When it comes to good things, they say moderation is best. Don’t eat all your candy at once, or so they say, but instead I’m diving right in for a Dick Smith Electronics kit overdose. This is kit four of five that I bought at the CCARC Wyong Field Day.
This particular kit is the K4309 Daytime Running Lights for Cars based on the Silicon Chip August, 1999 edition. This kit belongs to their automotive series, coloured green and arriving in the same sized box as the other kits constructed so far. I don’t see it written, but I’m pretty sure a car is not included. So I guess what I need is a car. Maybe in kit form. I’ve never seen one on the shelf at DSE though …
The main feature of this kit is the ability to automatically power the headlights at a reduced 80% brightness as soon as the engine starts and turns off with the ignition. It also has a dusk sensing mode which puts the headlights to full brightness when the cabin is dark. It can handle up to 200W load, and comes with components, PCB, hardware and a diecast metal case – a premium touch especially due to the harshness of the environment in automotive electronics and potentially a desire to shield stray emissions from what appears to be a PWM style controller.
Part of the reason I left this kit towards the end to construct was that I felt it was probably going to be a bit difficult in comparison. For one, some metal-working would be involved and for another, I’m sure that inductor needs to be wound. The transistor mounting could get tricky too. Maybe it’s a fear of failure, or a desire not to screw up – in truth, it’s probably somewhere in-between.
As promised, the parts are supplied separated into categories. This must have been an expensive kit – the retail die cast case, the inclusion of an automotive relay in the “auto parts” box, specific ferrite ring for winding a inductor filter and a significant amount of thick gauge insulated copper wire would certainly have added to the cost. Some ICs are used including the venerable NE555 which is used to generate the PWM waveform, an optoisolator (suggesting there would be some floating circuitry) and a logic gate used as a MOSFET driver. Very interesting. Note that no IC sockets were supplied with this kit – the obvious reason is to avoid any unreliability due to vibrations causing the chips to become unseated.
As usual, the paperwork includes the Assembly Manual, Guide to Kit Construction, Quality Control Card and Disclaimer.
The PCB itself is also very modern, with a very special shape including rounded corners and notches. The copper layer bears the Silicon Chip logo with a limited amount of guidance text. It is still very much a basic single-sided PCB without any silkscreening or solder resist. The pad shapes and sizes, however, have been quite optimised for soldering with some thicker traces for higher current connections.
The other side of the PCB doesn’t have anything much to show.
Likewise, the assembly manual is very much in the same laser-printed monochrome stapled format that the other manuals were in.
Once we open the case, we can see why the PCB was designed in this way – it will clear the rails inside the die-cast box and secure to the casted holes in the base of the case. This is the best way to mount it, as it requires less metalwork – otherwise we might be drilling holes into the rear in a strict alignment, assembling a mount out of nuts and spacers.
As usual, construction on through-hole boards has become rather routine and straightforward, in terms of populating components and soldering them down. This board does have a few traps, including a jumper wire underneath an IC that makes construction order critical, and many closely spaced holes which can make incorrect connections a possibility – I had to desolder one connection that was due to a resistor being populated in an adjacent pair of holes than the one intended. In any case, the below image shows the construction of the PCB at an advanced stage with most components mounted.
Almost everything that needs to be mounted has been mounted, save for one PCB pin and the inductor. The board is fairly densely packed, with the optoisolator impinging on the electrolytic capacitor next to it. It’s interesting to see an audio transformer repurposed as an isolating transformer of sorts.
I used my own fluxy Multicore solder, but to be honest, soldering this board was quite enjoyable as it didn’t have any particularly nasty design issues or oxidation on its surface.
Now that the PCB mounted components are mostly completed (I missed one PCB pin above), we need to wind the inductor. The inductor is made of twelve turns of thick enamelled copper wire, which needs to be made to cover about half of the ring. The wires need to be cut to length and tinned before soldering to the PCB pins, then mounted to the board with some cable ties.
They were somewhat generous with the wire, as they usually are. Winding was not a big issue, although mounting it to the PCB and pins proved slightly difficult as the wire is stiffer than the pins and needed to be bent into good compliance during mounting to minimise residual stress.
In the end, I managed to get the inductor mounted to the board along with the missing pin.
The coils had to be shifted away from the PCB board edge, as a warning is given in the manual that the enamelled wire could come into contact with the die cast casing and rub through the insulation over time. As automotive systems generally use a negative-ground system, this would mean that the output of the circuit would be shorted to ground which would be bad news.
As a result of shifting the coil, it turns I cut the lead slightly short, so I was a little generous with the solder to compensate. Not my finest moment, I’ll admit, as I had an excess on the other leg. But it should still work.
This is what it might look like when mounted to the case, showing that inductor clearance being rather critical. To complete mounting, there are a few critical steps that need to be completed first.
One of them is to mount the ground wire that should be attached to the chassis of the vehicle to the case. An M4 bolt is used with serrated washers and ring crimp terminals to attach a thick black wire to the outside of the case and to the inside where a short tail would connect to the PCB. In my infinite wisdom, I drilled right in the centre of one of the PCB rails … thus the internal ring is basically contacting a reduced area. I figured since the bolt was long enough and this connection did not have to carry much current, I wouldn’t drill another hole and spoil the case. The other thing, of course, is that you need a crimp tool to make proper joints. Luckily I had a ratcheting crimp tool of sorts, but my crimp placement could have done with some improvement.
The next thing to be concerned about was the cable entry/exit grommet. This plastic piece holds the wires tightly to prevent them being pulled out and prevents chafing of the insulation of the wires due to movement and sharp edges of the case. This required a 14mm hole with two ends filed to 16mm. Unfortunately, the largest drill bit I had was only a 10mm and my tapered reamer maxed out at 12mm.
I got somewhat creative, using the reamer off-angle to enlarge it slightly. I got to 13mm and that was about it. So I got out the cordless drill and a 6.5mm drill bit at high speed, using it sort of like a milling machine to try and enlarge the edges of the holes, going around the box. This gave me a little more size, at the cost of losing the circularity of the hole and a lot of noise. Finally, I got close enough that I was able to make it just large enough with a set of small hand files. This part of the project took almost as long as everything else combined.
The LDR used for a light sensor in the cabin to tell between light and dark is attached to the figure eight wire. The heatshrink tubing is provided and goes over the connection. As the two solder joints are not otherwise insulated inside, I cut the LDR leads staggered as well as the cable – so that the two joints cannot be shorted together by crushing.
From there, it was a very tricky task to tin the wires and attach them to the circuit board as the thick wires were stiff and imposed significant stresses on the PCB. The transistor also had to be mounted to the case in an insulated way with a mica washer, plastic grommet and thermal paste. Then the PCB is to be mounted to the case using four self-tapping screws and the grommet used to fix the wiring into place.
At this stage, I am almost ready to close the lid, but one thing needs to be tested – the insulation between the case and the MOSFET tab. Unfortunately, the first time, I had continuity which indicated a short somehow, despite carefully ensuring everything was correctly positioned.
After removing everything, cleaning off the mica and trying again, isolation was achieved. I suspect a small shard of aluminium in the thermal paste from the hole drilling may have spoiled the insulation the first time.
One strange design issue is the brightness/darkness trimpot is inside the case in a position that is hard to adjust due to proximity to connecting wires and orientation. While one could potentially drill a hole through the casing to adjust it – I felt it easier to test and adjust before closing the case.
In order to close the case, some weatherproofing foam beading needs to be inserted into the channel of the lid, then the lid is aligned with the base and four Philips head screws are used to close the case for good. Despite this, the design is not waterproof, and instead the use of silicone sealant is recommended to secure the inductor internally, as well as waterproof all penetrations through the casing. That being said, there was a recommendation to mount the case inside the vehicle to limit exposure to heat and vibration for longevity.
Bonus: Zung Sung AR-502 20/30A Auto Relay Teardown
As the kit included a 20/30A auto relay, I thought we’d take a little look at it. Because this is used to drive the tail-lights at full intensity by having the PWM drive the coil of the relay directly, it’s not something that I felt was worth wiring in for testing, so I kept the relay aside.
The relay comes in a brightly coloured cardboard box, with the Zung Sung branding. It merely says “Auto Parts”, so I suspect the box may have been used to sell light globes and other parts … or it has some deep and dark secret about its identity that it doesn’t want us to know.
Inside is a fairly standard automotive 20/30A relay, which has a 12V coil and a normally closed and normally open contact, single pole. These were frequently used to switch lighting loads especially in older vehicles, mounted by its “tab” to the chassis somewhere.
Unlike some of the better quality relays I’ve encountered, this relay seems to just “snap” into its shell and has no sealant or adhesive along its seams. As a result, it can be broken open for a teardown with a flat-blade screwdriver.
