tschak909 Posted December 19, 2019 Author Share Posted December 19, 2019 The ESP32 can do TLS quite comfortably, so at the very least, we will be able to make TCP (TLS) and UDP (DTLS) connections at some point. We will also be able to do SSH. Adding this to TNFS will need some coordination with those people on the ZX Spectrum side, as well, but it is doable. 2 1 Quote Link to comment Share on other sites More sharing options...
AtariGeezer Posted December 19, 2019 Share Posted December 19, 2019 TLS 1.2 is currently supported in the SSL libs that also works on the NodeMCU boards. Soon after I finish the Full TNFS Client (should just be on the Atari side with the N: Handler and demo program), my next objective is to make a TNFS Server that could be ran on the FujiiNet Board with the built-in SD Card. So it shouldn't be much of a problem adding a SSL wrapper on that The Great and Powerful mOZz has graciously sent one of his boards my way too 5 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 19, 2019 Author Share Posted December 19, 2019 Most excellent. Yeah, it's all doable, and it looks like interests are converging. This is good. We'll get a lot more done even quicker now. I am feverishly trying to get the next pass of Diskulator (Multilator) done, hopefully either tonight or tomorrow I will call it ready. (pushed regardless) I'm hoping with the commands that I've put in so far, we can minimize any collisions. -Thom Quote Link to comment Share on other sites More sharing options...
AtariGeezer Posted December 19, 2019 Share Posted December 19, 2019 There's always room for change on all sides Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 19, 2019 Author Share Posted December 19, 2019 Indeed. Something amazing is happening here. One thing I want to ask from everyone. If you see my code, and see something wrong with it, fix it. send a patch. Point it out. give reason why. It'll allow us to get this all done even faster. If I may be bold: What is happening here is the single largest jump in capability for an Atari 8-bit system in the last two decades, possibly ever. The amount of interest being generated is nothing short of staggering. We have people who are going out of their way to build their own boards, and people asking to come in and help. I am slack-jawed. In my three decades of contributing to open projects, I have never been a part of such an intense and consistent build-up of interest, and it is pushing me to try my best to keep this ball rolling. -Thom 8 2 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 With Test #24, it is possible to mount multiple disks across multiple virtual drives, across multiple hosts, local or remote. This demo shows me doing a sector copy of Jumpman from one internet TNFS site, to another, transparently. 7 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 am doing a multilator test: d1: sector copier on irata.online d2: jumpman on fujinet.online d3: dest.atr on a google cloud vm, writing D2 to D3 and it's going like a champ. -Thom 3 1 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 Connected the #FujiNet Rev3 Mars Bar to my Ben Heck Rev2 XE Laptop: 3 Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted December 21, 2019 Share Posted December 21, 2019 Just curious, since I see a 1200XL sitting there: any issues with using the FujiNet on one of those? I realise that SIO power is probably out of the question without modifying the XL, but does everything work OK with the OS? Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 The fujinet uses the 5v ready line. Not the 12v line. You should be ok. There are no issues with the 1200XL OS, either. 1 1 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 Here is the #FujiNet Mars Bar (rev3) connected to my daughter's XEGS. No external power needed, it just works. https://imgur.com/gallery/RIytOqe 2 Quote Link to comment Share on other sites More sharing options...
+mytek Posted December 21, 2019 Share Posted December 21, 2019 3 minutes ago, tschak909 said: Here is the #FujiNet Mars Bar (rev3) connected to my daughter's XEGS. No external power needed, it just works. https://imgur.com/gallery/RIytOqe Looking good ? . So has your daughter started charging you rent for the use of her XEGS? Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 21, 2019 Author Share Posted December 21, 2019 haha. She and I have an understanding. I use it in the morning for my aerobics (I do the Atari Personal Fitness. I am not kidding.), and she and I use it in the evening for games and her diary. -Thom 3 Quote Link to comment Share on other sites More sharing options...
