SlotForum banner
1 - 20 of 133 Posts

·
Administrator
Joined
·
10,264 Posts
Discussion Starter · #1 ·
OK, I've been head down on this for a while trying to bottom the issue.
Here are my findings, and a solution.....

To start with lets dispell a few myths:

Axle proximity has nothing to do with this. I've tried with the chip on the roof and it still happens. Equally proximity to the track has no effect.

Missing ferrite man is not the cause. My best "faulting" car has both in place at guide and motor.

Powerbase is not the cause, I've had this with 4PB, stock V1.5 6PB, and PB-Pro-SHA

So a few thousands of laps later what have I found?

With no LC in the circuit you have no problem. So we do have an electrical noise problem. Trials with the chip isolated shows it is not an inductive effect from axle, passing LC coils etc, it has to be noise that is conducted up the wires, i.e. picked up from the rails.

First thoughts were that a corrupt data packet causes the runaway or stopping. As the occurance of the fault always happened immediately after crossing an LC I thought it was possibly the break in the current at some critical point in the read sequence. Or the other thought was that the processor was getting a reboot and then defaulting to zero or full throttle.

Wrong.

I lashed up a DPR with a 470uF capacitor across the 5V supply to the chip.


That made a bit of difference to the time to the fault, but it would still happen. And reset would take a 5 second lift off the track. So we can be sure that the processor still had juice and didn't brown out.

A discussion with Adrian Norman pointed towards the XLC exit being part of the issue. Sure enough the braids on my Porsche 997 were quite spread. I pulled them in close to the guide. Magic! problem disappeared completely.

Now as I had a PB-Pro SH I had no worries about shorts so I splayed the braids out wider and ensured a good short on every passing. Enough to dim the display. I could now get very consistent faulting behaviour from my car. With the extra capacitor mounted it allowed the car to run for 20 laps rather than 6, but it still only postponed the inevitable.

So where is the problem? On the exit of the LC is a place where the joining lane rail gets very close (about a mm). Solution? Insulate the last 5mm of the rail by painting with a black enamel paint at the positions I have marked in red on the photo. Or keep your braids very very tidy.



Riko
 

·
Mike Calcutt
Joined
·
452 Posts
Riko,

I'll let everyone else test this before I buy another one, but great news if you have indeed cracked it...Well done mate.

be interested to know what Adrian makes of your findings.

Mike
 

·
Greg Gaub
Joined
·
15,795 Posts
Awesome! I feel MUCH better about buying some DPR cars now. I can solder OK, but if I can avoid it, I sure will. I'd much rather put some paint or tape on those rail points.
 

·
Premium Member
Joined
·
1,165 Posts
Magnificent Riko, thanks for all the hard work.


Collected this thread into SSD 101 already, hope everyone will post their results. I'll order in a DPR tomorrow and have a go myself, the Chapparal has just been waiting. Mike's experience scared me off DPRs completely!


Similar problem on the clcs?

Paul
Circuit TrustChrist
 

·
Martin Kay
Joined
·
728 Posts
Nice work Riko, am relaying my track at the moment so a golden opportunity to try out your suggested fix. A customer (separate from this issue) recently asked if he could just generally prolong the life of braids (particularly badly affected by drifting) by soldering the trailing end, and I wonder if this might not also work for this. Did you try with SCX copper braids, they tend to splay out less and might also be a good solution.
 

·
Premium Member
Joined
·
4,882 Posts
But....and it's a big but as zipp has asked, why does the DPR chip act in this way and not the Saloon or F1 chip


That's the question I have tried to answer and failed on several times.


Rich
 

·
Banned
Joined
·
8,038 Posts
Nice work Riko,
Do you think more likely hardware or firmware Rich? Interesting problem. Im assuming same circuitry has been used.
 

·
Premium Member
Joined
·
4,882 Posts
The "real" solution to this one has eluded me. Have on two occasions now spent several hours on it.

The boards are electrically identical. There is one small capacitor missing on the DPR board but fitting it makes no difference.

I even resorted to swapping processors between a Saloon & a DPR board. Again no difference so it's not firmware.

I have hacked about the DPR board to isolate the input & output wiring from the board again no difference. Remember that the F1 board has a track adjacency problem that makes it more susceptible to motor interference than the Saloon board.

So unless I am missing something that leaves a layout problem, that was not fixed by taking the input & output tracks off the board.

One day when I have nothing else to do, could be a long time, I will get to the bottom of it. I feel it must be something to do with the input signal routing that my hacking did not isolate fully?

Why on earth Hornby don't just put an engineer on it and sort it I do not know? The DPR boards rightly have a very bad name.
Not fixing it is leaving the whole DPR concept in jeopardy. Not because of the DPR concept which is sound, but because of some silly electrical problem.

So well done to Riko for the fix, but with respect it's not a solution. It's like have a make of car with very harsh suspension and fixing by making all the roads smoother.


Rich
 

·
Administrator
Joined
·
10,264 Posts
It could be either hardware or firmware, although a little birdy tells me they have not changed the firmware.

As Rich says, all the components are the same as on the F1 and saloon boards, the only real difference is the layout and the proximity of the various components and tracks on the PCB.

