Quirks and oddities

[LuLuJo]

I have been using the new HM7000 system since its launch as a complement to my Railmaster dcc controlled layout. I bought one decoder and power bank on pre order to give it a try.

It's been far more useful than I imagined and so I now have 10 installed decoders all with power banks in a mixture of Hornby & Bachmann Locos and using 3 IOS devices( Iphone 6S, Iphone 11 and a 5 year old Ipad). I sometimes use all 3 devices at the same time. All operate happily with the dcc system.

Overall I am very happy with the system and its now my standard dcc decoder solution. Its very reliable but there are one or two issues I feel worth feeding back.

I have observed that:

The first installed decoders seem to have the most reliable bluetooth connection and connect first. Is there a set order that locos connect in?

That the most recently installed decoders have the least reliable connection.

That with use the decoders connection seems to become more reliable.

That the older locos with higher motor current needs and a power bank will often trip the decoder on first starting and once that's happened often will not reconnect. My solution to trickle charge the power bank with one press of the + key will usually work but needs several minutes to be effective. With one loco that's not so much of a problem but now with 5 its more of a pain and loses the switch on an go benefit.

I have had a clear example of the Ipad connecting to two locos not installed on it and operating them. I actioned sound and motion on a Stanier 2-6-4T and Hardwick installed on the Ipad but connected to two Fowler 2-6-4T's installed on the iphone and they responded. Supposedly this cant happen but it did. Is there an issue with a limit on the number of control devices?

Any comments welcome.

[Fishmanoz]

That’s great feedback LLJ.

Rob, making sure you see the last para.

[96RAF]

A lot of observations there, which the test team have seen and are up for fettling.

1. Decoder priority - static decoders are connected to first - I.E. HM6K modules or HM7040 dongle. These establish a solid node base for the mobile decoder nodes to connect with and thus expand the mesh reliability. Each node talks to the next and the next and so on to increase the mesh range and to help a lost soul back into mesh after say a derail beyond the capacity of the power bank to cover until re-railed. If you have no static nodes then more mobile nodes are better than a single loco. Polling of nodes is staggered to prevent a bottleneck in the app processing.
2. Power bank inrush stalling. The power bank is defaulted to charging only on the move, not stationary. This was introduced to cater for expected in-rush of the new steam generator, but with hindsight the current thinking is to provide an option to set it to stationary or mobile charging by way of a CV setting. Such a fix would be distributed by an OTA firmware update to the decoder from the app.
3. Multiple decoder connections to various control devices. This is more a local management issue at user level than a bug. If you have all decoders on track and powered up, they looking for a valid controller and the first control device to see them will grab any decoders registered to it previously - I.E active in their engine shed, even though they are not truly linked to that decoder and thus controllable by that control device. This unfortunately takes true control away from the prime control device coming on line afterwards . The answer is to ensure that any such decoders in the engine shed of a non prime control device are made inactive, then their prime control device will be able to pick them up and control them. In the event a first up control device picks up an inactive device in error, simply click refresh host on the non-prime device to release control. A bit difficult to describe but easier to see in practice.

Fettles in the test app versions will be made public in the iOS app hot update which is likely to coincide with the Android app launch, and that is getting very close.

[LuLuJo]

Thanks for the explanations which are helpful.

In point 1 is it the case that the sequence of polling of mobile nodes is based on past history as it certainly appears that the first two decoders I added to the system always connect first and in the order they were added?

In point 2 I welcome the idea that a fix will arrive in due course as it will become an increasing problem as the decoders spread across the market place. I have two Clans. The new one works fine all the time. The older one always has this problem.

The only answer which does not fit my experience is No. 3.

In this instance the ipad was the only device in use and the decoders it grabbed have never been used on that device and are not or ever have been in its engine shed. I have tried to recreate the circumstances once but without a repeat. I will try again and feed back if it occurs again.

Mike

[SteveM6]

I’ll take option 2 right now Rob.

[Wayne-in-Oz]

I’ll take option 2 right now Rob.

 

 

Me too :)

[96RAF]

@LLJ

I have no idea if the polling order is fixed by way say of the order the decoders were presented to the app or if it is random. Maybe a question for the techies.

Presumably it is picking these strangers up in a scan for other decoders to be linked to your iPad, and showing them as ready to reset, if so just ignore them. If not how are these other decoders on your system under control. If they are not then where are they. You cannot control a decoder on a new control device unless it has gone through due process of reset, link to and assign to a loco. More info would be good, say a step by step with screen grabs.

