Hi all,
Some good news for today. We've had a small fairwaves hackathon here in Moscow and we've done a good progress in UmTRX development. I'm quite exhausted, thus in short:
1. A software issue has been fixed in PLL tuning code. It works reliably now. Previously it didn't correctly handle edge cases when the best VCOCAP value was at the end of a range. 2. A software issue has been fixed in automatic LMS calibration code. Turns out a process described in the datasheet is not complete - you have read FAQ, question 4.7 for an amendment. 3. Now we correctly select Rx input. You have to change _two_ registers to do that - 0x75 and 0x25. This wasn't obvious from the brief reading of the manual. 4. A hardware bug was fixed - RF diversity switch required a bias to work correctly. Luckily we decided to move the switch from the main board to a frontend in the rev2 of UmTRX. So we don't have to re-make the UmTRX schematics.
Huge thanks to Andrey Sviyazov ("RF guy"), Andrew Karpenkov ("FPGA guy") and Srdjan Milenkovic ("remote LimeMicro guy") for going through all the issues.
Fixes for the code will be pushed to github in the next few days after I clean it up.
PS Another good news is that we should have first UmTRXv2 units on hands in the next few days, when we get them from customs. Can't wait to put my hands on them.
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
Andrew Karpenkov -- do you already have support for UmTRXv2 in the FPGA code? If we get our units from customs by Fri I want to test them during the weekend hackathon.
On Thu, Sep 13, 2012 at 2:40 AM, Alexander Chemeris alexander.chemeris@gmail.com wrote:
Hi all,
Some good news for today. We've had a small fairwaves hackathon here in Moscow and we've done a good progress in UmTRX development. I'm quite exhausted, thus in short:
- A software issue has been fixed in PLL tuning code. It works
reliably now. Previously it didn't correctly handle edge cases when the best VCOCAP value was at the end of a range. 2. A software issue has been fixed in automatic LMS calibration code. Turns out a process described in the datasheet is not complete - you have read FAQ, question 4.7 for an amendment. 3. Now we correctly select Rx input. You have to change _two_ registers to do that - 0x75 and 0x25. This wasn't obvious from the brief reading of the manual. 4. A hardware bug was fixed - RF diversity switch required a bias to work correctly. Luckily we decided to move the switch from the main board to a frontend in the rev2 of UmTRX. So we don't have to re-make the UmTRX schematics.
Huge thanks to Andrey Sviyazov ("RF guy"), Andrew Karpenkov ("FPGA guy") and Srdjan Milenkovic ("remote LimeMicro guy") for going through all the issues.
Fixes for the code will be pushed to github in the next few days after I clean it up.
PS Another good news is that we should have first UmTRXv2 units on hands in the next few days, when we get them from customs. Can't wait to put my hands on them.
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
Hi all,
The fixed LMS control code is now at github, branch fairwaves/umtrx: https://github.com/chemeris/UHD-Fairwaves/commits/fairwaves/umtrx
Note, that umtrx_ctrl.py command line interface has been extended and slightly changed for more consistency:
Renames: --lms-pa-on and --lms-pa-off => --lms-set-tx-pa --lms-set-vga1-gain => --lms-set-tx-vga1-gain --lms-get-vga1-gain => --lms-get-tx-vga1-gain --lms-set-vga2-gain => --lms-set-tx-vga2-gain --lms-get-vga2-gain => --lms-get-tx-vga2-gain
New commands: --lms-get-tx-pa --lms-set-rx-lna --lms-get-rx-lna --lms-set-rx-vga2-gain --lms-get-rx-vga2-gain
On Thu, Sep 13, 2012 at 2:40 AM, Alexander Chemeris alexander.chemeris@gmail.com wrote:
Hi all,
Some good news for today. We've had a small fairwaves hackathon here in Moscow and we've done a good progress in UmTRX development. I'm quite exhausted, thus in short:
- A software issue has been fixed in PLL tuning code. It works
reliably now. Previously it didn't correctly handle edge cases when the best VCOCAP value was at the end of a range. 2. A software issue has been fixed in automatic LMS calibration code. Turns out a process described in the datasheet is not complete - you have read FAQ, question 4.7 for an amendment. 3. Now we correctly select Rx input. You have to change _two_ registers to do that - 0x75 and 0x25. This wasn't obvious from the brief reading of the manual. 4. A hardware bug was fixed - RF diversity switch required a bias to work correctly. Luckily we decided to move the switch from the main board to a frontend in the rev2 of UmTRX. So we don't have to re-make the UmTRX schematics.
Huge thanks to Andrey Sviyazov ("RF guy"), Andrew Karpenkov ("FPGA guy") and Srdjan Milenkovic ("remote LimeMicro guy") for going through all the issues.
Fixes for the code will be pushed to github in the next few days after I clean it up.
PS Another good news is that we should have first UmTRXv2 units on hands in the next few days, when we get them from customs. Can't wait to put my hands on them.
-- Regards, Alexander Chemeris. CEO, Fairwaves LLC / ООО УмРадио http://fairwaves.ru
Hi all,
Another two days of intensive hackathon are over and OpenBTS now runs with UmTRX. This is thanks to the earlier efforts of Thomas Tsou and bugfixes from Thurs hackathon.
Few things which we planned, but haven't finished: 1. Dual-channel mode doesn't work yet. We nailed down this to the SR_TX_CTRL register position in FPGA, but haven't found a fix yet. To get dual-channel working we have to move SR_TX_CTRL to another place to free some space for the second channel registers. But when we do that, the first Tx channel stops working. 2. Sergey Kostanbaev started working on integration of essential LMS controls into UHD, but hasn't finished this yet. When done, this will greatly reduce number of manual steps to get OpenBTS running and will make it possible to use all those GnuRadio utils without limitations. 3. We still have to repeat results by Thomas and Sylvain where they get good quality GMSK from OpenBTS. What we get right now is quite crappy.
Many thanks to all participants who contributed their time and passion during these long days and to Sylvain for helping to tame our E4406E.
-
Alexander Chemeris