Project: Generic 18 Red LED Flashing Love Heart Kit (TJ-56-30)

Another day, another classic kit. It seems that LED heart-shaped displays are a timeless novelty, so after building a multi-coloured chasing heart and a breathing heart, why not a flashing one? For just AU$1.68 including postage, I managed to score this kit from eBay which is an upgraded version of the classic two-transistor flasher circuit.

The Kit

This kit looks like any other cheap eBay kit from China, in that it’s packed in a zip lock plastic bag. But already, from the beginning, we see an improvement – they included a leaflet!

The kit is built on a standard “boring” square PCB, being of the single-sided paper-resin variety. White silkscreening on the front gives clear indications as to how to mount the components and their values. The kit seems to have a designation of TJ-56-30.

The underside reveals lacquer-coated copper pads and green soldermask.

Included are the LEDs, transistors, resistors, capacitors and a set of terminal blocks for power connection. A quality control tag is added to the bag as well.

The included double-sided half-page of information is in Chinese (which, sadly, I cannot read). However, it does include the schematic, as well as the bill of materials and PCB design which is handy reference. We can already see that this is a standard collector coupled astable multivibrator design, albeit with three branches rather than the more usual two.

Construction and Testing

Lets do some quick sums to check how much work is involved in building this kit:

18 LEDs x 2 connections
6 resistors x 2 connections
3 capacitors x 2 connections
3 transistors x 3 connections
1 terminal block x 2 connections
TOTAL - 65 joints

This isn’t a particularly high number, which should make the kit quite suitable for a novice. In all, including photo taking and testing, it took me about 40 minutes from start to finish.

Overall, the construction experience was not bad, but could be improved. For once, I dislike lacquer coated paper-type boards for soldering, mainly because the lacquer can interfere with formation of the joints, thus requiring a little more heat and care when soldering. The type of board is also more sensitive to overheating, so this isn’t necessarily a good thing for a beginner to have to deal with. The fumes from this sort of board, while distinctive, aren’t exactly the most pleasant to deal with either. The pads were a little on the small side as well, making solder control critical. The shape and routing of the traces were almost haphazard, along with the directions of the LED polarities, making it highly probable that a mistake could be made necessitating extra work to resolve (increasing the chance of damaging the board). Finally, there were no spare parts and only exact quantities were provided. That being said, I suppose, given the low cost, it’s something to be expected and are only minor niggles.

The result is a heart which has two-thirds of the LEDs lit at any time, chasing around the perimeter in a way that is reminiscent of some of those LED restaurant signs. Note that the capacitors and terminal block don’t move in operation – that’s a side effect of perspective shift and post-processing to remove hand shake from the recorded video.

Finally, I simulated a simplified version of the circuit (omitting the parallel LEDs) in Analog Devices LTspice, which clearly shows the behaviour of the circuit (albeit substituting the transistors with “close enough” equivalents. If you want, the model file can be downloaded here.

Simulators often have problems (in my past experience) simulating these types of circuits as their stable oscillation only occurs when the components have some “mismatch”. It was interesting to see that, aside from the initial glitch, it seems like the simulator worked just fine.


This kit is nothing but a simple classic design upsized. The kit itself was complete, although the single sided paper-type PCB with lacquered copper was not the easiest to solder. The quasi-random directions of the LEDs and the odd trace patterns also were not helpful, however, it did come with Chinese instructions that included a schematic, adding to its educational value. For the price, it’s a decent amount of fun and soldering practice.

Posted in Electronics | Tagged , , , , , | Leave a comment

Teardown: Generic Wall-Mountable Passive PoE Midspan Injector

For a lot of modern networked devices, Power over Ethernet (PoE) is a big convenience, allowing the one cable to power a remote device (be it an access point, camera, etc.) and supply the data connectivity to it.

While there is a standardized PoE in the form of 802.3af/at, which specifies negotiation rules and a 48V nominal system voltage, many lower cost devices (especially older ones) cannot justify the increased cost of integrating the necessary logic and power conversion circuitry. These utilize an alternative solution which is termed Passive PoE.

Passive PoE does not involve any negotiation whatsoever and supplies power at all times. In its earliest and simplest incarnation, it used the two unused pairs from Fast Ethernet to carry the DC voltage. This was informally standardized as pins 4-5 as positive and pins 7-8 as negative, similar to 802.3af mode B. Later on, Gigabit Passive PoE used transformers to allow both data and power to share the same pins, although these also come in a few variations.

Most passive PoE installations use a variety of voltages, with 24V being a common compromise but even voltages as low as 12V can be practical depending on run length. Cheap passive adapters (endspan injectors and splitters) can be purchased to convert some 2.1mm DC jack based equipment to run off passive PoE although the cheapest adapters are fast Ethernet type. This is often good enough for older wireless APs.

However, whatever you do, you should not apply passive PoE to a device not designed to take it, otherwise damage to the Ethernet port’s magnetics may occur!