@steve & WiO

Yes the initial stalling irritates me especially if I am running an automated sequence from a program. I would like to see the option as a priority but getting the android app out is top of the list at present.

[SteveM6]

Yes the initial stalling irritates me especially if I am running an automated sequence from a program. I would like to see the option as a priority but getting the android app out is top of the list at present.

 

 

Couldn’t agree more Rob. The flurry of small updates we have seen in the last few days in testing have addressed many of the issues we have seen in testing. It certainly feels very robust now and appears to work well across a variety of devices.

Its definitely getting close.

From my list of required changes only one remains outstanding as you know. In my opinion that one is a ‘need to’ fix whereas the powerbank change is a ‘nice to’ amendment.

[96RAF]

@LLJ

Polling - Hornby says initial polling for the host is in order of decoder listing to the app, thereafter it picks them up by way of strongest signal strength first, thus that is dependant upon where the decoder is in distance from the app control device. Work continues to improve polling which could become problematic if there are hundreds of nodes to trawl through, albeit the process is in mS.

@All

One of the ‘fun’ bits of being on a test team is when something that was working pretty fine for days suddenly crashes out for no apparent reason, dashing hopes of an early release, then as more reports come in the problem escalates in importance from random glitch (finger trouble) to oh-dear (or similar expletive) and a rapid fix has to be found and deployed. Rather that the test team found these things out than gen-pop when they get it. Fingers crossed we are in good shape now.

[Moccasin]

@96RAF & @SteveM6 Sincere thanks to you and the rest of the testing team for all your hard work to help get the Android app ready for release and to further improve the iOS/iPadOS app.

I know you'll all be fed up with the "strident comments" here and in some even less forgiving modelling forums. It probably feels like a thankless task at times but am pleased to hear that you're feeling closer to a stable release. It really is a massive step forward, whatever sceptics and naysayers might say and I know many of us here are looking forward to the system's further development.

[SteveM6]

@Moccasin thee are many on the test team that have done much more in-depth testing than I have, particularly on iOS in the early stages (I did not have a compatible Apple device). Much of the hard work that they did at that stage, with a few of us as spectators, has carried over to the Android version.

When we were given a test version of Android, I approached it as if I were a total newbie (not too difficult) who hadn’t read the instructions as I figured it would be easier to uncover issues that the average user may encounter. All I can say is that at that level it worked first time, straight out of the box leaving me mightily impressed. Even the previously identified issues with some of the sound profiles failing to load did not show themselves. It’s only when you really start digging into the minutiae that the ‘quirks and foibles’ reveal themselves.

That’s not to say I went into testing ‘blind’ - I had taken note of some of the issues raised both on here and within the testing group and took care to eliminate or mitigate many of the variables that seemed to be causing issues.

For example, I made sure that I had the strongest possible WiFi signal in the layout room, I used a new tablet with no extraneous apps (to avoid interruption by notifications). I also had a dedicated programme track plugged in to the new HM7K PSU, spotlessly clean track, wheels and pickups. Once the initial programming was done, the locos were let loose on the layout which is powered from a 4amp Elite (ready for future dongle testing).

In some respects that’s the best advice I can give - read and understand what has been said on the various forums, eliminate variables and follow the recommendations given by Hornby from the very start.

[LuLuJo]

Update on from the issue of controlling locos not set up on a device. I have been able to recreate what happened and for a while control locos from two devices at the same time.

The set up is as follows. On my iphone I have 2 fowler 2-6-4T's 42314 decoder 6DD7 and 42355 decoder 6E02

On my ipad Stanier 2-6-4T decoder 652B and 790 Hardwicke decoder CF0B

I switched power to the whole layout which would access all the 10 locos except 42464 which was off the layout. I switched on the ipad and launched the app. No other controller was live.

Pressing the controls for 42464 controlled loco 42314. Sound and motion. Pressing the controls for 790 controlled 42355 with sound and motion. The screens for 42464 and 790 showed them linked to their own decoders. I ran them up and down for several minutes.

I then launched the app on the iphone. I could control both fowler tanks from their normal controls and they showed as connected. I then ran both locos using both iphone and ipad together with alternately commands from each device for several minutes.

I took screenshots at various points if they are of interest.

I decided to video what was going on and but by the time i had got set up suddenly normal operation was resumed. Hardwicke which was on the layout throughout then responded to the ipad rather than sitting quietly whilst I ran 42355.

