EEVblog 1418 - The Most EMBARRASSING Repair!
ฝัง
- เผยแพร่เมื่อ 13 ก.ย. 2021
- This "Repair" of a Keysight DSOX1102G oscilloscope was a tad embarrassing...
Forum: www.eevblog.com/forum/blog/ee...
Subscribe on Odysee: odysee.com/@eevblog:7
EEVblog Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
EEVdiscover: / eevdiscover
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
#ElectronicsCreators #Repair #Hack - วิทยาศาสตร์และเทคโนโลยี
Dave exercises his right to repair after exercising his right to screw it up.
But he owns the product and uses it as he sees fit. Lol
I sure hope he remembers to screw the board back down or it might mess things up.
@@randallneikirk6385 Probably about as well as he remembers what experiments he's done in the past! Hehe
Keysight have recently been a topic in the EEVBlog forums because they now appear to be refusing to provide supply, repair or calibration service for individuals and only providing service for companies. If they are not going to support hobbyists then I'm going to be avoiding them and will not be recommending their products to my clients. Seems like a bit of an own goal because hobbyists and students become equipment specifiers in the future.
Yes, it's stupid. But managment aren't going to go against what the lawyers advise. And it's some dumb arse laws in certain countries that have forced them to do this. I'm pretty sure they don't *want* to do it.
@@EEVblog I don't believe you for a second. They want the almighty dollar like every other corporate scum. I cannot see how certain country's laws have anything to do with sharing vital information on how to repair products. These scopes are not life or death, nor are they national security issues. Best case I can see is they don't want their IP stolen.
We need laws against this sorta crap. Users should be able to repair their own purchased products and when you purchase something, you own it! Consumer electronics should come with schematics (or be provided on request if there's a lot of them (cell phones, etc.))! Replacement parts should be available *from the manufacturer* at sane profit margins! We're killing our planet with this throw away consumerism bullshit -- it needs to change or we're all f'd!
P.S. Thanks again for partnering with Brymen to produce a kick arse multimeter; love it.
@@jimmio3727 We are talking about slightly different things. This isn't about repair info, it's about product availability in a certain country to consumers vs "professionals". IIRC even Farnell /Element 14 won't supply Keysight scoeps to "consumers" in one country. There is some silly law in that country that prevents dealers selling to anyone deemed to be a consumer. There are lengthy threads on the EEVblog forum about it.
If Keysight wanted to do this for "almighty dollar" then they would do it world wide. But they haven't, it's only in that one country(s) with some PITA quirk in the law.
It's just like how they can't give away scope in certain countries because of local laws.
@@jimmio3727 ahh I just wish I could afford one..
Was reading here. Are they trying to prevent just regular Mr. consumer from buying test equipment to try and stop anyone from doing there own repairs? Did I word that right? lol
this is why it makes sense recording yourself disassembling something, even if you don't upload it anywhere....you can always go through your archive and watch why you have done something
You should've soldered one side of the resistor to the board and left the other side flapping in the breeze so when you come back to it the original part is there :)
Yeah I've seen people turn them 90º so they know what goes back where.
@@Okurka. I commented too soon as I heard Dave mention exactly this...LOL
@@SidneyCritic that's also a good plan, maybe even heat shrink the end in case the envelope of "which"device get's turned to a side in which gravity could cause a short from as Dave says "flappin in the breeze". At min, it's good to keep it there no doubt.
@@jasonbrindamour903 gets*
I do this too
Back in the 70,s I worked for Tektronix as a service engineer. The problem you saw used to be down to corrosion/ galvonics in the "Y" input select switching. DC offsets. Oh what fun. Clean those wafer switches. :)
If I had to guess, I would say that different devices that have different hardware may need to have their offsets compensated for in software, and so by changing it so there is a mismatch between the physical hardware and what the software thinks is present, those compensations end up producing what appears as an offset.
I dunno, if I were the engineer designing this thing I'd have to leave a couple idiot traps lying around. Like one of those video games where if you play a bootleg version it'll act like it's working but it absolutely won't let you beat level 4. If you manage to set R174 to 4.7k it thinks it's a Fisher Price My First DSO.
LOL
No. Just no. The problem is that shit will break and legitimate consumers/clients will get bit in the ass by your genius plan, as they were already by those things in these clever DRMs...
Stereo amplifier repair 101: Use the good channel to trouble shoot the bad one. Have been doing that since my early teens. Of course at that age you often blow up the good channel trying to fix the bad one! My Dynaco 120 of the day is a good example. That was before I understood ultrasonic oscillation in baseband audio circuits
Licensing... on a piece of equipment you bought.