The concept of passive PoE is fairly straightforward and endspan injectors can be purchased in pairs quite cheaply (about AU$2 a pair). But what if you don’t want to inject and remove power at the ends of the Ethernet cable? Maybe you have voltage drop issues, or you have a more convenient source of power closer to the device? In that case, you need a midspan injector.

The Device

On eBay, for about AU$1.59 posted you can have a “wall mountable” midspan Fast Ethernet passive PoE injector.

The unit has a plastic body with a hole on the top for a power indicator LED. The wiring is printed on the casing for reference.

A 2.1mm DC barrel jack receives power from your power pack to inject to the cable. The PoE port supplies the data (two pairs used for Fast Ethernet) and power (over the unused pair) to a device expecting passive PoE or a splitter that you supply. The LAN port provides only the data and is not connected to power.

It is “wall mountable” owing to a small hole eyelet in the casing, although it does seem a little flimsy.

The unit is held together with one screw covered by a QC label I had removed. How does it look inside? Lets see.


The screw was the only thing stopping you getting inside.

The unit is built on a paper-type single-sided PCB to reduce costs. Inside, there are the Ethernet jacks, the power jack, an LED, a 10kohm resistor and two jumper wires. This is probably the second simplest design that could be used – the simplest would have omitted the power indication altogether. The PCB is marked XLY-POE2(20150717) suggesting a company by the initials XLY manufactured the unit, this is probably the second version designed 17th July 2015.

A quick calculation seems to show the resistor is well sized – at a maximum 802.11af voltage of 57V, about 55V is put over the resistor resulting in a current of 5.5mA (very safe for the LED). At a low of 12V, with 10V over the resistor, the current is just 1.0mA which should result in a dim but visible glow.

Unfortunately, the unit doesn’t employ any safeties – no polyfuse protection against short circuits, no diode protection against reverse polarity. As a result, be sure to use a limited-power supply which will only source a safe amount of power in case of a fault and confirm the polarity before using it. It would be trivial to add it with a knife and soldering iron though.

However, because of its simplicity, the unit could also be used as a splitter by plugging in the power+Ethernet lead into the PoE port and extracting the power from the DC jack (by making a male to male cable). You can also use it as an indicator as to whether PoE power is being delivered to a passively-powered Fast Ethernet device just by plugging the powered lead into the PoE port and seeing if the LED lights. Or you could use this as a crude Fast Ethernet only cable joiner in an emergency. For the most part, most endspan passive PoE kits can’t do this as they lack an LED.

A look at the underside shows green solder resist and very oddly shaped thick chunky traces. It might guard against delamination during soldering, but it might not have a good effect on the data signals.

Tracing the path of various signals, the positive power is in red, the negative power is in dark blue. The LED to resistor connection is in yellow. The remaining data pins are colour coded green, orange, purple and sky blue. The light-blue overlay shows where the LED, resistors and jumper wires on the other side are.

This is a mess. In fact, this is where endspan units might be better as they are based on a crimped RJ45 end and CAT5 or better cable soldered to a socket which isn’t particularly complex. However, as we can see in this design, the data lines are all different lengths. The traces are also shaped quite oddly with some data lines needing to jump over to the other side and back. This would probably cause both skew and impedance mismatches which could affect the data integrity slightly. Being unshielded, it could also introduce interference into the network in the case of being near high EMI/RFI sources.

Luckily Ethernet is extremely tolerant of dodgey cabling. Low rates of errors will be corrected through retransmission – I’ve even had success running 100Mbit/s Fast Ethernet for a while on CAT3 “voice grade” cabling which was only suitable for 10Mbit/s just on a whim. So I don’t think this design will cause problems per-se, but it’s not the best situation for an Ethernet signal. Of course, using two pairs for power in a dedicated fashion will preclude Gigabit Ethernet from working.


This inexpensive mid-span passive PoE injector is very much as minimalist as it can be, with the exception of the inclusion of a power LED. It can be useful in a pinch as an injector, a splitter (with your own male-to-male lead), diagnostic tool to check for PoE presence or to forcibly strip out any passive PoE to safely connect non-passive-PoE-aware devices or as an emergency Fast Ethernet cable joiner. The design isn’t exactly the safest, so ensure you are using a limited power supply with the correct polarity. The design also isn’t the best for ensuring the integrity of an Ethernet signal, but it should be “good enough” that it would work in most cases. Definitely nice to have as a spare in the toolbox.

Posted in Computing, Telecommunications | Tagged , , , , | Leave a comment

Project: Unbranded Generic NE555+CD4017 SMD LED Chaser Kit (YL-117)

I had a little time this evening, so I thought, why not crack open a kit? After all, I did buy a stack of cheap eBay kits to keep me amused and give me something to write about … so I decided to pick out a kit with a slight twist.

The kit itself is an NE555 timer and CD4017 counter based LED chaser. This is a pretty classic configuration and one I’ve seen from the Funway 2 days. It is also advertised under the name of “running flow LED light”, which is an interesting translation.