Inside, we can see the coil occupies a majority of the volume, with the return spring visible towards the top of the image. The current appears to be connected to the contacts in the bottom right by the means of a flexible copper braided conductor – this is quite thick to carry the “up to 30A” load.
THe contacts can be clearly seen from the front side.
It appears the frame of the relay is connected to the common terminal, with a rivet near the base. The copper braid appears to be spot welded into place to carry the bulk of the current, especially as the “pivoting contact arm” may not maintain contact with the frame during switching.
The coil windings are soldered down to lugs connected to the external terminals.
The proof is in the pudding … or testing, I should say. For this, I used a 12V halogen downlight rated at 50W as the test load. The power supply is the trusty Rohde & Schwarz HMP4040.04, using the first two channels in tracked mode. Channel 1 is supplying the “ignition” line to the box, which basically powers the circuitry in the box, while Channel 2 supplies the “lighting fuse” line to the box, namely the supply to the FET which is chopped to run the lights. This way we can monitor the currents independently. The grounds are commoned together by the stacked banana plugs, while my own Agilent/Keysight U1241B handheld digital multimeter is monitoring the voltage across the lamp.
Powering up the unit at 12V, there was no loud noises or smoke. A great start. The circuit is consuming 56.6mA which corresponds to a power of 679.2mW. As promised, the light does not come on until the engine is started. So lets start our engine … or pretend to …
Cranking up the voltage in 0.1V increments in tracked mode, I find that the lights come on as soon as we hit 13.2V. This is basically sensing the voltage rise due to charging by the alternator to determine that the car has started. In this case, I can tell the PWM is working, as the input voltage is 13.2V but the voltage across the lamp is 10.216V. The lamp is drawing 3.725A or a power of about 49.12W. But more than this, I know the PWM is working because I can hear it thanks to the vibrations from the inductor.
Shifting the LCD to be shadowed by the cable results in the whine stopping and “full brightness” applied to the lamp as if driving in night time. The voltage on the lamp rises to 12.94V and the power supply is putting in 4.467A or about 58.96W. The voltage drop is hence about 0.26V for a current of 4.467A or 1.16W dissipated. Considering some of this will be in the wiring, while the remainder will be in the MOSFET, this is a very acceptable performance and the MOSFET will likely remain fairly cool. The calculated resistance is just 58 milli-ohms.
Another promised feature is hysteresis – the lamp will not shut off until the voltage falls much below the initial turn-on point. In this case, it took reduction of the input voltage to 10.9V before the output was switched off. As a result, the circuit appears to work exactly as described which is a great result.
Looking back, I think I should have wound the inductor much tighter if possible, as that should reduce the coil “whine” that usually only is exhibited where magnetostriction causes the coil or cores to vibrate in synchrony with the changing current.
The Daytime Running Lights for Cars kit seems to be a practical add-on for older vehicles, intended to simply automate the process of running lights at reduced brightness at the daytime after the car starts, and at full brightness when it is dark, while also running the tail-lamps at full intensity through an extra automotive relay. To do this, it employs PWM techniques using a traditional NE555 as an oscillator in a rather “old fashioned” design with an N-channel MOSFET performing the switching and a hand-wound toroidal inductor as the filter. It seems to be very well thought out, although it needs a practical person comfortable with modifying automotive wiring to integrate into a car of the vintage. Construction of the kit was not as simple as other kits due to the need for metal working for the grommet holes and mounting holes, with some tricky points including the soldering of heavy gauge wire, winding of the toroidal inductor, soldering of enamelled copper wire and insulated mounting of the MOSFET using thermal paste and a fragile mica washer (where we might just use a silpad instead nowadays).
In the end, construction was completed successfully with a few minor touch-ups, but the usefulness of this kit has diminished as newer cars integrate some of these features as standard. In some sense, I do think it to be a little hazardous to trust something as important as the headlights of your car to a homebuilt circuit which might fail for a variety of reasons (e.g. broken solder joints from vibrations). But I guess that’s a risk the “do-it-yourselfer” would have taken into account before they bought the kit!
I’m glad that this turned out well in the end – I feared it would be a kit I might butcher, but now, I have the feeling of satisfaction and confidence that comes with seeing it working. Maybe now, I’m qualified enough to build the final kit from my haul … the one with the biggest box of the lot. The one that might defeat me in the end. Or will I put it off and chicken out?