Recovering Keysight U1273A from failed firmware update

[Zbig]

Hi

Today, after owning a Keysight U1273A multimeter for almost a year, I noticed they have released firmware version 3.01. They're stating the release date of 2014-10-09 but they have to had posted it no more than a month or so ago as I was checking it fairly often. Go figure. They're stating following changes over the previous version 1.95:

version 3.01
-       Add Filter setup mode option DC, DCAC and OFF
-       Bug fixed for incorrect LPF icon under dBm dual mode
-       Bug fixed for incorrect dipslay when AHOLD exit from LOG view under TEMP function
-       Bug fixed for incorrect Average auto range in AVG under ACV function
-       Bug fixed for incorrect Average auto range in AUTO
-       Change identity to Keysight

In the instructions on the webpage, they tell you to use the U1173B IR-USB cable. The update program tells you the same in red font. The thing is, I don't have it but I have the U1177 Bluetooth adapter so I went ahead and tried it anyway. Yes, I know. Don't judge me. As you could probably guess by now, I bricked the meter. Kind of. And then resurrected it. I don't encourage anyone to try updating using unsupported tools, like me, but I know of at least one case of bricking the meter using the recommended USB cable as well on this very forum: https://www.eevblog.com/forum/testgear/failed-firmware-update-on-u1273a-and-agilent's-response/
In my case, the update program detected the meter just fine. After I clicked "Update", it apparently put the meter in a bootloader mode, then said I should power-cycle before OK'ing the dialog box. That's when I got stuck with the "D...." (like the unlucky fellow from the thread I'm linking) and since when the update program refused to do anything useful further, failing to detect the meter and spewing various error messages instead. First, I remembered that I've changed the default baudrate of the Bluetooth adapter from 9600 to 19200. While I'm not sure that was critical for the eventual unbricking, it couldn't hurt either. Of course, it doesn't apply to you if you're using the USB cable you were supposed to in the first place. Set the slide switch on the adapter to the middle "Setup" position, connect to it from your favourite terminal application via the virtual COM port (I use PuTTY) and issue ATL1 if it responds to ATL? with anything other than "1". Now, the trick was to watch the BT adapter's green "Link" LED. I'm not sure whether the USB cable has some kind of activity light as well, otherwise there'd be more trial-and-error involved, I guess. So, I turned the meter off, re-launched the update software so it started its autodetect procedure. As soon as the link LED on the adapter went steady, I quickly switched the meter back on and that's what triggered the software to re-detect the unit properly and proceed with the actual update after I pressed "Update" for the second time. At last, an hourglass appeared after the "D..." and the progress bar on the PC program started to fill up terribly slow (whopping 64 bytes/s as the firmware itself seems to be 128kB in size) and some 2048 seconds later it greeted me with the Keysight (instead of Agilent) logo and new version number. Then you're supposed to perform a reset to defaults from the menu.

As a form of punishment, I'm hereby subjecting myself to the default startup tune for the duration of one month. I hope at least I could help someone.

Regards,
Zbig

[EV]

I think that the instructions are wrong also when using the cable U1173B. There is advised to connect the meter on before update. If you do so you get that "power recycle" note and the update does not begin.

I got it working when I connected the power on after the update button was pressed.

[Zbig]

I think that the instructions are wrong also when using the cable U1173B. There is advised to connect the meter on before update. If you do so you get that "power recycle" note and the update does not begin.

I got it working when I connected the power on after the update button was pressed.

Overall, I have very bad experiences with firmware update procedures where there is a reboot or re-enumeration designed-in as a part of the process. That is asking for trouble. The timing could go wrong, the device could get "snatched" by other application, the flaky drivers could fail, etc. If I was to design a firmware procedure, I'd resort to separate bootloader mode, enumerating as a different "updater" USB device, etc. only in an emergency. If you managed to connect to the thing successfully, hold on to it and don't mess around. Just push the bytes to the device and let its CPU deal with it. Even it that involved putting an additional memory buffer on the device - I don't care, I'll pay for that, just make it reliable.

