All posts by Tony

DHCP Server won’t authorize Error 20070

Built a brand new Windows 2012 R2 Domain the other day.  It was to replace a busted one elsewhere – busted? This happens when developers add anything they want to AD and there is zero management.   As this had to replace a 14 YO domain,  back in 2001 they had no concept of the issues around “Single label Top Level Domain zones” and the root domain was (and still is) as a SLTLD. This in itself is an issue as 2012 R2 simply refuses to create a  SLTD under any circumstances. The solution to that is break out you crusty old Windows 2003 R2 disks and use it to build the first server in the forest. You can then add target 2012 R2 servers to that domain and gradually work to removing the 2003 server and upgrading the forest to 2012 R2 functional level.  Stupid but it works.

Once I had it all built I then needed to add DHCP service to the child domain. Reason, in this organisation, the root is a placeholder and all the client based action occurs in the child domain so naturally I added DHCP to the child DC’s and set my IP Helpers to reflect that. This then revealed the following.

Problem: the DHCP server won’t authorize – returning error 20070 “The DHCP server could not contact Active Directory”. Not a lot of stuff on the internet but enough frustration to indicate that I’m not Robinson Crusoe on this issue. I bashed my head on this for about 4 hours and then did the right thing. I left the machine and went outside! It’s amazing how going for a walk, treating yourself to an ice-cream etc can clear stupid brain blocks.

Solution: Add the DHCP management tools to the root domain controllers and then authorise from there. Seems that the something has changed in the permission structure for 2012 R2. Thinking about it, the authorisation happens in the configuration partition of AD and naturally this would require Enterprise Admin rights.

I know it sounds trivial now but I hope this helps some other frustrated enterprise (sorry about the “Z” – In Australia its Enterprise not Enterprize) admin

Pug Pt3

Have populated most of the Speeduino and hooked it up to the Mega and TunerStudio (going to buy a license for this GREAT piece of software). I cut up the temporary extension,  and hooked up the “receptacle” to the board. Plugged it all in to the Pug and immediately blew the main fuse. Turns out I can’t count the CPS is pins 18 and 16 on the Pugs wiring, not 18 and 17.  Replaced the 25A fuse with a 15 (temporarily) checked the numbers one more time and got it right!

Powered it up again and bingo! After setting the TunerStudio calibrations I now have a zero to 100% TPS, a Coolant Temp (CLT) and IAS (Air temp). The coolant shows about 15C and IAS about 18 which is close enough for me.

Had to play around with the Crank Position sensor but I’ve figured out I have a VR (Variable Reluctance) sensor for the CPS vs the easier to use Hall Effect. Now need to get a VR conditioner which may take a little bit of time – I have a BOM to build one but would still need to source the LM1815 chip.

Still, its moving on considerably! Can’t proceed much further until I get the CPS working but I’m amazed at the progress so far.

Important note – The “electronic” solder sold at Bunnings is shit. I bought some of this about a year ago and spent most of my time blaming my 35 YO Weller iron for problems soldering. I purchased a new Jaycar TS-1390 soldering station and my soldering was still “average” – all the resistor solder joints look a little “dull” – anyway there was one joint that would not “wet” at all so I decided to go to Jaycar and purchase some “new” solder. WHAT a difference! All of a sudden I can solder again. Clean bright joints.  You get what you pay for I guess.

Pug Pt2

Started down the Speeduino path. Ordered the Mega and board from Josh which arrived quickly and downloaded the git repo. Fired up the Arduino 1.65 IDE and the Speeduino code loaded and compiled without error.  Downloaded a copy of TunerStudio and after replacing the TxRx dll – all good. TunerStudio found the board without problems.

Now need to order all the components to populate the board.

Getting all the bits from a single distributor is just about impossible – I’m trying to juggle availability and delivery costs for what is a small order. On top of this some suppliers have minimum order quantities but this can get ridiculous. I need a single 5.6V 2W zener but RS parts minimum is 50 items. They aren’t expensive individually (~20c) but what will I do with 49 spare diodes? I need a 22pin Molex connector but RS will only sell me 5 and the pins have a minimum order of 100 @ 14c so I have to buy $14 worth of pins….

There is also one or two parts that simply aren’t available locally so I’m stuck with $35 delivery for about $10 in parts. So it’s RS for whatever I can get locally and DigiKey for the rest.

In the meantime, I bought a Hantek 1008 8 channel USB (PC) oscilloscope on eBay. Each channel is about 1Mhz bandwidth which is plenty for a scooter. I gutted an old Pentium 4 motherboard and extracted the 20 pin Molex “receptacle”, used for the power, cleaned it up a bit and soldered back on the leads from the PC power supply. This makes a 20 pin extension cable with some exposed contacts for connecting the Hantek. I also need to strip out a single extra wire for Pin 22 (11 isn’t connected) – it comes from the Air temperature sensor cable and the ECU diagnostic comes on if isn’t connected.  Hooking it all up I now get this


This is exactly what I wanted. Yellow is the Crank Position Sensor. Clearly a 23 + 1 tooth wheel. 15 degrees between teeth. Blue is the injector. Purple is spark. The engine is 4 stroke but the CPS is driven from the crank and there is no camshaft sensor. Therefore the engine uses a “lost spark” and “lost injection” i.e. both the spark and injection fire every revolution. Makes the whole EFI somewhat cheaper to build. It would appear that the “missing tooth” on the CPS is about 6 pulses @ 15 degrees / pulse so 90 degrees BTDC. I can’t simply pull the cover off as the CPS and the alternator are not isolated from engine oil and I don’t have access to a gasket set should it tear during removal.

Now need to wait patiently for the arrival of parts and start to get the Speeduino assembled.

Current plan is to get the “input” end of the Speeduino up ASAP. This will be TPS, CPS, Coolant and Air Temp sensors. There aren’t any hard to get semi conductors in the input path so it should be achievable. I’ll have to power the Mega via USB temporarily too.

The Peugeot Jetforce 125 or How a scooter becomes an engineering project

Synerject Contact formThis is now getting to be a saga. I work in the “Parliamentary Triangle” in Canberra – you know the capital city of Australia not Sydney or Melbourne as some would believe – and in 2014 the ACT government in concert with the “National Capital Authority” (aka the NCA) decided to introduce paid parking. I have no real issues for charging for the land use and maintenance of the car parks but $12 a day in an area where there are no national tourist attractions and no shops is a bit rich.  $5 a day I would not have any truck with so I decided to fight back and but a motor scooter.  As the NCA uses a system on “pay and display” and as there is no way of securing a ticket on a motor cycle, they have exempted motor cycles, scooters included from the impost.

So there are several scooter dealers in Canberra and I happened to be close to the one in Mitchell (name withheld) where I was temporarily working, I went over and spied a very pretty red Peugeot JetForce 125 EFI. As I’d had a great experience with a Peugeot 407 in the past and we had a little family history with Pugs, I plonked down $2500 to become the owner. First mistake – zero research – the dealer for these went broke in the GFC in 2009 and there is no Pug scooter representation in Australia.  Second mistake – any simple research would indicate that these scooters have all sorts of reliability issues. Still I took it for a test ride and it seemed to go OK at the time.

Well, the issues started from Day 1. I rode it to work next day from North Lyneham and by the time I got to Commonweath Ave, the bike wouldn’t do over 60Km/h. I rang the dealer and organised to take it back that afternoon. By the time I got out to Mitchell the bike wouldn’t do 40Km/h and was positively lethal in terms of throttle response. The dealer put the bike on his “do when other stuff was done list” which was pretty surprising seeing how I had only bought it the day before so it took almost a week until I got it back. By then, it has magically fixed itself. The mechanic stated that he had cleaned the injector, replaced the fuel and the spark plug. He actually just got lucky.

The bike went OK for about another 3 months – well after any sort of warranty period would expire then the same trouble started all over again. The scoot would just go slower and slower until eventually it would just idle parked up in the gutter. I did find it would “go” for a small distance, maybe 500m by turning it off and restarting. Each time I did this, it would go again but a lesser distance. It quickly gets to the point where it will only start and idle.

Now the saga begins. Google is invaluable here. The scoot runs an EFI system originally designed by the Orbital Engine Company of Perth in Western Australia (a great co-incidence). The original Orbital Engine was effectively a dud as they could never seal or cool it effectively but they did develop and patent a 2-stroke compressed air injection system which required the development of a cheap to manufacture ECU (Electronic Control Unit). This ECU was also found suitable, with minor adjustments, for small 4 stroke engines.

