Different FujiNet performance on different machines

[mozzwald]

6 hours ago, SS said:

[12:35:39]CF: 70 db 00 00 4c
[12:35:39]sioFuji::sio_process() called
[12:35:39]NAK!

This tells me you're using an older firmware and you should update. This command was added in the beginning of November and is used for the weather app. Other than that, I see this error which looks like there was a duplicate SIO transaction.

7 hours ago, SS said:

[12:35:40]CF: 71 52 1e 04 e5
[12:35:40]sioNetwork::sio_process 0x52 'R': 0x1e, 0x04
[12:35:40]sioNetwork::sio_read()
[12:35:40]ACK!
[12:35:40]Read 1054 bytes
[12:35:40]TCP read 1054 bytes
[12:35:40]->SIO write 1054 bytes
[12:35:40]COMPLETE!
[12:35:46]
[12:35:46]CF: 71 52 1e 04 e5
[12:35:46]sioNetwork::sio_process 0x52 'R': 0x1e, 0x04
[12:35:46]sioNetwork::sio_read()
[12:35:46]ACK!
[12:35:46]Read 1054 bytes
[12:35:46]TCP read 1054 bytes
[12:35:46]->SIO write 1054 bytes
[12:35:46]ERROR!

I know absolutely nothing about the Atarimax 32 in 1. Looks like a way to choose from 32 different OS's? I'm guessing it does something at boot that allows you to pick? Maybe there's an issue with that, but I don't know enough about it.

[MrFSL]

Add me to the list of folks having this issue with the Sophia rev. C. on an 800xl. 

 

I just thought the Fujinet was in the early stages of development and so very buggy --- turns out it is RF interference.

 

One symptom not mentioned here that clouded this for me is that the Fujinet acted very slowly or intermediately on power up --- not even using wireless functionality. I suspect that has to do with the Unit trying to acquire networking when it starts. (I really wish there was a "start in offline mode" setting we could set as it does have an onboard SD card and could therefore be used with no network access thought the config tool doesn't allow me to escape it.)

 

I started doing a continuous ping to the device last night. I noticed if I lightly touched the top of plastic case of the Fujinet the networking was drastically improved. I thought perhaps this was a connectivity issue --- like the small amount of pressure was making a pin contact or something along those lines. Next I noticed that just laying my arms on each side of the Atari drastically cut latency on the ping responses. So at this point I started looking for devices near the Atari that could be producing noise. When I shutoff the monitor "bingo" everything was fixed. When I unplugged the DVI and powered on the monitor, same thing, everything worked.

 

Shortly after I found this thread. (NOTE - it might be helpful to rename this with Sophia in the title.) 

 

The DVI cable I was using had a damaged end:

And this was definitely causing more of a problem though a new cable does not fix this issue. (Still packet loss, still high latency but more consistently around 200-300ms then with damaged cable.) I decided to test with other DVI devices (Placing the fujinet near other DVI connections in various states of plugged in tightly, not tightly, damaged cable, etc. --- none of them cause interference with the Fujinet but that doesn't make this not a Fujinet probelm - or rather a Sophia only problem. I can put many wireless devices near the Sophia DVI and watch the signal to noise ratio and it doesn't increase and these devices run just fine. This tells me that the Fujinet is very susceptible to noise and interference.

 

I have read about USB 3 interfering with 2.4-5 GHz frequencies but I have never heard of DVI. Perhaps @Simius could shed some light here about the Sohpia and the noise its giving off. If someone has any information on DVI producing noise in this spectrum, I can't myself find it, so I think this is more of a, "the Sophia should not be doing this" rather than "you installed it to close to the SIO port" issue.

 

On a positive note @mozzwald I now see what people enjoy about the Fujinet. The device actually works pretty well when you mitigate this problem. I had, up until this point, only ever seen intermittent success and was waiting for that firmware update that would finally fix things. I also have had some issues with Fujinet bluetooth. Do they share the same antenna I wonder? I look forward to doing more testing now that the unit can be made to work better. In the meantime it might be worth making the Fujinet community aware that this issue exists, especially @MacRorie and @Gavin1968 as they might have customers purchasing from them that are then complaining that the Fujinet is faulty.

 

I think I will try to get a hold of a RF guys I know and get his take on this as well. He might at least have better equipment to measure the noise or suggestions on shielding and antenna design. 

 

I will let you know what I find out!

 

[Mark Simonson]

I got mine working reasonably well (almost no interference or none) by plugging the FujiNet into my 1050, which is positioned to the left of my 800XL, and then plugging an SIO cable into the FujiNet on one end and the 800XL on the other. This puts the FujiNet about a foot away from the DVI connection, and closer to my WiFi router. In other words, the DVI connector is not between the WiFi and the FujiNet like it was before. I also placed a little metal shield over the DVI connection just in case, but I'm not sure this is necessary. The SIO and DVI cables wrap around the right side of the base of the tv.

 

[mozzwald]

15 minutes ago, MrFSL said:

I also have had some issues with Fujinet bluetooth. Do they share the same antenna I wonder?

Yep, there is only one antenna on the esp32.