That's like buying a calculator and having to pay a license to enable multiplication and division. Utterly ridiculous! Why do people put up with it?
Because we don’t have a choice. Every company does this now.
Do you really want to go back to the times where they actually made 3 different low to mid range scopes? It will cost way more and you can not upgrade it.
This is your choice what product package you want to get in the first place. Why would you get the calculator with disabled multiplication and division in the first place if you need it?
More than half of the Australian population are still locked up in their homes with most on a curfew after 18 months? People are rediculously compliant. Sad
I bet anything you want that you own a smartphone that doesn't let you jailbreak it at all or without an absurd amount of extremely difficult steps.
an we won't even begin with Windows or EA's expansion packs ffs.
Why should I have to pay for multiplication and division when all I want to do is add and subtract? Are you suggesting I should have no choice of configuration? Or perhaps do you think it would be magically cheaper for a manufacturer to run multiple production lines for different models to cater for all the combinations of features customers want?
License code? Jesus, that's straight up evil.
It has multiple license codes. Sig. gen. option is just one of them, around 15 in total if I remember it correctly. If you want all the features then all the licences may cost you almost half of what the scope itself costs. And with some manufacturers it is even worse.
Welcome to the new world order.
Same thing for new cars. Oh, you want the full potential of your headlights? Pay a monthly fee or 3000$. This shit is crazy. Will never support this BS
@@Daniel-hd7gq At least once you've paid for your option it is all yours and you can use it as long as your device is alive. Some commercial software providers do not even give you that. They have switched to using time limited licenses only for their most recent products and you never know exactly how much it will cost you next year. Say hello to the new SolidWorks CAD packages for example. And with current cloud based services development it looks like many more are yet to come.
How embarrassing, but being naked behind the camera while filming reflective materials, did it not cross your mind this could happen?
What? Where
It happened, but we won't take away the fun of reviewing all of Dave's vids to spot it. 🤗
People do that deliberately with teapots etc for online selling.
@@justice83 that is way too many videos to sift through! At least narrow it down. I am guessing one of the dumpster dives particularly for the TV's or monitors
"I didn't change anything about it!" It's always the same line users give when their stuff stops working...
That or "only lightly used"
Repairer opens equipment and shows customer wires with pots dangling off them.
Customer: "I have no idea how that got in there."
What's the saying? Anyone who tries to make something completely foolproof is underestimating the abilities of complete fools!
Maybe the ID by voltage divider was a windowed range and you were out. Maybe your ID presented an ID for a product with a different hardware.
Keysight: "Oi mate! Ya got a loicense fer dat do ya?!"
You need to buy the Australian English license for that.
At least Murphy didn't get you this time :D
When I do similar modifications like you did, I usually solder those original resistor from one leg ( Just to keep them in place and not electrically connected ). You also might use any sort of electrical tapes and glue them to the oscilloscope's chassis.
Good tip! Will have to remember that one. Thanks!
@@LordPrecision In 2021, this should be standard operating procedure, yet how many of us don't....?
A Keysight BLT is gonna be delicious but very pricey ....cheers.
And you have to pay for the bacon license key.
@@EEVblog touché!
@@EEVblog Try the BLAT: A for avocado. It's a premium add-on however.
I can see how invalid strapping could lead to problems like this. At my job, regulations require handling "impossible" settings, but in most of the embedded world, they just aren't considered. After all, it can't happen. 🤔😉
A good visual inspection through a magnifier will find places where the solder has cracked around a through pin and even though it looks fine to a naked eye it isn't making electrical contact with the through component at _all_ anymore. Years of heating and cooling sometimes cracks the solder, especially if it was a little scant. My experience is old electrolytic caps and obvious things like this and solder balls account for a huge percentage of faults. Add in the occasional failed chip/diode (check the inputs they're all good, check the outputs, not so much), and you account for almost all. By the time you get that deep into it you'll be ready to find the occasional weird fault.
1:52 That's why they call it Keysight needing a license key to use certain stuff.
It's not so much Keysight as it is no key, no sight.
When it is first calibrated it probably stores those initial resistor divider values as part of the calibration data and if the hardware no longer matches the stored value of the resistor divider it just assumes that the calibration data is faulty and something is wrong so never loads the rest of the calibration data.
Re: the question about different configurations: the most reasonable explanation to me is that the analog side of the DC offset generation circuitry differs between models, or even hardware revisions. Those set resistors are supposed to tell the firmware which hardware it's currently running on. So with this explanation it's not so much that the calibration data is ruined per se, but that the stimulus to the DC offset generation is ass backwards because it's for a different hardware configuration.
