Hi, i got a little experience with the cPCI bus myself and i think i could give you some ideas to brainstorm in. In normal operation, no slavecard like the DMM should be able to force the bus itself to reset. A slave could only scratch on some bottom functions while starting up. In most issues where the bus is jammed, im 99 % sure, its not the interfacechip. Damn i changed alot of them in the past, bummer. There is more potential in investigate around the hotswap controller if the board has one externally. It could be, that either the bus is jammed by the card, or the card shuts down the power supply in case of a short somewhere. So i would try to figure out, what state the entire bus is on. Since their hotplugability, the cPCI Bus is tristated. That means, you got it either on High, Low or at a range between .8 to 1.2 volts which is kind like a pseudo high Z on the address and Data bus. To measure this, you should remove the Masterdevice. When no slavecard gets it's Masterclock clock signal (typically 33/66 MHZ), then the entire bus is tristated and all slavecards should stay in reset (this is actually part of the cPCI specification, dont know much about the PXI itself, is it a cPCI express?! Could be totally different. Never got in touch with that rubbish.) I think, that if the bus is jammed, specialy by an interface device (the thing you removed), you would find one of the Buspins in a strange state. Thats rather unusual. Too often i suspected... The Datapins handle pure parallel data, they are easy to measure with an MSO or LA simultaneously. Then, the entire bus has one reset pin. In normal operation (depending on the system), this pin should be on a logic high (3,3 or 5 v). The Masterdevice uses this signal to reset the slavecards simultaneously. I had a system, running 4 slavecards, where one of the slaves jammed this signal to low suddenly, this case forced the master to stay in reset. Finally, try to power up the card externally to ensure, there is no short on one of the voltage rails. Whats finally, should be the first action to do. ;)... The card needs GND 3,3, 5, 12 and presumably -12 volts to say hello. You could pick up some 2mm pitch pinheaders to do that. That really does the business in my former enineeringwork on the cpci. Now i have a all accessable backplane. Finally, i dont get a clue in what type of chip you soldered to the top site. Was it a flash Memory? Maybe the card is just stupidly staying in reset because there is no bootloader on it? Most of the Signals on the cPCI bus are just logic highs and lows. They are easy to investigate. But after i saw, how you reflowed the NI chip my guess would be, the card is obsolete. ;) Pick up a pinout for that on the web! It's fun and takes alot a time ;) Good luck. Geronimo Stade
Unfortunately, I have no idea what happened to the multimeter. I'm still a bit in awe over the fact that you can run all of these modules in such a small form factor. Thanks for sharing!
Love the video! I for one would like to see a video about using Matlab with the NI test equipment. In school we never got that far into Matlab. My exposure to The pxi test equipment is also very limited.
Another great video. Thanks Shahriar. More missing footage ... its the cat. Has opportunity. Still working on motive but I suspect that pooch is planing to release some competing video as "Cats Whisker".
Really enjoyed this. It would be interesting to revisit the 4071 bga re-ball and maybe put it under your xray machine? Thanks for so much interesting content.
My first guess would be that the new DMM memory needs flashing with firmware to allow the onboard processor to configure the PCI. It may be that the PCI master can see something is there, and is waiting for it to talk, or that the power-on defaults are conflicting with an existing device.
Loved the Video Shariar, looking forward to the investigation as why the DMM still doesn't work. Great Lab and an impressive cat too. I'm amazed you have in the past replaced a BGA without equipment to align it first. Did you use the silk screen as a guide? Hope not as that isn't guaranteed.
Soldering the BGA with a hot air gun alone is not recommended, specially if the gun is placed on a fixed position. It's worth mentioning that the solder melting temp. will concentrate only on a small area (smaller than nozzle diameter), after that the temp gradient will prevent the solder from melting and/or will stop the copper substrate from reaching the required temperature. This is further amplified by the inner layers which look like solid pour around the BGA. One possible solution is to use a uniform preheat source from the bottom , an electric heating plate could do the trick or a iron (clothing) at approx 120-140 C. This temperature should not damage the soft plastics (please don't take my word 100% on this). As an extra safety aluminium foil (food grade) must be used to protect the connectors by simply wrapping the connector. One way to go would be to preheat the bottom of the board until the top temperature reaches the 120-140 C and then "add" top heating @ 310 C. Top heating should be applied over the entire surface of the BGA not on a fixed spot ( due to the size 25 x 25 mm ??). Also flux is a must be used. Kapton tape use is a very good choice.