15 minutes ago, MrFSL said:

I think I will try to get a hold of a RF guys I know and get his take on this as well. He might at least have better equipment to measure the noise or suggestions on shielding and antenna design.

RF is out of my league so any insight would be helpful. I followed the Espressif design guide as best I could. Maybe adding makeshift shielding with copper or aluminum foil inside the fujinet case tied to ground would help. I would try myself, but there's no way for me to test it without a Sophia and I don't have a need for it.

 

30 minutes ago, MrFSL said:

On a positive note @mozzwald I now see what people enjoy about the Fujinet. The device actually works pretty well when you mitigate this problem

Glad you're able to experience the power of FujiNet now  

[MrFSL]

1 minute ago, mozzwald said:

Glad you're able to experience the power of FujiNet now

Yes me too! Now please --- Airplane mode!

[+MacRorie]

Hello all,

@mozzwald Let me know that there was a Sophia/FujiNet interaction issue.  I will add this to the equation: I have an 800XL at the front door to the shop (it's my "Greeter" ;-)) that has had a Sophia in it for at least a month now with FujiNet.  It goes on in the morning, I boot weather.atr and it stays on for 10 straight hours.  It polls for changes every 90 minutes and runs like a champ. I have attached pictures of the setup.

 

The 810 sitting there is a relatively new addition and is only turned on for quick testing occasionally. The 800XL is powered via USB.

 

Now, does my one outlier situation prove that all is a) hunky and b) dory between Sophia and FujiNet? No, not in the slightest.  However, it *might* indicate that it *may* be something else. 

[Mark Simonson]

@MacRorie Where is your WiFi router in relation to this setup? In my case, it seems to happen when the DVI (Sophia) connection is situated between the FujiNet and my WiFi router. When it's not, the interference goes away (or mostly goes away).

[MrFSL]

@Mark Simonson Mine it is not in between

 

@MacRorie If you have time, can you ping your fujinet and let me know what your latency looks like? If you put the Fujinet on a SIO cable and move it closer/farther from the Sophia DVI connection do you massive changes and/or drop outs from a continuous ping? I know that the weather app is supposed to pull ever X amount of minutes - but what would be the indication if it failed to do so? Does the weather app warn? Is there a way to test the amount of time it takes to pull that weather information?

 

I for one hate intermittent failures - they are so hard to diagnose. What I can say so far is that moving the fujinet towards or away from the DVI port is measurable, as is turning off the monitor or disconnecting the DVI port.

 

Also, what wireless channel are you running on?

[+MacRorie]

1 hour ago, Mark Simonson said:

@MacRorie Where is your WiFi router in relation to this setup? In my case, it seems to happen when the DVI (Sophia) connection is situated between the FujiNet and my WiFi router. When it's not, the interference goes away (or mostly goes away).

It is in the room that is adjacent to the entry hallway in the picture (on the far side of the building).

[+MacRorie]

1 hour ago, MrFSL said:

@Mark Simonson Mine it is not in between

 

@MacRorie If you have time, can you ping your fujinet and let me know what your latency looks like? If you put the Fujinet on a SIO cable and move it closer/farther from the Sophia DVI connection do you massive changes and/or drop outs from a continuous ping? I know that the weather app is supposed to pull ever X amount of minutes - but what would be the indication if it failed to do so? Does the weather app warn? Is there a way to test the amount of time it takes to pull that weather information?

If the weather app cannot pull data, it hangs.  I have seen it do that with poor connections in non-Sophia machines.  Time to pull: Dunno, didn't write it so no clue.

Quote

 

I for one hate intermittent failures - they are so hard to diagnose. What I can say so far is that moving the fujinet towards or away from the DVI port is measurable, as is turning off the monitor or disconnecting the DVI port.

 

Also, what wireless channel are you running on?

Huh.  I cannot say I have seen that behavior.  My impression was that the response was always the same no matter what machine I put it on and where I put the machine.  Now, I will tell you that during the middle of the day when everybody and their sister has their WiFi up, it is decidedly more difficult to connect, but I attributed that to a general WiFi issue and not to any one machine configuration.

 

Channel is set to Auto select on all WiFi clients that connect.

 

EDIT:  Did a ping of FujiNet with it attached directly and when on an SIO cable.  Did not notice an appreciable difference.

Edited December 20, 2020 by MacRorie
ADDED Ping results

[Atari8guy]

As a guy who’s expecting one shortly has anyone had this issue with the Sophia 2?


Sent from my iPhone using Tapatalk

[+MacRorie]

7 minutes ago, 8bitguy1 said:

As a guy who’s expecting one shortly has anyone had this issue with the Sophia 2?


Sent from my iPhone using Tapatalk

Like I said, the 800XL in the picture has a Sophia 2 in it now (it had a rev C before) and have never seen the issue.

[MrFSL]

15 minutes ago, MacRorie said:

Like I said, the 800XL in the picture has a Sophia 2 in it now (it had a rev C before) and have never seen the issue.