+Stephen Posted December 21, 2019 Share Posted December 21, 2019 2 hours ago, tschak909 said: haha. She and I have an understanding. I use it in the morning for my aerobics (I do the Atari Personal Fitness. I am not kidding.), and she and I use it in the evening for games and her diary. -Thom That's beautiful and heart warming. I still remember the time my mom spent with me on my Atari 400 36+ years ago. I am sure your daughter will be the same. It might even shape her career later in life, as it did with me. 2 Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 22, 2019 Author Share Posted December 22, 2019 Deep hack mode: Working on the very next #FujiNet test, can anyone guess what it is? Quote Link to comment Share on other sites More sharing options...
Kyle22 Posted December 22, 2019 Share Posted December 22, 2019 I wish I could see the pictures w/ more resolution. When I try to blow them up, I get fuzziness. Looks good, though. [no Rodney]. :) Quote Link to comment Share on other sites More sharing options...
+x=usr(1536) Posted December 22, 2019 Share Posted December 22, 2019 Well, I just officially abandoned this idea for STs in favour of #FujiNet. IMHO, FujiNet has the potential to offer a much more seamless experience to data access than the idea I'd proposed, and I don't see the sense in going down a different track when there's no clear benefit to doing so. So... Work is continuing on the 8-bit side, and it appears to be maturing *very* nicely. Has any consideration been given to the 68K side of things, or is it early for that? As I see it, three things (in broad strokes) need to happen: Hardware: wifi adapter. TNFS implementation for 68K. Interfacing TNFS with TOS' disk handling. Are there plans underway, or is that a bit further off? Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 22, 2019 Author Share Posted December 22, 2019 I have no cycles to do an ST port of any of this. the tnfsd could be ported using mintlib. As for a client, you'd need to interface with GEMDOS in the same way that certain emulators do so. -Thom 1 Quote Link to comment Share on other sites More sharing options...
billkendrick Posted December 22, 2019 Share Posted December 22, 2019 23 hours ago, tschak909 said: am doing a multilator test: d1: sector copier on irata.online d2: jumpman on fujinet.online d3: dest.atr on a google cloud vm, writing D2 to D3 and it's going like a champ. -Thom Thomas, you beautiful maniac 1 Quote Link to comment Share on other sites More sharing options...
ivop Posted December 22, 2019 Share Posted December 22, 2019 (edited) @mozzwald and @Mr Robot I finally have a new Atari setup near my main PC. Well, this is not near my PC, but while testing on the couch Looking through my parts bin and cupboard, I found this: So that's good. But I was unable to find a proper Schottky diode. While considering using a normal pn-diode, I looked up the NCP1117 datasheet and it seems it needs between 4.75V and 10V for a stable 3.3V. The DSS12 Schottky diode has, depending on manafacturer, a forward voltage drop of 0.45V-0.55V, so even with a steady 5V, Vin could be as low as 4.55V-4.45V. Apparently it still works, but isn't this way out of spec? And can we rely on it? Bass, how low can we go? Edited December 22, 2019 by ivop Quote Link to comment Share on other sites More sharing options...
mozzwald Posted December 22, 2019 Share Posted December 22, 2019 3 hours ago, ivop said: So that's good. But I was unable to find a proper Schottky diode. While considering using a normal pn-diode, I looked up the NCP1117 datasheet and it seems it needs between 4.75V and 10V for a stable 3.3V. The DSS12 Schottky diode has, depending on manafacturer, a forward voltage drop of 0.45V-0.55V, so even with a steady 5V, Vin could be as low as 4.55V-4.45V. Apparently it still works, but isn't this way out of spec? And can we rely on it? Bass, how low can we go? The Schottky diode connected from SIO5V to the Nodemcu VIN (power input) is needed to prevent back powering the Atari if USB 5V is supplied to the Nodemcu. The Nodemcu has an on board voltage regulator that takes the 5V (probably more in the range of 3.5~5.1V) input and supplies steady 3.3V to the ESP module. So, we are not using the diode to drop the voltage on purpose, it's a side effect. A Schottkty is used since it has a lower voltage drop, and it's just for protection. 1 Quote Link to comment Share on other sites More sharing options...
