Open Research Institute Weekly Reports, 2 Aug 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: ORI FPGA Meetup 2 August 2022

Description

encoder read/write problem *solved*

A

All right, hello: everybody welcome to the stand-up meeting for fpga for open research institute for the 2nd of august 2022.

A

So what we do here is to talk about what we've done over the past week, what we're going to be doing over the next week if we need any resources, if we have any road blocks, um so let's get started. uh Ancho go ahead.

B

Between finally, here the read is start, it read is sorted so what was happening uh when we are trying to read from uh register zero or the memory address of encoder, so the instruction was coming properly till interconnect and then from interconnect down to the encoder.

B

uh The address that we were using suppose it was four four: a b eight triple zero. It was taking the lower 12 bits as an offset, rather than taking this complete as an address and offset 0.. So it was trying to read some non-existing location and that was giving us a buzz error. So once I moved things around now, the encoder is at four four: a b: zero, zero zero zero.

B

So if, if I pass on the address, if I pass on four four a b zero zero zero it takes offset as lower four bit for lower 16 bits, which are all zeros and it reads the resistor properly.

B

So that's sorted now: jesd has moved to another location, four, four, a b, eight triple zero, so that's the jest location uh and now all the peak pokes to all the addresses and the registers for all the components connected to xe, lite and other buses is working properly. Now the next step is dma.

B

uh uh I remember you had one sample dma application. I will try to work on that so that we get one packet across to the radio out through the radio.

A

Yeah I had a some source code that was uh called dma test and that's just a sort, just a start. So all it had was the register reads and writes yes, um so I started to look at like what do you need to to read and write, or what do you need to write to what registers to do dma, but it's not completed yet.

B

A

I think it looks like that's the next step, so I'm very excited. This is good news. Thank you so much. I know it was a lot of hard work.

B

Yeah, it was basically to get that ila I like to get working and to understand and uh to learn how it works so yeah once that was done, it was pretty good and pretty good yeah. So.

A

Good um is there, is there anything, that's stopping the the way that we set up the repo from from working like? Is there anything that needs to change in the repository.

B

um I think so now the uh yeah the address map the address has changed. So I think I need to update the tackle script. Okay uh or I don't I'm not sure where the addresses are saved uh so whatever it is, I will see the diff and then yeah.

A

They're in the tackle script, that's under adrv 9371 and then I.

B

A

Common, so so that has the address in it, and that seems to.

B

Be where it gets.

C

A

Okay, so yeah, so it's a new address anything else in that script,.

B

uh uh Just a new address and yeah, I added some debug course uh which will be helpful in debugging, so I would let them uh remain there. So there will be some scripts or something related to that script.

A

So is that extra.

B

A

Right, like you, so you call.

B

A

And then okay yeah, because I've seen that done elsewhere, okay, cool that sounds like we can. We can do that.

B

A

You that's a that's a lot of good news.

B

Yeah, that's good, and now uh the next thing is, how can I get uh do you have any idea? How can I get sample to dma out to the radio board via jest link and then out? Where can how can I get that sample.

A

Yeah, that's a good question. Anybody got a comment.

A

Because I think what you're talking about is like a baseband baseband frame- yes, yes, okay! We we believe that the that we have a number of files um that we've been using for the beacon demos and that we think that those might work and if they don't work for any reason, uh because they should just be a file full of baseband frames. um Then we can go back to several uh people in the in the community and they should be able to help us.

A

So the transport stream files like from ron economos, they should be baseband frames and that's how we've been using them and they've worked pretty well for the the beacon work.

B

Right and even er error is tested successfully with pluto, so he should have some file.

A

B

Also tested with bluetooth, so he should also have something.

A

Yes, yeah that should work yep and if it doesn't we'll figure it out, yep.

B

Okay, thank you.

A

You bet cool, okay, uh hey paul. Do you have any comments.

C

Well, I have a question for actual um now that we know how to add ila instances and use them.

C

Is there some particular tutorial that we need to follow, or do we need to write down a few pages of of notes on how to do that, so that other people can follow along.

