Hi there. I have trouble connecting to the rs232 connectors.
Running an usb-serial (whick works)
I had connection the other day and the umanager said v0.8. (used puttytel)
It got "stuck" and I had to disconnect.
Now I am not able to connect at all.
Is it possible that the fpga is "out of order" or something?
Total cable length is about 1m.
Best Regards.
Running an usb-serial (whick works)
I had connection the other day and the umanager said v0.8. (used puttytel)
It got "stuck" and I had to disconnect.
Now I am not able to connect at all.
Is it possible that the fpga is "out of order" or something?
Total cable length is about 1m.
Best Regards.
Did you power cycle ? Just by going in and look can't break anything. And check you communication programs settings.
From our experiences: Hyperterminal just works. Most *unix programs works. Other might need to adjust settings.
Seems like there are problems downloading with some of those others, will work on making it more robust.
Yes did power cycle. Program settings seems ok.
Also tried Hyperterminal - also with recommended settings, still no luck.
Will look into other serial adapters when I have the chance.
Hyperterminal can be downloaded here :
http://digitizor.com/wp-content/uploads/2009/08/hyperterminal1.zip
From this site:
http://digitizor.com/2009/08/29/ins...al-client-on-windows-vista-or-windows-7-free/
Also tried Hyperterminal - also with recommended settings, still no luck.
Will look into other serial adapters when I have the chance.
Hyperterminal can be downloaded here :
http://digitizor.com/wp-content/uploads/2009/08/hyperterminal1.zip
From this site:
http://digitizor.com/2009/08/29/ins...al-client-on-windows-vista-or-windows-7-free/
Last edited:
OK so I've created a 1000 tap filter that is -6dB a 22050Hz in rePhase, saved this out as a C snippet, cleaned up the FIR parameters in excel, and moved the FIR parameters to the 1021filt.txt. After a bit of messing around I worked out that you need to have the 1021filt.txt in the same directory as MKROM.exe, and you need to cd to the directory to run or you get invalid filename errors. So now I have a new 1021filt.skr file.
I've connected to the uManager and there are no options I can see for uploading the new .skr file.
Code:
dam1021 uManager Rev 0.80 20150121 FPGA Rev 0.9 Press ? for help.
# ?
uManager Commands
?/Help show this help
set par=value set paramter to value, set alone to show
exit exit uManager
update update uManager firmware
download download and update system flash
idf show ID of flash
df [adr] dump flash content
#So how do we update filters?
cheers
Paul
I suspect Søren might be the only person who might be able answer that at present. I'd guess most people will have ordered the same tolerance for all their boards if they purchased more than one.
Not sure if this is correct but the answer seems to be:
Exactly the same way you update the firmware.
Code:
# download
Start sending file using XMODEM/CRC protocol.
CCCCCCCCCCCCCCCCCCCCCCCCCCC
### Send (X) 1021filt.skr: 10752 bytes, 0:02 elapsed, 3644 cps, 31%
File downloaded and programmed, size 010752 Bytes.The DAC still behaves so there are two possibilities a) it did nothing and the original filters are installed, or b) the upload was successful. Not sure how to confirm this.
Interested in your progress.
Either Soren will confirm the correct method, or you could build a filter with something ridiculous such as low pass at some very low frequency, upload it, then have a listen.
As others already suggested, the same download command, the .skr files have headers so the download command knows what to do with them....
When testing we do a FFT and measure distortion at -1db, we don't really see any difference between the different version as the initial precision is so good, even on less precise resistors.
But I expect long term the boards with more precise resistors will change less than the less precise resistors as temperature and long term drift are significant better with the more precise resistors....
Hi Paul,
For the 44.1kHz 8x filter you need to manually set the sampling rate to 352800 in rePhase (this will be included in the drop menu for the next version...)
Make sure you use a gentle windowing algorithm such as Blackman-Nuttall or Blackman-Harris to get a low ripple floor.
Set the number of taps to an odd value to get a better ripple rejection (there seems to be a bug there, sorry about that).
If you use iterative optimization make sure the optimization floor dB to a low value (default to -80dB, which means that iterative steps to fit the target might results in ripples rising up to that point during optimization...).
You will want to check the result curve in both large scale (for example -150dB) and a more closeup scale (like +/- 0.1 dB) to observe the behavior of the filter and decide what optimization and windowing parameters to use.
As for the output format, choose the '32 bits floats mono (.txt)' one, and you should then be able to easily replace the coefficient within Soren's example file with a simple text editor.
It might probably be a good idea to only replace the 44.1kHz filter in Soren's file for your tests.
If 1016 taps (use 1015!
