gain distribution and and test app

fairwaves / UHD-Fairwaves

Fairwaves version of the UHD drivers, tweaked to support Fairwaves UmTRX.

http://umtrx.org

25 stars 21 forks source link

gain distribution and and test app #7

Closed guruofquality closed 9 years ago

guruofquality commented 9 years ago

ensure that overall gain setting distributes to available gain elements and they actually get set
create an app to test set the gain and confirm that the values are as expected

guruofquality commented 9 years ago

Update commit with test application on umtrx_gains branch: https://github.com/fairwaves/UHD-Fairwaves/commit/4d7a11c21a09f0c049ef30386889a44223385f89
The test app will fail with 1dB errors due to uhd bug( I am trying to upstream a patch)
I also performed a loopback test on the hardware to confirm rx and tx gains

chemeris commented 9 years ago

Interesting. I indeed saw a 1dB error when setting the gain, but was thinking it's my problem.

chemeris commented 9 years ago

What kind of loopback test did you perform? Did you check the actual register values in LMS with get_tx_vga1gain(), get_tx_vga2gain(), get_rx_vga2gain() or with the umtrx_lms.py utility?

Could you try to run osmo-trx and check that it actually set gains properly on both channels? Please test with this commit: http://cgit.osmocom.org/osmo-trx/commit/?h=fairwaves/master&id=79ca9ae69450142abf51c8cbd4d4a39e33169c92 There is a later commit by @kluchnikov , which I think is incorrect, so please ignore it: http://cgit.osmocom.org/osmo-trx/commit/?h=fairwaves/master&id=48da6f97e3274f751a133365b526f4ca0f3b7ea1

PS Also see my comments for the patch.

guruofquality commented 9 years ago

To test that the registers were actually set, I changed set_tx_gain() and set_rx_gain() to return the actual gain values as readback through spi using the get_tx_vga1gain(), get_tx_vga2gain(), get_rx_vga2gain(), etc calls from lms6002d.hpp.

And I created a unit test app to check the set/get gain API calls on multi-usrp under several discrete values to check that the distribution algorithm is working correctly: https://github.com/fairwaves/UHD-Fairwaves/blob/4d7a11c21a09f0c049ef30386889a44223385f89/host/utils/umtrx_test_gains.cpp

The loopback test involved a simple transmit/receive app though leakage, and dragging the gain through the complete range and ensuring that the levels appeared to change correctly without jumps and strange behaviour: http://i.imgur.com/Mabcurx.png

chemeris commented 9 years ago

re: testing registers and leakage - great idea! Did you check for both channels?

Could you test with osmo-trx as well, though? We must be sure that the gains are set to optimal values, as described in my patch for osmo-trx: http://cgit.osmocom.org/osmo-trx/commit/?h=fairwaves/master&id=79ca9ae69450142abf51c8cbd4d4a39e33169c92

guruofquality commented 9 years ago

I believe that this issue is well tested and completely merged now. OK to close it?

chemeris commented 9 years ago

Yes, let's close it. We haven't done much testing ourselves, but I trust your testing app.

Please excuse typos. Written with a touchscreen keyboard.

Regards, Alexander Chemeris CEO Fairwaves, Inc. https://fairwaves.co On May 7, 2015 3:38 PM, "Josh Blum" notifications@github.com wrote:

I believe that this issue is well tested and completely merged now. OK to close it?

— Reply to this email directly or view it on GitHub https://github.com/fairwaves/UHD-Fairwaves/issues/7#issuecomment-99992071 .

guruofquality commented 9 years ago

Cool, closed