=== Let's Drone Out — Episode 428 ===
Title: LDO 428 - With Special Guest Un!t (Beta Flight/AM32)
Recorded: 2023-11-09
Hosts: Stephen "Belrik" Howell, Wayne "CurryKitten" Andrews, Andrew "Frank" Hyams, Jack "Brightontillifly" Nash
Guest(s): Eike "Un!t"
Total audience: 1,097
Platforms: YouTube (https://www.youtube.com/watch?v=nn0qTXJtgKg); SoundCloud (http://www.podtrac.com/pts/redirect.mp3/feeds.soundcloud.com/stream/1661499357-letsdroneout-ldo-428-with-special-guest-unt.mp3)
=====================================

Un!t's links:

Drone Community discord: https://discord.gg/g7EJEHDW

AM32 discord: https://discord.gg/5bJcRaS9

Let's drone out is:
Jack - https://www.youtube.com/user/brightontillifly
Stephen - https://www.youtube.com/user/ShinyVideoStuff
CurryKitten - https://www.youtube.com/currykitten
Koel -https://www.youtube.com/channel/UCqkF5Al06Bs_TtnqQqADMTg

#LetsDroneOut #FPV #Podcast #DronePodcast #FPV #BetaFlight #AM32

[0.0] Jack: Hello and you're listening to Let's Drone Out and special thank you to our lovely patrons that help donate so that we can pay the editor to remove the swear words and silly things we say. Massive thank you too.
[14.0] Jack: Carlos Campos.
[16.0] Jack: Art Faulkner.
[18.0] Jack: Sam Dahmer.
[19.0] Stephen: On with the show.
[21.0] Eike "Un!t": Let's Drone Out now.
[23.0] Stephen: Hello and welcome to Let's Drone Out.
[54.38] Stephen: With us this week is Curry Kitten.
[58.38] Stephen: Try again.
[59.38] Stephen: Everyone's favourite Curry Kitten.
[63.38] Stephen: Hello.
[64.38] Stephen: He answers to that. Nothing else. That's it. Unless Frank's here. Frank's can get away with it. I can't.
[69.38] Stephen: Frank can get away with it. You can't.
[71.38] Stephen: And Jack aka Brighton to Livefly.
[75.38] Stephen: No microphone there.
[77.38] Stephen: And also our guest Aiki aka Unit. Welcome to the show.
[84.38] Eike "Un!t": Yeah. Hello. Awesome to be here.
[87.38] Stephen: So I think we're going to be running through a few things you do. You've been quite an active developer recently with a couple of projects. AM32 and Beta Flight also.
[99.38] Stephen: And I think you were also judging the freestyle competition for Beta Flight as well, right?
[105.38] Eike "Un!t": Yeah, that's true. We did that together. That was awesome. That was fun.
[109.38] Stephen: Great. So we were just talking about a seven inch that you built. But before we go any further, I've got to share my own piece of irritating tragedy where this evening I managed to get something jammed where it should not be jammed.
[124.38] Stephen: No, we're not talking about that again. This time it's a 3D printer.
[129.38] Stephen: And the PTFE tube that's supposed to be, you know, like self lubricating and unjammable. Yeah, it's got a bit of filament jammed in it.
[137.38] Stephen: The rest of the print is fine, but the bit that's supposed to be all slippy got jammed. So yeah, that sucks anyway.
[147.38] Stephen: Moving on. So you were talking about a big seven inch that you're building with kind of a test capable ESC.
[158.38] Eike "Un!t": Yeah, it's this one. It's in the AOS 7 from Chris Vosser with a Foxia Reaper,
[167.38] Eike "Un!t": 60, no, 65 AMP 8S capable ESC flash to AM32. Right. And it was actually my first seven inch flight ever.
[179.38] Eike "Un!t": So I was very cautious and was not doing much. Also, yeah, the rates, I just put in my five inch rates for my five inch freestyle.
[191.38] Eike "Un!t": And yeah, they felt quite low on this. And yeah, well, I was just cruising around and testing out if AM32 is holding strong.
[201.38] Eike "Un!t": Later, I do like one or two light punch outs. And yeah, AM32. Pretty awesome. It works pretty well.
[210.38] Eike "Un!t": And whole number strong that those are 20, 20, 28 or seven on HQ prop seven inch three blades.
[218.38] Eike "Un!t": And they are drawing quite some current. And yeah, I was not expecting to work it work like that because I haven't tuned it any.
[229.38] Eike "Un!t": Anyhow, it's just plain default view of light, default fits, default features. It's just just flying very, very, very, very well.
[237.38] Eike "Un!t": A little bit of oscillations, which are expected when you don't tune anything. But yeah, that will get sold out later.
[244.38] Stephen: What got you interested in AM32 then? Because it's kind of a struggle. Sometimes the flashing process takes a lot of work.
[251.38] Stephen: You seem to have really gone into it 100 percent.
[254.38] Eike "Un!t": Oh, yeah. When I started my drone life with a fly egg, I got a F3 back in the days and started on that.
[267.38] Eike "Un!t": I destroyed it right away. And then I got into all the other stuff. And yeah, it was a BL Heli S ESC, which I flashed to BlueJ.
[276.38] Eike "Un!t": Then I got to know that there's BL Heli 32, yada, yada, yada. And I smoked some motors, some ESCs on BL Heli 32.
[284.38] Eike "Un!t": And it was closed source. And so I couldn't figure out why. And that's basically when I stumbled over AM32.
[292.38] Eike "Un!t": And yeah, at first, as you said, it's a little bit of a process to get a ESC to AM32. But yeah, I got used to it.
[302.38] Eike "Un!t": And now every BL Heli 32 ESC in any of my quads is AM32. And yeah, then I got to know Alka as well. Pretty awesome.
[313.38] Eike "Un!t": He is quite a brain when it comes to ESCs. And yeah, we got to chat. And then I offered that if he wants, I could join the development team.
[323.38] Eike "Un!t": And now we are together in the team and developing on AM32. Yeah.
[330.38] Curry: So speaking as someone that doesn't, I've not really experimented with like alternate ESC firmwares.
[338.38] Curry: And most of my stuff is, it's what it came with, which is normally BL Heli S or BL Heli 32.
[346.38] Curry: What is a draw to AM32? And what might I be able to do with it that I can't do on the firmware I already have on my quads?
[357.38] Eike "Un!t": What you can't do, you can't smoke a motor that easy. That's one big thing. What changes if the stuck motor protection on AM32 is really, really awesome.
[367.38] Eike "Un!t": It's top notch. I have not seen anything work that well. I've never smoked anything with it. And I've flown right into concrete.
[377.38] Eike "Un!t": And yeah, that's the ESC and the motors were sort of fine. Okay, I bent some motors and then they are trash after that, but they wouldn't smoke.
[386.38] Eike "Un!t": Yeah. And what AM32 also does is it has a way better torque curve. Chris Rossa already covered that when on his prop and motor test bench setup, he uses an AM32 and he showed the prop, the deep torque curve on BL Heli 32 and AM32.
[409.38] Eike "Un!t": Also, what I saw that you have in both ends, you have way better energy performance, which means I'm on my three ends, which I first flown on BL Heli 32.
[424.38] Eike "Un!t": I got when I was like motor freestyling, I got around, I would say four to five minutes and on AM32, I got one or two extra minutes when I was even pushing a little bit harder.
[440.38] Eike "Un!t": And that got me, that really catch me where I get more performance when I want it and I also get more flight time when I want it.
[451.38] Eike "Un!t": And yeah, that's my own experience. I can't really say it's on every quad you get like 10 or 20 percent more flight time.
[461.38] Eike "Un!t": But yeah, when I tested it at pretty much same lipo, same quad, just different, even the same ESC, but just on a different firmware.
[474.38] Stephen: Yeah, so Chris Rossa tested a few different firmwares to see what would happen even through FedEx as well. And the output was that there's generally a much smoother torque curve with AM32.