) are available then there is probably plenty of room for improvement compared to the existing 383 taps windowed brickwall.You might try minimum-phase filters, or "intermediate" phase filters by combining a linear-phase and a minimum-phase LP filters (similar to what sox resampler does). You can also set a minimum-phase L-R filter and partially linearize its phase with the "phase linearization" tab...
When using non linear-phase filter you might want to set the centreing to "energy", or do it manually with percentages (something like 1% or a bit more for purely minimum-phase filters, and 50% for linear phase-filters, and any proportional ratio in between for intermediate ones...).This will ensure you make the best use of the available taps.
You can also play with EQ to adjust some specific areas.
Lastly, you might want to remove something like 1dB (in the "general" tab) to ensure to saturation will not take place...
Last edited:
Not in my opinion....
But if you insist I would recommend to remove the opamps instead of damaging the board. It's pretty easy to cut the opamp pins close to their housing using a sharp scalpel.
Hi Paul,
For the 44.1kHz 8x filter you need to manually set the sampling rate to 352800 in rePhase (this will be included in the drop menu for the next version...)
Make sure you use a gentle windowing algorithm such as Blackman-Nuttall or Blackman-Harris to get a low ripple floor.
Set the number of taps to an odd value to get a better ripple rejection (there seems to be a bug there, sorry about that).
If you use iterative optimization make sure the optimization floor dB to a low value (default to -80dB, which means that iterative steps to fit the target might results in ripples rising up to that point during optimization...).
You will want to check the result curve in both large scale (for example -150dB) and a more closeup scale (like +/- 0.1 dB) to observe the behavior of the filter and decide what optimization and windowing parameters to use.
As for the output format, choose the '32 bits floats mono (.txt)' one, and you should then be able to easily replace the coefficient within Soren's example file with a simple text editor.
It might probably be a good idea to only replace the 44.1kHz filter in Soren's file for your tests.
If 1016 taps (use 1015!
You might try minimum-phase filters, or "intermediate" phase filters by combining a linear-phase and a minimum-phase LP filters (similar to what sox resampler does). You can also set a minimum-phase L-R filter and partially linearize its phase with the "phase linearization" tab...
You can also play with EQ to adjust some specific areas.
You might want to remove something like 1dB (in the "general" tab) to ensure to saturation will take place.
When I looked at rePhase I didn't see any "intermediate phase" setting ? Or how to combine ? I would be nice if I don't have to mess with the SoX source code....
You don't need to worry about saturation, the filters coefficients floats are converted to 2.30 fixed format.
The filters appear to "stack" and are combined when you generate the filter.
Thanks POS for all the hints, they are very helpful for getting a handle on the process. Still just stumbling around with what can be done but it's a start.
Add: I keep reading back over POS' post and finding more information that I didn't pick up on the first pass. It's going to take me a while to digest everything I think.
I'm basically just having a play around at the moment - as I'm not sure what does what
I've attached a .skr with the 44.1 filter replaced with a minimum phase filter, linkwitz riley "shape" and 12dB filter slope and corner frequency of 159994.8Hz. My numbers are probably off but I'm setting the corner frequency at 0.4535*fs where the fs is 44100 * 8. (I re-read POS's post and corrected from 384 -> 352.8, oops)
I'm not sure why, but the rePhase filter is about 6dB quieter than the one Søren made, so
******* Beware, Warning, Actung, Alert, Danger *******
The 44.1khz filter is significantly quieter than the official filters. If you change to a higher sampling rate without adjusting volume it will be significantly louder. Be warned!!
*************
Anyway don't expect perfection, because I'm no expert in filter design, but might be of interest to someone!
And a question... is there a way to visualise impulse response in rePhase?
cheers
Paul
Add: I keep reading back over POS' post and finding more information that I didn't pick up on the first pass. It's going to take me a while to digest everything I think.
I'm basically just having a play around at the moment - as I'm not sure what does what
I've attached a .skr with the 44.1 filter replaced with a minimum phase filter, linkwitz riley "shape" and 12dB filter slope and corner frequency of 159994.8Hz. My numbers are probably off but I'm setting the corner frequency at 0.4535*fs where the fs is 44100 * 8. (I re-read POS's post and corrected from 384 -> 352.8, oops)
I'm not sure why, but the rePhase filter is about 6dB quieter than the one Søren made, so
******* Beware, Warning, Actung, Alert, Danger *******
The 44.1khz filter is significantly quieter than the official filters. If you change to a higher sampling rate without adjusting volume it will be significantly louder. Be warned!!
*************
Anyway don't expect perfection, because I'm no expert in filter design, but might be of interest to someone!
And a question... is there a way to visualise impulse response in rePhase?
cheers
Paul
Last edited:
