This is quite interesting. A friend of mine did something similar with a HDD enclosure. He took the drive out of the enclosure, covered some pins and he now has an expensive drive that he paid a fraction of the original price for because he bought it in an enclosure. Thank you!
that pin in the drives is used for staggered spinup so for a server backplane to keep the drive powered off so it can power on the drives gradually and not all together to not overload the power supply
SMBUS devices have a hard coded address, it sounds like the card responds to an address that conflicts with something on the mother board that is critical to the POST.
Came to say basically the same. The question of course is WHY would a PCIE device have an SM bus connection used? Poking around it looks like it has to do with setting up ACPI, so it's possible that by covering these pins some power management stuff won't work right, not sure
@@nalinux And then? Some I2C devices are hard coded. With most they can be changed, but only by physically moving some jumpers that probably aren't populated. Tape is a straightforward solution
It is pricey for a job where masking tape works. I bet you'd also tell me I was wrong for using 3M packing tape to tape mod my C2Q but it is still working, years later. Also, any kapton tape I've ever seen wouldn't stick half as well. Would probably just get bunched up during installation and not work anyway.
Tape is apparently a great fix. I have an 8TB backup hard drive in my main PC running 24/7 and it has pins taped to make it work. I pulled the internal HDD out of the WD usb enclosure, and they have proprietary info that goes out one of the unused 3.3v pins, and you have to tape it to get the drive to show up outside of its expected WD enclosure.
Correct me if I'm wrong but I believe taping the 3.3v pin is only necessary if you're using an older power supply. I have a few shucked WD 14TB connected to a corsair RM850x and didn't need to tape the 3.3v.
The SATA spec changed, so one of the 3.3v pins was unused and was switched to be a "reset" pin. If you don't tape it on drives that use the newer standard with an older PSU/connector, the 3.3v will hold the drive in reset rather than letting it power up.
I love that a commenter helped you, that's so cool! Good job, guys. This is very interesting, thanks for sharing! In simple terms, this means that taping these pins will interrupt communication between the bus and the network card.
If you're careful, you can just cut the tracks on the board. Did it on a Dell RAID controller to make it work on consumer boards. It still worked in the original server too. Edit: oh well you covered it anyway
Perfect timing for a video. This literally reencountered this issue 2 hours before you posted this. I have a OptiPlex 5060 and a Precision 3430 with the same CPU. The Mellanox connect3 cards I have work in the OptiPlex but not in the precision unless I take the ram stick out of dimm slot 1. I'm going to have to do the tape fix on it.
Yeah, that's hit or miss. You don't always have that option. Considering this is an Optiplex, my gut says it isn't happening. I've had enough experience with several generations of them to say OP was out of his mind trying to use an Optiplex for that in the first place. They always have SMBus issues. I can't get more than 6GB of DDR3 to run on my C2Q Optiplex. It can't be a dead slot because it straight up refuses to post. If it was a dead slot, it probably wouldn't see it at all. I have verified by getting it to post with the RAM it does like in that slot. What it does is see a stick of RAM that doesn't comply with their narrow standard and just refuses to post. It does the same thing with this NIC. Pretty bad when you have to hard mod anything just to get it to work in a standardized slot. I tape modded my C2Q but that isn't the same thing.
It’s crazy how people figure this out but hey! Great PSA video and I love how you try to figure out what is going on. I will definitely keep this in my back pocket if I run into problematic server hardware myself!
Oh the irony I just got one of these at work and was unaware that it would only work in my server. Had planned for it to go in a workstation I use for editing. Well now it can. Thanks to you .
This is fascinating as electronics engineer, will have to look into the smbus protocol. Seems like jtag/i2c compatible bus, i2c used for communication between nvme wifi cards and even between arduino & b&w oled display while jtag used to program fpga or arm chips. From the snippet it video, seems to be just for inputing data to the card, so even if standard might have something to do with feature set available only in server side hardware or just business use case like how ecc memory is/was.
I've worked with SMBus primarily for battery packs and it's basically I2C with some more strict timing specifications, ACK/NACK requirements, and address resolution. It's a pretty interesting protocol overall and an interesting approach to making I2C more robust
@@FowlerAskew battery packs? Never would've thought of that of all things but I suppose it's critical info to prevent boom. Looks like eSPI & SMBus are intels updates to SPI & I2C. Idk if I3C owuld catch on but getting popular. SMBus for pcie seems to be for getting temp data for the card along with other things, so my initial assessment is probably correct that consumer chipsets just doesn't know what/hot to respond to the things the card asks for and is mapped to something else. Stopping it from communication just makes it behave like normal pcie card which is why I suppose it works
@@arnabbiswasalsodeep I've come accros this very recently, bought an SMBusUSB converter in an attempt to revive a dead battery, I'll have to talk to the chip to reset it. And I'll probably fix another battery in the process, which has bricked itself when the cells overdischarged, I charged them but once a fault is detected it'll never supply power again, and the computer is telling me I need to replace the battery ... even redirecting me to a shop where I buy a new one🤣 Some smart people reversed engineered the more recent chips used nowadays in a lot of batteries, specifications were never released for it (why allow people to fix their stuff themselves). I had some BMC modules I saved "just in case" or for future DIY projects but they won't work either since cells can't be replaced anymore due to these "safety features", once it detected a cell is removed it'll stop working (one is still trying to balance the cells I wired back, but doesn't get power on the connector). Old fashion ones weren't that picky, I replaced cells in the past with some success.