[488.38] Stephen: And I think it was coming out of a background of like RC crawlers and things where people would notice that the torque of a motor, whereas I think in quads, we don't generally immediately notice the torque.
[500.38] Stephen: We just notice is an ESC smooth or not, and that can be down to motors and all sorts of things.
[507.38] Stephen: So, yeah, he did a fair bit of testing to look into these torque curves and was seeing that, you know, you're getting very high maximum torque for a longer time with AM32 and it's kind of gradually increasing.
[521.38] Stephen: Whereas BL Heli32 is a bit of an odd burden and kind of jumps around, doesn't really give you consistent torque.
[529.38] Stephen: And I think that consistent torque is perhaps reflected in sometimes the current that goes into the motors, right? So you couldn't end up putting more heat into the motors without intending to.
[541.38] Eike "Un!t": Yeah, imagine you have a torque curve that's jumping on different RPM levels or RPM sections. The flight controller has to compensate.
[552.38] Eike "Un!t": If you want a flip or anything and it's wobbly, then it has to push in more power and flips around a little bit.
[560.38] Eike "Un!t": You can see that sometimes. And if you have a smooth and well-rounded torque curve, it can actually work very well with that.
[572.38] Eike "Un!t": And it's just, yeah, it's my experience. I feel it has more power on the top end and you can have more and more energy reserves on the low end if you fly in general.
[584.38] Stephen: For all of your listeners, I was flushing up some of Chris Ross's graphs there, showing the kind of current graphs and torque graphs that he did.
[592.38] Stephen: That seems to be what people like. And it's kind of the acid test for an ESC and indeed a motor when you do snap rolls and fast racing turns where you have to stop two rotors completely dead and you know you're damping to break those.
[607.38] Stephen: And then you're getting the motors from zero up to full speed as quickly as possible. And to do that without causing lots of excess heat or problems, vibration, that's what you want to achieve.
[621.38] Eike "Un!t": And I felt like that's better on AIM32. I would even say that it's better on BL Heli S or in BlueJ than on BL Heli 32.
[636.38] Stephen: Well, yeah, they're still developing BlueJ, aren't they? If you see those guys talk about it, they've got a little GoFundMe or something and they've got so many sets of cheap motors to test and they'll just like thrash these things and be like, what does burn out the motor?
[650.38] Stephen: Right. Let's find out. These are the settings that burn motors and then we keep repeating until we understand all the causes of motor burning and we code around them.
[659.38] Eike "Un!t": Yeah, that's one thing that I like about AIM32. It's fully open source. You can read every line of code and the bootrotors open source, the firmware itself is open source.
[669.38] Eike "Un!t": The thing with BL Heli, no, not BL Heli S, BlueJ, even though it's not readable by the normal developer because it's assembler, but yeah, it's open source.
[680.38] Eike "Un!t": And that's what I like about stuff like that because you depend, not your life, but your drone on it. And to fly something that actually only a few people have control over is, yeah, I don't know.
[695.38] Eike "Un!t": I just don't like it. I like open source.
[699.38] Stephen: It sounds like you're going to make AIM32 a bit more approachable right now as well because Alka has been doing this one-man army and you're kind of going in there and reorganizing the data a little bit, making it easier to find stuff.
[712.38] Eike "Un!t": Yeah, that's right. Also what we are doing, the last weeks, last month, is making it easier to develop because right now we have a repository for G071, for F051, for the AT32 chips.
[736.38] Eike "Un!t": It's all separate repositories. So if Alka does a fix or perform the improvement and one repository has to kind of push it to every other.
[747.38] Eike "Un!t": And that's what we are doing lately and we are making a multi MSCU repository, which is already finished. We are in the last bits of publishing it.
[757.38] Eike "Un!t": The plan was to actually make it public this week. I'm not sure if we can make it, but yeah.
[764.38] Eike "Un!t": ExpressLRS always says it's always two weeks. So yeah, we will be finished in two weeks. And then we have a repository where a lot, I would say really a lot of performance improvements are in and also where you have every firmware from every repository in one.
[784.38] Eike "Un!t": So it's easier to find, easier to maintain, and I hope this will improve even the project even further.
[793.38] Stephen: Brandon Bean asks, regarding FETEC and KISS, the proponents of these firmwares are quite happy with the performance they achieve.
[799.38] Stephen: But whenever it's analyzed and tested compared to BetaFlight or BlueJ or AM32, it always looks worse. Why is this? I don't know if you've looked at FETEC and KISS much?
[811.38] Eike "Un!t": No, I don't. I don't.
[814.38] Eike "Un!t": That's to say it's religion then.
[816.38] Eike "Un!t": Yeah, I can't really answer.
[820.38] Eike "Un!t": Yeah, I have actually not touched any FETEC, KISS or KISS Ultra or anything. I have no gear in my home from this area.
[836.38] Stephen: Yeah, I've got a FETEC FC somewhere. It's never actually been flown. It's just kind of sitting in a box. I got it to get a 10S ESC actually and 10S motors.
[847.38] Stephen: And I flew those a couple of times. The ESC died very quickly. And I actually plan to replace it with an AM32 model to reduce the risk of burning stuff.
[858.38] Stephen: Hopefully, the motors survived because they're hard to find 10S motors, like 22S drive and 10S.
[866.38] Eike "Un!t": Yeah, I actually have only 6S motors in my 7-inch. They are 6S, 1,300 kV. And I will just try to run them on 8S will be 70, 75.
[879.38] Stephen: There seems to be some disagreement about what a 6S motor is because we were talking about this before.
[885.38] Stephen: Some people say, oh, yeah, 1,500 kV, 6S, 7-inch. Some people are like, oh, 1,750 kV on 7-inch is fine, 6S.
[893.38] Stephen: And other people will say, no, that's a 4S motor. What are you doing? You're crazy.
[897.38] Eike "Un!t": I have flown my 3-inch on 3,500 kV on 6S. So there is actually no 6S motor. It's just what works with it.
[906.38] Eike "Un!t": And if it's producing too much RPM on full throttle, then, yeah, it just burns or the prop just goes poof.
[914.38] Eike "Un!t": And yeah, what works, it works. So if you take it, you just calculate the RPM and then see what current it goes.
[923.38] Eike "Un!t": It will draw. And then if that works, you can try.
[929.38] Eike "Un!t": If the manufacturer rates it at, yeah, like 6S, 1,300 kV, 6S, then it will most likely be perfect from 0 to 100% throttle.
[939.38] Eike "Un!t": But yeah, if you go for it with 8S, yeah.
[945.38] Stephen: I think this is probably why I was getting problems because I was trying to run 8S with EMAC's Eco motors.
[951.38] Stephen: They're not bad motors, but they are budget motors and they were 1,300 kV.
[956.38] Stephen: Like, okay, 7-inch, like 7x4 blades, dry blades, and I'm going to put like 42 volts into you.
[963.38] Stephen: We're going to see what happens. Not 42, would it be like 36 volts for 8S?
[968.38] Stephen: Yeah.
[969.38] Stephen: And see what happens. And yeah, they got to like the top of the throttle.
[975.38] Stephen: I could get to like 85% throttle, but beyond that, things would get sketchy and it would get kind of twitches.
[981.38] Stephen: You know, you get twitches because it's desinking.
[985.38] Stephen: So I'm hoping better motors fit.
[987.38] Eike "Un!t": What also must take into consideration is that the ESCs, if the fat is rated for 8S,
[996.38] Eike "Un!t": they tend to have lower internal resistance, but higher gate charges, which translate into,
[1005.38] Eike "Un!t": they're not capable for so much RPM. So if you take the back EMF and everything into consideration,
[1013.38] Eike "Un!t": yeah, that's something I don't fully understand it.
[1017.38] Eike "Un!t": Alka is the brain behind it, but I got a little bit of that to understand.