I have gotten several PXI modules also for cheep, not $50. Included is a PXI 4461 audio / vibration analyzer. To my surprise the National Instruments audiosoftware is way over the top expensive. I do have LabView 2014 running on my computer. Is it practical to do FFT and THD , and frequency response testing with the basic version of LabView? Tom
Did you ever have a look inside the WEP858D? In mine the creapage was so horribly small that the mains arced over across the thyristor. There was charring and scratchmarks on the pcb (so this must have already happened during assembly and testing and was "repaired"). I basically had to reverse-engineer and rebuild the whole thing and put the mains switch on a seperate board.
The obvious answer is that there's config data in that missing memory IC, though I'm sure that occurred to you. My next step would be to try reflowing the chip again (BGA are the worst). After that, either find a PXI extender card or put it in a different mainframe so you have access to the board while powered up. Check supply voltages to the interface IC you replaced, maybe see if the flyback transistors are doing anything. Probing the bus is a possibility, but I don't know if that will give you any useful information.
It's kapton tape, just like regular sticky tape but with a high melting temperature. I've learned the hard way that if you heat up a board to rework some components, some neighbouring parts tend to fall off unexpectedly. The tape helps keep heat away and hold things where they belong until the solder solidifies.
why would you mess with that bga if you had a missing memory chip. What kind of memory was it? from the package i'd say it was a flash memory that contained the board's firmware .
+The Signal Path Blog I wonder if you could have used that second connector space, maybe by moving around the wires on the relay. If the common is in the same place (and it looks like the unused connections are the same) that could have been a better looking fix. But the important thing is that it is fixed of course.
With PXI-2596 wouldn't it be easier to solder the connector onto good footprint near to it, maybe resolder some transistors on empty footprints and put some wires between old transistors' pads and new? That will give more structural integrity to the connector itself when it will be sitting on all pads connected to the PCB. That memory chip on PXI-4071 is definitely a Flash ROM. You didn't tell anything about it's content. Did you ordered programmed chip or just a blank one? According to its proximity to that problematic BGA it probably holds some startup code for it... And that BGA just doesn't start from the flash ROM and thus hangs the bus.
+Макс «Fagear» Крюков The EEPROM is blank. There is more than one on the board actually. The other connectors on the PXI-2596 aren't actually blank, they are routed to other points on the PCB. It would have been a conflict.
The Signal Path Blog I thought that those connectors are routed directly to transistors nearby and are not connected anywhere else. Then control pins of transistors are routed to the IC. So it could be easy to reroute control paths from IC to transistors (cut some traces, solder some wires) and thus no firmware modification will be required and you'll get proper mechanical integrity of the connector. Maybe I'm wrong on that, I guess.
Furry boss approved your switch and DAC/ADC repairs. That's why DMM module fail, no approval ;) On a serious note: 1. Do you have firmware which goes that NAND memory which was missing? It could be as simple as missing firmware, so NXI controller cannot run it's config and locking up bus 2. Another way to diagnose bus problem is to try measure resistance on those bus resistors. If you find one or two out of wack (way different resistance than others) - perhaps that line is shot/shorted. 3. Really curious which voltage reference under the can, perhaps LM399, as I cannot see precision resistors needed for LTZ1000-based ref. Also would be cool if you could post some hi-res photos of modules on your site as well, for us, volt-nuts :) Good work, thanks.
+Greggan The workbench must be cat heaven. So many humming and slightly warm spots to lie on. ;) +Riskteven I'm sure Shahriar knows when something is dangerous himself. Or did you see the cat run around when the heat gun was on? And you don't "catch" cancer, a definitely not by walking around. Cancer is one of the main diseases for old dogs, even the happiest farm dogs.
