startup time to 1 meter of WS2811 LEDs. It was powered from cheap 12V power supply (for WS) and through Pololu DCDC 5V to the chip. Even with proper caps on DC converter it locked up over time or had a startup problems. Adding a single bypass cap on AVR cleaned up power rail and make it stable. So as you said it all depends on the circuit but it is better to be save than sorry. Switching noise was visible even on Rigol DS1054Z.
I noticed Nintendo was always guilty of Muntzing when opening up their controllers and such out of curiosity. There's an outline for a capacitor, but it's not there. Yep, the capacitance of the parallel wires in the cord must have been enough... I guess...
On the last revisions of the SNES mainboard, there's still the holes for the rather large filter capacitor nearby the 7805 regulator, but of course they left that out. Didn't do a whole lot of good anyway on the first revision board, because the +5V line was tortuously routed around on its double-sided PCB. By the time the power got to the video's RGB drive transistors, it was an absolute droopy mess. Hence being able to see the slightly dimmer areas called "jailbars" and the darker spot where the DRAM is refreshed every video scanline. Probably 95% of the people in 1991 used the awful RF adapter, which adds so much more noise into the picture than the suboptimal trace routing. So, good enough!
whickervision If like I, you remember the issues of blown capacitors when those employees decided to turn coat into corporate spies, which in reality could have happened at any time, I think you'd appreciate any functioning device produced with ghosted parts, or muntzed caps to be specific. Otherwise we'd have a hole in Arizona right next to where the E.T. Game cartridges were buried, for all the Super Nintendos.
Excellent way of approaching this from every angle. Awesome video!
Always wondered about this, excellent excellent excellent in depth look at it and how things change. Well done, I'm sure it took a lot of time to do! Good fun video to get us to think about things from different ways. Thanks
That was interesting Dave, thank you. I love to see what happens under 'not recommended' conditions.
Learned about Muntzing by reading Bob Pease in electronics design. Enjoyed the video.
Interesting topic, Dave. Great explanation.
Adding multiple decoupling caps on IC supply pins to cover the "gaps" was always SOP when I was designing PCBs, but my boards weren't going into low-end TVs either, so cost wasn't an issue. I bet the Gigatron sans-caps makes a really great broadband radio jammer, as I recall my old C=64 used to really interfere with my HF ham radio reception when it was powered up.
I would like to see a test that shows bypass caps close to an IC is better.
I always like vids where you learn stuff, and practical experimentation is the best. Great vid.
Yes, I should do another practical demo on the (good) effect of bypass capacitors.
Would require a ground plane board and some controlled impedance traces. Can maybe even use my magnetic current probe to show actual ground paths in the plane.
One of my early forays into board design (after previously only doing the firmware) was to reduce the size of our main board for a different application. Boss wasn't too pleased that I had the decoupling capacitors over the other end of the board to the ICs. Fortunately easily fixed and a lesson learned.
Holy moly! I remember very well watching tv in my grandmothers living room who had a Muntz tv. I remember changing the big tan channel knob and the loud 'ca-chunk-a, ca-chunk-a' noise it made when it was turned. Also loved watching the tv repair guy coming over to repair it time to time and was fascinated watching the repairs. I think I annoyed him with my endless questions though. "..and what does that tube do?, and what does this tube do?🤣🤣
I don't know the backstory behind this "Gigatron" thing, but it seems clear that it's a low volume hobbyist product for an enthusiast audience. As such, it could easily be "over-built" and many of these caps may truly be superfluous. If I was buying or building something like this, I'd want to have a cap on each chip just because it makes me feel better, if for no other reason. People who buy a niche product like this appreciate build quality and rightfully consider the cost of "extra" parts to be negligible.
If this was a mass market consumer device sold in 6+ figure quantities for minimum cost, and which the customer would never open and judge it's quality, then things would be different. It would be worthwhile to look critically at how many of the passives were actually needed, and also shrink the PCB. A reasonable degree of "Muntzing" would be in order, and it's likely that some of these caps might be eliminated (probably not all of them of course).