[1024.38] Eike "Un!t": And if you design a ESC, you pick the parts for what you want to run and you have the limitations.
[1034.38] Eike "Un!t": Either you have low internal resistance, which means you can run high currents,
[1039.38] Eike "Un!t": or you have a low internal, no, the gate charge is very low, which means you can switch it on and off faster.
[1048.38] Eike "Un!t": And you have to match both. You can't have both very low.
[1056.38] Eike "Un!t": There are fats that have both lowers, but neither of them are very low.
[1061.38] Eike "Un!t": It's just a balance. And if you're done in 8S ESC, you'll have to...
[1069.38] Eike "Un!t": Also, the highest voltage is, yeah, it's also in the equation.
[1078.38] Eike "Un!t": Yeah, there's definitely a lot of factors at play there.
[1080.38] Stephen: It was a lot of fun to fly around and I didn't crash it, so it never got bad, but it would twitch.
[1087.38] Eike "Un!t": What I liked the most was that you have so much inertia.
[1092.38] Eike "Un!t": When you do something, it's not like a 3-inch or even a 5-inch, low weight, a 5-inch.
[1100.38] Eike "Un!t": If you do something, it's so fast from here to there in the flip and you go upside down and it goes slowly back down.
[1108.38] Eike "Un!t": If you have the 7-inch, it's just so beefy.
[1111.38] Eike "Un!t": If you could do a fast turn and you know it wants to drag out, that's something so different than what I was used to.
[1120.38] Eike "Un!t": That's very fun.
[1122.38] Stephen: Plus, with a really tight tune in that AOS frame, it's still got that tight feeling of the 5-inch.
[1129.38] Stephen: But you just do the punch and you're over the trees and you think, oh, I'll do a quick invert.
[1133.38] Stephen: And you invert and you just carry on going up like a cannonball.
[1139.38] Eike "Un!t": Yeah, I have not done that because I was a little bit afraid to crash it and destroy it.
[1149.38] Stephen: But you have been pushing your racing rig quite hard, right?
[1153.38] Stephen: I gather you've been saying before the show.
[1157.38] Eike "Un!t": That's this one.
[1159.38] Eike "Un!t": It's a light switch from 533 with a Foxy Reaper Mini ESC, 65Ms as well.
[1169.38] Eike "Un!t": And it's got AM32 as well.
[1171.38] Eike "Un!t": And this was actually a proof of concept because we wanted to know what we can actually push.
[1179.38] Eike "Un!t": And I ended up with pretty insane settings.
[1183.38] Eike "Un!t": And if you go to the AM32 configurator, you can set up a KV slider.
[1191.38] Eike "Un!t": The KV slider is not really setting the KV.
[1196.38] Eike "Un!t": What it actually does is an amp protection.
[1199.38] Eike "Un!t": So if you go to zero, you have zero amp protection.
[1201.38] Eike "Un!t": If you go to max, I don't even know what max is.
[1204.38] Eike "Un!t": I don't know, 9000 or something like that.
[1206.38] Eike "Un!t": It has a lot of amp protection.
[1209.38] Eike "Un!t": And on this thing, normally if I start my AM32 tuning, I set it to the double of the KV of the motors.
[1219.38] Eike "Un!t": Those are 2010 KV excess flying motors.
[1225.38] Eike "Un!t": And I set it to 4000.
[1228.38] Eike "Un!t": And then I just keep decreasing it and pushing and pushing and pushing.
[1232.38] Eike "Un!t": And I ended up with a stable build at 1000 KV, which is pretty insane,
[1237.38] Eike "Un!t": which means it has all new amp protection.
[1241.38] Eike "Un!t": Yeah.
[1242.38] Eike "Un!t": And it's holding strong.
[1244.38] Eike "Un!t": I wasn't expecting that.
[1245.38] Eike "Un!t": I was like, yeah, around the ballpark of 1500, 2000 KV, it would desync.
[1251.38] Eike "Un!t": But I haven't gotten one desync yet.
[1253.38] Eike "Un!t": And it's pretty insane.
[1255.38] Eike "Un!t": I haven't expected it because it's running the new performance improved firmware version.
[1261.38] Eike "Un!t": And we had so many issues.
[1264.38] Eike "Un!t": We had reports of desyncs.
[1266.38] Eike "Un!t": And we still have reports of desyncs with one particular MCU.
[1270.38] Eike "Un!t": And do we have time for it?
[1273.38] Eike "Un!t": I will explain why this particular MCU is actually not good for ESCs in general.
[1280.38] Eike "Un!t": It's a good MCU, but it's not good for MCUs for one particular reason.
[1284.38] Eike "Un!t": But to finish the first, yeah, we wanted to know if we actually solved the main issues with AM32 concerning desyncs.
[1295.38] Eike "Un!t": And we did.
[1296.38] Eike "Un!t": I would say we did.
[1297.38] Eike "Un!t": And, yeah, this racing quad was the quad that I can drain in under a minute.
[1303.38] Eike "Un!t": The MCK best racing, best racing lipo.
[1308.38] Stephen: 1300, you said?
[1309.38] Eike "Un!t": Yeah, 1300, 500 milliamp hours, 6S, .com, MCK, V2.
[1315.38] Eike "Un!t": I don't know exactly about that beefy, beefy lipo.
[1319.38] Eike "Un!t": And if I really push it and going back and forth, back and forth, it's empty in under a minute.
[1325.38] Eike "Un!t": Which I find pretty good for a ESC firmware that was known for desyncing on some ESCs on high amperage.
[1337.38] Stephen: I remember Limon was saying, another racer was saying, oh, yeah, I tried this two years ago.
[1342.38] Stephen: And I was seeing desyncs and you were like, okay, right.
[1346.38] Stephen: So this type of thing is like, oh, you can't go in a straight line and, you know, just keep a full throttle and do a turn.
[1353.38] Stephen: 180 degree turn, a little desync.
[1355.38] Stephen: Yeah.
[1356.38] Stephen: And you're like, no, I can do that for a minute until my battery is smoking and it's good.
[1361.38] Eike "Un!t": Yeah, that's why actually Foxy has sent me some.
[1365.38] Eike "Un!t": I will flash them to AIM32 and send them.
[1369.38] Eike "Un!t": I will send one 30x30 to Superfly.
[1375.38] Eike "Un!t": He will go in to test it on his sim mode, which is my 5mm freestyle.
[1382.38] Eike "Un!t": I finally can put in AIM32s in my sim modes.
[1386.38] Eike "Un!t": I'm so happy.
[1387.38] Eike "Un!t": And then I will flash the Foxy Ripper Mini and send them to Limon and he can race test them.
[1396.38] Stephen: You can give it hell and fly around concrete and smash it to pieces.
[1400.38] Stephen: Yes.
[1401.38] Stephen: I mean, what else would you do?
[1403.38] Stephen: What else? Yes.
[1404.38] Stephen: So you're about to say there's one MCU that isn't good.
[1408.38] Stephen: We talk with Jack about the history of ESCs and Jack's always very keen to remind people of the Little B and all these old chips.
[1417.38] Stephen: So maybe we can kind of continue our ESC lineage.
[1421.38] Stephen: So we talked about the Little B.
[1423.38] Stephen: There's like, I think, Silicon Labs, the Busy B series that you started out with the BB-1.
[1430.38] Stephen: And there was some stuff before that.
[1432.38] Stephen: But BB-1 was where BL Heli S appeared, right?
[1436.38] Stephen: And then the BB-2 and BB-21, BB-51, what we've got now with BlueJ.
[1443.38] Stephen: And that gives us RPM filtering via lots of kind of very clever assembly code, right?
[1449.38] Stephen: But then we moved on to BL Heli 32.
[1452.38] Stephen: And from that, we took those MCUs forward.
[1454.38] Stephen: Do you want to do the kind of the rundown or should I?
[1459.38] Stephen: I can cover them if you like.