Yeah, but that's the wierd part, there is no difference in the front end hardware between models.
@@EEVblog Another possibility is that there was supposed to be a different model with changes in the hardware that was never released. That sort of thing (left over code for variants or features that never made it to market) happens all the time.
I'm glad to hear I'm not the only person who allows projects to age like a fine wine before getting around to them.
Did anyone else notice that the "CANINE_MODEL_NAME: MARSUPIAL" was below the BLT_PRODUCT_CONFIG?
That's probably normal for a scope shipped to 'straigha.
Actually the behaviour at the beginning seems to be just a setting in the menu!
Under Utility -> Options -> Preferences you can select between "Expand about Ground" or "Expand about Center". I think your scope was set to the second option, so when you changed vertical scale you also changed the position of ground on the screen.
That feature moves the ground reference indicator as it changes the scale. This unit was moving the waveform but leaving the ground point at the same place.
@@donreid358 True, my bad.
I thought there would not be anything a scope could do which I couldn't fix... until I was presented with this video. LOL
I do wonder, if this would have still happened, if you had used the "reset to factory settings" option after changing the configuration resistors. You probably did something outside what the software developers anticipated :)
I could imagine that data left over from before the change + new data saved afterwards, might cause a problem.
Hey Dave. Thanks for making me feel better about mods that I have totally botched.
My old DS1052E is still doing it's 1102 impressions well.
Nice one Dave 👍
If that scope isn't of any use to you now that it has been "amateur modified" ;) then do feel free to send it my way. I badly need something to replace my old Philips 100MHz.
Send it to me... i have a handheld scope from maplin, or a very scary crt unit from 1958.
My most "embarrassing" repair was when I was desoldering the SMT pins on an OKI 900 with a Weller 12 watt. I thought they were wick'd clean and hot and impatiently yanked. Lifted a few pcb traces. :(
Patience is the highest virtue. I'm about 22 years older than I was back then. Entropy has justly slowed me down. :-) Dave rules!!!
...see, this is what happens when you have 20+ scopes
I recently had a very bad trim pot. Wanted to use a variable buck converter to charge a Li-Ion cell, but while slowly turning the pot up, my meter did not say the expected 4.0 - 4.1 - 4.2V increase, but instead the voltage jumped around 4.1 - 4.3 - 3.9 - 4.2V. And that was even repeatable: turning the pot up (towards max voltage) would lower the voltage in certain positions.
I only ever know a small % of what you're talking about but I always enjoy your videos. Someday I'll understand EE more.
You can have various product codes for minor hardware changes and firmware will have no means to suspect HW/SW mismatch. So yes, it's possible to set product code that works with some quirks.
As for software being involved in this error, you likely found an edge case that was not compensated for. I've had these kinds of issues before while developing things and trying to fully define the boundary of a configuration. Also, I don't think any developer would look at this particular part of the system and think that anyone would get in there and start mucking around. They probably had the specs from the hardware engineers and didn't account for much, if any, variation from those specs as it would be reasonable to not expect those values to be in an undefined state..
I would have at least had the software throw up an undefined mode error message or something.
The strange bit is why the mode even effects the DC offset of the channel, when everything else seems to continue working.
@@EEVblog That may just be a byproduct of the unknown/floating state. Now, a proper software shop wouldn't move beyond the logic building stage until the team had found all possible output boundaries of all possible inputs. Wait, that's right! I forgot that no business includes that level of overtime! Lol
Maybe it thought that it was a completely different model, where the analog front-end is constructed slightly differently, and the configuration it was running would work with that model but would be incompatible with the hardware present.
Wow, the signal generator, despite the hardware being all fully present, is not enabled....
It's pretty scummy when companies cripple features of their devices with software lockouts, and charge extra for them.
I don't know how but I kind of guessed you'd say you regret selling the COS6100 scope. And sure enough you said it. Those old bearded virgins knew what they were up to. FWIW, I'd never heard of one until you mentioned it here.
Don’t be too hard on yourself; We all shit in our own bed from time to time!
23:00 bulging capacitor at the upper left?
From a programmers point of view, I guess that the different modes can result in that different data structures are expected in some sort of write-once memory with rough hardware-calibrations, so that after assembly and product specification, the different offset and characteristics values are brought into the ballpark for user calibrations to work. Using this logic, it could be that the hardware channel DC offsets are no longer read from the correct place, and just gets garbage instead.
"It's a software induced fault".....nooooo it's a Dave induced fault. I am glad that I am not the only one to fluff-up 'smart ideas' 😆
I wonder if they apply different loads to the divider to measure output resistance as well as Thevenin voltage... a security precaution.
Don’t turn it on, crank it until it pops! 🤣
We still talking about scopes? This [is] a family oriented channel :O) Cheers WS
It would be interesting if you ran a poll on people who did your original hack to see if they get the same offset issue you have. I think the takeway is if you want to trust your test gear then best not to modify it.
i work with tablet testing machines and all models that we sell use the same mainboard. and are just configured to act as different models. the hardware of some models need different config files that will be loaded from memory depending on which model the board currently identifies as. so changing what the device thinks it is can missmatch the hardware and software. sometimes this mismatch can cause the software to crash and your device wont even boot and sometimes everything appears to be normal until you hit that one line in the config that needs to have a different value. then you got two options: either the software crashes at this point or you get outputs that are weird.
I can remember the old Tektronix 7603 Cro - had to adjust the DC balance in the 7A26 (and others) Vertical plug in
The cal data is likely specific to the hardware model. Other models could need additional data or different values for their hardware. If the stored cal data that is found is not for the hardware indicated by the jumpers then the software would think it couldn't use it...
I wonder how many people 'rooted' their scopes by following the modding video!!
Now apparently they want to put fixed panels above roads instead of in roads.
Maybe different product configurations have different calibration requirements, so if you change the product, the calibration that exists doesn't "fit" anymore.
Like, sometimes the hardware differs within the product range, right?
And if you try to run user cal when the software thinks there's hardware that isn't actually there, it seems reasonable that it could fail.
Looks like that KeySight came from the "EEVBlog Dumpster" better give it away for safety.
Hang on. Dave stores his resistors based on the E series position?? 47R with 470R with 47kR? Maniac.
Yep, incredibly efficient way to do it.
That's how I sort my smd resistors and caps. Super Friggin handy
Why would you not? Sub-dividing by multiple is easier to index than integer value. MSB vs LSB
I guess it depends on how many values you're holding, how they're stored and how they're used. A fair few of my stocked resistors are through-hole and from big dumb packs which are all over the map E-wise, and often they're deployed more as "I need something about... -yeah this one's good enough for 'straya..." I'll certainly admit it makes sense for reeled SMD parts, especially if you're only holding say, the E12 or 24 set.
Haha, my nickname is DCOffset, and it was SO STRANGE to watch this video )))
Dave 2 has long since left.... so blame him.
Dave, it's common practice to use identical software and identical mainboards for multiple product SKUs with different features, and identify the specific SKU with resistors on multifunction GPIO pins so you don't need to rework and retest everything for every configuration every time you make a change to something. (You can test disabling individual features once, and then test a feature rich build to be sure that software changes to specific features don't step on each other).
If you get those GPIOs messed up enough and the software is looking up product features in a table, who knows what combination of features it thinks your scope has... it could go right off the end of a feature table and use some garbage configuration that Keysight doesn't even sell.
Yes, I know. But in this case it's not hard to throw up an error message if your *product model* configuration is not one of a set number of valid configration. Remember this is not the feature configuration bits.
"Breaking security label voids warranty..." Ha ha
It's a classic case of the Murphy-Jones effect
If I stack my three COS5060A 3 channel scopes, will I have a triple screen 9 channel scope?
Also a Keysight fail, as it seems quite nonsensical to need a license to activate the signal generator.
This is why I don't hack things I need. Obviously, there's tons of other scopes in that lab... so while I can't have this sort of fun with my equipment, fun to watch :)
Well I think what it showed is the steps leading up to that mod or the mod produced that outcome. After that it is a debate. The most obvious things I could think of is the settings have a hardware string or something in there that has to match or be rejected. Also there could be some small hardware somewhere that does not get along with the different model number, or the memory size of the settings could differ for the 2 models.
I didn't hear " is good enough for Australia " this time..
Question.. Low battery voltage? I work on old radios/transceivers. I have found certain makes and models along with other types of equipment will act all funny, not boot right, other artifacts due to a low battery.
I see the brass repair tool in the background ;)
I have the new version of this and I like the Black as it does not show Fingerprints and crap and it works Excellent about as close to my tek 2465A, but getting thru menus is a bit different
The black case really screws up my camera exposure, I might use this now fixed unit in videos from now on.