The build quality of a home video game console, for example, tends to be a bit dodgy. Just good enough to be "good enough". You probably couldn't eliminate many parts from those without having issues.
A good example of products that have demonstrated sensitivity to bypass caps would be PC motherboards. Most mainstream PC motherboards from the late 90s to mid-2000s or so had high rates of failure caused by their caps failing to a high ESR state, meaning they no longer functioned as bypass.
this is classic case of modern measurement tools. The machine works fine without anything , but our measurement tools show us the signal isn't clean, thus we treat that with additional components even tho it makes no difference to the end product
You should take a look at a Muntz TV underside of the chassis and compare it to a television from a reputable manufacturer such as RCA GE zenith philco of the same era. Muntz TVs were known as gutless wonders. They worked reasonably well close to the transmitter, but had little chance of working in the far suburbs or in a fringe area. There were even stories that some months models were spray-painted underneath to obscure all the parts values so you had to go back to Muntz for service. Urban legend? I don't know. I have heard of the spray-painted chassis but never seen one in person.
I actually commented on this in the video but edited out that part to get the time down. It's in the links.
minelab a metal detector company, covers its boards in white epoxy paint.
I would like to see u parallel a second scope into this board just to see how different scopes and probes affects the signal.
Excellent video
So, this video is about bypass caps and you didn't probed the power rail just one time? :(
LOL, good point.
Thank you, I didn't knew these were not as important as I thought, I removed (out of 150 decoupling caps) 60 capacitors and added ferite beards and myproject still works flawlessly !
Very nice and educational video!
I still want to see that thing overclocked, tho.
Thank you Dave. I was wondering what was the use of the springs with my Rigol's probe. They're similar to the ones found on a Keyhin carburetor, so I couldn't figure out what use...
This is the second time I have seen an article in Extreme/Tech online Magazine about this issue lately and it answers a lot of questions about the question of a warranty being voided by just opening the case.
Those “warranty void if removed” stickers have plagued many a tinkerer over the years, but the FTC said last month that those warnings are illegal. It told six companies, including Microsoft, Nintendo, and Sony, to knock that off. The full text of the notices has been revealed, and the FTC isn’t fooling around. It gave the companies 30 days to comply with the law or the agency would pursue “legal action.”
The original notices went out to Asus, Microsoft, Nintendo, Sony, Hyundai, and HTC. The FTC believes all six companies are violating some aspect of the 1975 Magnuson-Moss Warranty Act, which controls what companies can and cannot restrict in a warranty. Specifically, the act states that no product costing more than $5 can have repair limitations in its warranty. That applies even if you have a third-party repair your device or you just want to open it up to blow out the dust.
Motherboard got its hands on all six letters through a Freedom of Information Act request. Each one details what’s wrong with the company’s warranty. In some cases, the company threatens to void warranties if certain repair procedures aren’t followed. The most troublesome provision cited by the FTC (and the one that’s gotten the most attention) is the use of warranty stickers. The FTC says you can’t void your warranty by merely removing one of these stickers, which are intended to make sure you haven’t opened up your device.
All the companies were given 30 days to fix their warranty issues in the letters dated April 9. So, we’re coming up on the deadline. The FTC letters note that investigators have made copies of the original warranty information on each company’s website and will be comparing that with any new language provided.
According to the letters, Microsoft is in hot water for restricting the right of consumers to have third-parties repair their hardware. Meanwhile, Nintendo’s warranty required consumers only to use Nintendo-certified accessories and software. Sony really went for it, designing its warranty to become void if the buyer used non-Sony peripherals or removed the tamper sticker. HTC and Asus had similar provisions to Sony’s. Hyundai required the use of Hyundai parts to keep the warranty intact.
These violations probably represent just a tiny fraction of the sketchy warranty terms out there. Hopefully, the FTC follows through on its promise of legal action if the companies don’t comply. That might scare everyone else enough to get in line.
Aloha from Maui.