[1461.38] Eike "Un!t": Yeah, go ahead.
[1463.38] Stephen: Okay, so we have like the STM microelectronics, like with flight controllers.
[1467.38] Stephen: Everyone was quite used to programming these.
[1469.38] Stephen: So we have STM32F051, which was on a lot of the early BL Heli 32 ESCs.
[1476.38] Stephen: And it's more a sedate speed.
[1479.38] Stephen: I think it's something like roughly the same speed as a BB-21.
[1483.38] Stephen: It's not actually that fast.
[1485.38] Stephen: Then we started seeing the G071 from STM32 as well.
[1489.38] Stephen: That increased the speed a little bit, which made things interesting.
[1492.38] Stephen: Because the commutation of the ESC and the telemetry and everything has to account for a slightly different MCU speed.
[1499.38] Stephen: And then the big chip shortage hit us and everything just went sideways.
[1503.38] Stephen: And suddenly, instead of these being a couple of dollars, they're like $10 or more.
[1508.38] Stephen: And suddenly, you don't want to put four of them on an ESC.
[1510.38] Stephen: So there started to be a lot of diversification.
[1513.38] Stephen: And people were looking around at something that behaved like an STM32 part, but was considerably less.
[1520.38] Stephen: And we started seeing Giga devices, GD32, so GD32E230.
[1527.38] Stephen: And then there was a GD32F350, which I think is the cursed one.
[1532.38] Stephen: And then we saw people trying to go for performance in their marketing materials.
[1537.38] Stephen: And they went to Artury Tech with AT32.
[1540.38] Stephen: And we've got like AT32, F421, and I think the F415 as well.
[1547.38] Stephen: And then I think what we're about to see appear some chips from Airbot, which are the F4A,
[1552.38] Stephen: that take those F421s and they add like diodes and gates onto them.
[1559.38] Eike "Un!t": A fat driver. It's a combination with currently the current version has the AT32F421 and a fat driver inside.
[1571.38] Eike "Un!t": And yeah, I have heard that there might be a version in the future with diodes inside, but yeah, not yet.
[1579.38] Stephen: And this kind of ties to the STM32 spin stuff that was always kind of overpriced and people didn't really like it.
[1586.38] Stephen: I never got into that many products, right?
[1590.38] Eike "Un!t": Okay, so...
[1591.38] Eike "Un!t": It's a great MCU, the STM, but yeah, AIM32 actually would work on it, I think.
[1600.38] Eike "Un!t": I'm not 100% sure. I won't continue because, yeah, I'm not 100% sure.
[1605.38] Eike "Un!t": But it's a great MCU, but yeah, it has drawbacks and also I think it's quite expensive.
[1612.38] Stephen: Yeah, but the important thing is we've got a cheap part and I think with the F4A stuff,
[1616.38] Stephen: which is the culmination of this, it now takes this 32-bit ESC and gets it to a footprint that's comparable
[1623.38] Stephen: with what you've got with BlueJ and the old Busy B chips, right?
[1627.38] Stephen: So you're finally in that ballpark.
[1629.38] Stephen: Correct.
[1630.38] Stephen: Now, what about the cursed stuff?
[1632.38] Eike "Un!t": The cursed stuff, yeah. The Giga device, the 230.
[1637.38] Eike "Un!t": It's very simple. Alka actually told me just today
[1645.38] Eike "Un!t": why it's so doomed and that's pretty easy.
[1648.38] Eike "Un!t": It's meant for running from minus 25 degrees Celsius to up to 85 degrees Celsius
[1657.38] Eike "Un!t": and 85 degrees is not a lot for an ESC.
[1660.38] Eike "Un!t": If you like punch, if you do flips and rolls, it easily reaches 100 degrees.
[1665.38] Eike "Un!t": And then what actually happens, the MCUs run and find, but the cooperators inside of the MCU is stopping.
[1671.38] Eike "Un!t": And then we don't have commutation. We don't have zero crossings any more.
[1675.38] Eike "Un!t": So the ESC doesn't even know that the balance is spinning.
[1678.38] Eike "Un!t": So it just stops because what kicks in is the stuck motor protection, which, because we don't see anything spinning anymore,
[1685.38] Eike "Un!t": because we don't have back EMF detection anymore, the stuck motor protection kicks in and shuts down the ESC.
[1693.38] Eike "Un!t": And yeah, that's why this MCU just doesn't work on a M32.
[1701.38] Eike "Un!t": I would say I'm neither, because yeah, you will exceed this temperature limit quite easily on an ESC.
[1711.38] Eike "Un!t": And yeah, it's really a shame because we have two very, very great ESCs from Skystars,
[1719.38] Eike "Un!t": the KM55 and the AM60, which are really, really awesome ESCs, but they have this MCU, so you can't push it.
[1729.38] Eike "Un!t": And that's really, really a shame.
[1731.38] Stephen: Yeah, I've got a couple of those. That's kind of a shame.
[1734.38] Eike "Un!t": Yeah, because you can't really do anything. We can't fix it in software.
[1737.38] Eike "Un!t": We can't say, okay, yeah, we are around that temperature limit because if you do really fast flying,
[1743.38] Eike "Un!t": it reaches so quickly that you can't really account for it.
[1747.38] Stephen: At least they ship with a very big heat spreader on them. So that's something.
[1751.38] Eike "Un!t": Yeah, but that doesn't help. If you have like 200 amps going through, it can't get away fast enough.
[1760.38] Stephen: But the 200 amps isn't going directly through the MCU. It's just near the MCU, isn't it?
[1766.38] Eike "Un!t": Yeah, right. But the heat is pretty near. And also, yeah.
[1773.38] Stephen: So you'll get away with it for a bit before the heat saturation.
[1777.38] Stephen: Because heat sinking is kind of like you've got a body and you put the heat into this metal and into the board.
[1785.38] Stephen: And it takes a while to fully soak the whole board in heat, right?
[1790.38] Stephen: So you'll have like a window a few seconds where it's good, maybe 15 seconds.
[1794.38] Stephen: But if you're a racer and you're pushing it for a whole minute, I guess that's when you get the problem.
[1800.38] Eike "Un!t": Yeah, you could do one test. If you have a heat gun, you put the ESC on a bench, put a motor on it,
[1809.38] Eike "Un!t": just spin it a little bit, point the heat gun at it. And when it's around 100 meters, it'll just stop.
[1816.38] Eike "Un!t": And if you go in a quad, if you do a flip or two flips, it could already be enough for one ESC to reach.
[1826.38] Eike "Un!t": Because at 100 meters, it shuts off. But way before that, it gets sketchy.
[1830.38] Eike "Un!t": So the zero crossings aren't getting detected good enough anymore. So that's more.
[1836.38] Eike "Un!t": When it goes too high, it just doesn't work very well anymore.
[1841.38] Eike "Un!t": So you can go cruise with it. So put it in a long range quad and just cruise around. That will work.
[1849.38] Stephen: Someone's saying it's 100C for some of them. I don't know if that's a variation of the chip or what they're getting at.
[1856.38] Eike "Un!t": 100C is 100C if the ESC is made in China, like many of them.
[1861.38] Eike "Un!t": Oh, the SkyStores, in 30K, SkyStores are producing them in China.
[1869.38] Eike "Un!t": What do you actually mean?
[1871.38] Stephen: I think he's implying that the comparators might be a little bit more heat tolerant on some of them.
[1878.38] Eike "Un!t": This MCU is not meant for high temperature. We are exceeding the limits of the documentation of this MCU.
[1887.38] Eike "Un!t": So it's kind of expected to fail.
[1891.38] Stephen: Don't. That's why you should always read the datasheet like three times before you buy something, right?
[1899.82] Eike "Un!t": Especially the limits, yes.
[1902.82] Stephen: Victoroid is asking about the new Airbot Fenix ESCs as well.
[1909.82] Eike "Un!t": Oh, yeah. Most of them are using the F4A chip.
[1915.82] Eike "Un!t": And yeah, I've tested all of them and they're actually pretty great. So yeah, go out buy them.
[1922.82] Eike "Un!t": They don't have those limitations. They were great. They have the F4A chip.
[1929.82] Eike "Un!t": And it doesn't have any of those issues.
[1932.82] Eike "Un!t": Actually, Alka reported that he can get his F.
[1937.82] Eike "Un!t": It's actually another MCU, but the F4A is similar.
[1944.82] Eike "Un!t": He tested the F051 with a heat gun as well.
[1950.82] Eike "Un!t": And it just stopped at 150 degrees, which is quite higher than 100.
[1958.82] Stephen: What about the F41? Is that that's happy at 150 as well or we're not sure?
[1963.82] Eike "Un!t": I would say yes. I have pushed it really, really crazy.
[1967.82] Eike "Un!t": I mean, if you take the Foxy Reaper I have here in my racer, it's actually the same MCU and the same fat driver just in one MCU.
[1979.82] Eike "Un!t": So I would say yes, they would hold up strong at that degree.
[1984.82] Stephen: Cool. Well, yeah, it's going to be good to see those because I think for racers and anyone who wants an AIO as well,
[1991.82] Stephen: they're going to allow a lot more in smaller space.
[1994.82] Stephen: So you can either have a smaller board or you can put big effects on the same board or you can make the board.
[2000.82] Stephen: So there's stuff not close to the edge of the board to make it stronger. So only good things, right?
[2007.82] Eike "Un!t": Yeah. If you take some of the AIOs actually, you have the Fenix G4 AIO,
[2016.82] Eike "Un!t": which has a little dollar board on the ESC side, which is actually the whole ESC.
[2022.82] Eike "Un!t": So you only place the pads around and then you have to make the traces to the starter board.
[2028.82] Eike "Un!t": And then you have complete ESC, which will fit on every AIO.
[2034.82] Stephen: So Liviu FPV replies, meaning that most of the ESCs use the cheapest parts they can use.
[2040.82] Stephen: And the heating is a direct result of that. You have to really stress the hell out of a good ESC to reach 100C.
[2046.82] Stephen: So I guess they're saying if it's a cheap ESC, it'll heat faster.
[2050.82] Eike "Un!t": But if you make punch, there is a really, really short time where the fat is, I don't know, pushing 100A.
[2062.82] Frank: Each fat, each ESC is pushing 100A around each other or something like that.
[2093.82] Stephen: We should probably take a minute and talk about this initiative that Lemon has sparked,
[2097.82] Stephen: a kind of good-natured joke amongst racing pilots, right?
[2101.82] Eike "Un!t": Oh, yeah. We all are shit pilots, right?
[2105.82] Stephen: Do you want to get up to the camera with the shirt there?
[2109.82] Stephen: Yep.
[2110.82] Stephen: Yeah.
[2111.82] Stephen: Some brand recognition for shit pilots the world over.
[2115.82] Stephen: Oh, yeah.
[2118.82] Eike "Un!t": I'm a proud shit pilot.
[2122.82] Eike "Un!t": Lemon gave me the permission to come and save the pilot.
[2125.82] Eike "Un!t": That's good.
[2126.82] Stephen: Well, you've got to check. I want to check.
[2129.82] Eike "Un!t": Okay. Now I forgot where I was.
[2132.82] Stephen: Sorry. We were talking about heating ESCs and I think Liviu FPV and the good ones don't heat as much.
[2140.82] Eike "Un!t": Yeah. I mean, if it's thick, then it can handle the heat way better.
[2145.82] Eike "Un!t": And cheap ESCs, especially AIOs, can't be that heat resistant.
[2152.82] Eike "Un!t": And you reach the temperature limit in a small part where the MCU is quite faster.
[2162.82] Eike "Un!t": And not the whole ESC is 100 degrees Celsius, but maybe that part alone.
[2167.82] Eike "Un!t": And even if the E230 gets near that temperature limit, like 70, 80 degrees, then it already starts to behave strange.
[2178.82] Stephen: So you just got to hope that if you do have one of those ESCs, there's a good thick slice of copper through it.
[2184.82] Stephen: So it's not going to get too hot because the cheap ESCs, they want to use the minimum amount of copper
[2191.82] Stephen: so they can make them cheaply as copper is expensive.
[2194.82] Stephen: And it gets higher resistance and higher resistance, it's hotter.
[2198.82] Eike "Un!t": Yeah, that's true.
[2200.82] Curry: So my question in this is I'm looking at the AM2 site and I see there's a list of supported hardware.
[2207.82] Curry: Is this extensive? Because this is my first time of even thinking about what the MCU is on a board I have.
[2216.82] Curry: And I've got apparently a BB21 MCU, if that makes any sense.
[2223.82] Eike "Un!t": Yeah, that's a busy B.
[2225.82] Eike "Un!t": That's an 8-bit MCU and AM32 is 32-bit.
[2231.82] Stephen: Ah, it's old.
[2233.82] Stephen: So the busy Bs are the range that work with BlueJ and BL Heli-S.
[2239.82] Curry: I see, right? But you're looking for 32-bit MCUs basically.
[2244.82] Curry: It's like a lot of my lab busy B.
[2248.82] Eike "Un!t": Yeah, Livia said that the Fox Weeper telemetry doesn't send out temperature. I don't know.
[2253.82] Eike "Un!t": Thus, on AM32, I have temperature readings on my...
[2258.82] Eike "Un!t": Is that on the serial port or EDT?
[2262.82] Eike "Un!t": Oh no, EDT extended telemetry.
[2264.82] Eike "Un!t": There you are. That's probably it then, right?
[2267.82] Eike "Un!t": Oh yeah. I'm not sure if... Livia, you're using Kiss Ultra, right?
[2272.82] Eike "Un!t": I don't know if they have implemented EDT from detail.
[2277.82] Stephen: So with EDT, instead of a serial bus for all of the MCUs,
[2282.82] Stephen: it uses the DSHOT channel to each MCU for a dedicated telemetry feed from that.
[2288.82] Stephen: And it's much faster turnaround as well.
[2292.82] Eike "Un!t": Because you basically get every DSHOT packet, you get a return, the telemetry package,
[2298.82] Eike "Un!t": the RPM telemetry package back, and there you have some free bits.
[2304.82] Eike "Un!t": And that's the extended where you get temperature and maybe in the future more.
[2310.82] Stephen: Yep. So I think currently you have to switch it on, but it's kind of handy.
[2313.82] Stephen: If you have issues with your voltage or current sensor,
[2316.82] Stephen: I've got one board where the ADC just did not play ball, just some cheap FC.
[2322.82] Stephen: And I turned on EDT and boom, you know, read out the ESC and get voltage for all corners.
[2330.82] Eike "Un!t": I love EDT. I actually think I have flashed the latest 4.5 master yesterday and EDT was on by default.
[2339.82] Eike "Un!t": I must double check it. I'm not sure. It could be.
[2344.82] Stephen: It's handy. It's an easy way to get temperature for your ESC as well, which is kind of useful to have.
[2350.82] Eike "Un!t": And you don't lose anything, right? So if you have it, you don't need the ESC telemetry wire.
[2356.82] Eike "Un!t": And yeah, but yeah, the temperature is not calibrated. I'm actually on the Fox here. It is.
[2362.82] Eike "Un!t": I flashed both the Reaper Mini and the Reaper on AIM32 and I got pretty accurate temperature readings.
[2369.82] Stephen: I guess it depends where they buy the MCUs from then, perhaps, or if they're doing any calibration after the fact.
[2379.2] Eike "Un!t": I don't know if they can do the calibration themselves. I'm not sure if they have to be done at the factory where they produce the MCU.
[2390.2] Stephen: I would have thought it would be a writable value, wouldn't it? How they calibrate it?
