New 100D Angelia Protocol 2 Firmware Topic

[rainbow]

Thank you very much N1GP.

[DL8LAQ]

Dear friends,

I have got a 100D on my bench for checkup and testing firmware versions. I use a spectrumanalyzer at tx out (J3) of the Angelia board to check the spectrum from 10 to 144 MHz (the CMU200 starts at 10 MHz).

With v12.1.6 the spectrum is cleaner than that from v12.0. The pictures show a 28 MHz tune signal after some minutes of warm up.

Now comes the interesting part: With v12.1.6. and cold spray on DAC U5 I can get the noise down to be almost as good as with P1 V6.0!
Since there is no feedback from DAC to FPGA I would assume that this noise is not caused by timing closure.

I would like to get some feedback from the audience before I go further and replace the DAC (AD9744ARUZ, 19€ at Mouser). Does anyone have experienced DAC failure modes similar to this one?

I checked the 3.3 VDC at F3, there is only a 30 mV drop from 3.105 V with P1 to 3.086 V with P2, both measured at tx.

[w9mdb]

Not clear to me that you're lowering noise or just the noise floor.
Are you using a signal generator for your reference level? That level went down 1dB

[DL8LAQ]

Not clear to me that you're lowering noise or just the noise floor.
Are you using a signal generator for your reference level? That level went down 1dB

The noise floor of the spectrumanalyzer is just below -40 dBm without signal. The tune signal at 28.050 MHz is around +23dBm. The noise / spur at 10 MHz did show up to 10 dB difference, but it could be that I checked that at different times (temperatures) after flashing another firmware.

I should have provide you with pictures before / after cold spray. This is how it looks with v12.1.0_pre2_5550:

[w-u-2-o]

Now comes the interesting part: With v12.1.6. and cold spray on DAC U5 I can get the noise down to be almost as good as with P1 V6.0! Since there is no feedback from DAC to FPGA I would assume that this noise is not caused by timing closure.

The situation is exactly the opposite from that. This is clearly a timing problem in the P2 firmware. Indeed, it is a classic example of the sort of results you get with poor clock alignment with one, many or all data bits. This can also occur with the ADC, as many of us have seen when the panadapter starts to display a very high noise floor. It's not necessarily a timing closure problem, but it is almost certainly a clock to data timing alignment problem.

In addition, you have proven that there is nothing wrong with the DAC because you say it works fine with P1. Since the clock and data rates to the DAC in both P1 and P2 are identical one should expect the same DAC performance with both versions of firmware.

Finally, because freeze spray on the DAC fixed the problem, clearly there is nothing wrong with the data coming over the GigE link. That's actually a very good thing because moving the MAC implementation from 100Mb/s to 1Gb/s has proven to be the hardest part of P2. And it's some really excellent data for Rick, @n1gp

[DL8LAQ]

In addition, you have proven that there is nothing wrong with the DAC because you say it works fine with P1. Since the clock and data rates to the DAC in both P1 and P2 are identical one should expect the same DAC performance with both versions of firmware.

Finally, because freeze spray on the DAC fixed the problem, clearly there is nothing wrong with the data coming over the GigE link. That's actually a very good thing because moving the MAC implementation from 100Mb/s to 1Gb/s has proven to be the hardest part of P2. And it's some really excellent data for Rick, @n1gp

Ok, Scott, thank you for clearing things up. I do not really understand what's going on here. I am just an analog and rf guy.

It is a tiny bit of cold spray what makes the signal so much better and I was sure that I found the bad device on this Angelia. Still hard to believe, but like you said and I have seen that too, the DAC works perfect with P1.

Let's see what Rick can do with this information.

If there is anything I can check or measure on my bench please let me know. A feedback from Rick would be nice.

[IK2LRN]

Hi,
the last version Metis_Angelia_Protocol_2_v12.1.6.rbf does not work with my Angelia : the second receiver interfere with the first.
The previus version seams works fine.
73
Luca IK2LRN

[DL8LAQ]

Hi Scott,

I just did some search & read on the net and I believe to understand now. The internal delays are changing with temperature and now clock and data from the FPGA get a better match within the DAC.