This reminds me of my GCSE project, a house alarm.. mostly made with 74 series logic and some 555 timers for things like entry/exit delay. I'd never heard of bypass capacitors at that point or what they were for. My circuit worked, but it would occasionally do odd things such as arm/disarm itself, or trigger the siren for "absolutely no reason". I found that whacking a 1000uF capacitor on the board across the supply line "fixed" it most of the time.
Now I know that what I should have done was put 100nF across all of the 74 series chips and the 555 timers :)
Peter Mulholland
Allo, mate. :p
I just left a reply to somebody about my early screw-ups where I lazily left the caps off of boards, like on PIC chips. lol
I then found it impossible to flash the chips, so there must have been a lot of switching noise there, and / or a current peak.
(this was before I had an o'scope, but I still only have a crappy Hitachi V-525 50MHz one, with no digital storage.)
Alright Ash :)
If I remember right the earlier PICs had some sort of internal step up converter for the Flash.. this was of the days when Flash chips normally needed a higher programming voltage I think. Probably the noise due to that
What is the opposite of Muntzing? I tend to add capacitance and ferrites to everything. More is better. :)
You can never be enough on the safe side :)
Paranoia, over-engineering, wastefulness, etc.
I recall a story Jerri Ellsworth tells about the C64TV, she shipped the design to the factory in China, who promptly removed ALL the decoupling caps, they were confused as to why it wasn't working, she went to the factory and said "WTF" and had to start sticking stuff back in.
I was muntzing as an early teen when I cut the capacitor off my scalextric slot car motot LOL, few years ago now! Thxs for the vid..
Love it! In most of our prototype builds we didn't bother with caps to be honest :-)
But we wouldn't have wanted to see you review our board if it had come without any of them included. #859 is a great reference.
This would be a great option for all those complaining about the price. You could offer a "non-deterministic operation feature" option for $2 less!
Nick - to reduce the sale price I would suggest selling it without the fancy wooden box as option 1 and selling just the PCB and programmed EPROM as option 2. I can see many people already having most of the components or even wanting to build it with 74LSxx, etc. instead. I can also see people discussing their results in the forums when using the different families, etc. I used to have a lot of 74 series logic including pulled parts that one would like to put to good use.
beopstek . I know enough circuits that will be totally unstable without proper rail voltage filtering.
Though the circuit may work fine, I'd imagine that it's probably backfeeding noise into the power source, and also producing more RF interference. So it's not so much about the circuit's own stability, but how it might affect other equipment too.
Dave, do you have a video on the Fet probe already? if not, can you make one? it would be an interesting video comparing traditional probes vs the active ones.
I suspect the steps in the leading and trailing edges are reflections on the blue trace around 31m20s are due to un-terminated clock lines. Obviously not a problem here due to the nice slow clock. The positive going half-step is the source switching on and effectively charging up the transmission line, the second is when the incident wave comes back. Similarly with the negative going edge. I would like to see a "proper" investigation on ICs that are actually doing some work shuffling some charge around, not just 6.25MHz.
It may work without the bypass caps but try keying a low power RF transmitter up anywhere near the board, I bet that would bring it to its knees.
In the lab those transmitters are everywhere not to mention personal items like phones etc.
Joel Caulkins True, my thought to even before he startet and what about radiating noise.....
It would be interesting if you mentioned what kind of power supply you used. I recently had some issues that only occurred after I switched from my linear Rigol DP831 supply to a normal/low-quality switching supply.
It might work on the engineering bench, wait and see what happens when you start shipping them, better staff up the warranty dept... Ask me how I know. We had a DSP board that would randomly lock up in the feild, never happened in the lab, turned out it was a power decoupling and current sourcing problem, we thought for a long time it was the firmware, very very difficult problem to solve, and for audio gear that had a live audience of a few hundred to thousands of people attending, not a good thing... Better to use way more power supply than you think is necessary. When you lock up, it’ll usually be a fast glitch not noise, so very hard to find especially when it takes days to weeks to happen. Over the years this has happened to so many people that we all just use tons of caps everywhere. Like Nike says; Just do it.