[2395.2] Stephen: Because it just gives you a raw reading and you have to match up the number to a lookup table that says this number means 20 degrees, this number is 30 degrees.
[2404.2] Eike "Un!t": What they do is they have a chamber. They put it in the chamber. The chamber is flat on 30 degrees and then they burn that as 30 degrees.
[2414.2] Eike "Un!t": And then, yeah, the temperature center has a curve and then it gets adjusted to that this point is 30 degrees.
[2423.2] Stephen: We've got Aleksandr Fedorov of Ultra asking about DSHOT2400 as well, which I think was added to the firmware from some QSUS requests.
[2436.2] Eike "Un!t": I think it should work the same. I mean, it's the RPM telemetry package.
[2443.2] Eike "Un!t": You have to implement RPM telemetry and it should just work like that on DSHOT2400. I don't think why it shouldn't.
[2459.2] Stephen: So long as the timing works out, I guess, but I'm not sure. I remember when you got some really crazy fast rates, there were some subtleties.
[2468.2] Stephen: I think Alka mentioned something about some stuff working at 1200 and not 2400, but I think it's all documented.
[2478.2] Stephen: And I think that the BlueJ crowd actually maintain a document that explains all the timing values and the encoding and everything.
[2486.2] Stephen: So that's a good place to start. We should also mention you've done a whole bunch of stuff for beta flight as well.
[2493.2] Stephen: So there's some new features that you've really pushed for. Do you want to talk about any of those?
[2501.2] Eike "Un!t": Yeah, actually my first contribution to beta flight was the new page. That's where I got actually invited.
[2513.2] Eike "Un!t": It was pretty funny. I did the page with Bitroid. We together, we coded basically the whole webpage from beta flight.
[2524.2] Eike "Un!t": And then over the night, the next time I woke up and then I was a beta flight developer. That was pretty fun.
[2532.2] Eike "Un!t": And then my first contribution was I did the DSHOT bitbank code for the AD32 MCU.
[2543.2] Eike "Un!t": So we still haven't discovered the bug and why it's not working, but PWM DSHOT doesn't really work on AD32.
[2553.2] Eike "Un!t": We can't just get it to run. It's weird, but yeah, it doesn't work. And we don't actually have the time to get it done because the bitbank DSHOT code just works.
[2565.2] Stephen: I've had this on some SGM32 ESCs back in like 4.3 days that just didn't work. Yeah, move on.
[2575.2] Eike "Un!t": Yeah, I mean, it works. The bitbank code is just fine. So why bother? And then I got into the GPS code and implemented the M10 API into beta flight.
[2591.2] Eike "Un!t": Yeah, that's my two big contributions. Yeah, rather big, but my two contributions to beta flight.
[2599.2] Stephen: Yeah. And you pointed me to some code snippet. I was experimenting with this little quad here, which has like one GPS at the back.
[2608.2] Stephen: And for good measure, another small GPS attached via this terribly unsafe wire that I haven't secured yet, which I was testing out this ANA stuff that was supposed to improve the fix.
[2620.2] Stephen: And I think some some FCs are just noisy. And yeah, I can't get a GPS fix.
[2625.2] Eike "Un!t": The ANA stuff doesn't really do anything about the connection itself. It's what it does is I must disclaim. I have no idea how GPS works. I just did the API code and how the communication with the M10 unit.
[2643.2] Stephen: Yeah, I mean, what I'm saying is I've got a similar unit on other quads and it works fine on this one. It's just driving me a little crazy.
[2651.2] Stephen: But but I did find on a different quad when I added those lines in to enable the ANA feature instead of waiting for a couple of minutes, it was just right away.
[2660.2] Eike "Un!t": Yeah, that's pretty insane. It's the anonymous mode, which basically means that the GPS is building up the the almanac on its own and pretty fast.
[2672.2] Eike "Un!t": So the normal normal way is it would wait for the satellite and then pull data, pull data, pull data and then building up slowly, slowly, slowly.
[2682.2] Eike "Un!t": And with the anonymous mode, I have no idea why and how, but that that process has done way, way faster. And yeah, that's actually the reason why the M10 units and some M8s are working so, so well.
[2699.2] Eike "Un!t": In my opinion, we should not do this or run the GPS unit without the ANA feature.
[2705.2] Stephen: So I guess the outcome of your work around GPS really has been the kind of accelerated support for M10 GPS, right?
[2713.2] Eike "Un!t": Yeah, we fully support the full M10 API. So what got this started is that setting up an M10 unit on beta flight was just a mess.
[2726.2] Eike "Un!t": You have to go to your center, do some stuff there, then write it into flash, then put it back to beta flight and then set the board rate correct, which is another part because the board rate setting is a mess in beta flight.
[2739.2] Eike "Un!t": Normally you would say, okay, when I set it to auto in the reports tab, then it would auto-detect the current board rate and then set it and then it goes on.
[2752.2] Eike "Un!t": But it actually does the code on the GPS units. It's setting the board rate you want.
[2760.2] Eike "Un!t": So when you set it to 1155, then it will detect this body, which is the cool though, I think default on M10 is 38200 and then it would detect that board rate and then send the code so it would run on 1152 and then would go on.
[2779.2] Eike "Un!t": And that's so counterintuitive because we also had a switch which got removed because I simply commented out the code was the auto board rate, which could get you to think that it then would auto-detect or auto maybe highest possible board rate or something.
[2800.2] Eike "Un!t": I don't know. What it actually does is when you lose connection to the GPS unit in flight, it reduces or it increments the board rate index and then it would set that new board rate.
[2812.2] Eike "Un!t": So in the worst case, you are on 1155 and after auto board rate hits, you are on 9600 board rate, which is not possible to run.
[2822.2] Eike "Un!t": So confusing.
[2826.2] Stephen: I think there was a phase where like Chris Thompson was pretty much the only person who really understood GPS and he was kind of slogging away and then you've joined him.
[2837.2] Stephen: And between the two of you, it's gone from being this complete black box, like everyone's scared of GPS to, oh, wow, my quad just landed at my feet.
[2849.2] Eike "Un!t": That's his part. I just did that the units behave well and we can auto-config M10 units.
[2857.2] Stephen: I think you got the project out for scrape because people had actually started selling quads with M10 GPS before they were confirmed as working, right?
[2865.2] Eike "Un!t": That was pretty hilarious because nobody on the beta flight development team have ever seen an M10 unit.
[2874.2] Stephen: You've got to love manufacturer relations like this.
[2877.2] Stephen: People start selling stuff and you're like, wait, we don't support that. That doesn't work. And they're like, no, it's good. It's in the mail.
[2885.2] Stephen: We sent it to 10,000 people. It's going to be fine.
[2888.2] Eike "Un!t": Yeah, then then 5,000 or so 10,000 come to us and say this doesn't work.
[2893.2] Eike "Un!t": And we have to explain that.
[2895.2] Eike "Un!t": Yeah, you're lucky if it works because some of the M10 units are backwards compatible or work backwards compatible.
[2902.2] Eike "Un!t": And that's why some of the commands works, but not all.
[2907.2] Eike "Un!t": And then you get a unit which is not probably contributed because not all units are flash or the BBR, the little...
[2916.2] Stephen: I guess Alexander's pointing out he's dealing with a very small sample size from the ultra flyers, obviously.
[2921.2] Stephen: So he just gets lucky, I guess.
[2923.2] Curry: Yeah. So when was support for M10 because I was just messing with one today and again, I didn't have to do anything.
[2930.2] Curry: It just worked. So I wanted to test it.
[2933.2] Eike "Un!t": Oh, that's... I don't know. I would say half a year prior.
[2937.2] Curry: It's probably one of the four or four somethings.
[2941.2] Eike "Un!t": Actually, I developed it on 4.5, so it got back ported. So, yeah.
[2948.2] Eike "Un!t": Okay, cool.
[2949.2] Eike "Un!t": Oh, yeah. And 10s are the best. It's insane. I got one big GPS unit from Fox here.