I stumpled upon this myself with a Broadcom 4 port NIC. The interesting thing was that the last time I tested it in that system it was working. Eventually I found the same instructions and then I realized that I installed two more RAM sticks since the last time I used the card. So yeah, SMBus with RAM can be a problem with certain setups obviously. To be fair, that was server-grade hardware that I've been trying to use in a consumer system, so maybe the consumer-grade parts are to blame here.
This works because of a few reasons vendor lock is one. And data the other. You need to use thin tape or the slot will scrape off the tape when you put the card in. Also not goopy becuse you dont want glue on the contacts. You only usually need one pin covered and it will work.
Just used this on a Dell branded 4 port 1gbit nic in a Lenovo M92p. At first when it turned on it played a little musical tune (aka POST beep code).... remembered about watching this and applied some Kapton tape.
This actually corruputed the BIOS on one of my motherboards. Luckilly I had spare chips around, since the BIOS chip is in a socket. The tape did the trick and it's still working after about 5 years.
I got a HP Elitedesk 600 G4 SFF a few months back intending to build a router out of it and while I waited on the "router parts" to show up I upgraded everything else; CPU, RAM, NVME & SSDs, even some HP cages & I/O stuff... Probably $500 in stuff I had laying around (probably could have just bought a REALLY nice router). When the "router" items, a dual port 10G card and a quad port 2.5G card finally arrived NEITHER one would work in this little HP (but both worked fine in other machines). Really?? !!@#$!! I screwed around with it for a week or so before the whole thing went in a garage box along with the rest of my other nope/fail projects.🙄 Now I found your video and "fixed" both cards in about 5 minutes with some packing tape... 🤣The Elitedesk 600 G4 SFF router has been officially pardoned from the garage! Thanks!!!!👍
I fixed a printer from goodwill with the same kind of tape. There was a physically rotating part (I don't remember what it was called, an actuator? A piston?). After time, the foam spacer on it deteriorates and sticks so it no longer rotates properly. Putting tape over it makes it work again, so it doesn't stick. Tape is such a versatile fix for electronics! I spent $10 and got an awesome printer!
There's that old saying that You can fix everything with WD40 and Duct Tape : If it moves, but shouldn't move, use duct tape. If it doesn't move, but should move, use WD-40. But you've proven it's not always true, since the other way around worked best in your case😁
You should definitely use Kapton tape instead of any other type of tape, they can leave residues and/or react with the traces. Some cards also don't work because of missing PRSNT link, fix is to wire A1 to B(last-1). I do this often to convert x16/x8 cards to x1 without any modification to the motherboard, for example to add an old video card to x1 slot and have the x16 slot available.
Kapton tape is one of the weakest adhesive bonds you're gonna get. If anything is just going to bunch up and slip off when you insert the card, it will be Kapton tape. Even some high grade 3M packing tape is a better choice. It is what I used to tape mod my C2Q and it still holds up to what is pretty much a needle, constantly pressing on it. Let's see kapton tape do that. Kapton tape is designed for heat shielding ICs when doing SMT rework. It is barely even sticky. Also, those traces are gold plated so nothing is going to react with them :P
If you paint scale models, or wargaming miniatures you might be familiar with Tamiya tape. It' just paint masking tape, but you can buy it in super thin sizes, you can skip the exacto knife step
unironically many years ago I used to have a Dell PC with a core 2 duo, I wanted to overclock it but the bios didn't allow me, so I used duct tape of all things to cover a couple of specific pins and overclocked it from 2.6Ghz to 3.2Ghz. Edit: I didn't get to the part of the video at the time of this comment showing the tape CPU overclock, my b. still cool that it was possible.
I just clip the pins off the card since I'm never, ever going to need the functionality they enable, and I kept finding that my tape would slip off. Had to do this with some dirt-cheap two-port 10G NICs I found on Amazon that were for sale from parted-out servers, and in my case they were HP, so definitely not just a Dell problem.
I ran into an issue like this last year with an LSI card when dropping it into an eMachines. Experimented with this a bit by covering the SM bus pins and then the issue went away. Might be more related to the older BIOS driven computers of that time period. Never quite figured out why this behavior is hit and miss but it's gone now.