Looks like some of the noise you were measuring was from crosstalk from adjacent signals. It would be interesting to see if you could measure adjacent traces to see if you could determine if one or the other was the aggressor.
Also the hips on the channel 2 trace of the clock looks like a reflection due to impedance mismatch or, you may have been probing at mid-trace where the reflections are most prominent. Perhaps another subject for a video.
Keep 'em coming Dave
I use to have two Commodore VIC20 and one of them lost a lot of capacitors, like most of them and was still working. Also his keyboard driver was plug and play :)
Love you Dave..
Hi Dave, your studio/lab is a fairly innocuous environment, I'll wager. Likely near a comfy 22 degrees C? So this is an interesting inquiry, but likely not the final say-so for whether bypass caps are ever needed. This was a great controversy stirring video and thanks for producing it. I wasn't familiar with Muntz or Muntzing, but I did learn long ago, an apocryphal story about how US military contracted electronic designs were treated. Similarly, a contracted design was presented to the department of defense and they removed one part at a time from the circuits until it stopped working. If they removed 50 parts, they had a working, least expensive design, with 49 potential field revisions. If at all true, I'm sure this was done many years ago when electronics were expensive and not ubiquitous, and all hand assembled, and designed to be serviced, and labor was cheap, and, and....
Dave, Muntz would have been proud. I can remember him from the old days but I never heard him used as a verb. LOL
Shouldn't you rather find it's maximum working clock frequency first and then compare it to the maximum achievable one after removing some caps!?
Or otherwise show how it is failing with conditions fabricated, that demand using those bypass caps.
Good idea
Prehistoricman Good idea if he could. He doesn't know their exact layout, so how would he increase the clock frequency? If it's related to the power supply then sure. But otherwise this is infeasible without contacting the creators at the very least.
You don't need a layout to be able to spot the only clock crystal on the board. Plus he said in this video that he was scoping clock lines, so... I think it's fairly obvious to him what's generating the clock.
Also there is a section of the board labelled "clock generator"
Prehistoricman I don't recall there being a discrete clock crystal on the board nor any mentioned in the build video, but it should be just as easy to look up the CLK pin on any chip's schematic, without relying on silk screening print accuracy, and if you've got this spec sheet, it would also publish your maximum CLK freq.; any further experimentation into higher frequencies should open up jellybean considerations from much overclocking experience, which our host had NOT YET demonstrated in any of the vids of his, IMO.
About the layer order thing at 12 minutes...I get that you can use it as capacitance, but wouldn't it be nice in terms of EMI to keep signals inside of two ground/power planes?
I remember lashing up a 74LS163 TTL counter to a power supply without any decoupling and it actually refused to count. A 100n across the power rail soon made it start working. Doing Muntzing on a complex design is a bit fraught because there's so many variables. Who's to know if a certain sequence of instructions fails, due to a build up of edges overshoot.
Muntz was probably removing parasitic suppression on TV's that had no sequential states. Nowadays, one spurious glitch caused by a rare pattern can make havoc on all subsequent states. The only ways to fix these is with expensive test equipment or shotgunning. Better to keep the protection overhead using good design practices.
Pretty cool video dave! Maybe I can try something like that on some different products.
Your scope looks triggered.
Bloody Millennial scopes!
you sir have won two internet points. Well done!
Late response but that was brilliant! Everyone someone says 'triggered' I have to ask what kind of scope they are.
:-) I'm happy that we could all have a laugh at Dave's perpetually triggered scope. Go ahead and meme the hell out of that. I doubt that the perpetually triggered would even know what a scope is because, you know, being an "activist" is more important than getting an actual education.
This needs to be on t-shirts!
The glitch in the middle of the transition at 30:52 looks like the classic effect of an un-terminated transmission line. It would be interesting to know the path of the track carrying this signal and the gates connected to it
Yes, but the path is likely very ugly in terms of chopped up ground plane. Hopefully I'll have another video on this soon.