ivop Posted December 22, 2019 Share Posted December 22, 2019 (edited) 1 hour ago, mozzwald said: The Schottky diode connected from SIO5V to the Nodemcu VIN (power input) is needed to prevent back powering the Atari if USB 5V is supplied to the Nodemcu. The Nodemcu has an on board voltage regulator that takes the 5V (probably more in the range of 3.5~5.1V) input and supplies steady 3.3V to the ESP module. So, we are not using the diode to drop the voltage on purpose, it's a side effect. A Schottkty is used since it has a lower voltage drop, and it's just for protection. I understand the reason why there's a diode in the first place, but the NodeMCU has an AMS1117/NCP1117 voltage regulator. Datasheets for both specify they need at least 4.75V for a stable 3.3V output, so not in the range of 3.5-4.75V. Still, it works with 5V - Vf == ~4.5V. A standard pn-diode has a Vf of 0.6-0.7V, so perhaps that will also work, even though it is way out of spec of the onboard voltage regulator. Edited December 22, 2019 by ivop typo Quote Link to comment Share on other sites More sharing options...
mozzwald Posted December 22, 2019 Share Posted December 22, 2019 13 minutes ago, ivop said: I understand the reason why there's a diode in the first place, but the NodeMCU has an AMS1117/NCP1117 voltage regulator. Datasheets for both specify they need at least 4.75V for a stabe 3.3V output, so not in the range of 3.5-4.75V. Still, it works with 5V - Vf == ~4.5V. A standard pn-diode has a Vf of 0.6-0.7V, so perhaps that will also work, even though it is way out of spec of the onboard voltage regulator. I had not looked at the AMS1117 datasheet (I did not realize how crappy the part is ). It lists the typical dropout voltage of 1.1V so, you should be able to get stable 3.3V from input of 4.4V (or worst case 4.6V) up to 800mA (then it gets worse). Quote Link to comment Share on other sites More sharing options...
tschak909 Posted December 23, 2019 Author Share Posted December 23, 2019 A quick status update: * I took the midimaze sketch that @mozzwald put together, and tossed it into a fully functional test program that sets up a Midi Maze connection between two end-points. A few SIO commands were added to support the required setup, and I passed the result off to @mozzwald. The code is here: https://github.com/FujiNetWIFI/atariwifi/tree/master/esp32/tests/midimaze * @mozzwald has been taking my code and tweaking it to work. ESP32 uses a totally different API to do PWM clocks, and even with the requisite changes, (and a proper clock signal), we are having timing issues, causing the game not to work. The same code works just fine on the 8266. * I have written a new cio test program to exercise UDP communication, still to be debugged, it's here: https://github.com/FujiNetWIFI/atariwifi/tree/master/esp32/tests/cio6-udp. We have noticed that the ESP32 serial timing is slower than the 8266. There are definite delays incurred in transmitting data that are having wide ranging effects from slow read performance (write, somehow is faster!), to the fact that midimaze isn't running. We definitely need more help to try and track down these timing issues, as they are adversely affecting reliability (and when you least expect it!) So if you have any ESP8266/ESP32 experience under Arduino, please look at what we have to see what can be done better. -Thom 2 Quote Link to comment Share on other sites More sharing options...
ivop Posted December 23, 2019 Share Posted December 23, 2019 (edited) 23 hours ago, mozzwald said: I had not looked at the AMS1117 datasheet (I did not realize how crappy the part is ). It lists the typical dropout voltage of 1.1V so, you should be able to get stable 3.3V from input of 4.4V (or worst case 4.6V) up to 800mA (then it gets worse). Yeah, the AMS1117 is pretty shitty Have been looking for better schottky/blocking diodes, and this is what I came up with so far: 19TQ015 $0,70 from China (min. 10pcs) or $1,70 at Farnell (1pc) Very low Vf for low currents! This one should keep SIO's 5V input well above 4.6V, probably even above 4.8V, before entering the AMS1117. Edited December 23, 2019 by ivop 1 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.