I suspect this IS a DELL thing - I have an elderly friend who got ripped off buying an ancient Dell laptop from around 2012. It was a POS even when new, but 12 years later . . . but anyway my point is that while some components were upgradeable (HD for SSD and some extra RAM etc) when I tried to install a cheap secondhand wifi card in the available slot it refused to POST because DELL whitelists specific parts numbers on a per country basis, so if your part is not on the list it just won't boot. Thing is that specific part is actually listed in the DELL repair manual as 'compatible' - which it is in EVERY other regard than just not being on the BIOS whitelist which is totally inaccessible by design. No idea if this tape hack might have worked - I suspect not because it's looking for a positive confirmation.
SMBus is related to I²C and has a similar enough protocol Linux uses the same driver code for both. I²C has a block of addresses that a device can live on, and the way it's supposed to work generally is each device gets its own address. If your first memory module uses address 50, the next one should be address 51, etc. How these things negotiate what address they live at if they can't have their "primary" for some reason or how they know that they can't have their primary is … there, but I don't know the details. But because SMBus is basically I²C except … I suspect there's a lot of barely-compliant/slightly-non-compliant devices mucking things up. I have a Gigabyte motherboard that Will Not recognize drives on my LSI SAS HBA. Nor will they work with my cheap little ASMedia HBA. The board boots, the card's BIOS loads, but no drives are seen. I suspect this might be the same SMBus issue. I don't think blocking SMBus is what I want in this case because I want monitoring to work. I'm switching motherboards to an ASRock board, hopefully that will solve the problem. If not, my old ASUS board worked, so it's a fair bet I could try one from ASUS, but I didn't see an ASUS B550 board with the right PCIE arrangement. Need two PCIe slots that aren't x1.
I wonder if this would wok in my one system. The problem is any PCI-e Graphics card that needs supplemental 12v power in refuses to be detected. GTX 1050 ok GTX 1070 no go.
Some Dell and HPE HBA card like this. i just cut the trace since I don't wan any stuff inside my PCIE slot. the HBA cost me nothing so cheaper way from my logic.
I'm using the one-port variant of that same NIC in my Windows machine. I didn't have this kind of trouble, instead the metal bar going across the card blocked my front panel USB header on my motherboard, but I got everything working with a right-angle adapter. This is good to know, though, if I decide to move it to a different machine for some reason.
I have done similar with WiFi card in notebooks, which would be disabled by default and thus wheren't working notebooks with the matchin 'enable' key combination. Also I once had issues with some real crap notebooks (Dell), which blacklisted any internal hardware that wasn't originally supplied with the notebook. There you had to isolate the pins when powering up but after POST they needed to be reconnected again so the hardware would work with Windows, so I put in an arudino that did the (dis)connecting at the right time. Modern hardware is utter trash, because vendors actually are putting more energy into preventing owners from using their hardware than in innovating their products.
Dell has a history of making strange compatibility issues on their workstation hardware to prevent people from experimenting or adding third party stuff to their machines. From non-stardard versions of standard connectors like fan and power supply ones to different BIOSes to lock compatibility of certain features.
FWIW... even you're 600% sure you won't move the card elsewere or sell it... *don't scrape off contacts*! If you're lookint for a permanent yet reversable workaround, consider conformal coating. You can find cheap one that can be converted with ultraviolet light, but can be later scraped off without damaging the underling traces using solvent (buy the right one and please test it on a spare board before). It's fare easy to remove coating than create contacts again.
I loved your Crafty video on how to setup a minecraft server locally. Do you think it's financially worth it or should I just buy a server from a service online like Apex?
If it's the x540-t2 then you might want to enable CSM and disable PXE-boot. My card just straight locked the boot, I couldn't go past the first boot screen with UEFI logo.
Also that 10GB card you had looks to be an old fibre channel storage network card. So technically, it counts as a storage controller. I used to use those in an old role managing a SAN. I recently bought one for same idea as a 10gb network adapter.
For me, I prefer using packing tape when I tape pins. It's thinner but doesn't come off as easily. Better for multiple inserts. And if one asks me how do I do this so much, it's because I used to and can still overclock Pentium II processors.
I got a CISCO UCSC-PCIE-IRJ45 Intel Quad Port Adapter Network card I350-T4 74-10521-01 from eBay. It didn't post like you shown in the video. Even with B5-B6 covered with electrical tape, the PC still doesn't post. Any idea?
I don't get the sudden craziness around 10 gig rothen copper ethernet? I mean it starts with actually getting a proper Cat6a/Cat7 patch panel that really properly likes speeds of 10Gbs. Also these 10Gbs Rj45 switch(routers) are hard to come by and on the expensive side for what they are. On the other side all the SFP+ hardware for Fiber like OM3 or so is kind of cheap and way more capable all the copper stuff. There are a lot of almost 10 year old SFP+ stuff out there for way cheaper that all the proper 10 gig RJ45 stuff. The only downside of Fiber (currently) is if you insist to terminate all your fiber yourself but fiber with all plugs spliced in is most times cheaper as properly terminated Cat7 with all the properly rated plugs. IMO if someone really has the need (or wants' to have) 10 gig networking they should solely look towards fiber (maybe DACs in racks) and SFP+ gear. That is more future proof and currently with older hardware but good to use in a homelab, cheaper. Also you can do 100 gig with OM3/OM4 in the future when this equipment isn't costing an arm and a leg. I am not so sure you will ever do this with copper over longer distances that a few meters.