[2958.2] Eike "Un!t": I don't know for how long, but it must have been there for years or so.
[2963.2] Eike "Un!t": I don't know how long, but I plugged it in.
[2965.2] Eike "Un!t": And the first time it was a really beefy one. It was really big.
[2969.2] Eike "Un!t": And I plugged it in and I got a lock, like 10 satellites in 30 seconds. It was insane.
[2976.2] Curry: Oh, I was testing this today. I had my curious setup.
[2981.2] Curry: I'll show you, actually.
[2984.2] Curry: It's almost like I've got... This is not redundant GPSs.
[2988.2] Curry: It's like my little test rig. So I could go outside, plug one in, test it and see how long it took
[2994.2] Curry: to get some satellites and then unplug and plug that one in and retest it.
[2998.2] Curry: And, yeah, it's faster. And I've got a particularly bad place for GPS in my garden.
[3004.2] Curry: So the MA got up to eight in five minutes, which is amazing.
[3010.2] Curry: Normally it gets like five, but the M10 got 12 in about three minutes
[3015.2] Curry: and it got a very fast lock and it was a complete cold start straight out of the box.
[3020.2] Eike "Un!t": Yeah, that's really... One of my best M10 units I was developing on
[3025.2] Eike "Un!t": and it was sitting near me in my room behind concrete walls.
[3030.2] Eike "Un!t": And after 10 minutes, I got 32 satellites inside.
[3035.2] Curry: I didn't get any inside. I've got a magic roof.
[3037.2] Curry: No GPS signal can penetrate it, would you?
[3041.2] Eike "Un!t": Yeah, well, I mean, initially when I started the development,
[3045.2] Eike "Un!t": I was always putting the GPS unit near my window because I wanted to have some...
[3050.2] Eike "Un!t": Yeah, then when I got into it and the commands are flowing and everything works.
[3055.2] Eike "Un!t": Yeah, I just could have stood it near me because it just works.
[3059.2] Eike "Un!t": Cool.
[3061.2] Eike "Un!t": Yeah, that was fun. That was really fun.
[3063.2] Eike "Un!t": I had quite some calls with Chris Thompson.
[3066.2] Eike "Un!t": We chatted back and forth and yeah, that was awesome.
[3071.2] Eike "Un!t": That was what a really cool collaboration and I think we did a good job because...
[3077.2] Eike "Un!t": Yeah, it's nice to hear that someone just plugs it in and it just works.
[3083.2] Curry: Just worked. If I have to test it again because I forgot to turn on...
[3087.2] Curry: What's it called?
[3089.2] Curry: One of the things that was written on the back and it wasn't switched on.
[3094.2] Curry: I thought, I'm going to have to test it again.
[3097.2] Curry: GLONASS. I noticed I didn't... I forgot to turn it on and be applied,
[3102.2] Curry: so I'll retest it and see if it makes a difference.
[3106.2] Eike "Un!t": Actually, you have the Chinese satellites, the Russian satellites, the USA,
[3113.2] Eike "Un!t": and that's just enabled you to just increase the number of possible satellites.
[3119.2] Eike "Un!t": Yeah.
[3121.2] Eike "Un!t": That's pretty funny and I have no idea why.
[3124.2] Eike "Un!t": I have one GPS and 10 GPS unit from iFlight
[3128.2] Eike "Un!t": and one can enable all satellites simultaneously and one can.
[3134.2] Eike "Un!t": I have no idea why.
[3136.2] Stephen: I think what you can't do with these U-Blocks units is flash them to a newer firmware, right?
[3142.2] Stephen: Oh yeah, no, no. You are stuck on what you got.
[3145.2] Stephen: Exactly, so you've probably got one version of U-Blocks firmware on one unit
[3149.2] Stephen: and then two weeks later, a new version of U-Blocks firmware.
[3153.2] Stephen: Yeah, that's true.
[3155.2] Stephen: That's kind of a problem with that. They're nice, they work really well,
[3159.2] Stephen: but it's very much controlled and locked down firmware.
[3163.2] Stephen: I don't think we'll probably ever see an open source GPS, so just have to work around it.
[3168.2] Eike "Un!t": I mean, they work well, so if we have some limitations, it's acceptable, I guess.
[3174.2] Stephen: And it looks like the NMEA stuff is probably going away soon,
[3177.2] Stephen: which is a little bit sad because I've got a few quads with them, but they never work that well.
[3182.2] Eike "Un!t": Yeah, I'm glad that I only had to look at the U-Blocks code because...
[3188.2] Eike "Un!t": Yeah, and Mia is...
[3192.2] Stephen: It's old and confusing and they're all binary messages instead of easy-to-read stuff.
[3200.2] Eike "Un!t": Yeah, it works just fine, you're right, but it's not good.
[3206.2] Stephen: And it's like one of these standards where there's a main standard
[3209.2] Stephen: and then there's 101 different manufacturer-specific pieces on it as well, I think.
[3216.2] Eike "Un!t": I can't say neither.
[3219.2] Curry: I thought the NMEA stuff was quite human-readable.
[3222.2] Curry: Here's some ASCII text, but you have to decode it to determine what it's doing.
[3226.2] Curry: Although if you have to debug it, you can at least look at it and say,
[3229.2] Curry: oh, right, this is how it works.
[3232.2] Eike "Un!t": I looked at it once at the code and it's weird.
[3238.2] Eike "Un!t": I don't know, if you take U-Blocks, it has a pattern, it has bytecodes, you can decode it.
[3243.2] Eike "Un!t": I don't know, it all makes sense.
[3245.2] Eike "Un!t": And for NMEA, it doesn't make sense.
[3247.2] Eike "Un!t": Maybe it's just NMEA, it didn't make sense to me.
[3250.2] Stephen: NMEA is like a very old standard and people have tried to improve it,
[3254.2] Stephen: but the standard itself never got updated after the 1980s or early 1990s.
[3260.2] Stephen: So a lot of the things that make it better only work for a handful of units
[3265.2] Stephen: and the rest of them don't work.
[3268.2] Stephen: I've got these ATGM336H that are budget NMEA units and they worked okay,
[3275.2] Stephen: but they relied on some fixes that only work on that one model of GPS
[3280.2] Stephen: and don't work for any of the rest of them, which made them.
[3283.2] Stephen: But it's kind of a risk putting a fix in like that
[3286.2] Stephen: because you're just blindly assuming everyone's gone and bought that model.
[3289.2] Stephen: And if they haven't, what does it do?
[3292.2] Stephen: Undefined, which isn't great.
[3295.2] Eike "Un!t": Yeah, that's not something you want in a unit that you rely your quad on.
[3302.2] Stephen: So we did have a formation flight on a while ago and I was talking to them
[3309.2] Stephen: and people appear to have kind of started to put more emphasis
[3314.2] Stephen: on the iNav MSPv2 sensors for GPS.
[3318.2] Stephen: That's something I was chatting on Discord earlier today.
[3321.2] Stephen: And I'm not sure if, do you think that might happen in 4.6?
[3325.2] Stephen: We're going to see like MSP capable GPS or is that not really interesting?
[3330.2] Eike "Un!t": Just recently, I can't really say much about it.
[3335.2] Eike "Un!t": So I won't confirm or deny anything
[3338.2] Eike "Un!t": because I really don't know.
[3340.2] Stephen: No, no, I wasn't looking for a hard commitment.
[3342.2] Stephen: I'm just wondering if it's an interesting thing, if it's interesting to look at or not.
[3348.2] Eike "Un!t": Yes-ish.
[3350.2] Eike "Un!t": If we take the Ubox code, if you go in and read it,
[3356.2] Eike "Un!t": it's still not good because we, I mean, what would say good, it's limited
[3363.2] Eike "Un!t": because if you take an F7 and you want perfect GPS rescue,
[3372.2] Eike "Un!t": you have to run 4K loop because the GPS task is taking too long.