Datasheet AD9744:

DAC TIMING
Input Clock and Data Timing Relationship
Dynamic performance in a DAC is dependent on the relationship
between the position of the clock edges and the time at which
the input data changes. The AD9744 is rising edge triggered,
and so exhibits dynamic performance sensitivity when the data
transition is close to this edge. In general, the goal when applying
the AD9744 is to make the data transition close to the falling
clock edge. This becomes more important as the sample rate
increases. Figure 30 shows the relationship of SFDR to clock
placement with different sample rates. Note that at the lower
sample rates, more tolerance is allowed in clock placement,
while at higher rates, more care must be taken.

[w-u-2-o]

Exactly, Norbert. Table 3 in the data sheet (and refer to figure 2 as well) defines data setup to clock of 2ns minimum, and hold after clock of 1.5ns minimum. Given that the DAC is clocked at 122.8MHz, which is a clock period of 8.14ns, meeting the setup and hold time requirements does require some careful attention, as the data needs to be valid and stable for 3.5ns minimum, and properly aligned as well, but still also needs some time to transition to the next data word.

[n1gp]

Hi Norbert,

I don't have a very good spectrum analyzer, I use an SDRPlay RSP with a spectrum analyzer app.
Seems OK.

I cannot reproduce your images with it though on my Angelia board. Both FW version look pretty much
The same.
However I did do some analysis of the v12.1_pre4 source code and the v12.1.6 and found some peculiar differences when it came to the DACD data and it's latching clock.

I made some changes that should agree with setup and hold times better. Could you please
try:

https://www.dropbox.com/s/cl54cp3wb4c9non/Angelia_Protocol_2_v12.1.8.rbf?dl=0

Tnx

Rick / N1GP

[DL8LAQ]

I made some changes that should agree with setup and hold times better. Could you please try:

https://www.dropbox.com/s/cl54cp3wb4c9non/Angelia_Protocol_2_v12.1.8.rbf?dl=0

Tnx Rick! I will try that "tomorrow"! It's just before real midnight and I am just back from late shift. Bed time now

EDIT: Morning Rick! The v12.1.8 seems to work perfect regarding the spectrum purity! Transmitting for more than 20 minutes and the signal is perfectly clean from the beginning (flashing). The temperature at the FPGA heatsink is 60°C (140°F) and DAC is 40°C (104°F).

Thank you Rick, well done! We owe you a beer!

The power reading of 25 dBm is without any calibration, but added attenuator values into the CMU200 settings. I just use cables out of the junk box and well used attenuators!

[n1gp]

That's great news Norbert. Tnx for testing that out.

I notice in your signature you have a 7000 as well. Have you, or could you
also test your 7000 to check how clean the TX spectrum is with the latest
firmware?

viewtopic.php?f=32&t=3192

[DL8LAQ]

That's great news Norbert. Tnx for testing that out.

I notice in your signature you have a 7000 as well. Have you, or could you
also test your 7000 to check how clean the TX spectrum is with the latest
firmware?

viewtopic.php?f=32&t=3192

Hi Rick,
I use 2.1.18 daily since release and have never seen any problem. The spectrumanalyzer shows the same clean spectrum like the Angelia with the latest release.
This measurement was not done with the maximum power available from Orion MKII!

[n1gp]

Tnx agn Norbert!