Without trying to measure something like the latency of ping responses you might have interference changing the signal to noise ratio but not enough for you to notice for the application you are using. Like I said, I can change the latency consistently by where I place my hands/arms on the desk near the Atari for instance.

 

Anyways - I am not trying to point any fingers here anyhow - like I said I have a friend who is a EE working in RF who said he would have a look so I will have more information hopefully soon. The only thing he has said to me so far is --- "think of it like a water leak, for some reason the area of that DVI connector is spraying a bunch of noise and I don't yet know why." 

 

And in the meantime if someone complains about the fujinet and has a Sophia - have them run it off a SIO cable and move it away and see if it helps them would be my suggestion.

[Jochen]

On 12/19/2020 at 8:26 PM, MrFSL said:

continuous ping

sorry

Edited December 22, 2020 by Jochen
wrong thread

[MrFSL]

Well the final word I have heard back is --- it seems that the Sophia might give off some noise but it "shouldn't" interrupt the wifi signal. The problem seems to be therefore with my fujinet more than the Sohpia. The device might possibly have some antenna problems making it extremely susceptible to interference. I did some different testing and posted information in another thread starting here:

 

That is the best that my testing, and my friend's work, yielded. I was hoping for something much more amazing with graphs and spectrometers and oscilloscopes --- but alas not everyone is interested in spending their spare time playing with Atari.

 

@MacRorie I saw this on your site and really appreciate you making that disclaimer:

 

That is super cool of you to take proactive steps. I think the problem is actually with my Fujinet. The Sophia "noise" does not interfere with other WIFI or Bluetooth devices and isn't detectable using multiple WIFI adapters. Whether the interference is in spectrum, or a harmonic thereof, the Fujinet that I own should handle it better and as you point out, yours has no issues. I only suggest that anyone building these look at the network latency as part of your normal QA processes. 

 

With a tremendous amount of babysitting and the tips described here (and not too little the application of aluminum foil LOL) I can get the Fujinet working "better" than it was and that opens a bunch of avenues for me to play and have fun.

 

Thanks everyone.

 

 

 

[+MacRorie]

4 minutes ago, MrFSL said:

 

@MacRorie I saw this on your site and really appreciate you making that disclaimer:

 

That is super cool of you to take proactive steps. I think the problem is actually with my Fujinet. The Sophia "noise" does not interfere with other WIFI or Bluetooth devices and isn't detectable using multiple WIFI adapters. Whether the interference is in spectrum, or a harmonic thereof, the Fujinet that I own should handle it better and as you point out, yours has no issues. I only suggest that anyone building these look at the network latency as part of your normal QA processes. 

 

With a tremendous amount of babysitting and the tips described here (and not too little the application of aluminum foil LOL) I can get the Fujinet working "better" than it was and that opens a bunch of avenues for me to play and have fun.

 

Thanks everyone.

No worries.  Am glad you got it working for you and we shall all keep a closer look at it.  This is what means to be on the bleeding edge of technology (in a retro 80s sort of way!)

[MrFSL]

Just now, MacRorie said:

This is what means to be on the bleeding edge of technology (in a retro 80s sort of way!)

LOL! Absolutely!

 

I would add - This is what it means to be part of an enjoyable online community of collaborative enthusiasts. 

 

Merry Christmas Ya'll.

[_The Doctor__]

Quality cabling and putting the shielding back on an 800xl helped fix this issue for a fellow in town... Other suggestions that seem to help were ferrite and toroid ...  just like back in the day... the oldies are still delivering.

[MrFSL]

3 hours ago, _The Doctor__ said:

the oldies are still delivering.

This actually works remarkably well:

[+x=usr(1536)]

On 11/29/2020 at 4:38 PM, SS said:

This pretty accurately describes the behavior that I see with my SOPHIA machine too.  Sometimes it just times out and I cannot get a full load at all.  I got the cable idea when I was trying the FujiNet with a 600XL with a 1064 attached because everything simply would not fit next to each other.  The extra distance does seem to help sometimes though.  

Just as an addendum to this for anyone searching: being in the same 600XL + 1064 boat as @SS on this one, my workaround has been to daisy-chain the #FujiNet to the 600XL via a 1010.  The 1010 doesn't need to be powered; +5V will pass through from the 600XL just fine.  Also, the #FujiNet's LEDs are still visible over the 1010's case and the buttons remain accessible.

 

One caveat in all of this: I'm not using original Atari power supplies.  All machines are fed via 8bitclassics' / The Brewing Academy's micro-USB power cables, which means that they're all being powered by 2.5A RasPi PSUs.  I also haven't tested with the 1010 powered to see if that makes a difference or not.

 

Other SIO devices would likely work for this as well, but it's probably not a good idea to attach the #Fujinet to anything not immediately connected back to the computer.  I'm reading 4.694V at its SIO port when connected to the 600XL via the 1010, which compares fairly well with the 4.741V I see when directly-connected to the 800XL.  Things are still stable at this level, but if it's at the end of a 1010-1020-1050-1050-850 chain, it might be a different story

Edited April 11, 2021 by x=usr(1536)

[Mark Simonson]

Works fine plugged into a 1050 connected to the Atari.