Way way back in the stone age of computing, "we" use to use that tape trick to copy Radio Shack Colour Computer game cartridges to cassette tapes and game swap. Sure hope the Statute of Limitations has expired on this crime :-)
I fixed my Core 2 Quad with a piece of tape. It wasn't running fast enough. SMBust issues have been a problem with Dell for a long time. This is why I can't seem to find another 2GB of RAM that works in my C2Q Optiplex and it is stuck at 6GB. That's plenty for Windows 7 but still... A PC should never be that picky about RAM. The 486 I built in 1995 wasn't that picky and that's back before we had asynchronous RAM. Your problem here is wasting your time and sanity messing with an old Optiplex to try and do something useful. Unless you want a basic PC, they are useless. Although, I did manage to run a mining GPU in a Core 2 era Optiplex for a year....That was just random luck, really.
that card is kernel locked and the pin you taped off stop reporting to the pcie e lanes that its locked it use for some xeon servers and its becuase it used for data centers old cards that was locked out and they where sold at a decent price when they came out but the companie want to limit it when there 10 gig cards where not kernal locked you can try to flash a random bios to it and by pass it but there a small chance of find one to work with it
Hold on. I have tried this trick on the same exact card on my HP z2 g9 and it did not work..... I had to buy a brand new 10GBaseT card in order to get 10G 😥
Update : I tried again, and it worked. I think it might have been a driver issue all along. Time for me to return the other card, and to buy a quieter fan because the stock one screams !
This is quite interesting. A friend of mine did something similar with a HDD enclosure. He took the drive out of the enclosure, covered some pins and he now has an expensive drive that he paid a fraction of the original price for because he bought it in an enclosure. Thank you!
Yes , you remove the 3.3v pin, if the hdd sees 3.3v it won't spin up...
that pin in the drives is used for staggered spinup so for a server backplane to keep the drive powered off so it can power on the drives gradually and not all together to not overload the power supply
I ran into this with a RAID card one time. I guess ex-server hardware just needs this literal patch sometimes.
Haha "literal patch", that's great
Real patch 😆😁
Yes I applied the tape method to a old Dell H310 HBA a while ago.
Do you still get SMART data from your drives if you do that? I imagine the HBA no longer reports temp of the card, but I guess that's okayish.
@@knghtbrd SMART still worked fine for me.
SMBUS devices have a hard coded address, it sounds like the card responds to an address that conflicts with something on the mother board that is critical to the POST.
Supposedly there _is_ an address resolution protocol. Supposedly. For two of the same device in the system, if nothing else.
Came to say basically the same. The question of course is WHY would a PCIE device have an SM bus connection used? Poking around it looks like it has to do with setting up ACPI, so it's possible that by covering these pins some power management stuff won't work right, not sure
Wake-on-lan, maybe? @@repatch43
It should be quite easy to check the address with an Arduino.
@@nalinux And then? Some I2C devices are hard coded. With most they can be changed, but only by physically moving some jumpers that probably aren't populated. Tape is a straightforward solution
Kapton tape would be my reccomendation for electrical isolation
Ditto; it will last a bit longer especially if you ever plug/unplug the card more than once or twice :)
Yeah it's great for things like this:
- doesnt leave a goey residue like electrical tape.
- resistant to high temperatures.
Side note, "Kapton" is a brand name, the more widespread name would be "polyamide" :D
@@pixel_vengeur391Just to one-up you on being nitpicky 😉, I'd say polyamide is the more general name, but Kapton is the more widespread name.
It is pricey for a job where masking tape works. I bet you'd also tell me I was wrong for using 3M packing tape to tape mod my C2Q but it is still working, years later. Also, any kapton tape I've ever seen wouldn't stick half as well. Would probably just get bunched up during installation and not work anyway.
Tape is apparently a great fix. I have an 8TB backup hard drive in my main PC running 24/7 and it has pins taped to make it work. I pulled the internal HDD out of the WD usb enclosure, and they have proprietary info that goes out one of the unused 3.3v pins, and you have to tape it to get the drive to show up outside of its expected WD enclosure.
Correct me if I'm wrong but I believe taping the 3.3v pin is only necessary if you're using an older power supply. I have a few shucked WD 14TB connected to a corsair RM850x and didn't need to tape the 3.3v.
Yup. Taped over the pin on 3 diff shucked WD White drives. After limited warranty period, I just pulled em out.
Yeah that is the first thing I thought of, I have a couple shucked drives that needed this fix.
The SATA spec changed, so one of the 3.3v pins was unused and was switched to be a "reset" pin. If you don't tape it on drives that use the newer standard with an older PSU/connector, the 3.3v will hold the drive in reset rather than letting it power up.
it's not proprietary, it's a newer Sata spec that uses that pin for staggered spinup
I love that a commenter helped you, that's so cool! Good job, guys. This is very interesting, thanks for sharing! In simple terms, this means that taping these pins will interrupt communication between the bus and the network card.