Oh and besides whenever you say "very little money" I would like to know numbers instead... btw. for the bga, maybe you can locate a dentist or vet for a cheap xray? or walk over to mike from @mikeselectricstuff
Now get 3 more cats and your oscilloscope will have CAT IV protection.
+ReaxI0 Also, 3 more cats would get full use out that Rigol 4-channel cat-warmer.
+Simon Fitch I'm pretty sure that's the latest Rigol Oscillocat
Hi, i got a little experience with the cPCI bus myself and i think i could give you some ideas to brainstorm in.
In normal operation, no slavecard like the DMM should be able to force the bus itself to reset. A slave could only scratch on some bottom functions while starting up. In most issues where the bus is jammed, im 99 % sure, its not the interfacechip. Damn i changed alot of them in the past, bummer. There is more potential in investigate around the hotswap controller if the board has one externally.
It could be, that either the bus is jammed by the card, or the card shuts down the power supply in case of a short somewhere.
So i would try to figure out, what state the entire bus is on. Since their hotplugability, the cPCI Bus is tristated. That means, you got it either on High, Low or at a range between .8 to 1.2 volts which is kind like a pseudo high Z on the address and Data bus.
To measure this, you should remove the Masterdevice.
When no slavecard gets it's Masterclock clock signal (typically 33/66 MHZ), then the entire bus is tristated and all slavecards should stay in reset (this is actually part of the cPCI specification, dont know much about the PXI itself, is it a cPCI express?! Could be totally different. Never got in touch with that rubbish.)
I think, that if the bus is jammed, specialy by an interface device (the thing you removed), you would find one of the Buspins in a strange state. Thats rather unusual. Too often i suspected... The Datapins handle pure parallel data, they are easy to measure with an MSO or LA simultaneously.
Then, the entire bus has one reset pin. In normal operation (depending on the system), this pin should be on a logic high (3,3 or 5 v). The Masterdevice uses this signal to reset the slavecards simultaneously. I had a system, running 4 slavecards, where one of the slaves jammed this signal to low suddenly, this case forced the master to stay in reset.
Finally, try to power up the card externally to ensure, there is no short on one of the voltage rails. Whats finally, should be the first action to do. ;)... The card needs GND 3,3, 5, 12 and presumably -12 volts to say hello.
You could pick up some 2mm pitch pinheaders to do that. That really does the business in my former enineeringwork on the cpci. Now i have a all accessable backplane.
Finally, i dont get a clue in what type of chip you soldered to the top site. Was it a flash Memory? Maybe the card is just stupidly staying in reset because there is no bootloader on it?
Most of the Signals on the cPCI bus are just logic highs and lows. They are easy to investigate.
But after i saw, how you reflowed the NI chip my guess would be, the card is obsolete. ;)
Pick up a pinout for that on the web! It's fun and takes alot a time ;)
Good luck.
Geronimo Stade
probably the mem ic have some dsp instruction or fpga code in it.
i'm very excited about the network analyzer repair tough
Unfortunately, I have no idea what happened to the multimeter. I'm still a bit in awe over the fact that you can run all of these modules in such a small form factor. Thanks for sharing!
Instructions Clear Enough. Successfully repaired PXI-2596,PXI4461 and PXI4071.
Thank you.
Love the video! I for one would like to see a video about using Matlab with the NI test equipment. In school we never got that far into Matlab. My exposure to The pxi test equipment is also very limited.
Always so jealous of the hardware you have. Really like watching your videos!
Another great video. Thanks Shahriar.
More missing footage ... its the cat. Has opportunity. Still working on motive but I suspect that pooch is planing to release some competing video as "Cats Whisker".
It feels a bit crazy that there is need for an FPGA for a simple relay controller.
Due to the PCI bus. But still, that does not require FPGA, but rather larger CPLD.
Excellent video. Thanks!
the missing part seems to be be the memory for the fpga. where will you get the data to put it in?
Really enjoyed this. It would be interesting to revisit the 4071 bga re-ball and maybe put it under your xray machine? Thanks for so much interesting content.
What voltage reference does the 7.5 digit DMM use? LTZ1000?