[3379.2] Eike "Un!t": So if you run 8K, 8K, it would mess, sometimes mess up with the pit loop
[3385.2] Eike "Un!t": and you don't want that.
[3387.2] Eike "Un!t": So you have to run 4K, 8K, 4K.
[3390.2] Eike "Un!t": This is why you need an H7 problem solve.
[3393.2] Eike "Un!t": Yeah, sort of.
[3395.2] Eike "Un!t": But that's why the code isn't perfect.
[3399.2] Eike "Un!t": So if we go and do GPS over MSP, this might be a solution to this performance issue.
[3409.2] Stephen: Because that's more, you push the GPS coordinates
[3412.2] Stephen: rather than having to take all the GPS data and manually translate it yourself.
[3417.2] Stephen: You can get coordinates in the format that you want them.
[3421.2] Stephen: We'll see.
[3423.2] Stephen: I've promised to test a 500 hertz GPS task firmware that Chris Thompson did
[3430.2] Stephen: as a response to me asking stupid questions a while ago.
[3434.2] Stephen: And yeah, I'm still on my list.
[3436.2] Stephen: I still need to go and do that to make sure it works or doesn't work.
[3440.2] Eike "Un!t": I compiled a one kilohertz GPS task version and it doesn't work very well.
[3448.2] Eike "Un!t": I wanted to answer to a federal commander who said, yeah, a 1K lumen works,
[3455.2] Eike "Un!t": but nobody wants to fly on 1K lube.
[3459.2] Stephen: Yeah, we have people saying, why can't we run at like 32K?
[3463.2] Stephen: And the reason for that is, well, you get more noise from the gyro than useful signal.
[3467.2] Stephen: But 8K, you get more signal than you do at 1K.
[3472.2] Eike "Un!t": But I would say that we have nearly a perfect experience on 4K.
[3482.2] Eike "Un!t": So having to reduce it on 4K to rely on GPS rescue is something I would do every time.
[3489.2] Eike "Un!t": So that's nothing I would do.
[3491.2] Eike "Un!t": I would get the H7 and just try and make it work.
[3496.2] Eike "Un!t": There are other issues with H7.
[3500.2] Eike "Un!t": So I'm flying G4s and F7s mostly.
[3509.2] Eike "Un!t": The G4 is an amazing MCU.
[3512.2] Stephen: Why G4 and not F4?
[3515.2] Eike "Un!t": Because the G4 is more or less the successor to the F411.
[3520.2] Eike "Un!t": It has no DMA park.
[3522.2] Eike "Un!t": It has plenty of AO.
[3525.2] Eike "Un!t": So we don't have to use soft serial.
[3527.2] Eike "Un!t": And also it's a low.
[3532.2] Eike "Un!t": What's the name?
[3533.2] Eike "Un!t": It doesn't get hot.
[3535.2] Eike "Un!t": It's really, really cold.
[3536.2] Eike "Un!t": So when you compare the G4 to an F7 or even an F4,
[3544.2] Eike "Un!t": the G4 idles at like 40 degrees.
[3547.2] Eike "Un!t": And the F7 reaches 50, 55, 60, pretty easy on idle.
[3552.2] Eike "Un!t": And what heat does is it increases noise.
[3557.2] Eike "Un!t": And if you have all of the cooler board, it will improve flight time.
[3562.2] Eike "Un!t": One of my favorite AROs is the Phoenix G4 AO from Airboat.
[3569.2] Eike "Un!t": And that's only because it's just running really, really clean.
[3575.2] Eike "Un!t": I had neither issue with anything.
[3579.2] Eike "Un!t": I have tested the gyro noise on my oscilloscope and it's perfect.
[3584.2] Eike "Un!t": It's just running.
[3586.2] Eike "Un!t": And even if it's getting hot, there's nothing that's inducing any noise anywhere.
[3590.2] Stephen: That's interesting.
[3591.2] Stephen: I guess I was kind of a snob.
[3593.2] Stephen: And I thought kind of looking at the chip lineup, just looking at the specs,
[3598.2] Stephen: I was like, oh, people are just using G4 because they can't get this other part.
[3602.2] Stephen: So that's why they're using it.
[3604.2] Stephen: And it's a more budget MCU.
[3606.2] Stephen: So yeah, I was kind of a snob about it.
[3609.2] Eike "Un!t": I think at first, that's actually not false.
[3614.2] Eike "Un!t": It's true.
[3615.2] Eike "Un!t": But then when you get to use it and test it, there are actually a lot of benefits.
[3620.2] Eike "Un!t": So if I want to, I must choose, I would always go for an G4 or an F7.
[3629.2] Eike "Un!t": If you need more IO, you have to go for the F405.
[3632.2] Eike "Un!t": But when you don't need that, you can easily go with the G4 because it's running 8K 4K
[3640.2] Eike "Un!t": on bi-directional D-shot pretty, pretty decent.
[3643.2] Stephen: Is that the one where we had to enable all these low power UARTs to get more IO on the G4?
[3650.2] Eike "Un!t": Yes, that's exactly that one.
[3651.2] Eike "Un!t": Okay, cool.
[3653.2] Eike "Un!t": Yeah, okay.
[3654.2] Eike "Un!t": I actually have to go.
[3657.2] Eike "Un!t": No problem.
[3658.2] Stephen: I think we're getting towards the end of the stream anyway.
[3660.2] Stephen: So if people want to find the work that you've been doing, where can they find you?
[3666.2] Eike "Un!t": On the Betaflight Discord or on the best drone community Discord ever out there, drone community.
[3674.2] Eike "Un!t": That's where my journey started.
[3676.2] Eike "Un!t": So that's why I wanted to call them.
[3679.2] Eike "Un!t": Send us links.
[3680.2] Stephen: We'll put them in the show notes.
[3682.2] Stephen: Thanks very much.
[3683.2] Stephen: Thanks for joining us.
[3684.2] Stephen: Unit, aka IKEA.
[3687.2] Eike "Un!t": Yes, IKEA, yeah.
[3689.2] Eike "Un!t": Thank you very much.
[3690.2] Eike "Un!t": Thanks for being here.
[3691.2] Eike "Un!t": Thanks for the invitation.
[3693.2] Eike "Un!t": Yeah, awesome to be here.
[3695.2] Eike "Un!t": And I hope it was informative.
[3698.2] Eike "Un!t": Great.
[3699.2] Eike "Un!t": You got to.
[3700.2] Stephen: I think we're all wiser and smarter now.
[3702.2] Stephen: So yeah, excellent.
[3703.2] Stephen: Awesome.
[3704.2] Stephen: Thank you.
[3705.2] Stephen: Right.
[3706.2] Stephen: You have been joined by everyone's favorite, Curry Kitten.
[3709.2] Stephen: Goodbye.
[3710.2] Stephen: Jack, aka Bright Until I Fly.
[3713.2] Jack: Thank you, guys.
[3714.2] Jack: We can hear you.
[3715.2] Jack: Thank you, Jack.
[3716.2] Stephen: We can hear you.
[3717.2] Stephen: Sorry we couldn't hear you earlier.
[3718.2] Stephen: No, it's my fault.
[3719.2] Stephen: Sorry.
[3720.2] Stephen: And the man who's probably got to dash soon.
[3723.2] Stephen: Thank you very much for your time.
[3725.2] Stephen: Unit, cheers.
[3727.2] Eike "Un!t": Thank you for inviting me here.
[3729.2] Stephen: Oh, and we should thank all our wonderful Patreons.
[3731.2] Stephen: Please support the show.
[3732.2] Stephen: Keep this train wreck going.
[3734.2] Stephen: You're wonderful.
[3735.2] Stephen: You're awesome.
[3736.2] Stephen: Thank you, Patreons.
[3737.2] Stephen: Telemetry lost.
[3739.2] UNKNOWN: Thank you.