Hi, my pleasure!! You really did the video....well done!!Looking forward for next videos!!
YAY glad you saw it! And yeah I did haha. This is proof that I do in fact take viewer suggestions seriously 😂
Thanks again!
@@HardwareHaven I'm also passioned about computer hardware and operating systems. Maybe we should get in contact to discuss projects and ideas ;)
@@Mihai-Marian-Cenusa feel free to shoot me an email! That's usually what I check most often
If you're careful, you can just cut the tracks on the board.
Did it on a Dell RAID controller to make it work on consumer boards.
It still worked in the original server too.
Edit: oh well you covered it anyway
Wow,what a fix.
I love nifty little fixes like this, it’s the kind of thing that once you have the tool in your toolbelt, you will never want to forget it.
I'm so happy you made a video on this, i spent so much time trying to figure it out, i hope your video helps people having this issue
Perfect timing for a video. This literally reencountered this issue 2 hours before you posted this. I have a OptiPlex 5060 and a Precision 3430 with the same CPU. The Mellanox connect3 cards I have work in the OptiPlex but not in the precision unless I take the ram stick out of dimm slot 1. I'm going to have to do the tape fix on it.
Ruth's footage does make up for the short video 🥰
Hahaha I’m glad
This was way longer then it needed to be lol!
If there's an option in bios, you can disable smbus in bios too.
Yeah, that's hit or miss. You don't always have that option. Considering this is an Optiplex, my gut says it isn't happening. I've had enough experience with several generations of them to say OP was out of his mind trying to use an Optiplex for that in the first place. They always have SMBus issues. I can't get more than 6GB of DDR3 to run on my C2Q Optiplex. It can't be a dead slot because it straight up refuses to post. If it was a dead slot, it probably wouldn't see it at all. I have verified by getting it to post with the RAM it does like in that slot. What it does is see a stick of RAM that doesn't comply with their narrow standard and just refuses to post. It does the same thing with this NIC. Pretty bad when you have to hard mod anything just to get it to work in a standardized slot. I tape modded my C2Q but that isn't the same thing.
It’s crazy how people figure this out but hey! Great PSA video and I love how you try to figure out what is going on. I will definitely keep this in my back pocket if I run into problematic server hardware myself!
oh.. nice transition when inserting the ram.. very nice!
Oh the irony I just got one of these at work and was unaware that it would only work in my server. Had planned for it to go in a workstation I use for editing. Well now it can. Thanks to you .
Love those videos - soooo interesting and insightful. Has inspired me to tinker around at home! Thank you
This is fascinating as electronics engineer, will have to look into the smbus protocol. Seems like jtag/i2c compatible bus, i2c used for communication between nvme wifi cards and even between arduino & b&w oled display while jtag used to program fpga or arm chips. From the snippet it video, seems to be just for inputing data to the card, so even if standard might have something to do with feature set available only in server side hardware or just business use case like how ecc memory is/was.
I've worked with SMBus primarily for battery packs and it's basically I2C with some more strict timing specifications, ACK/NACK requirements, and address resolution. It's a pretty interesting protocol overall and an interesting approach to making I2C more robust
@@FowlerAskew battery packs? Never would've thought of that of all things but I suppose it's critical info to prevent boom. Looks like eSPI & SMBus are intels updates to SPI & I2C. Idk if I3C owuld catch on but getting popular.
SMBus for pcie seems to be for getting temp data for the card along with other things, so my initial assessment is probably correct that consumer chipsets just doesn't know what/hot to respond to the things the card asks for and is mapped to something else. Stopping it from communication just makes it behave like normal pcie card which is why I suppose it works
@@arnabbiswasalsodeep I've come accros this very recently, bought an SMBusUSB converter in an attempt to revive a dead battery, I'll have to talk to the chip to reset it.
And I'll probably fix another battery in the process, which has bricked itself when the cells overdischarged, I charged them but once a fault is detected it'll never supply power again, and the computer is telling me I need to replace the battery ... even redirecting me to a shop where I buy a new one🤣
Some smart people reversed engineered the more recent chips used nowadays in a lot of batteries, specifications were never released for it (why allow people to fix their stuff themselves).
I had some BMC modules I saved "just in case" or for future DIY projects but they won't work either since cells can't be replaced anymore due to these "safety features", once it detected a cell is removed it'll stop working (one is still trying to balance the cells I wired back, but doesn't get power on the connector).
Old fashion ones weren't that picky, I replaced cells in the past with some success.
Very cool. Kapton tape would probably be better for your long term pin covering needs.
The socket would certainly appreciate the thinner tape.
I stumpled upon this myself with a Broadcom 4 port NIC. The interesting thing was that the last time I tested it in that system it was working. Eventually I found the same instructions and then I realized that I installed two more RAM sticks since the last time I used the card. So yeah, SMBus with RAM can be a problem with certain setups obviously. To be fair, that was server-grade hardware that I've been trying to use in a consumer system, so maybe the consumer-grade parts are to blame here.