[Muttley Snickers]

Nice work....

I just hope that the fellow at the bottom of this thread hasn't thrown his meter away, he couldn't get the meter to recover, your guidance here may help him.

https://www.eevblog.com/forum/testgear/howto-u1272a-firmware-update-with-u1177a-bluetooth-adapter/msg312441/#msg312441

[Zbig]

Nice work....

I just hope that the fellow at the bottom of this thread hasn't thrown his meter away, he couldn't get the meter to recover, your guidance here may help him.

https://www.eevblog.com/forum/testgear/howto-u1272a-firmware-update-with-u1177a-bluetooth-adapter/msg312441/#msg312441

Thanks Yeah, I hope that too. I shot him a PM already.

[rdubelectronic]

I am the unlucky guy who's meter WAS bricked

I used the cable and was stuck with D...
Tried everythinh, nothing worked. I also have the BT adapter. Tried many things, nothing worked.

THEN, I wote to keysite.
I could send the meter to their service center in Germany. I live in Holland.

THE SEND ME BACK A FULLY WORKING AND FULLY UPDATED EXCHANGE MODE, cannot be distinguished from new.
 

[miro_gt]

So here's an update that might help somebody out there ..

Long story short - bought u1273a at a discounted price, then found out about its interference issues, then decided to update it as my unit is apparently from 2012. It had the first issued firmware v.1.64 . Quick search and I found the 1.95 and the 3.01 firmwares available .. though I decided to do the 1.95 as I didn't want the Keysight logo on my unit since it was labeled as Agilent, as the 3.01 update seemed to only update that part (as another "improved averaging" didn't cut it for me). The bluetooth adapter I got on discounted price as well Lets just say that the meter + adapter cost me less than $200

How I got to the issue: I opened the meter to see what I got inside (rev.003). Since the case opened kind of hard on the top side, and I needed it open for further actions (will post later), I decided just to hold it closed (without fully closing it on the top side) with my hand as to be able to do what I thought would be a quick firmware update.  Well .. I knew that this was not a good idea ... and of course after 10 min of agonizingly watching the slow bar graph, my hand started to hurt .. and I took my chances of changing hands to hold it closed with the other hand .. and that's where the screen blinked for a moment and the sand clock disappeared. And so I got stuck with the "D...". Ouch.

How I got it to work: Well I tried with the suggested programs up top, and then with virtual ports and what not .. but nothing. The issue was that the bluetooth adapter was connected to my laptop however the firmware update software didn't activate the "firmware update" button no mater what I did. It didn't want to get the meter's serial number and the loaded firmware number anymore, and the "D..." was just sitting there on the meter itself. The link LED on the bluetooth adapter would just blink and that's it. Fast forward 2.5 hours (or from 12:30am to about 3am in the morning, cough .. dont ask) I noticed that the link LED would get steady for about a second when I had the v.3.01 update software running and searching for ports (not the v.1.95), which meant that for a moment that software would connect to the meter (and then nothing again), but that gave hope! And then I came up with the idea: since all I needed was to make the v.1.95 update software believe that there was a meter connected, why not use that moment when the v.3.01 actually connects to it. And guess what, it worked I put the v.1.95 to search for ports, and at the same time started v.3.01 to do the same. Right at that moment of a link between the v.3.01 and the meter, the v.1.95 activated the "firmware update" button! Hell yessss. Before it would only find ports (COM3 and COM4), but no active button. And this same button wasn't active on the v.3.01, but now it was on the v.1.95 ! So I closed v.3.01 and tried it out over v.1.95. It didn't go over COM3 but over COM4 the meter showed the sand clock again, and after another 30min or so there was a nice beep and there it came back to life, with v.1.95 loaded . This time with fully closed meter of course, not holding it by hand anymore .. duhh.

Hope that helps somebody out there.

Best Regards