Bill,
How can you dismiss such an obvious improvement! There was one professional publication I get that announced 64 bit audio converters. For some reason they never thanked me when I pointed out at 64 bits of output above the absolute minimum noise level required the destruction of a galaxy or two!
I am more impressed he could actually measure such a difference reliably!
How can you dismiss such an obvious improvement! There was one professional publication I get that announced 64 bit audio converters. For some reason they never thanked me when I pointed out at 64 bits of output above the absolute minimum noise level required the destruction of a galaxy or two!
I am more impressed he could actually measure such a difference reliably!
Last edited:
Yes Ed, people forget the scale, NASA uses pi to 16 bits, they reckon it only gave a 2" error on a 1.5billion mile jouney for a satellite.
Richard Cabot, Bruce Hofer and Robert Wright were the force behind Audio Precision. Bob was quite helpful discussing circuitry before he retired. Still e-chat with Rich and a little with Bruce. Bruce did a bit of follow up on my bit on resistor distortion.
I presume that's 16 decimal digits, not binary.
Also, there's the thread in recent years on the "32 bit DAC." Wait, I looked, there are several threads...
pi sort of 3.141592653589793238 so the error at 16 decimal places is .000,000,000,000,000,038 x 1,500,000,000 x 5280 x 12 = .000361"
Seems it is not decimal digits!
Seems it is not decimal digits!
Last edited:
Pi - Wikipedia
In high school we were taught a mnemonic which Plato used (mentioned by Plutarch in “Questions”):
(see attachment. Count and write down the number of letters in the sentence, it’s 3,14159)
(translation: Eternally, The God The Great designs by Geometry)
George
In high school we were taught a mnemonic which Plato used (mentioned by Plutarch in “Questions”):
(see attachment. Count and write down the number of letters in the sentence, it’s 3,14159)
(translation: Eternally, The God The Great designs by Geometry)
George
I can't imagine any major scientific software not using floating point math anymore, this is probably some anecdote from 1968 or so. The only reference to 64 bit converters I can find are file conversion packages that include the IEEE 64 bit double precision format.
Last edited:
True for the ground computers and software, but once you step into the embedded computers/software that leave the planet on board of deep space probes you may have a big surprise.
I suppose it's been hard to find rad-hard and space qualified Cortex M4F MCUs until very recently 🙂.
^^^
Are you aware that NORAD still relies (or it did, about 6-7 years ago when I last heard about) on 1.44Mb floppy disks to carry coordinates around? USB and thumb drives? Naaahhh, those are for wusses 😀. In the early 90's when I've last seen a PDP-11 installed in an ahem place, it was running a special version of RSX-11 from a huge sized (10 cu.ft.) single plate disk unit of 5Mb. My 486 PC at the time had a 450Mb 5" IDE hard drive (was a biggie then, though).
And that military Agilent/HP T&M (GHz scopes, logic analyzers, etc...) critical gear relied on Windows 95 until ~2007 because that was the only qualified and approved OS? I have a 54832M 1GHz scope (M stands for military) that was decommissioned in 2009 while still running W95. The first thing I did was to replace the (carefully erased, could not resist peeking) 8GB hard drive (also military approved) with an 120GB SSD plus a $10 IDE-to-SATA adapter and set up Windows XP, as in all commercial 54832B scopes. It now boots x5 faster and did not crash once in the last few years.
Are you aware that NORAD still relies (or it did, about 6-7 years ago when I last heard about) on 1.44Mb floppy disks to carry coordinates around? USB and thumb drives? Naaahhh, those are for wusses 😀. In the early 90's when I've last seen a PDP-11 installed in an ahem place, it was running a special version of RSX-11 from a huge sized (10 cu.ft.) single plate disk unit of 5Mb. My 486 PC at the time had a 450Mb 5" IDE hard drive (was a biggie then, though).
And that military Agilent/HP T&M (GHz scopes, logic analyzers, etc...) critical gear relied on Windows 95 until ~2007 because that was the only qualified and approved OS? I have a 54832M 1GHz scope (M stands for military) that was decommissioned in 2009 while still running W95. The first thing I did was to replace the (carefully erased, could not resist peeking) 8GB hard drive (also military approved) with an 120GB SSD plus a $10 IDE-to-SATA adapter and set up Windows XP, as in all commercial 54832B scopes. It now boots x5 faster and did not crash once in the last few years.
