I don't really see why you are making such a fuss about this, if you have one you leave it powered on anyway (reference oven and all that), the front button is only a standby button really and then it takes not even half a second to turn on.
Let me get more technical then.
1) they use a softcore mapped in an fpga that can barely tick 20Mhz...
2) They need a massive amount of flash and ram to support this.
3 all a multimter dos is switch some relays around , read an adc and do some simple maths to compensate offset / gain and show the result on a display.
A friggin braindead PIC16 series can do that and it will sleep half of the time.
This is the ' we need 5 million transistors to turn on an LED' problem
All that effort and silicon had better be spent making a more precise meter. It adds nothing to the capabilities of the machine. Overkill, overhead , do not want.
And they do crash sometimes...
It is absurd to have to resort to a softprocessor in fpga with operating system for something so simple.
They are more focused on the flashy Pixel display ( why is slow by the way as to write those
Arge digits quite an amount of data has to be blasted . The starbursts that agilent and keithley , or alpha 7*5 for keithley 2002 , use are much faster in that aspect )
The only reason they did it is for the networking. So they could avoid having to write teir own network stack. Buy one then. Vxworks, or freertos will happily provide.
Why didn"t they use an off the shelf soc ?
Its a strange design. Those fpgas are more expensive than a soc . It's like this thing was thrown together and is not a real fluke design..
The older fluke benchmeters like the 45 and 88xx had a real asic to do the analog. They used a hitachi h8 had starburst displays, were fast and reliable.
This thing is slapped together from a bunch of user programable fpga's and discrete components.the software is written lazily by using a full blown operating system. The actual meter code is probably 20 kilobyte. With 2 megs overhead for the operating system....
It simply does not line up with any of the other machines fluke builds. It is awkward, strange and out of line.
None of the other meters out there use such an aproach.
It's not so much that it is running loonix, its that it does it in a strange way , for strange reasons. The whole design feelws uncomfortable.
There's other things in it that are 'weird'... Stuff is stored in battery backed ram (cr2032 coincell)... Why not eeprom or flash ? There is at least 3 flash chips in that thing... And they need a coin cell ? That is almost not user replacable ( need to break cal seals to swap it... )
They make the inguard outguard using discrete leds and phototransistors ?
The front/ back is not a true switch but some sliding contacts on the pcb held together with crummy
Plastic.
It's like they spent so much money on fpga they had to cut corneres elsewhere to save money , and cut corners in development time by resorting to a resource hungry os so they could use codemonkeys as opposed to paying real programmers.
Half of the parts in that instrument are dedicated to running the os and have nothing to do with measuring...
Very weird.
I'll see if i can get a few pictures of the guts of that thing. There's a department that has one. Maybe over linchtime i will rip it open and write a teardown of it. It's a very weird contraption.
Oh, andi t just doesnt feel right to have to wait that long for it to start.. Yes we power down fully setups over the weekends. Besides linux can boot in under a second. But you need someone who knows how to build such a beast.... Fluke clearly didn't . Pre-chewed , slapped together duct-tape engineering.
Home
Help
Search
Login
Register