This works because of a few reasons vendor lock is one. And data the other.
You need to use thin tape or the slot will scrape off the tape when you put the card in. Also not goopy becuse you dont want glue on the contacts.
You only usually need one pin covered and it will work.
Just used this on a Dell branded 4 port 1gbit nic in a Lenovo M92p.
At first when it turned on it played a little musical tune (aka POST beep code).... remembered about watching this and applied some Kapton tape.
Covering pins tends to fix many an issue. Had the same issue with a Dell perc raid card some years ago
This actually corruputed the BIOS on one of my motherboards. Luckilly I had spare chips around, since the BIOS chip is in a socket. The tape did the trick and it's still working after about 5 years.
NGL, I'm pretty sure this would do NUMBERS as a YT short because of how quick it is to summarize and how generally unknown the info is.
Knock me down with a feather!👍
Tape to cover pins for a bona fide fix? Golly, that is so amazing.
Thank you.
I'm an IT engineer and never heard of it.
Thanks for the video!
3:20 I was about to comment that this reminds me of OC'ing LGA 775 CPUs on non-OC motherboards.
This is a good shout....Have done similar with various dell perc cards for years to add them to none dell systems.
Thanks, this fixed my SAS HBA!!
neat. i'll be saving this for future troubleshooting.
I got a HP Elitedesk 600 G4 SFF a few months back intending to build a router out of it and while I waited on the "router parts" to show up I upgraded everything else; CPU, RAM, NVME & SSDs, even some HP cages & I/O stuff... Probably $500 in stuff I had laying around (probably could have just bought a REALLY nice router). When the "router" items, a dual port 10G card and a quad port 2.5G card finally arrived NEITHER one would work in this little HP (but both worked fine in other machines). Really?? !!@#$!! I screwed around with it for a week or so before the whole thing went in a garage box along with the rest of my other nope/fail projects.🙄
Now I found your video and "fixed" both cards in about 5 minutes with some packing tape... 🤣The Elitedesk 600 G4 SFF router has been officially pardoned from the garage! Thanks!!!!👍
Flex Tape's eternal enemy....Blue Electrical Tape
It's for painting, not electrical stuff. It's designed to come off easily.
I fixed a printer from goodwill with the same kind of tape. There was a physically rotating part (I don't remember what it was called, an actuator? A piston?). After time, the foam spacer on it deteriorates and sticks so it no longer rotates properly. Putting tape over it makes it work again, so it doesn't stick. Tape is such a versatile fix for electronics!
I spent $10 and got an awesome printer!
There's that old saying that You can fix everything with WD40 and Duct Tape : If it moves, but shouldn't move, use duct tape. If it doesn't move, but should move, use WD-40.
But you've proven it's not always true, since the other way around worked best in your case😁
another great use for painters tape. that stuff has literally hundreds of uses!
Most server cards talk directly to memory controlers and because home board do not use ECC memory (well some do) this can be a problem on many boards
You should definitely use Kapton tape instead of any other type of tape, they can leave residues and/or react with the traces.
Some cards also don't work because of missing PRSNT link, fix is to wire A1 to B(last-1). I do this often to convert x16/x8 cards to x1 without any modification to the motherboard, for example to add an old video card to x1 slot and have the x16 slot available.
Kapton tape is one of the weakest adhesive bonds you're gonna get. If anything is just going to bunch up and slip off when you insert the card, it will be Kapton tape. Even some high grade 3M packing tape is a better choice. It is what I used to tape mod my C2Q and it still holds up to what is pretty much a needle, constantly pressing on it. Let's see kapton tape do that. Kapton tape is designed for heat shielding ICs when doing SMT rework. It is barely even sticky.
Also, those traces are gold plated so nothing is going to react with them :P
If you paint scale models, or wargaming miniatures you might be familiar with Tamiya tape. It' just paint masking tape, but you can buy it in super thin sizes, you can skip the exacto knife step
unironically many years ago I used to have a Dell PC with a core 2 duo, I wanted to overclock it but the bios didn't allow me, so I used duct tape of all things to cover a couple of specific pins and overclocked it from 2.6Ghz to 3.2Ghz.
Edit: I didn't get to the part of the video at the time of this comment showing the tape CPU overclock, my b.
still cool that it was possible.
I just clip the pins off the card since I'm never, ever going to need the functionality they enable, and I kept finding that my tape would slip off. Had to do this with some dirt-cheap two-port 10G NICs I found on Amazon that were for sale from parted-out servers, and in my case they were HP, so definitely not just a Dell problem.
Short tip videos like this are perfectly acceptable!
I ran into an issue like this last year with an LSI card when dropping it into an eMachines. Experimented with this a bit by covering the SM bus pins and then the issue went away. Might be more related to the older BIOS driven computers of that time period. Never quite figured out why this behavior is hit and miss but it's gone now.