Through various purchases and divestments of capital, the SynerJect company bought a percentage of Orbital and took over the IP and marketing of the small fuel injection market.  How did I find this out? I rang Orbital and one of their engineers actually talked to me! I’m still grateful that they would bother to do this. Synerject, on the other hand has a somewhat hostile message on their site which states that unless you are an engine OEM we won’t talk to you.  The two stroke version is called “DITech” – Direct Injection Technology I assume and is reasonable common on 50cc 2-stroke scooters from Peugeot, Aprillia and I think Suzuki has one as well. Peugeot then used the Synerject 4-Stroke system on a number of models mine included. The components seem to sourced from all over the world including Mitsuba for base electrics, Phillips actually make the ECU, Bing for the throttle body etc.

I emailed Peugeot in France and they don’t have any capacity to talk to some random owner in Australia where the dealer doesn’t exist. Spare parts / diagnostic support – forget it – so I’m now in the grey area of people who “might” have some parts or “might” know something about this puppy.

First thing I did was replace the plug and change the fuel. No idea how a machine that recovers temporarily would recover from fuel problem but it was a start.  During this period my research showed that the diagnostic computer was actually a Nintendo Colour Gameboy with a specific cartridge. I’d been to “Motorini” Scooters in Phillip where the owner Nico and his mechanic gave me the final judgement on Peugeot Scooters – just say no.  Their words were “If you had of asked we would have said just don’t buy it” – that said Nico lent me his Aprilia diagnostic cartridge and $5 gameboy. I took it home but the diagnostic connector is different and I wasn’t about to “hack” his cartridge. The cartridge is almost certainly a DITech (2-Stroke) not the 4-Stroke

Couldn’t get it going so I loaded it on the trailer and took it 300Km to my real home in Sydney. Got it into the shed and noticed the battery was a bit weak so I charged it overnight and for some reason, it decided to run properly. I fanged it up and down the road outside and it ran as good as new. I ordered and fitted an ebay $12 voltmeter to the scoot just so I could keep an eye on the charging system.

Loaded it back on the trailer, 300Km back to Canberra and it lasted 3.5 days. Exactly the same symptoms but I thought the battery recharge was significant (it wasn’t) and bought in turn and new battery ($80) and regulator ($65 from E-Bay).   The voltmeter actually told me so but I was getting desperate.

Trailered back to Sydney – the trailer is doing a lot of Km and its now back in the shed. I’m determined no to just “junk it” so here is where I’m at…

– When I disconnect the lead from the TPS (Throttle Position Sensor), nothing changes. i.e. the bike acts like there is no TPS. It will start and marginally throttle up to 5000 RPM then dies back to idle. I suspect that the minor throttle up is due to the opening up of the throat leaning the idle mixture up enough to increase the RPM.  I’ve removed the TPS from the bike to see if it was the issue. The ohms across the device 2K (2040 actually) and the wiper moved linearly from 2K down to 1.1K – there is no bumps and wiggles in the response to it seems OK – problem is – I don’t know what the correct values are and there seems to be zero information. I’ve emailed the “Bing” company in Germany (they make the TPS) and asked – currently awaiting a reply. I actually think the problem is in the ECU as there isn’t anyway I can think a simple variable resistor would have issues around power being re-applied unless it was overheating (Voltage Regulation issues?)

Further update. I measured the voltages on the TPS. With the plug disconnected I get +5V across the resistor (4.99 actually) so the 5V regulator in the ECU is OK. When I connect and back-probe the wiper to earth I get a smooth +0.7V to ~4.4V. Would look like sane numbers. Zero Volts on the wiper would be a fault condition (Open Circuit) if I was an ECU. 5V would be almost short. The fickle finger of blame is now firmly on the ECU input circuitry.

The nice people from Bing have responded and I now have the full technical data of a Type 72 Throttle Body and TPS. The specifications match exactly what I measure on my TPS so now its 100% the ECU. I can provide the TPS specs if anyone is interested.

Decided to attack the ECU last weekend. Takes 1 minute to get it out of the bike but it is encapsulated with silicone rubber. Had a go at cutting some of it out but it is almost impossible to remove down at the board level. 90% of the board components are surface mount so any attempt to remove the rubber mechanically will certainly damage something. Without a circuit diagram, if I chipped off a resistor etc, I would be totally stuffed. I did find out that the interface connector is a 22 pin version of the ATX PC (20 pin) connector. Molex 5566 series so I can order a male 5566-22A from Element14 (or someone) and not totally stuff the bikes harness.

Have ordered the V0.3 board so looking forward to populating that and getting stuck into building a “New” ECU. It should be a lot of fun.

Lightning! You just can’t beat it.


Here is my collection of SPA-3102 VOIP Gateway devices that have been fried by lightning over the last few years.  @ $ 55 a pop, it’s still cheaper to use VOIP but it gets a little boring after a while trying to keep up with the replacements! I’ve ordered some new lightning arresters to try and cure this but I’m not hopeful.

Combined with the one that is in service… This makes 7!

By the way Pennytel, can you stop sending a new GE handset out each time… I already have 2 of them! More than one is one too many.



About 18 months ago, worried about privacy concerns I exited Facebook. I was never a very prolific poster, again I used to think “this will be here forever” so I also worried that I was a but of a “lurker” and not participating in the “spirit of the game”.

Well, I was walking down the stairs the other day and I finally determined why I don’t want to Facebook any more.

Its a personal thing, so you can ignore this post, but the whole idea of knowing everyone’s minute by minute status, regardless of the fact that these updates are (in general) provided by the owner, in some way feels almost voyeuristic or just a touch sleazy. The feeling is exacerbated by not being an active poster too. Cue Peter Sellars in”The Party” ……. “I like to watch”

I also don’t need the “noise” that comes from this. The signal to noise ratio on Facebook is particularily low.

Perhaps journalling your life in a blog is the same but I feel there is slightly more control here.

Tight Arse JBOD

A mate of mine is working on a DIY Hadoop cluster, for learning. Rather that souce “X” servers to build a cluster, and the fact its a learning device, he has cobbled up a couple of G5 DL380’s and for disk, built a “TAJBOD”

Here are his comments (Name supressed!)

Thought you may appreciate the low tech engineering approach here for a tight @arse jbod array: standard sata drives housed on a made from scrap jbod shelf ( 10 minutes of careful routing to make the slots in the chipboard ) and sata extension cables to join the dots. Power for the drives is directly from the server and the RAID is setup by the internal P400 controller, as far as the server is concerned these are just ‘internal’ disks. Best of all it just works!

Shelf detail


Congratulations CERN!

Epic, magic, creative, wonderous – There probably aren’t enough adjectives to describe what has been going on in the world of high energy particle physics.

I watched, with wonder, the CERN webcast on the 4.9/5.0 ? results on the discovery of a new boson at 125 Gev. It’s the Higgs of course! Watching those hundreds of dedicated physicists, engineers, technical and support staff being riveted to the presenters and the data being shown was amazing. Peter Higgs losing it with the emotion of the announcement was just icing on the cake. This is up there with Chadwick discovering the neutron and Rutherford finding the proton.

From a small, insignificant watcher of the amazing world of science – Congratulations! and congratulations to the thousands of physicists who have laboured tirelessly over their blackboards and apparatus over the last 100 years.

Thank you CERN for letting us all watch this incredible journey unfold in real time. It’s almost like being at the Cavendish is 1931!

Tony’s rules of data

I’ve been copying some large data sets around lately and have formulated 2 rules of data

1. Data will ALWAYS expand to fill any available storage
2. Moore’s Law is irrelevant as data is ALWAYS conspiring to bring the apparent speed of a system back to that of a 386.

There you have it – 30 years in the IT industry condensed into 2 statements

Chucking stuff out 2

My wife reported our little Sunbeam coffee maker was no longer working. It wasn’t pumping water etc. Here we go again I thought. Another unrepairable appliance!

We have had this Sunbeam for about 5 years and it made OK coffee, as long as you gave it plenty of time to heat up. There was also a lingering problem with puddles of water under it but I always assumed (that word again) that the problem was related to the removable plastic water tank on the back.

Totally ignoring the warnings “no user serviceable components inside” I pulled the back off and immediately noticed rust on several components –

Now that wasn’t cool. Rust doesn’t happen the moment something gets wet and this looked long term. I could understand rust on lower components getting there from a leaking valve but the bracket holding the pump (blue component on right) was rusty. Could see anything spectacularly wrong so I plugged it in and watched. Once I started the coffee cycle, the problem was obvious almost straight away. The union on the hose above the pump sprayed water over most of the innards of the machine. Why it hadn’t tripped a earth leakage breaker was unknown.

That accounted for the leaks and the rust. Once the coffee cycle starts the pump starts a rhythmic thumping and this had loosened the union.

20 seconds with a spanner, 2 minutes to put the back on and its as good as new. Made my day as for once, I didn’t have to “Chuck it out” Yay!