Forum
Please or Register to create posts and topics.

NOW AVAILABLE: FPGA RAM based IQ-Generator

So far the SPECTRAN V6 IQ signal generator gets the data via USB stream.

We are just working on a FPGA upgrade to implement basic generators (CW, sweep, ramp, step, chirp etc.) within the FPGA. This will reduce the USB data rate drastically and will give you more headroom for time relevant missions.

This feature was implemented in the RTSA-Suite PRO 1.5.152.8641.

Now you can select an additional Transmitter Mode called "Pattern Generator" within the SPECTRAN V6 Board Config -> Signal Generator config. This will load the selected signal generator to RAM and replay it.

Since the Tx no longer needs the USB data transfer it is possible to use both Rx without any data limit together with the Tx:

RAM Memory based IQ Signal Generator

Is there any way currently or in the future to load arbitrary I/Q data into RAM on the SpectranV6 (as opposed to RAM on the PC) for playback out of the VSG without taking up USB bandwidth while playing it (like hooking up a File Source block to the SpectranV6 block does).

This was an idea from the beginning (we even have the RAM on board) but since USB bandwidth is no poblem (and we have 2 USB which can operate indepenent for Rx and Tx at the same time) with todays hardware we skipped it to offer more flexibilty (and to save FPGA space for more usefull features).

Thanks for the response.

I am a little confused, isn't the point of this pattern generator feature described in this thread, and this statement in your post above from 12/4/2021: "Since the Tx no longer needs the USB data transfer it is possible to use both Rx without any data limit together with the Tx", that the USB bandwidth is a constraint?  I have certainly been finding the USB bandwidth and/or buffer sizes to be an issue in a lot of cases (see USB overflow and underflow warnings frequently even on single channel RX for 184 and 245 MHz rates).

The use case of an arbitrary IQ playback from "On Spectran" RAM to the VSG that does not use USB bandwidth so it is available for RX seems to apply in the same way as this non-arbitrary IQ from "pattern generator" applies over use of the "signal generator" option in the TX drop down or the IQ Signal Generator block with Stream dropdown option.  I've found that when using "Signal Generator" dropdown option or the "IQ signal generator" block that it seems happy right up until I also try and do any RX on even a single channel at the same time.  As soon as I do RX as well I see a "glitches" in the TX output and USB warnings.

Even better in the future would be "On Spectran" modulators for TX so that binary data could be sent over USB or stored in "On Spectran" RAM rather than IQ data, and the bits to IQ conversion happened in FW on device.

It would be really amazing if blocks like IQ modulator, IQ demodulator, and IQ signal generator (or any other ones that would significantly ease the PC processor load and/or USB BW if run "On Spectran") got automatically pulled into the SpectranV6 FW rather than running on the PC if they are hooked up to the SpectranV6 IQ ports in the block diagram.  I know this is probably a ton of development, but the potential is there I think.

Maybe another thing to think about would be to use Thunderbolt rather than USB for future devices, as I understand it that allows for DMAs and other features that USB does not support that would unburden the PC and improve data transfer/buffering performance.

CorNic East has reacted to this post.
CorNic East

Big ECHO!  I currently am having issues with the transmit function and would love to see a (deep) ARB/Pattern RAM implemented (in addition to USB-C Lightning).

Thanks for the response.

I am a little confused, isn't the point of this pattern generator feature described in this thread, and this statement in your post above from 12/4/2021: "Since the Tx no longer needs the USB data transfer it is possible to use both Rx without any data limit together with the Tx", that the USB bandwidth is a constraint?  I have certainly been finding the USB bandwidth and/or buffer sizes to be an issue in a lot of cases (see USB overflow and underflow warnings frequently even on single channel RX for 184 and 245 MHz rates).

No its not. This feature simply gets more out of the unit. Since the unit can't generate a higher IQ rate then the 245 MHz via USB the RAM would not help to overcome this limit.

With the right setup and hardware you should never get any USB underflow or overflow issues. Which cases do you mean?

The use case of an arbitrary IQ playback from "On Spectran" RAM to the VSG that does not use USB bandwidth so it is available for RX seems to apply in the same way as this non-arbitrary IQ from "pattern generator" applies over use of the "signal generator" option in the TX drop down or the IQ Signal Generator block with Stream dropdown option.  I've found that when using "Signal Generator" dropdown option or the "IQ signal generator" block that it seems happy right up until I also try and do any RX on even a single channel at the same time.  As soon as I do RX as well I see a "glitches" in the TX output and USB warnings.

Check your setup e.g. follow the instructions at https://v6-forum.aaronia.de/forum/topic/file-source-block/#postid-1282

Any other problem please create a NEW thread under "bugs" so we can look into it.

Even better in the future would be "On Spectran" modulators for TX so that binary data could be sent over USB or stored in "On Spectran" RAM rather than IQ data, and the bits to IQ conversion happened in FW on device.

It would be really amazing if blocks like IQ modulator, IQ demodulator, and IQ signal generator (or any other ones that would significantly ease the PC processor load and/or USB BW if run "On Spectran") got automatically pulled into the SpectranV6 FW rather than running on the PC if they are hooked up to the SpectranV6 IQ ports in the block diagram.  I know this is probably a ton of development, but the potential is there I think.

This might be implemented in the new SPECTRAN V6 "PRO" generation.

Maybe another thing to think about would be to use Thunderbolt rather than USB for future devices, as I understand it that allows for DMAs and other features that USB does not support that would unburden the PC and improve data transfer/buffering performance.

Sure as soon as chipsets are available we will go for USB 4 but the last 4 years showed no progress in this. But we see light on the horizon for 2023...