Seen and done a similar thing with wifi cards on laptops when the laptop makers won't let you use a different wifi card.
kapton tape is the best option vs painters tape for longer term use cases
Had to do the same thing to a Dell H310 raid car. If I'm not mistaken it's due to the address of the smbus
I suspect this IS a DELL thing - I have an elderly friend who got ripped off buying an ancient Dell laptop from around 2012. It was a POS even when new, but 12 years later . . . but anyway my point is that while some components were upgradeable (HD for SSD and some extra RAM etc) when I tried to install a cheap secondhand wifi card in the available slot it refused to POST because DELL whitelists specific parts numbers on a per country basis, so if your part is not on the list it just won't boot.
Thing is that specific part is actually listed in the DELL repair manual as 'compatible' - which it is in EVERY other regard than just not being on the BIOS whitelist which is totally inaccessible by design. No idea if this tape hack might have worked - I suspect not because it's looking for a positive confirmation.
SMBus is related to I²C and has a similar enough protocol Linux uses the same driver code for both. I²C has a block of addresses that a device can live on, and the way it's supposed to work generally is each device gets its own address. If your first memory module uses address 50, the next one should be address 51, etc. How these things negotiate what address they live at if they can't have their "primary" for some reason or how they know that they can't have their primary is … there, but I don't know the details. But because SMBus is basically I²C except … I suspect there's a lot of barely-compliant/slightly-non-compliant devices mucking things up.
I have a Gigabyte motherboard that Will Not recognize drives on my LSI SAS HBA. Nor will they work with my cheap little ASMedia HBA. The board boots, the card's BIOS loads, but no drives are seen. I suspect this might be the same SMBus issue. I don't think blocking SMBus is what I want in this case because I want monitoring to work. I'm switching motherboards to an ASRock board, hopefully that will solve the problem. If not, my old ASUS board worked, so it's a fair bet I could try one from ASUS, but I didn't see an ASUS B550 board with the right PCIE arrangement. Need two PCIe slots that aren't x1.
This reminds me of the fix when shucking drives and they were using the reset pins.
I think the LED in the on off button flashes a pattern to indicate why its not posting. Similar to the old beep codes.
Tips and tricks, always appreciated.
I remembered there´s some laptops that would requiere that to upgrade the wifi cards (On some models doing the tape mod would fix the whitelist issue)
had something similar 7 years ago with the old style ac7260 pci wifi card.
I fixed my Vizio 55" TV by placing a 1mm sliver of cellophane tape over one connector to a single failed processor. No joke.
diagnostic led code 2 flashes
I wonder if this would wok in my one system. The problem is any PCI-e Graphics card that needs supplemental 12v power in refuses to be detected. GTX 1050 ok GTX 1070 no go.
Some Dell and HPE HBA card like this. i just cut the trace since I don't wan any stuff inside my PCIE slot. the HBA cost me nothing so cheaper way from my logic.
I love this kind of fixes and I want more please.
I'm using the one-port variant of that same NIC in my Windows machine. I didn't have this kind of trouble, instead the metal bar going across the card blocked my front panel USB header on my motherboard, but I got everything working with a right-angle adapter. This is good to know, though, if I decide to move it to a different machine for some reason.
I have done similar with WiFi card in notebooks, which would be disabled by default and thus wheren't working notebooks with the matchin 'enable' key combination. Also I once had issues with some real crap notebooks (Dell), which blacklisted any internal hardware that wasn't originally supplied with the notebook. There you had to isolate the pins when powering up but after POST they needed to be reconnected again so the hardware would work with Windows, so I put in an arudino that did the (dis)connecting at the right time.
Modern hardware is utter trash, because vendors actually are putting more energy into preventing owners from using their hardware than in innovating their products.
Years ago I had a 64mb ddr geforce AGP card that was incompatible with my motherboard. One sliver of tape on a pin and it then worked.
Dell has a history of making strange compatibility issues on their workstation hardware to prevent people from experimenting or adding third party stuff to their machines. From non-stardard versions of standard connectors like fan and power supply ones to different BIOSes to lock compatibility of certain features.
FWIW... even you're 600% sure you won't move the card elsewere or sell it... *don't scrape off contacts*! If you're lookint for a permanent yet reversable workaround, consider conformal coating. You can find cheap one that can be converted with ultraviolet light, but can be later scraped off without damaging the underling traces using solvent (buy the right one and please test it on a spare board before).
It's fare easy to remove coating than create contacts again.
I loved your Crafty video on how to setup a minecraft server locally. Do you think it's financially worth it or should I just buy a server from a service online like Apex?
If it's the x540-t2 then you might want to enable CSM and disable PXE-boot.
My card just straight locked the boot, I couldn't go past the first boot screen with UEFI logo.
I also had to do this with an intel nic on an asus mobo
Had the same thing on Dell and HP SAS cards. Did the same thing with painters tape.
Also that 10GB card you had looks to be an old fibre channel storage network card. So technically, it counts as a storage controller. I used to use those in an old role managing a SAN. I recently bought one for same idea as a 10gb network adapter.
