[halftone]

After its 2 month winter holiday on axle stands, my 147 alarm is playing up intermittently. The other day it was OK for almost 24hrs. Yesterday it went off twice so I locked the car on the key. This morning it it went off within 2 minutes of locking the car on the fob. Then again a few minutes later. Tedious. Every time, the reported cause was 5 flashes = tilt/movement. At least today it went off whilst I was watching it and was able to see there wasn't a passing vehicle or gust of wind, so it looks like an electronics issue.

There seem to be a lot of threads with similar, and a lot of uncertainty about where the tilt sensor is, and frustratingly none reach any resolution. Some say it's in the roof module along with the ultrasonic transceivers, others that it's in the main alarm/siren unit under the passenger wing, and others that it's a sensor unit bolted to the front suspension.

Since it's dead easy to pry the ultrasonic panel from the roof, I started there. I pulled the unit from its housing and removed the PCB. There is what looks like a single miniature non-mercury tilt switch. It's the little metal can about 10mm x 4mm dia, with 2 solder connections and siliconed to the board.

No matter how I positioned this, it was open circuit when tested with a meter, which of course a tilt switch shouldn't be.

As a test I pulled it away from the silicone and twisted it so that the leadout wires shorted each other, which meant it now is 0R regardless of position and stuck it back in the car. The alarm armed OK, now with a second beep that signals a system check failure. But I have had no further false alarms.

However, the existence of a check failure with the sensor shorted says to me that the system expects a resistance value xR (ie not ~0, with the sensor shorted as it is now) which it then takes as baseline. If R varies sufficiently, the system knows that tilt has occurred and the alarm goes off. I think it just can't cope with open circuit at all. That doesn't give an error beep, but the system has no reference R from which to determine tilt. Maybe any voltage change, eg due to temperature, will set off the alarm in that condition.

I am sure the sensor isn't a simple on/off switch, but gives a range of resistance values depending on its attitude. Of course I don't know the range. If I was more patient, I'd try substituting different values of resistor for the tilt sensor, and find out what the system doesn't think is an error, or patch in a variable resistor to figure exactly what is needed. But where's the fun in that? I've ordered one of these, which varies between 5R and 10M, and physically fits. http://www.ebay.co.uk/itm/121743222769 I'll follow up and say whether it worked or not. If it doesn't, I'll try different resistors to see what the system accepts as a non-fault, and forget about having a tilt sensor.

 

[edraven71]

Nice bit of detective work there.
If you do find a resolution should be pinned up in the stickies for future reference:thumbup:

 

[Egon]

nice to see someone getting into component level diagnostics.
I might have a spare one in the garage to compare R values with, if its any help?

 

[gwinter]

Are you sure that's a tilt sensor? Maybe I'm wrong but that part looks a bit like a crystal oscillator for the circuit to me..

 

[Paddy OPlastic]

Are you sure that's a tilt sensor? Maybe I'm wrong but that part looks a bit like a crystal oscillator for the circuit to me..

Or perhaps a capacitor.

In any case the tilt sensor is the white square thing at the top left of the board in the first picture.

 

[Paddy OPlastic]

I got out a magnifying glass and gwinter is correct - it's a 40MHz crystal oscillator.

The tilt sensor itself has perpendicular arrows labelled X and Y etched on the top which is a bit of a giveaway. It's an NS 25/B2 if you want to look it up.

 

[halftone]

I was thinking 'well, it's broken anyway'. I have a black ultrasonic unit, which come up rarely s/h, so initially took it apart to see whether the PCB could be swapped from the more common gray/ivory cases. It can. But to me it stood out as a tilt switch, similar to others I have come across, and the silicone mounting encouraged me to think so.


For now the alarm now activates without further false alarms, and disarms OK, which is an improvement, and no smoke has come out. However, simply unplugging the ultrasonic unit has the same effect.

Paddy is right. Further digging shows it's an NS25/B2 2-axis sensor mounted to the board by 10 soldered pins, DIL style. Turns out to be a an electrolytic-fluid based thing made by HL Planarteknik Gmbh, that relies on fluid moving between plates according to inclination. Whether this still works after 10 years is a moot point. Doesn't appear to be available anywhere now. Welcome to Willow Technologies lists it, but the datasheet says refer to the mfr. Likely superceded, NS25/E2 seems current. I might try and find out.

And what I thought was a tilt switch is a 40,000Hz crystal oscillator/resonator. Oops. I might just try replacing that, since the leadouts are crusty and are the only bit of corrosion on the board.

Still, at least I established that a PCB from a more common ivory or gray ultrasonic module can be transferred to my black casing easily, even if it will probably fail in the same way.

 

[halftone]

^^ Crossed in the post. At least we're saying the same thing.

 

[dgmayhead]

Does holding the lock key down for a long press close your car but not activate the alarm? This could make ownership more tolerable.

 

[halftone]

I've not tried, but that isn't mentioned in the 147 owner handbook. I can lock it using the key, which doesn't arm the alarm, or on the fob. With the ultrasonic module now unplugged I get a second 'beep' a few seconds later to say the system has noticed the missing module, but the rest of the alarm system is still active - eg if a door is opened, the alarm sounds. It disarms without issue. It's tolerable like that. I am not getting any false alarms, and it's better than no alarm.

