General > General Technical Chat
The Rigol DS1052E
Ferenczyg:
Then, what approach is the rigth one? USB and the National Instruments Tool?
/F
Mark_O:
--- Quote from: drieg on April 08, 2010, 08:16:26 am ---Using serial cable... is very risky... In some cases it corrupts part of internal flash memory. Especially if you accidentally hit the ENTER instead of sending correct <LF> code, you can bet that you have just corrupted some data. Sometimes I've experienced these corrupted data even if I sent correct <LF>.
--- End quote ---
Drieg, thanks for raising a flag on this. Your second point I find most alarming.
--- Quote ---Internal flash holds some important factory calibration data (different from Self-Calibration) which are unique to every unit.
--- End quote ---
Yes. People should think of it like this. The precise values of the components in every Rigol unit is different. So during the initial factory calibration process, they fed in specific reference standard values in each range, and save the correction factors in NVM. These are per channel, per voltage range, per frequency range base/offset values. Those are what the scope uses when it later does it's own self-cals. It has to have an externally supplied and internally remembered reference to calibrate to. Normally, once set, these would never be changed, and thus the scope has a solid reference to rely on. But when backdoor mods are done, it's possible to change more than what you were planning on. Once overwritten, they're gone. :o
You mentioned "internal flash". Do you happen to know where this non-volatile data is being stored? I had assumed EEPROM.
Lastly, is sending these commands via USB any safer? I.e., isn't sending too long a string still a potential (though more easily avoided) problem? I would think if done properly they would be 100% reliable and safe, since that's how it's done in the factory.
- Mark
Mark_O:
BTW, the downside of all this is that it's essentially destroyed the resale value of used Rigols. :o It's the Law of Unintended Consequences.
Why do I say that? Because now there's no way to know whether a used unit was tampered with. Even if it's set back to 1052E, you don't know what's been done with it. And so far, if a unit is modded improperly, there's no way to recover vital data. It still displays a trace, but there are strange and unexplained behaviors. The bottom line is that there is invisible data that's critical to proper Rigol operation. Normally that data is secure, but a side-effect is that this data is vulnerable and can be undetectably destroyed. :(
One thing that no one here has done is perform a full set of base measurements with their unit (reference set), and then compare them with the readings after the mod. That would be extremely time-consuming, and here was no reason to think it was necessary. Yes, several have done some spot and range checks, which were indicative of good results (mostly at the high end of the frequency spectrum). But we don't know WHICH reference data is located immediately after the strings involved. So the problem might be only on, e.g. Chan2, 200-2000 mV, in a certain frequency range.
Who'd like to buy shafri or dimlow's unit now (sorry, guys), and try to use it for meaningful measurements you can rely on? Now, potentially any unit you come across could be like this. The mod has certainly been publicized widely enough. Until we find a solution to their problem (if one exists), it's now a problem for everyone. While shafri's problem may be very obvious, due to his making repeated mistakes when sending strings in, others may be much more subtle. I'd really like to hear more from Drieg on the corrupted data he's experienced when sending the proper <LF> terminator.
Rigol may even need to get involved with this, since it could completely undermine their integrity as an instrument that can be trusted. And it was their oversight that left the door open to this damage, so ultimately it's THEIR responsibility. Yeah, let's blame it on them. ;)
- Mark
EEVblog:
--- Quote from: drieg on April 08, 2010, 08:16:26 am ---@Dave: I'm sorry to say, but I don't think it was a good idea to make a video encouraging people to do this...
--- End quote ---
Err, so I shouldn't have told anyone about it at all?
What is a safer method? USB and a linux script? maybe, but plenty of people have had success with the serial port, with only a few failures it seems, and they probably didn't follow the instructions.
If you follow my instructions you shouldn't corrupt your firmware.
Dave.
Simon:
I can't see what the hype is about over calibration etc, the 1052 and the 1102 are the SAME machine and work in the same way, the only difference is that the input filter on the 1052 is told to lower it's bandwidth and the firmware will allow you to go to 2ns/div. I don't quite see here. of course if data that is required for calibration is overwritten then well it's an issue but frankly if changing the model in the same way as rigol clearly do it can have such adverse effects who is to say that rigol are not selling scopes that are already defective, their firmware should be more solid, this is after all a feature that they put in themselves. really this is no "hack" this is how the machine was made, it was meant to do this but we were never told
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version