I can understand using the same code across all their scopes (with model specific tweaks), but its interesting that instead of building the code for an individual scope, the whole firmware is the same across all the scopes. I guess that means you could buy the most basic scope and bodge on a bunch of connectors and you would get the expensive model... 🤔
19:00 man it really irritates me when hardware guys complain about software. In this case, you told the software that some hardware module was present. So the software tried to calibrate it and couldn't.
Err, no. All the harwdare is present in this machine. There is no reason why a model bit change should impact the DC offset.
"Most embarrassing repair"? Shit, you found the two NES consoles I tried to get working again, didn't you?
Thanks for sharing this. Helps to know even the pros make some gaffs once in awhile.
As far as the programming - it could be just as simple as an inaccurate error message. That is, if the waveform generator can't run for ANY reason, it displays "invalid license", and for whatever reason the waveform generator can't run when the system is uncalibrated.
Is that thing using a hardware identification and config circuit with voltage dividers? Analog coding versus digital?
As a programmer I can say we don't develop for the possibility that somebody solders off the hardware.
Given the Product is set be resistors, they don't allow for someone using trim pots that might be half way between two different Product ID's. I don't think they would see a need for it given they set the resistor values in the factory. This really was a hardware fault as the resistor values were wrong. :)
Lost calibration - check the lithium coin cell visible at 6:20.
On balance, people learn more from their failures than their successes.
"Ah yeah look, look look look look look"
I'm looking! 🤣
So they just hard coded the voltage divider value and when it detects any abnormalities in the divider value it would mess up dc level and disable wavegen
Glad you screw up occasionally Dave, it makes the rest of us who screw up regularly think there is still some hope for us :-)
Does it POST and test its own calibration or does it just rely on a set testing date and say its out of cal when that date passes???
Don't be hard on yourself. I remember all the videos when you fixed something the whole two of them. 🤣
We've all heard it before, "Don't tamper with the running system." But we like to do it anyway.
Good one.👍👍
I think Dave voided his warranty... :-)
Hey sir have you tried to unhack it and power it on again?
For you South Main Auto fans... "There's your problem lady." On the wave gen - is it possible the clock reset put it back to some free hours of license? Just a thought.
If it's a software fault, wouldn't re-flashing/re-programming the FPGA make some difference?
When he showed his hack I was expecting him to say "allright, just remove these hacks, restore the thing to it's original form and Dave's your uncle".
Nice thumbnail, hardly any green
Well Dave, at least you gave it a try, which makes this a Parker Square of a hack.
Dave, did you check the battery ?
From a programming pov, it really doesn't make sense to make a device "robust" enough for that. It likely was an invalid product configuration, something that for all intents and purposes will never ever happen (unless its a hardware failure), thus always assuming the product id is right and consistent is the most economical to do. For why this happens, I would speculate the IDs are used as offsets into some configuration data structures, and just get wrong/unfilled ones, or interpret a slightly different structure diffferently. For the licensing I would guess that the license code/id/whatever takes into account the product ID and doesn't consider the license for one product valid for another.
Then why not put up an error message is an invalid model configuration is detected? That seems pretty trivial.
Did you root your scope trying to root your scope?
i noticed the freq counter on ch1 was jumpin around like crazy..where ch2 was rock solid...
but geeze Dave... ya shoulda marked it as hacked when ya did it.. ya got too many scopes to remember which one ya stuffed up! :P
Literally too many, yes!
Rooted by Dave ™
How about intentional, simulated defective unit, after detecting that the unit has been hacked. That should convince anyone hacking to desist such activity
Hi, how to connect the serial communication whit a pc?
It should be illegal to charge the customer to install hardware in a product, and then charge you for the privilege to use it in software.
I'm a calculator nut. Could you show us all your calculators one day?
I take it you are aware of BenHeck?
welcome to everbodys favourite segment, mailbag!
There are different programming techniques. Personally I always prefer to programmatically validate and sort of "close the loop", meaning never allow your code to get into a situation that was unplanned for. That can be as simple as validating every piece of data and redirecting to a certain message if anything is outside of the expected range, even if that situation is extremely unlikely. However, as you can probably imagine, this takes a lot of extra time and can delay go to market dates which delays revenue from a product. Like anything else, this tends to depend on the management style at the company. Time is money and therefore a departmental compromise usually gets developed at a company to allow for certain assumptions. For instance, we'll assume that hardware flags (e.g. resistor values) will be what we expect since customers most likely won't be disassembling and messing around with changing their values. That eliminates the time required to code, develop tests and validate the information as well as program space in memory. It's all a trade off. What trade offs are made tend to be developed over time and are different at different companies.