However, the twist is that this time, I’ll be building the kit using mostly SMD components. For the princely sum of just AU$1.00 including postage, you can challenge yourself to some non-trivial amount of SMD soldering. I didn’t even know they still made 555 timers in SMD packaging!

The Kit

Say hello to my new friend! Zip-lock bags, loose components and no instructions. Nothing new here, but that’s just the realities of low-cost kits.

As the project uses SMD components and 3mm LEDs, the physical size of the PCB is not very big. It is a quality fibreglass board, double-sided, tin plated with blue soldermask and white silkscreen on the top. It is marked 1918m and YL-117 which might have something to do with the creator. The components are pretty well marked, although there are a few subtle issues – the LED symbols have mostly been lost to the pads, the resistor numbering is pseudorandom, the text for the NE555 may have you mounting it upside down and the CD4017 is labelled as a CD4B17. There are also some dark stray marks on the rear which may be due to abrasion in sorting.

Aside from that, you get your SMD resistors, capacitors and ICs in cut-tape format, which is fairly standard. The LEDs are in their own zip-lock bag, and the trimmer and header pins are packed loose. That’s all there is to it – but admittedly, that’s quite a lot for AU$1 posted!

Construction and Testing

A quick compute shows the total number of joints needed to complete the kit, noting that a lot of them are surface mount joints:

10 LEDs x 2 connections
12 resistors x 2 connections
2 capacitors x 2 connections
1 CD4017 x 16 connections
1 NE555 x 8 connections
1 trimpot x 3 connections
1 power x 2 connections
TOTAL - 77 joints

Construction is probably the more interesting part of the kit, so I won’t bother reverse engineering the kit – after all, it’s the same as the Dick Smith Funway 2 LED Chaser which consists of a 555 timer as an oscillator driving a counter chip with each output hooked to an LED and resistor.

It makes most sense to start with the surface mount resistors first. As the row of LED resistors are the most numerous and equal-valued, I started with them first. Taking care, I first tinned the pads on the board with a little solder (as I don’t have any solder paste). I then extracted the resistors from the tape (carefully, so as not to lose them) by peeling off the plastic and shaking them out one by one. With a fine set of tweezers, I placed each resistor in approximately the right position, and then waved a heat gun slowly over the whole board with the airflow set low so as not to blow the resistors away. Slowly, you see the resistors “twitch” into their position, although some might need a subtle nudge.

Voila, the resistors are mounted. Not the most pretty job and I’ll be the first to admit I used a little too much solder and left the heat on long enough for the solder to start oxidizing. It’s not bad enough to be fatal by any measure though, especially seeing as I’ve never been trained to do SMD.

I repeated the same process for the capacitors and resistors at the top. Then, next comes the chips. Instead of using the “reflow” type process, I decided to solder these by hand using a fine iron and fine solder. This was because the legs were wide enough to permit this and I felt more comfortable doing it this way …

… although the result does look ugly with too much solder (again) but not too bad. The chance of bridging with SMD chips is quite high especially without careful control of solder and a fine tip, so reflow may still be preferable. Do note the text orientation on the silkscreen versus the actual pin 1 orientation of the NE555 chip could be a trap for inexperienced constructors!

After that, getting the through-hole components mounted completes the kit. I found that some of the through-holes are a big on the large side, resulting in solder flow-through the board which varied depending on how long you heated the joint. The last LED’s footprint was also slightly chopped off, suggesting just how “tight” the design is and possible limitations in quality control during manufacturing. It is, however, just a minor defect.

In the whole kit, the only spares you get are one of each type of resistor. I guess LEDs are a bit too expensive to give you spares of … but that’s fine.

In the end, total construction time was 40 minutes including time for photographs and distractions. The result is an LED which chases from one end to the other, with the speed configurable using the trimpot.

As with other versions of the kit, if you wanted fewer states, you could hook-up the next state to the reset line on the CD4017 to truncate the count. This would have sufficed to create the “windmill” kit constructed earlier in a way that would not require a microcontroller.

A quick analysis of the 555 shows it has an astable vibrator configuration that results in:

C = 1uF
R1 = 2kR
R2 = 10kR - 60kR
f = 65.591Hz - 11.828Hz

Chase Rate = 6.5591Hz - 1.1828Hz


For AU$1 posted, it’s a good skills tester or practice kit to get some experience doing SMD soldering while building what I consider to be a classic kit. The board quality is quite good and all the parts were included with a spare resistor of each sort (in case one pings out and gets lost). It’s best done with a hot air gun and tweezers at the minimum, but it’s also possible to build it with two soldering irons or even one if you’re willing to accept a slightly less pretty look. I was quite surprised to find the venerable 555 timer available in SMD form and fairly pleased that I was able to complete the kit straightforwardly. I wasn’t that confident starting out, but look at the result!

In the meantime, I’ve still got quite a few more kits on the pile for when I have some time to build and analyze them … so you’ll probably be seeing a few more in the near future.

Posted in Electronics | Tagged , , , , , | Leave a comment