Project, Reverse Eng: Hex3653 AV2B Stereo FM Receiver Kit

When I built my last radio kit, I lamented that there were some kits out there which were mainly black-box which don’t offer that much of an opportunity to learn about what is actually going on. While I was dismissive of its educational value, I thought it might be a nice way to occupy an hour of an afternoon just to see what its performance was like.

As a result, I ordered the Hex3653-based kit, a stereo auto-scanning FM receiver based around a single chip with push button volume, scan and power controls operating off two AA batteries. The kit cost AU$6.95 including postage, which wasn’t what I’d consider expensive. It arrived today, so I decided to build it right away. It wouldn’t take long …

The Kit


When I received the package today, I was a little surprised. An anti-static bag of PCB and bits, and a two-AA battery holder. That’s it. No instructions.


Inside, you have a tin-plated PCB with solder mask and silkscreening on one side. The component “type” is indicated with polarity, but the component values are notably absent.


It’s a double-sided PCB with solder mask on the rear but nothing else. Of note was that the pads themselves for some through-hole components were also a little on the small side for comfortable hand-soldering, so probably not the best “introduction” to building electronics especially considering the SMD IC as well.


The rest of the components, as sorted out. All the diodes are one type, all the electrolytic capacitors are one type. One of the ceramic capacitors was 33pF, the other two were 0.1uF. Three resistors, all 100k ohm, an LED, an inductor (in green), a 32.768khz watch crystal and a 3.5mm headphone jack. Aside from that are a row of header pins and a jumper shunt. There’s not many components, and I suppose, if you were to use process of elimination, you could probably work out almost all of the components.

As with all kits, you’ll need to BYO solder, soldering iron and side-cutters.

The Build and Testing

The seller offered helpful hints on the listing – their suggestion was to start with the SMD chip first and I think that’s a good idea. I started on the buttons first because I was concerned about the orientation (which turned out not to be a major issue as the legs were rectangular in arrangement), which resulted in limited access to the SMD chip and hence an ugly solder job on one side. Still not too bad. I mounted the components as I saw best and used some of the seller images for guidance, bending the crystal over in the opposite direction to avoid shorting the pins on the chip.

C3 and C5 were both 0.1uF, with C4 being 33pF. C1 and C2 were both electrolytic 100uF. R1 was 100k, as was R3 and R4. That’s all pretty self-explanatory … but …


then I came to a problem. Namely that R2 was supposed to be a 2.2kohm resistor according to the sellers’ image, but I had only a 1N4148 diode remaining. This is another case of Chinese kit sabotage – missing or incorrect components. This would probably be quite frustrating for a beginner.

I contacted the seller and he agreed to send me AU$0.48 so I could buy it from Jaycar (based on their prices), so I rummaged my junk box for one. Thankfully, I had some resistor strips left over from a pack … so I completed the kit in the end.


Note that there isn’t a need for the solder to flow through the plated-through-holes entirely, although in most cases, it had flowed through nicely.


Not a bad job, I thought, even if some of the pads were a little on the small side.

I stuffed some batteries in, and powered it up. It worked first time with no fuss, and tuned the FM band completely. The audio actually sounds very good for quality, which is nice, and had decent sensitivity and auto-muting of weak signals so no harsh static. From my testing, the kit consumes 39mA from the cells when running with output to the headphones at the lowest level.

However, there were a few niggles. The unit has a loud pop-on/pop-off because of presence of DC on the audio outputs, and the output is quite loud even at its minimum volume (which is a common annoyance on digital volume control based audio products). Other issues include the scan being a little “slow” locking into a station, the LED being extremely bright (almost blinding), the included battery holder being weak with the ends somewhat bending under the stress of the springs and the lack of casing which means that the unit is more of a module rather than a full radio and isn’t as durable or attractive as it could otherwise be.

I suppose that’s all par for the course when it comes to cheap and cheerful kits. The fact that it’s simple and straightforward to construct means that the chances of success are high which is a good thing. That is counterbalanced by the need to solder an SMD component and the smaller pads makes things difficult, as well as the lack of instructions. The educational value of building this “black box” receiver is mostly limited to the experience of soldering and inserting components, but the result is somewhat usable which is nice.

Reverse Engineering

To try and derive a little more knowledge from the design, I sat down and took some time to draw out the schematic. It took me three goes to get a good readable drawing by hand – note that I didn’t spoil the surprise by looking for the datasheet until after just to make sure I had it right.

The first step to doing it was to actually analyze both sides of the board at the same time. My favourite way of doing this is to take the original image of the board, flip the underside image and align it. Then I invert the colour of the underside of the board while mixing it with the top side at a lower opacity so that traces on the top come up one colour and traces on the underside come up another colour. This makes it easier to work out the connections as this PCB uses “spider” style connections from pad-to-pad top and bottom.