B

uh I can write it down, uh but the the guidance that I followed, especially basic, basically the docs from amd. They explain it steps uh step by step. So that's the only official documentation, that's available.

A

Can you link it in.

B

Yeah, of course,.

A

Yeah that I think that should be sufficient as long as it worked for you to make this happen, then linking the the correct documentation, I've tried to look up ila before.

B

A

It'd be nice to know if I'm looking at the right stuff, okay,.

C

Yeah because it's going to be a valuable tool throughout. Oh yes,.

B

Yes, it's it's really useful.

A

Yeah, I think the the scheduler, you know what what organizes the the baseband frames. I think that's going to be in firmware or more on the software side, um and that's that's something. That's that's coming right up uh as soon.

C

A

We can can manipulate this over the air. The next, I think, hardware part, is the the uplink receiver so and- and that's definitely going to need some ila, because we already know that the code from theseuscourse has some some to do's.

A

They, their code is great and we've demonstrated it before by itself using usrp and and rfnoc in gnu radio, but we do know that the code base has some limitations and they want those limitations to be resolved, and so ila is going to be pretty key for that too, so anything that we can use any tool that we can have to to make that process a little easier, it'll be good.

A

So thank you. This is a big deal all right all right. Any other reports.

C

I have a null report from the orient lab west: nothing, nothing new. There.

A

I don't know, that's uh that's good news, so everything's working, okay, I haven't had any major upsets, or I know that we had a reboot, but that was uh that was pretty ordinary. It went well.

C

Yeah, that was just more or less a scheduled reboot without the scheduled part.

A

C

Far, we've only been using much using it a whole lot recently, so nothing new has been broken or discovered, broken.

A

Well, it did I've been using it quite a bit trying to get the adi build to make. um So so I'll. Give I'll give my report, uh so I've been using remote labs, but only in the sense of the vm on the on chunk and what I've been trying to do is to to make our encoder and the um the hardware reference design from adi to pull both of those in as sub modules, um and so that we can pick the exact right commit level from these two different repositories.

A

So using them as sub-modules is pretty pretty neat, and I set that up. But the make has not worked ever since, and so I've reached out to engineer zone from adi to ask for help and I'm pretty sure that this could be solved. um It seems to be hung up on a particular clock uh signal called or a signal called ref underscore clk, so reference clock, we don't use it in our code that we're bringing into the reference design and when you make the reference design by itself, the reference design makes fine.

A

So I'm not really sure what to do here, but I reached out and asked for help and if we can solve that problem, then I think that the with the changes to the address map, then we should have a repository where all you need to do in order to use our code is to type make at the command line. So that would be that'd be nice to see happen, so I've been using the remote lab, but not the nothing over the air. Yet that's going to change, I think very very soon.

A

Another thing that I've been doing is is looking at some matlab code that demodulates fsk. So it's a matlab model that uses dfts to to demodulate fsk, and this will be on the uplink receiver. And what I'd like to do is take the matlab code and prepare it for hdl coder in matlab that will produce the hdl code and then we can try it out on the zc706.

A

So that's kind of the goal there. It's moving forward gotten a good good start there um that combines up with the work from theseus course that I I talked about a few minutes ago. um So that's coming. The the really big win is that the encoder is behaving a little better in the downlink section.

A

I don't have any roadblocks except the usual. You know. I think we we all are dealing with a pretty steep learning curve and this is ambitious work. um I think stepping back just a little bit that we have an excellent opportunity to put this into a proposal for a high elliptical orbit project that we're going to present to jamsat as soon as we're done and anything working or any prototypes that we can show make that proposal better. So that's some good news from from the wider community.

A

All right, hey ken! Do you have any comments? I know we didn't deliver the simulation that you asked for yet, but we did use the integrated logic analyzer. It sounds like to solve the problem with reading and writing to the custom ip, uh but anyway any uh comments or questions from you.

C

Oh just listening in.

A

Okay, thank you so much all right and then, let's hear from.