Great fun. Thanks!
Outstanding.
So, could one do a inner ground plane with a positive rail fill on the top layer, to produce the required capacitance? Possibly?
Be interesting to see the numbers run on that. The layers are probably not close enough but it might be something that board manufacturers could build in (probably not worth the extra expense though)
The bypass caps keep the power rail clean. Why ur not measuring on the Power Rail and show us if its clean as before ?
do you really have to worry about clean power line if the device still works? That's the whole point. In audio amplifier it would be a fail if removing the bypass cap around opamps would cause even small increase of hum or noise on the output, but here in digital world you have only 1 and 0. It works or it doesn't :)
EDIT. Of course we all know that you prepare your product for the worst conditions it will be operating in and we know that sometimes saying "it still works" is not good enough.
Kriss Olshaq . Yes you do. Maybe not for your arduino. But if you develop equipment that needs to meet certain EMC criteria, removing one tiny cap can ruin your day. In a MOSFET snubber circuit for example. Or circuit for high sensitive sensing, measure equipment, equipment for medical purposes, the list goes on and on. Not to speak about reliability issues on a production scale.
i've taken it in count in my comment hence
"it still works" is not good enough.
This is a 2 layer board. I bet it will fail EMC spectacularly with the caps sniffed off.
Do you have some viable reason, other than a very unlikely personal recognition of THE particular capacitor being used, to suspect that the setup is technically unviable? OR would your highness simply love to see Mr. J's scope draw more monochromatic stick figures for the rest of us to ogle? Hate those Bagginses forever, I will.
I met a technician that worked in a TV factory fixing the new ones that didn't work.
When he had no work, the factory manager told him to remove parts from the TVs
to determine all the parts that could be omitted to save money. Then more and more parts were omitted on the production line! (BACK IN THE 50'S)
Probably the TVs worked OK until they were in a weak signal area, or ran into some interference they were no longer able to cope with.
Have you made a video on the subject of PDN?
Nelson Muntz-ing?! Ha-Ha!.. looks like Dave was expecting the bypass caps to have an effect...i sorta did too!..i guess we both deserve the nelson ha-ha!...nice info on the probes!
I've thrown together quick circuits with nothing more than a PIC microcontroller using the internal oscillator (and, I think, a pullup resistor). Wouldn't do it for anything important but it worked just fine. This for battery supply and USB 5V so reasonably controlled voltage.
Dave, do I need per component bypass caps? Lets say my MEMS microphone needs a 1uF on its power input and my ldo power supply for it needs a 1uF on it's output, do I now need a 2uF or just one 1uF?
New here to the term "Muntzing" but Jerri Ellsworth said in an article that her first big project got an un-requested deletion of bypass caps by a stingy Chinese contract manufacturer. She said she was about to flip out because the circuits, needed urgently for a toy being made, weren't working. She figured out what they'd done, straightened the people out, and, problem solved!
Muntz made TVs for urban dwellers who lived right next to the station broadcast towers, those were the components he removed. Otherwise his sets were crap, which is why no one but large cities ever had his TVs.
But the bypass caps also provide a burst of power when the chip speeds up or its demand increases. So without them they’ll experience issues at higher loud?
Any idea why the through-hole DIP packages have the power pins so far apart? Not the best thing in the world to keep the tracks to the caps short.
I love the idea!
A product without bypass capacitors? Hell yea! Can't believe you said no or maybe!
I didn't see the rubber feet in the last 2 vids about the GTron!
Maybe Meticulously Muntz it Mate till it's More-or-less Munted?
You can remove your roof when it is not raining. So what? Wait until it rains! There is no point to discuss whether the roof is needed when you don't pour rain on it!
That is the essence of it.
I don't think this board will work even with bypass caps if you leave it in the rain for a while
The biggest take-home from this video for me is "pfteenth" as a unit of capacitance. I've already got away with using "half a bee's dick" at work, so the next challenge is to sneak that one in.
Check out my probes... But wait there's more
Great video, great info! Thanks Dave
Say tuned for modern-day Muntzing...
Can you do this on an analogue machine, or something where it has a more dramatic effect and really matters?
It's mostly a thing with high slew rate digital signals. But anything that gulps current benefits from bypassing to some extent.
It would be nice to see the difference between crap scope earthing and low inductance probes.
Interesting video. Maybe a variation on muntzing is to deliberately break various components on a working board to demonstrate different kinds of failures and their symptoms.
I ran out of decoupling caps to snip...
Thanks Dave. An interesting way to demonstrate the effect of filter caps.
Wow, you gave the MDO3000 an airing, have you made peace with its UI yet, or like me, do you still find it a bit of a fight?
Okay do this again but with a spectrum analyzer so we can see how the emissions vary as you remove the capacitors.
Maybe also include an overlay that estimates the total capacitance remaining on each voltage bus?
So what's not revealed until later is that you were using averaging on those probed clock signals, which hides any random glitches that might occur due to the bypass caps being removed. So it's not really surprising that the traces didn't change much! The impression I get is that most chips will seem to work fine without bypass caps most of the time, but they will be more prone to random hard-to-reproduce failures when the peaks of the additional noise (that was hidden by your averaging trace on the scope) happen to line up. They might also die sooner from being overvolted or reverse biased by glitches more often.
I would love to see the EMI radiation before and after removing the caps. The garbage we see on the signal will radiate out. (on the other side: the caps also allow sharp switching of the signals, so with caps, the should have more HF-Radiation and without, we should have more radiation in general)
A more interesting version would be when the computer was built up with CMOS technique. Here the effective load impedance by the TTL-technique is quite low so I assume that any ringing is well attenuated. This would not be the case for CMOS-components.
But how much of a difference is the scope ground lead clip?
I think this is a great thought exercise. If you can cut it out and it still works then you should be able to explain why you actually still need it. No, we wouldn't really rip out a bunch of stuff and ship it (well there was that Chinese water boiler stick thing...).
True. And if you are optimising a design for space or cost or something else then they are genuine questions.
*💀should?* that's impossible! If we snip a thing out and it still works, then it's not so critical, so I must say we should be able to explain why we might WANT it; but for "need" it, we had proven that we 'do not'.
I disagree with the comments about long-term "glitches"; Dave covered his arse @ 15:18!
I'm wingy tonight! "Wingy": pronounced win-gee (win-G), Australian for someone who is in the mood to complain about stuff:)
If there were termination resistors at the ends or source of all those transmission lines I wonder whether madman would be spinning in his grave if you snipped them off.
If it saved a cent and still worked he wouldn't have cared!
A much more innocent time all round, where people didn't care so much when a program crashed every few hours or there was a bit of noise on their TV channel, or maybe their Ford Escort 1100 didn't start when it was cold! (Cold? What's cold? I hear all the Australians and Texans say). These days if you can spend 5 cents on caps and save a lot of legal problems in 2 years time, you do (assuming you aren't in China and so immune to prosecution)
Zadster which is why I'm avoiding a trip to Texas if at all possible, but the Chinese aren't immune to persecution, as seen in the difference between seagate's HDD being less endowed than Malay's, and people opening up their tamper resistant RAID cluster boxes to see which of the two they have, and sending the Chinese-outfitted kits back to the PoS/retailer as 'unused'.(unwanted)
What would happen if you did this to, say, a high frequency boost converter?
If it works at all, it would probably create voltage spikes large enough to damage any sensitive circuits it's powering.
It would also make a great RF noise source. I was working on one a few days ago that had 5Vpp ripple and was spewing noise from about 100MHz to 350MHz because the idiot who designed it made the trace to the bypass capacitor go half way across the board.
Mad Mason Muntz seems to have been right, but the AND Gate is really great :-)
Could you order boards with thinner prepreg between the pwr and gnd planes for higher capacitance?
Yes, you can specify the pre-preg thickness you want. Not at the cheap shared panel proto houses though.