The ultrasonic units come up all the time on eBay so I may see if I can get a working one, but it wouldn't surprise me if many are defective by now. I suppose the Alfa one has done well really. My previous Renault had an (Renault PLIP) alarm that was replaced 3 times before I gave up. New ones only ever worked for 6m before dying, and the ultrasonics were never reliable enough to go a week without triggering for no reason.

 

[dgmayhead]

Maybe the long press is a 156 thing. It worked for me when my car was being recovered. There's also a way of locking pets in the car without them setting the alarm off. Something to do with turning the ignition on and off.

 

[Paddy OPlastic]

I was thinking 'well, it's broken anyway'. I have a black ultrasonic unit, which come up rarely s/h, so initially took it apart to see whether the PCB could be swapped from the more common gray/ivory cases. It can. But to me it stood out as a tilt switch, similar to others I have come across, and the silicone mounting encouraged me to think so.


For now the alarm now activates without further false alarms, and disarms OK, which is an improvement, and no smoke has come out. However, simply unplugging the ultrasonic unit has the same effect.

Paddy is right. Further digging shows it's an NS25/B2 2-axis sensor mounted to the board by 10 soldered pins, DIL style. Turns out to be a an electrolytic-fluid based thing made by HL Planarteknik Gmbh, that relies on fluid moving between plates according to inclination. Whether this still works after 10 years is a moot point. Doesn't appear to be available anywhere now. Welcome to Willow Technologies lists it, but the datasheet says refer to the mfr. Likely superceded, NS25/E2 seems current. I might try and find out.

And what I thought was a tilt switch is a 40,000Hz crystal oscillator/resonator. Oops. I might just try replacing that, since the leadouts are crusty and are the only bit of corrosion on the board.

Still, at least I established that a PCB from a more common ivory or gray ultrasonic module can be transferred to my black casing easily, even if it will probably fail in the same way.

I'd be interested to know the result if you do replace the B2 with the E2 variant. After replacing the original unit once I'm putting up with the second beep since the error doesn't trigger the alarm.

 

[halftone]

OK, I'll at least try and find out if it's possible. HL-P haven't bothered to change their E2 datasheet diagrams which still show the B2, so it seems quite likely - if it's available to anyone except OEM's.

 

[halftone]

And yes - ON-MAR-ON-MAR-OFF is supposed to disable ultrasonics, but only until the next time the alarm is disarmed. So it'd have to be done each time the car is locked on the fob.

EDIT: Not quite right. I'm still at the 'RTFM stage with this 147, which is quite new to me. I noticed the manual gives the procedure as ON-OFF-ON-OFF to disable the volumentric alarm until the next time the ignition is turned on. Whereupon it reverts to normal arming.

 

[halftone]

Just to update... I've been in touch with Willow and they are looking into backward compatibility of the current tilt sensor and availability of the bare module and low volume pricing. I assume this means they've had to refer to HL-P, the German mfr. I think the bare sensor module is normally OEM only, but eventually all the AR alarms
using the module will apparently fail so I hope they might stock it or even suggest someone who might re-manufacture the roof modules. Willow say they'll let me know the outcome ASAP.

 

[halftone]

Further info, with thanks to the extremely helpful Carlos Mendes at Willow Electronics who has put a fair bit of time and effort in trying to assist.

Basically, HL-P, who manufacture the NS25/B2 tilt sensor, say the later /E2 version is "not comparable". They seem to be intending to cease production of the E2, probably because there are now far smaller and cheaper tilt detection mechanisms on the market than a decade ago. The B2 is already discontinued. HL-P don't normally supply either 'bare', only incorporated into their own control electronics boards.

Moreover the 'bare' version made for TRW, who manufacture the Alfa ultrasonic & tilt module, is non-standard. It incorporates a liquid electrolyte of a different viscosity to the standard B2 to modify its properties. Nobody except TRW knows the specifications, nor how their circuitry works, nor whether a defective B2 can be successfully replaced (eg without calibration procedures).

HL-P do have 50 pieces of TRW-spec B2 NOS, which Willow could supply to us, but HL-P will only sell as a batch so the cost would be almost £2,000. I don't see any prospect of a group buy raising that sort of cash, especially given the unknowns regarding replacement. There's also the consideration that these tilt sensors don't seem to last long anyway.

There are a couple of other possibilities. Other similar X-Y sensors do exist, and it may be possible to adapt them to the TRW board.

My son works in prototyping for automotive mfrs and reverse-engineering of CANbus comms has been part of his job. I asked him if it might be possible to make an entire replacement ultrasonic/tilt node, perhaps using Arduino, which is low cost and can do CANbus. He thinks so, provided the CAN is decipherable and he is keen to play with Arduino as a prototyping tool, but he thinks it just isn't worth the effort. Of course he's right.

I'm going to get a spare module and do some measuring. Just bypassing the flaky NS25/B2 so tilt is disabled and there are no false alarms, and the CAN doesn't see a fault, might be simple and cheap.