The resulting schematic looks like this:


To try and make it a little more readable, the Vcc lines are coloured pinky-red, the GND lines are coloured blue, the antenna signals are coloured orange and the two audio channels are coloured lime-green and turquoise. Some fixes were made to my drawing in Photoshop, as I did make some mistakes and omissions during my hand drawing which I later realised.

Lets start first by analyzing the switches – the Volume -, + and Seek -, + are all pull-down inputs. The switches merely connect the corresponding lines to ground to activate the function. The power switch is a little different, with the pin being “held” by a capacitor, which is shorted to ground through a 100k resistor. This design is probably so as to result in the power-on signal being a particular length of time (as the capacitor charges) to ensure reliable sensing and avoid stray interference potentially triggering the input.

The antenna input is routed through the ASW header which allows you to select the antenna. In fact, the antenna input is on pin 2 of ASW, with pin 1 connected to a pad labelled A (for external antenna) and pin 3 connected to the shield connection on the 3.5mm headphone jack for using the headphone lead as an antenna. By using a jumper shunt over 1-2, you can select the external antenna, and 2-3 allows for the selection of the headphone lead. Pin 4 is connected to ground – its function doesn’t seem to be strictly necessary or useful except possibly in the case when connected to a balanced antenna.

The antenna input goes through L1 and C4 which acts as a filter (although I suspect the values might be a bit suspect), and the two diodes “back to back” which acts as a crowbar to shunt transients (e.g. inducted lightning) and limit the antenna pin voltage to ~0.3v to protect the chip from damage.

The frequency reference for the system is put in via pin 9, and driving the oscillation is handled internally, as is all the IF, synthesis, mixing, and stereo decoding. Hence there’s really not much to see.

Audio comes out of pin 12 and 13. This is loaded down with a 100k resistor, and passed through a 100uF electrolytic capacitor for DC blocking. It appears that the outputs have DC bias on them as the chip is operating “single ended”, so that the negative part of the audio signal can be produced.

This effect seems to be exploited for the LED which indicates the unit is on – the LED is hooked to Vcc directly, with the negative end connected through an SS8050 NPN transistor to ground. The transistor’s base is driven by the left audio channel’s DC bias through a 2.2k resistor – when the chip is on and the output is silent, you’d expect the DC bias to be present on the output and this activates the transistor. The “turning on” of the bias also explains the “loud” pop that occurs during turning the unit on – as the transient causes the electrolytic capacitors to “charge up” through the external impedance. In which case, it would probably be nicer if the 100k resistors were on the other side of the capacitor. I’m not sure if that’s a design mistake, or whether the 100k loading is even necessary for the chip to maintain a stable output.

The B-header allows for use of the module in an integrated way. Power is handled by the first two pins, followed by audio (right, left, ground).

A broken copy of the Hex3653 datasheet was found, which wasn’t particularly helpful. That said, I was able to find another site where the pin-out for the chip was described and it is consistent with the schematic I’ve drawn, confirming that my understanding seems to be correct.


This kit is another cheap-and-cheerful kit from China, and is one provided without any instruction. It’s built around a system-on-a-chip with no real adjustments or insight into how the system works, which makes the kit somewhat limited when it comes to educational value. The board is nicely finished, although some of the pads are a little small and the SMD component makes the kit less suitable to absolute beginners. The biggest disappointment was incorrect component packaging, meaning that the kit could not be completed just with the supplied components. The resulting product produces good audio quality with good sensitivity and noise-muting, however, does have an overly-loud output, overly-bright LED, pop-on/pop-off noise, slow scanning, weak battery holder and limited robustness without a physical casing.

For the price, it’s actually not bad considering the price is inclusive of postage. It kept me entertained (and slightly frustrated) for about an hour, and then another hour as I tried to work out the connections. I hope this page is of help to whoever may end up with such a kit, as the sellers’ images were out of date, showing an earlier revision of the board with slightly different layout and components. Maybe it will help you learn something about the design of the circuit.

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Electronics and tagged , , , , , . Bookmark the permalink.

3 Responses to Project, Reverse Eng: Hex3653 AV2B Stereo FM Receiver Kit

  1. Philip says:

    All the resistors are 10 K, I placed a 470 resistor in series with the LED.
    Works well, and at £2.40 a bargain.

  2. Philip says:

    Looks like the earphone lead works as the antenna.

  3. James says:

    Thanks for your blog, I was able to assemble it based in your guide rather than having to wait for the Chinese ebayer to email them to me. I also had a missing 2.2k resistor :(, but after adding one all is good. Radio sounds pretty good.

Error: Comment is Missing!