A

James, I hear low audio or no.

A

Audio, okay, still nothing! I don't hear any audio from james, but we'll catch up later, because we have to talk about remote, lab south and grant application there for uh to the fda.

A

Oops, I made it sorry any last comments or questions uh before we close all right.

C

Thank you. Everybody.

A

Oh go ahead. Sorry.

C

um Yeah I've been working on this uplink uh essentially uh simulation, but the c plus was implementation of it and starting to get an excuse me, a notion of all the tracking loops and things that need to operate, and this is started a little background process in my brain, trying to figure out how this is going to go into hardware for a massively multi-channel receiver.

C

I would like to have a confab with you, ken and you michelle uh here on the whiteboard, maybe and uh see. If I understand where that's gonna gonna have to go to make sense. Architecturally.

A

Okay count us in.

C

I see some chat messages.

A

Yeah james is having trouble with audio, but we'll we'll figure it out later. I might have to. We can shut this one down and start up a new one for him to to talk about the fda grant and any other reports from remote lab south anything else. You want to share about the uplink progress.

A

There's been plenty.

C

Well, it's not directly on topic for a pga, but might as well share some good news. Well,.

A

It's got to be received by.

C

The fpga so we'll have to be re-implemented by the fpga. Yes,.

A

C

So I'm working on this opulent voice, uplink and down uplink receiver and transmitter, and it's been a conversion process, changing the m17 code into opulent voice code and of course that means breaking it and then fixing it and I'm in the fixing it stage now, mostly um as you, if you I've, been posting progress reports on the opulent voice channel on slack. But I've got to the point where I can actually go over the air over an rf, coaxial cable.

C

um It could be antennas, but cable is a little more reliable and that brings up a lot of new issues because all of a sudden you're, not you, know the same clock you're, not in synchronization for free and so of course there are problems, but two different computers, one I'm transmitting the other one receiving, and this is mostly working now.

C

I can hear a good quality voice with some glitches which I'm trying to blame on the on the usb audio output side. I'm going to run a test here shortly on that and see, if that's a plausible theory, but this means a lot of mechanism is working, we're acquiring symbol, timing, we're acquiring frame timing successfully with a few missed sync words which need to be investigated further.

C

Although there's enough state machine mechanism to paper over that missed sync words, so they don't cause any problem at this level, but they shouldn't be happening. So we have to find out why and then the integrating the the vocoder turned out to be pretty easy, just find the right apis to call and call them correctly and perfect or near perfect voice comes out. It's just remarkable.

C

If I play back the original file and the file that's gone through the whole system side, but not not simultaneously, but one after the other. It's really hard to tell the difference. If I play them in both ears of my headphones, I can hear which one is has been further processed and it's worse, but not very much worse. It's really good audio quality.

C

If we can make this work over the air in a real system and with uh enough robustness to to be tolerable, then it's going to be a giant step forward in in digital voice for ham, radio.

A

Good, thank you, I'm very much looking forward to it, because voice is the product for these sorts of systems. You know, even in a very data-centric world, you know you've got to treat voices the product and make it sound as good as it possibly can. So. Thank you really looking forward to demonstrations coming up uh very soon, uh defcon and and also at kiso, today, ham expo in september uh throughout the fall. Actually, that's that's our.

A

We have some of our best opportunities to to do demos, um and I think once people hear it and get to play with it. If we can pull it, pull this off on a hack, rf plus support a pack so that people can install it themselves and and play around with it locally. Then it'll get some some good traction so very excited about that. Anybody.

C

Wants to hear the audio quality for themselves right now, there's samples posted on the on the occult voice slack channel before and after uh listen to them and see. If you don't agree that it sounds great.

A

Yeah, maybe we should make a web page about it and and put that up. I think it might be a good idea at this point.

C

One of these days, yeah.

A

Well, I can add it to the to the current announcement for opulent voice on the on our website: cool. Okay, thank you. So much all right. We are going to close down the fpga stand up and we will move towards the bacteriophage and fda work. So thank you, everybody and see you at the white board.