-
Video Repair of PXI-2596, PXI-4461 & PXI-4071 Modules
Posted by
Hugoneus
on 04 Oct, 2015 03:44
-
-
#1 Reply
Posted by
TheSteve
on 04 Oct, 2015 05:36
-
Another great video Shahriar - thank you for posting it. Do you have Pooch helping when you solder stuff, an extra set of paws can be pretty useful.
Shame the DMM didn't come back to life. The memory chip you replaced is ram I assume. Have you spent any time to see if the module is killing the power supply or is it the data bus itself? You've likely done more troubleshooting off camera - maybe walk us through what you have all done.
-
#2 Reply
Posted by
bktemp
on 04 Oct, 2015 06:22
-
In the video you said the missing component looked similar to the one on the other side. And this looks like a Flash memory, most likely containing some firmware, or the FPGA bitstream.
If you replace it with a new one, the firmware is still missing. If this firmware controls and initializes the PCI controller this could explain why the bus is not workling.
-
#3 Reply
Posted by
Hugoneus
on 04 Oct, 2015 16:12
-
In the video you said the missing component looked similar to the one on the other side. And this looks like a Flash memory, most likely containing some firmware, or the FPGA bitstream.
If you replace it with a new one, the firmware is still missing. If this firmware controls and initializes the PCI controller this could explain why the bus is not workling.
It is possible that some firmware is missing, unfortunately I don't have the original firmware...
-
#4 Reply
Posted by
TiN
on 04 Oct, 2015 16:33
-
Perhaps you can read firmware from your other module and try flash that one. There is little chance PXI side of things is unified, so firmware from other module would work.
Or maybe NI can send you the correct one?
-
#5 Reply
Posted by
lukier
on 04 Oct, 2015 16:45
-
Do you have high-res photos of these cards by any chance?
I'm curious what ADC and DACs they use on the PXI-4461. 24 bit DAC is actually something spectacular, so I strongly suspect they use cheapo audio AD/DA! These things have horrible DC specs, so often they don't even mention INL/DNL and ENOB in the datasheets (not to mention gain/offset temperature drifts).
-
#6 Reply
Posted by
Hugoneus
on 04 Oct, 2015 22:35
-
Perhaps you can read firmware from your other module and try flash that one. There is little chance PXI side of things is unified, so firmware from other module would work.
Or maybe NI can send you the correct one?
I am not sure if this is easy to do... My other PXI modules don't use the same PCI bus IC...
-
#7 Reply
Posted by
TheSteve
on 05 Oct, 2015 02:27
-
Well if it is a flash chip I'd say you're much stuck unless you can find another DMM module and remove/read the flash chips to copy the images. I would have thought a card with a blank flash would not cause problems on the bus. Perhaps it reads some code from the first flash and then gets stuck looking for the second one.
-
#8 Reply
Posted by
Vgkid
on 05 Oct, 2015 03:19
-
What is the voltage reference in the 4071?
-
#9 Reply
Posted by
Hugoneus
on 05 Oct, 2015 17:28
-
What is the voltage reference in the 4071?
LTZ1000
-
#10 Reply
Posted by
guido
on 05 Oct, 2015 17:32
-
I would have done the repair of the relay module differently:
- put the connector on the third, unused, section
- connect the transistors (or FETs, can't remember) to the equivalent transistor pads going to the connector
Connector mounted sturdy on the pcb and just six (or seven with gnd?) easy connections.
Just my 2 cents
![Tongue :P](https://www.eevblog.com/forum/Smileys/default/tongue.gif)
Edit: too bad that was not possible. Looked that way but i guess you checked.
-
#11 Reply
Posted by
Hugoneus
on 05 Oct, 2015 18:01
-
I would have done the repair of the relay module differently:
- put the connector on the third, unused, section
- connect the transistors (or FETs, can't remember) to the equivalent transistor pads going to the connector
Connector mounted sturdy on the pcb and just six (or seven with gnd?) easy connections.
Just my 2 cents
![Tongue :P](https://www.eevblog.com/forum/Smileys/default/tongue.gif)
That was not possible. The empty connectors aren't actually empty. The connections go to the other MOS switches which are already occupied. The connections are designed with some other type of microwave switches in mind.
-
#12 Reply
Posted by
Vgkid
on 05 Oct, 2015 19:10
-
Snip
Thanks, that is what I assumed it was.
-
#13 Reply
Posted by
AF6LJ
on 05 Oct, 2015 19:52
-
Good video.
Good luck with the DMM module, but I am not holding my breath.
-
#14 Reply
Posted by
BFX
on 05 Oct, 2015 22:39
-
Thank you Shahriar for another instructive video. It's very helpful for lot of people here I think
![ThumbsUp :-+](https://www.eevblog.com/forum/Smileys/default/icon_smile_thumbsup.gif)
Did you try thermal camera scanning with power of this board ? Only to be sure if isn't bug somewhere in power lines.
-
-
PXI ? SET IT ON FIRE !!
#$%(@*#$% overpriced craptional instrucrap crap.
-
#16 Reply
Posted by
Monkeh
on 06 Oct, 2015 12:07
-
Another showing of the rare and underappreciated oscillocat
-
#17 Reply
Posted by
Hugoneus
on 06 Oct, 2015 20:32
-
PXI ? SET IT ON FIRE !!
#$%(@*#$% overpriced craptional instrucrap crap.
Haha! Yeah... that DMM is listed for $3.3k.
-
#18 Reply
Posted by
rx8pilot
on 06 Oct, 2015 20:39
-
PXI ? SET IT ON FIRE !!
#$%(@*#$% overpriced craptional instrucrap crap.
I like PXI and LabVIEW.
Sent from my SAMSUNG-SM-G890A using Tapatalk
-
#19 Reply
Posted by
c4757p
on 06 Oct, 2015 20:41
-
It's so overengineered! That microwave switch module had what - an FPGA, a PCI interface chip, like three memories, some unidentified SSOP, some microcontroller-looking thing... did I miss anything? All to switch relays! Even if they want it to be PCI only, it doesn't take that much to slap something onto a PCI bus, pretty much a lone FPGA with no external memory other than the main flash can do that. The FPGA can also generate the relay control signals directly, store the actuation counts back into the flash, etc...
-
#20 Reply
Posted by
rx8pilot
on 06 Oct, 2015 23:19
-
It's so overengineered!
I won't argue that. I will say that as I have been researching and learning a lot about automated T&M for design validation and production and National Instruments has been a great thing. If you look at it from the perspective of someone like myself that will use LabView with PXI hardware and stand-alone T&M - the integration is hard to beat. I can rather quickly setup and automate a test that was taking 30min to do and required me or some other qualified engineer to do and smash it down to 1 minute and require next to no skill at all. The time savings of automated and integrated tests that I can have housed in a single chassis is enough to justify the cost of a PXI system.
In my previous careers, I have always 'over purchased' what many called over engineered. It made me and my companies the quickest most reliable in our field - and the most profitable. I am a relative newbie in electronics design and manufacturing, but I can already tell that buying the high-end pays for itself. I have not yet purchased my PXI system, but I am confident the $ that goes in will be $$$ that comes out.
It may be never that I need a 26Ghz relay on the PXI system.....
-
#21 Reply
Posted by
c4757p
on 06 Oct, 2015 23:23
-
And yet the exact same functionality can be implemented with significantly less faff. What does any of that have to do with whether or not the modules are overengineered?
I'm not saying anything against the PXI system existing. It looks quite nice, I'd love to have one. It just looks like they made it cost about 10x what it has to.
-
#22 Reply
Posted by
rx8pilot
on 06 Oct, 2015 23:49
-
The over-engineered part raises the price of the modules. While I wish they were less expensive - I voted to pay extra to have a single, compact, totally integrated test system that is easily operable by a low-level technician. Having an immobile bench with a stack if 8 instruments all connected to a laptop is not nearly as friendly.
You say....
It's so overengineered!
The most glaring problem with over-engineered T&M is that the price makes it scary. My point was - who cares if it is over engineered if it solves all of my problems and I don't have to worry about it. The problem with the traditional route is the significant increase in labor costs since it needs a much better technician or engineer to operate. The over-engineering buys me a solution that cost very little to operate every day. I can put a PXI chassis on a cart with a bed of nails, roll it over to the end of the P&P line and have the operator running testing and programming of the PCB's as they come out. At that point I don't care if the relay card is controlled by FPGA's - I am happy that I don't have a $75k/year technician/engineer testing PCB's.
Hopefully that makes sense.
-
#23 Reply
Posted by
c4757p
on 06 Oct, 2015 23:51
-
Sure, of course, it can be overlooked, and clearly is, by many people. Just saying, all that stuff to drive some relays is a bit comical.
-
#24 Reply
Posted by
rx8pilot
on 06 Oct, 2015 23:58
-
I agree.
There is a company that I used to buy very expensive video terminal equipment from. They had many general purpose designs that could be used for many purposes. What you ended up with was a product that was totally appropriate if it was the most sophisticated in the series. When you look at the entry level configuration - it was hilarious. Sometimes a few FPGA's were doing what could be done with a single microcontroller. The I/O architecture required all three FPGA to be doing a teeny tiny bit of work.
The justification was the the cost of the extra components was less than the cost of whole new dedicated and optimized design. Volume was rather low, so I believed that explanation.
-
#25 Reply
Posted by
Bud
on 07 Oct, 2015 01:00
-
Speaking of the relays, at 8:20 it was said they have remarkable repeatability because they are mechanical, not solid state, which i find hard to believe (that mech ones are better)
-
#26 Reply
Posted by
tec5c
on 07 Oct, 2015 01:59
-
PXI ? SET IT ON FIRE !!
#$%(@*#$% overpriced craptional instrucrap crap.
-
#27 Reply
Posted by
rx8pilot
on 07 Oct, 2015 04:53
-
Speaking of the relays, at 8:20 it was said they have remarkable repeatability because they are mechanical, not solid state, which i find hard to believe (that mech ones are better)
At 26Ghz....a very carefully designed and constructed contact probably works better than silicon. It is rather far out of my understanding, perhaps Shahriar can help decode that one. I have worked on many projects (as mechanical engineer) but was watching over the shoulders of the EE's dealing with 3Ghz digital and the delicate nature of that work. At 26 Ghz, its hard to imagine how delicate that is.