+dollerstorehack Yes, that is what is being used.
Great video!
My first guess would be that the new DMM memory needs flashing with firmware to allow the onboard processor to configure the PCI. It may be that the PCI master can see something is there, and is waiting for it to talk, or that the power-on defaults are conflicting with an existing device.
+Zadster I don´t know if such a thing exists (or can it be designed?) but say he plugs in a PCI bus scanner to verify your theory?
Loved the Video Shariar, looking forward to the investigation as why the DMM still doesn't work.
Great Lab and an impressive cat too.
I'm amazed you have in the past replaced a BGA without equipment to align it first. Did you use the silk screen as a guide? Hope not as that isn't guaranteed.
Soldering the BGA with a hot air gun alone is not recommended, specially if the gun is placed on a fixed position. It's worth mentioning that the solder melting temp. will concentrate only on a small area (smaller than nozzle diameter), after that the temp gradient will prevent the solder from melting and/or will stop the copper substrate from reaching the required temperature. This is further amplified by the inner layers which look like solid pour around the BGA.
One possible solution is to use a uniform preheat source from the bottom , an electric heating plate could do the trick or a iron (clothing) at approx 120-140 C.
This temperature should not damage the soft plastics (please don't take my word 100% on this). As an extra safety aluminium foil (food grade) must be used to protect the connectors by simply wrapping the connector.
One way to go would be to preheat the bottom of the board until the top temperature reaches the 120-140 C and then "add" top heating @ 310 C. Top heating should be applied over the entire surface of the BGA not on a fixed spot ( due to the size 25 x 25 mm ??). Also flux is a must be used. Kapton tape use is a very good choice.
I have gotten several PXI modules also for cheep, not $50. Included is a PXI 4461 audio / vibration analyzer. To my surprise the National Instruments audiosoftware is way over the top expensive. I do have LabView 2014 running on my computer. Is it practical to do FFT and THD , and frequency response testing with the basic version of LabView?
Tom
What did you end up using?
I think the cat was trying to provide his own diagnosis!
Did you ever have a look inside the WEP858D? In mine the creapage was so horribly small that the mains arced over across the thyristor. There was charring and scratchmarks on the pcb (so this must have already happened during assembly and testing and was "repaired"). I basically had to reverse-engineer and rebuild the whole thing and put the mains switch on a seperate board.
Great video! Any resources out there for how to change out a backplane connector?
Did you try the NI DMM in the same slot each time?
+Justin Bell No, moved it around in different slots.
Any reason you did not take one of the unbroken connector pads for the relay card and run some wires to them? Should give better mechanical stability
Is that chassis capable of passing RF on the bus? If not, how do you set up automated oscilloscope functions with it?
The obvious answer is that there's config data in that missing memory IC, though I'm sure that occurred to you. My next step would be to try reflowing the chip again (BGA are the worst). After that, either find a PXI extender card or put it in a different mainframe so you have access to the board while powered up. Check supply voltages to the interface IC you replaced, maybe see if the flyback transistors are doing anything. Probing the bus is a possibility, but I don't know if that will give you any useful information.
Dumb question, but what is the yellow tape that you use in your repairs, and what's the purpose?
It's kapton tape, just like regular sticky tape but with a high melting temperature. I've learned the hard way that if you heat up a board to rework some components, some neighbouring parts tend to fall off unexpectedly. The tape helps keep heat away and hold things where they belong until the solder solidifies.
+Lawrence Billson Ah, thanks for that, good to know. I can see it coming in very handy when trying to solder components on a double sided board.
why would you mess with that bga if you had a missing memory chip. What kind of memory was it? from the package i'd say it was a flash memory that contained the board's firmware .
pxi4071 missing chip is another firmware chip ( maybe for ni pci controller )- look pics of 4071 in internet , its atmel or something
What SHARC Processor are they using?
Video production Question...How come you never use video overlays ? i.e take a zoomed snapshot of the reworked IC.
+Yaghiyah Brenner Mostly because of limited time for editing. I'll consider it for the future. Thanks!