I have a dual port qnap switch that doesn't work in a computer I have and it won't post as well. Might try this
For me, I prefer using packing tape when I tape pins. It's thinner but doesn't come off as easily. Better for multiple inserts. And if one asks me how do I do this so much, it's because I used to and can still overclock Pentium II processors.
bsel mod used to be a thing. Good times.
I got a CISCO UCSC-PCIE-IRJ45 Intel Quad Port Adapter Network card I350-T4 74-10521-01 from eBay. It didn't post like you shown in the video. Even with B5-B6 covered with electrical tape, the PC still doesn't post. Any idea?
I have never had issues taping pins on both sides.
This has been floating around for 10+ years on many raid controllers. I've done it on numerous cards, but nail polish works best.
been through this with a dell hba card
What's the 1/2 length card @ 0:17? I only know the red TP-Link cards that small...
What network card is that? Looks like it can fit in a P330 tiny as well, the fan is a bonus.
Anyone who brings up the BSEL mod is a friend of mine.
I'd use kapton instead of painters that can be a mess to remove from header if it happens to stick.
I need to get some!
i think HP raid card had similar fix due to issue when card was trying to contact ILO.
Reminds me of WD White label HDD (shuked drive) where you need some time to do the same on the 3.3V
Is "fix it with tape" the opposite of "kill it with fire"?
OMG that's amazing
good to know, i have 2 1Gb cards i havent touched yet.
Did this for my laptop wifi card
It's gonnna fall off and cause a headache for the next guy who works on it. Be sure to document it because the next guy will probably be you
I don't get the sudden craziness around 10 gig rothen copper ethernet? I mean it starts with actually getting a proper Cat6a/Cat7 patch panel that really properly likes speeds of 10Gbs. Also these 10Gbs Rj45 switch(routers) are hard to come by and on the expensive side for what they are. On the other side all the SFP+ hardware for Fiber like OM3 or so is kind of cheap and way more capable all the copper stuff. There are a lot of almost 10 year old SFP+ stuff out there for way cheaper that all the proper 10 gig RJ45 stuff.
The only downside of Fiber (currently) is if you insist to terminate all your fiber yourself but fiber with all plugs spliced in is most times cheaper as properly terminated Cat7 with all the properly rated plugs.
IMO if someone really has the need (or wants' to have) 10 gig networking they should solely look towards fiber (maybe DACs in racks) and SFP+ gear. That is more future proof and currently with older hardware but good to use in a homelab, cheaper. Also you can do 100 gig with OM3/OM4 in the future when this equipment isn't costing an arm and a leg. I am not so sure you will ever do this with copper over longer distances that a few meters.
Way way back in the stone age of computing, "we" use to use that tape trick to copy Radio Shack Colour Computer game cartridges to cassette tapes and game swap. Sure hope the Statute of Limitations has expired on this crime :-)
Dont copy that floppy!
Ruth is beautiful. I had Bostons while I was in the military in the early 90's great simple fix. nice video
She is, but she's also a little turkey haha.
More videos of Ruth please.
Cutting tape on a hard drive like come on! :D
Don't worry, that drive is long gone lol
I fixed my Core 2 Quad with a piece of tape. It wasn't running fast enough. SMBust issues have been a problem with Dell for a long time. This is why I can't seem to find another 2GB of RAM that works in my C2Q Optiplex and it is stuck at 6GB. That's plenty for Windows 7 but still...
A PC should never be that picky about RAM. The 486 I built in 1995 wasn't that picky and that's back before we had asynchronous RAM.
Your problem here is wasting your time and sanity messing with an old Optiplex to try and do something useful. Unless you want a basic PC, they are useless. Although, I did manage to run a mining GPU in a Core 2 era Optiplex for a year....That was just random luck, really.
dubios yea
Its due to an smbus address conflict
I've tried on my chinese RX 580 and... NOT WORKS!
that card is kernel locked and the pin you taped off stop reporting to the pcie e lanes that its locked it use for some xeon servers and its becuase it used for data centers old cards that was locked out and they where sold at a decent price when they came out but the companie want to limit it when there 10 gig cards where not kernal locked you can try to flash a random bios to it and by pass it but there a small chance of find one to work with it
Hold on. I have tried this trick on the same exact card on my HP z2 g9 and it did not work..... I had to buy a brand new 10GBaseT card in order to get 10G 😥
Update : I tried again, and it worked. I think it might have been a driver issue all along. Time for me to return the other card, and to buy a quieter fan because the stock one screams !
Thanks.
Welcome.
Next time use some kapton tape. Much safer.
Probably a better choice if I had some. I'm not sure to what extent this isn't "safe" though haha
Hello
If you “can’t” believe it, it simply shows your ignorance and non mastering of the PC subject.
And that was exactly my opinion.
I find it amazing that people figure this stuff out! - Such an AMAZING fix!!!
It shocks me. I wouldn't even know how to start troubleshooting something like that