It was interesting being able to control an engine from 2 devices. However think for clarity one control device is better.

I only found this issue because its my practice to sound the whistle of each engine when the app is launched and suddenly they came from the wrong engines. It would sometimes be unfortunate though to set off the wrong loco. I expect this glitch is rare and I can easily get round it. I also feel its likely I can recreate it.

If any more information or the screenshots are useful let me know.

[96RAF]

Document the whole thing again please and report with screen grabs and video to George at HM.customercare@hornby.com making sure you tell of which devices app is loaded on and their rev states, etc.

I cannot simulate the problem as I can only use my iPAD Pro until the iPAD Air fettle is enabled, else it is a mix of iOS and Android and that mix does not repeat the fault.

[Fishmanoz]

My suspicion is it’s not the app as such, rather an iOS feature that synchs between multiple Apple devices.

I’ll see if I can reproduce it. I only have one decoder and it’s linked to the app on my iPad. I’ll try installing the app on my iPhone, try that sequence and see what happens.

[LuLuJo]

I have emailed George as requested.

[96RAF]

He is on it and may ask further info direct peer to peer.

[LuLuJo]

Update on the transfer of control between devices. I have tried several times to recreate this without success until this afternoon. I had a variation on the experiences so far. I switched on the ipad first as usual There were 11 devices on the layout and I blew 790 Hardwicks whistle and instead 42314 replied. So with three of the locomotives allocated to my ipad I was able to control locomotives linked to my iphone So 42462 ran 42355, 790 ran 42314 and for the first time 45697 ran 72006. Then I switched on the iphone and 42314 ran 790. I took around 9 minutes of video and some stills so overall the oddity lasted for well over 10 minutes before suddenly normal operation resumed.

Watching it back I realise I should have narrated the video more to explain what I was doing and what was happening. I will pass this all onto George. I find this all much more interesting than irritating.

[RB51]

Good to hear you are taking it in good heart LuLuJo. Information of this sort is gold dust to the developers. R-

[96RAF]

It makes me wonder if the unique MAC used by the app has similar foibles to the cross addressing seen in DCC in the past before HM7K whereby multiples of 5 had proven difficult.

[Peachy]

My suspicion is it’s not the app as such, rather an iOS feature that synchs between multiple Apple devices.

I’ll see if I can reproduce it. I only have one decoder and it’s linked to the app on my iPad. I’ll try installing the app on my iPhone, try that sequence and see what happens.

 

 

I think you might be correct @Fishmanoz

When I was having issues downloading profiles on my iPad Pro I wanted to eliminate any possible interference so went back to my iPhone and opened up the app. I was still logged on and 2 locos that had never been installed were showing (when I say showing it was just the name and address # but no photo or profile).

So I logged out & deleted the app from my phone.

The very next profile that I loaded onto the iPad worked first time.

[LuLuJo]

Hopefully a final update on this issue of the uncontrolled swapping of decoders between IOS devices and control of one decoder by two IOS devices at the same time.

These issues have continued and gradually worsened until 10 out of the 12 decoders and all 3 IOS devices were impacted. There was a pattern of which locos paired together . It also latterly happened every time the system was switched on. I have been feeding information to George taking photos and video and by agreement making no real attempt to solve the problem so that there was a test bed available.

Then last Friday my IOS phone11 was updated to the just released latest IOS operating system from 16.5 to 16.5.1 so I updated the Ipad as well to 16.5.1 and the Iphone 6S to the latest available version 15.7.7

The problem has apparently now gone. The last 10 attempts to load up all have worked as they should. Hopefully it wont return although I feel a little sad we have no idea why having spent a lot of time on this.

My experience shows the wisdom of one control device linked to a decoder. My TXS part of the system was effectively unusable when you don't know which loco is being controlled by which device.

I think that I will now rationalise it so that I have only 2 IOS devices covering all the decoders. This should give a stronger connection and simplify knowing what loco is on which control device.

[Fishmanoz]

You actually get a stronger mesh if you are using more devices. Then given the crosstalk has been banished for good, you can spread your locos around as you see fit.

[LuLuJo]

Do you mean decoders or IOS devices?

My impression was that the more decoders on a mesh strengthened it so 6 decoders on two IOS devices would be stronger than 4 decoders on 3 IOS devices

[SteveM6]

The system is designed to create a mesh between the host device (singular), connected decoders and, in time, a dongle.

[LuLuJo]

But that does not answer the question is fewer meshes with more devices better than more meshes with fewer devices?