I would say that as the XLC can cause shorts it is that error that is the root of the problem, I'm not sure that any software or hardware changes will cope with the sparking type noise pulses that must shock the system. If anything it is surprising that the other chips do not suffer.

Anyway, I gave up with component changes when I identified the very "real" problem on the XLC and an easy fix.

This is one of those walk the talk things. Do the change. Believe the results......

Riko
 

·
Lars Ole
Joined
·
1,635 Posts
QUOTE (RikoRocket @ 2 Mar 2009, 11:20) <{POST_SNAPBACK}>This is one of those walk the talk things. Do the change. Believe the results......

Riko

still it's stunning to see it having been sent to market, don't tell me that they havn't experienced this flaw themselves !!!!
so much for inhouse testing :-(

cheers
 

·
Administrator
Joined
·
10,264 Posts
A pound to a penny they were just using CLCs in their test circuit.... Or their braids were brand new and tidy.

Nice simile Rich:

QUOTE (RichG @ 2 Mar 2009, 10:07) <{POST_SNAPBACK}>So well done to Riko for the fix, but with respect it's not a solution. It's like have a make of car with very harsh suspension and fixing by making all the roads smoother.


A bit like taking GT-R down a potholed road? And the solution? Don't go down a potholed road!
But seriously they could not have designed the system to cope with shorts like this. It probably takes out V1.2 6PBs as well. I would say this solution is the same as filling in the potholes - they were never intended to be there....


The important thing is ..... it works


Riko
 

·
Premium Member
Joined
·
4,882 Posts
I am sure that changing track pieces that cause braid shorts is a good thing all round. Remember in the early days with a 1.2 Powerbase the same problem existed on the CLC's and caused the Powerbase to overload. I fixed this with some Sellotape, without which the set would have been in the loft by Boxing Day.


However there is still a major problem / difference with the DPR chip. I assume you are running with the ferrite man on the slot guide? Try taking it off, you won't need to run over a CLC to see the car take off.

So yes making a change to reduce shorts / sparks is a good thing but the Saloon & F1 board cope well with them wheras the DPR chip does not. Must be worth finding the difference?

Rich
 

·
Administrator
Joined
·
10,264 Posts
Discussion Starter · #16 ·
Well, we know it is a conducted noise issue, so not having Mr Ferrite in place is only going to increase susceptibility. And the only thing left is to re-layout the board. Yes we can do this, but - hey - I want this working and if stopping the short is 100% effective that will do me.

OK as a solution technically it would be nicer if we were to say solder on an additional component, or reprogram the chips. But practically to the guys who won't want to go near the PCB with a soldering iron the option to rid the problem with a little bit of paint seems far more eloquent
 

·
Ian
Joined
·
99 Posts
That'll be me then!

You's guys really are "Crazy" in the nicest possible way, Riko how many laps did it take you to find this information
If it was me I'd have just stuck in a saloon chip and have done with it. I must admit it's a great find and I may even consider buying my first dpr chip some time this year, now you have sussed this, lets face it this fix won't do any harm to the other chips either win win !

See you soon
 

·
Administrator
Joined
·
10,264 Posts
Thanks Bleep, thousands of laps. Good job PB-Pro has pacer car setting, I could go into the lounge and watch TopGear whilst listening to the drone of the car going around and around.....

QUOTE (martinkay @ 2 Mar 2009, 08:13) <{POST_SNAPBACK}>Nice work Riko, am relaying my track at the moment so a golden opportunity to try out your suggested fix. A customer (separate from this issue) recently asked if he could just generally prolong the life of braids (particularly badly affected by drifting) by soldering the trailing end, and I wonder if this might not also work for this. Did you try with SCX copper braids, they tend to splay out less and might also be a good solution.

Hi Martin,

I'm not sure that would cut it. If you solder the ends you wont get the intimate contact that you need for current flow, it might lead to more severe noise and problematical behaviour. But I've not tried it so maybe I'm wrong???? I've only tested so far with standard Scalextric braid

The other thing to note is the DPR cars all have the braid plate guide set-up. If you look at this you see the braids are actually pitched quite wide compared to any conventional guide. You really need to angle the braids inwards to get them away from the danger area.
 

·
Premium Member
Joined
·
1,165 Posts
QUOTE (RikoRocket @ 2 Mar 2009, 12:53) <{POST_SNAPBACK}>A pound to a penny they were just using CLCs in their test circuit.... Or their braids were brand new and tidy.

Riko

Probably new braids. You got me curious, Riko, so I had a look back at the Scaley Nurenburg thread and their demo circuit did have an apparently unmodified xlc in it:



Got to wonder how much Scaley is just free riding on Rich, Riko, and other SF guys' engineering skill.
Oh well, that's my cynical lawyer side.

The neat thing for all of us is watching you guys work to make digital better!


Paul
Circuit TrustChrist
 

·
Registered
Joined
·
842 Posts
nice idea but dont believe its solution either
have not had problem since
taped man at front n back
while installing tube over front axle

as for other mystery
missed counts
wondering if position of track
screwed to base
can be adjusted to correct angle
 
1 - 20 of 133 Posts
Top