I saw some weird video compression at 4:27 just after the scene cut. No where else. Don't know why.
Now try doing same with some precision analog stuff, like opamps etc.
Or even any analog circuitry as in Muntz days.
The AP notes at 33:46 is a bit misleading. Who would parallel 26 pieces 0402 capacitors to de-coupling one pin?
From my time in the TV trade, I can say that you can't give a definitive answer on whether bypass caps are needed. Quite often some samples of a given model showed instability, and the manufacturer advised fitting extra bypass caps. Which cured the problem. Other samples didn't show any instability though. Probably due to tolerances in stage gain.
Which is why it's better to overdo the bypass caps somewhat, as it's preferable to having post-production reliability problems which cost more to rectify and damage your brand's reputation.
Muntzing is what chinese vendors do with stuff like powersupplies... you often find fuses and line filters designed into circuits, they even make it to the pcb-design, but instead of the components there are just some wires. I imagine that they have some muntzing-specialist who removes all "unneeded" components. He makes all the money-saving decisions like removing strain relief on power cables and saving a cent on the enclosures so they don't need to put screws in them, or glue or ultrasonic welding. ;)
You need those for the special edition you send in for EMI certification.
First class scope and kit.
150$ worth of kit. Ought to be first rate if it's built right.
You need to look at the noise on the power rail which will change significantly. You also need to look at conducted and radiated emissions from the board. Without bypass capacitors the board would probably not pass a EMC test. I have come across some badly designed commercial systems that had no decoupling of the 5 volt rail following a 78L05 voltage regulator. The 78 series regulators are notorious for self oscillating at a few hundred KHz and one particular board would only work with CMOS chips from one manufacturer without proper decoupling. This was probably due to the chips exhibiting a different amount of capacitance across the 5 volt rail. One particular system was used to trigger halon release in a fire protection system and this could have been a potentially deadly situation! More and more power supply systems pay no regard to EMC and omit common mode chokes and decoupling components. Whilst these will still work they are just adding to the sea of electromagnetic interference. It is good engineering practice to keep loop area as small as possible and to decouple circuits with multiple capacitors (and sometimes RF chokes) of different 'appropriate' values and to look at the noise on the power supply line in both the time and frequency domain to ensure that both conducted and radiated noise is kept to an absolute minimum. The other problem with omitting decoupling can be a lack of immunity to local RF (EMC is a two way street). Once again a circuit may work without decoupling but placing it in the proximity of an RF field from say a mobile phone may cause it to fail or behave in an erratic manner.
You should repeat the test, but insert a series resistance in the power source ;)
Those diodes made my OCD happy.
Should do some muntzing on a modern PC. There is a lot of bypass caps around the CPU, GPU, PCH etc...
I know some caps get knocked off on the CPU substrate by people attempting to de-lidding for better overclocking and they seem to work fine with some SM caps broken off.
Standard SMD 100n bypass caps are so cheap, there's no reason to even think about removing them.
It's an interesting experiment/video to show what they are for though.
I bet you don't even need all those expensive chips either. I'm sure some have been added just for looks, I mean consider how much it cost and how little it really does.
Err, there are no expensive chips in this, it's all standard cheap-as-chips HTC logic.
Absolutely - the makers of those counterfeit USB drives discovered long ago you don't really need most of the memory at all until you try to actually fill up the drive with data... :)))
so bypass caps on a usb charger needed ?
Of course there is the EMC argument to be had too. Plenty of stuff can work just fine whilst disrupting other stuff around it.
Even plane to plane capacitance is relevant. Consult with Mark Montrose. I have attended his lecture about EMC/EMI compatibility...
The are supposed to protect the circuit from spikes which would normally destroy the ic.
Check out the Pentagon 128 from russia ;) ~100 TTLs + your average Z80 configuration - with no individual decoupling caps in the layout (2-3 bulk capacitors tho). Usually works stable enough to get the machine to boot - any further action usually results in a crash tho.
Cool video by the way :-)