Where do you get your RF cables? Do you make them?
Would it be possible to relocate the mosfets and IDC connector to the third unused locations?
+Rob B The card firmware must be changed in that case.
+The Signal Path Blog I wonder if you could have used that second connector space, maybe by moving around the wires on the relay. If the common is in the same place (and it looks like the unused connections are the same) that could have been a better looking fix. But the important thing is that it is fixed of course.
Wanna see PXI-2596 mechanical co-axis switch (SP6T) internal structure, can you disassemble it?
I think the sp4t is made by radiall
Clock for the PCI Bus?
Can you get a PXI extender card? Then you can probe the bus at por. Given your investment in PXI cards, an extended might be a good buy.
👍👍
Hot air gun is very bad for reworking of such bga chips. You need infrared station with preheat plate. Try it so with another one ic.
Update on any of the Modules?
Is C7 knocked off the microwave switch board: 13:43
Cheers,
Mark
****************
With PXI-2596 wouldn't it be easier to solder the connector onto good footprint near to it, maybe resolder some transistors on empty footprints and put some wires between old transistors' pads and new? That will give more structural integrity to the connector itself when it will be sitting on all pads connected to the PCB.
That memory chip on PXI-4071 is definitely a Flash ROM. You didn't tell anything about it's content. Did you ordered programmed chip or just a blank one? According to its proximity to that problematic BGA it probably holds some startup code for it... And that BGA just doesn't start from the flash ROM and thus hangs the bus.
+Макс «Fagear» Крюков The EEPROM is blank. There is more than one on the board actually. The other connectors on the PXI-2596 aren't actually blank, they are routed to other points on the PCB. It would have been a conflict.
The Signal Path Blog I thought that those connectors are routed directly to transistors nearby and are not connected anywhere else.
Then control pins of transistors are routed to the IC. So it could be easy to reroute control paths from IC to transistors (cut some traces, solder some wires) and thus no firmware modification will be required and you'll get proper mechanical integrity of the connector.
Maybe I'm wrong on that, I guess.
Cat-on-scope.
My microwave doesn't even have switches. It's all those crappy soft buttons....
Furry boss approved your switch and DAC/ADC repairs. That's why DMM module fail, no approval ;)
On a serious note:
1. Do you have firmware which goes that NAND memory which was missing? It could be as simple as missing firmware, so NXI controller cannot run it's config and locking up bus
2. Another way to diagnose bus problem is to try measure resistance on those bus resistors. If you find one or two out of wack (way different resistance than others) - perhaps that line is shot/shorted.
3. Really curious which voltage reference under the can, perhaps LM399, as I cannot see precision resistors needed for LTZ1000-based ref.
Also would be cool if you could post some hi-res photos of modules on your site as well, for us, volt-nuts :)
Good work, thanks.
Your cat is going to add some enharmonic distortion to your read waveform on the oscilloscope.
I hope your cat pays attention to ESD safety!
That Cat has static dissipative fur.
I hope you've sprayed the cat with anti-static spray :)
+Greggan
The workbench must be cat heaven. So many humming and slightly warm spots to lie on. ;)
+Riskteven
I'm sure Shahriar knows when something is dangerous himself. Or did you see the cat run around when the heat gun was on?
And you don't "catch" cancer, a definitely not by walking around. Cancer is one of the main diseases for old dogs, even the happiest farm dogs.
+Riskteven Pooch is perfectly safe in the lab. Whenever there is anything that could be potentially dangerous, he is locked out of the lab.
+The Signal Path Blog And of course he's wearing a tin-foil hat :)
I wish my Rigol had a cat tail cursor :D
Your cat is just trying to help with your TH-cam ratings.... Everybody luvz cat vidz on the interweb...
before I start.....is there a cat?
You say "straight forward" in every video xd
Ha! Pooch!
Oh and besides whenever you say "very little money" I would like to know numbers instead... btw. for the bga, maybe you can locate a dentist or vet for a cheap xray? or walk over to mike from @mikeselectricstuff
+Dennis Lubert I got each unit for $50. :)
Hello, how i can contact with you?
Email - listed on my website.