Last edited:
Its a little harder to get a Trojan or malware on a 1.44 Mb floppy, especially if you don't know the language or OS or even what hardware its talking to. Seems pretty smart in retrospect. So does pre-automation utility hardware that just did simple stuff. Much harder to muck with it when you need to roll a truck to get it to do something.
All true, but I'm not sure if it was smart or just lucky that some of our critical infrastructure is air-gapped as a side-effect of neglect.
We are having a growing problem as time goes on. All kinds of devices, even in regulated industries, are using embedded Linux and Windows and will either never receive updates, or there are no patches that can be backported easily. This is definitely bad news for internet connected devices.
On the other hand, I recently witnessed a war between the product, regulatory, and the cybersecurity groups of a corporation over a device a few months ago. The product is not and cannot be connected to a network, but the cybersecurity guys insist the embedded Linux kernel needs to be updated to fix a TCP stack vulnerability because they scanned the Git repo and it showed up. They don't appear to understand that if someone has physical access to the device then it's game over anyway.
We are having a growing problem as time goes on. All kinds of devices, even in regulated industries, are using embedded Linux and Windows and will either never receive updates, or there are no patches that can be backported easily. This is definitely bad news for internet connected devices.
On the other hand, I recently witnessed a war between the product, regulatory, and the cybersecurity groups of a corporation over a device a few months ago. The product is not and cannot be connected to a network, but the cybersecurity guys insist the embedded Linux kernel needs to be updated to fix a TCP stack vulnerability because they scanned the Git repo and it showed up. They don't appear to understand that if someone has physical access to the device then it's game over anyway.
Pretty funny. I guess if you pay enough money Agilent will make sure their software runs on an ancient OS.
We had a demo recently of a LeCroy WaveRunner that ran Windows 10. It's amusing to see they didn't even bother customizing the splash screen and you can see ESET Antivirus load before their GUI does.
I guess that's better than the Keysight MSOS404A that showed up last year with a watermark in the bottom-right corner complaining Windows wasn't activated 😀.
I was trying to use a cash machine while visiting a UK University. It was not working, so I pointed it out to the staff. They unplugged it and plugged it back in.
First thing that happened was W7 booted up. Which is very scary - that a cash machine relied on a Windows operating system in the first place, and how was it updated in the second place....
First thing that happened was W7 booted up. Which is very scary - that a cash machine relied on a Windows operating system in the first place, and how was it updated in the second place....
This was a recent article about the Juno satellite that made a 1.5 billion kilometer journey through space, and they were saying that the 16 digits of pi only gave a 2" error on the final position, so they were doing everything in fixed point maths.
Juno Mission was selected by NASA Jun 2005. Budget accounting began same year.
Planetary Science Budget Dataset - The Planetary Society - Google Sheets
16 bits is too much for relative simple guidance and control ground orders/capsule response (deep space complex location identification and orientation calculation is done within the vehicle by constantly identifying star patterns).
The mass data flow through the communication to earth channel is required for telemetry data and downloading of on board scientific instruments read output, which Juno has plenty and quite advanced
https://web.archive.org/web/2011072...gov/dspace/bitstream/2014/40566/1/07-2266.pdf
Instrument Overview – Juno
>edit. This one too: https://descanso.jpl.nasa.gov/DPSummary/Juno_DESCANSO_Post121106H--Compact.pdf
George
Planetary Science Budget Dataset - The Planetary Society - Google Sheets
16 bits is too much for relative simple guidance and control ground orders/capsule response (deep space complex location identification and orientation calculation is done within the vehicle by constantly identifying star patterns).
The mass data flow through the communication to earth channel is required for telemetry data and downloading of on board scientific instruments read output, which Juno has plenty and quite advanced
https://web.archive.org/web/2011072...gov/dspace/bitstream/2014/40566/1/07-2266.pdf
Instrument Overview – Juno
>edit. This one too: https://descanso.jpl.nasa.gov/DPSummary/Juno_DESCANSO_Post121106H--Compact.pdf
George