73's

-Rick / N1GP

[rainbow]

Excuse me: can you release the qar (sources) of

- v12.1.8
- v12.1.6
- v12.1_pre4

thank you very much
/davide

[n1gp]

I had shared - v12.1_pre4 with you back in Apr 2021. For some reason it got deleted off
of Dropbox. I'll try and track it down. Here's the others:

https://www.dropbox.com/s/kbx73c6xby8s9v9/Angelia_Protocol_2_v12.1.6.rbf?dl=0
https://www.dropbox.com/s/cl54cp3wb4c9non/Angelia_Protocol_2_v12.1.8.rbf?dl=0

73s

-Rick / N1GP

[rainbow]

In fact, I was sure to have a copy but I can't find it. Anyway, don't worry too much since was just to have it has chronological backup.
The two dropbox files are rbf not a qar archive.

thank you

[n1gp]

Ooops

QARS:

https://www.dropbox.com/s/qewoiiq7mbynvel/Angelia_Protocol_2_v12.1.6.qar?dl=0
https://www.dropbox.com/s/k19k20ula4fol22/Angelia_Protocol_2_v12.1.8.qar?dl=0

If you come up with any fixes you think would be relevant to Angelia please let me know.

Tnx

[rainbow]

for sure. thank you very much!

[Segrest]

The Angelia_Protocol_2_v12.1.8.rbf and Angelia_Protocol_2_v12.1.8.rbf files don't appear to be available in the dropbox any more. Are they available somewhere else?

Bob, KO2F

[w-u-2-o]

The Angelia_Protocol_2_v12.1.8.rbf and Angelia_Protocol_2_v12.1.8.rbf files don't appear to be available in the dropbox any more. Are they available somewhere else?

Bob, KO2F

Look at the first post of this topic.

[John K2OX]

I finally bit the bullet and updated Thetis to work with my custom hardware. My prior s'ware was ms ps 3.3.7 from 2017.

Thetis with P1 firmware has allowed me, for the very first time since buying Angelia in 2016, to run Pure Sig! I tried running the latest release of P2, but no luck. Then I found this site and downloaded all the test P2 versions.

After trying eight of the ten versions listed, I hit paydirt. Firmware v12.1_pre2_5500 works very well compared to all others. I see a 'seq' flash about once a minute at 768k. PS works. CPU about 4% on rx with an old i7-3770 CPU @ 3.40GHz.

I also changed the reg key for throttling. Not sure that it has made any difference.

A big thanks to Rick / N1GP. Without your work I'd only have half the radio.

73,
John K2ox

[w-u-2-o]

Well done, John!

[rainbow]

Just as reference I recovered the sources of 12.1pre4
https://www.icloud.com/iclouddrive/03aD ... ocol_2_v12

[PD3LK]

I heared from at leasts one OM that he uses the latest 100D fw in his 7000 DLE MKII ...
(Metis_Angelia_Protocol_2_v12.1.8.rbf)

Are there more people who did this and does the radio work 100% well?
If so, what are the benefits over fw Metis_Orion_MkII_Protocol_2_v2.1.18.rbf ?

73, Leon

[w-u-2-o]

I can't see how that would work at all. The two FPGAs are completely different and there are many other hardware differences. This sounds like a fairy tale. Use the correct firmware designed for your hardware.

[PD3LK]

Thanks that was my thought to, hence the question here. Yesterday i discovered there was a misunderstanding between me and a other OM. We where e-mailing about the 7000 fw but he mingled the conversation with info about his 100D. My brains couldn't make that jump so fast.

But i'm still puzzled... the fw version in his 7000 setup says 2.1.18 - Protocol 2 (v3.9), which fw version is that? And where to find it, he doesn't know that anymore, the 7k was possibly delivered with that fw.

Thanks for any help

btw my fw is V2.0.0 Protocol 2 (v3.8), i can't find any info or link on this forum about his version.

[oe3ide]

First post in the 7000/8000 firmware thread: viewtopic.php?f=32&t=3192#p8917

73 Ernst

[PD3LK]

Thank very much Ernst! So stupid that i coulnd't find it, but then again i'm not the only one having problems to find fw or Thetis releases. It's a jungle out there.... especialy for new comers or 65+ OM's.

I know that this forum has it's limitations but imho there must be a better way to keep software releases (read only maps) and discussions about them, better separated.

[w-u-2-o]

If I had access to the TAPR Github repository I'd happily put them there. Alas I do not.