The Arduino had a big hand in making people play with electronics and create their own little devices to solve their problems. A desire for low-cost wireless connectivity has led to many people purchasing low-cost 433.92Mhz ASK/OOK modules to create their own low-cost wireless data links. Many of these users, unaware of the caveats, assume that it would be just like having a cable, or using Wi-Fi, but in reality, such links can be finicky to set-up and difficult to ensure they operate reliably.
The whole 433Mhz “low interference potential” devices band is actually quite a nasty place to operate. It’s inside the amateur radio 70cm band, and can suffer direct jamming as a result of these much more powerful transmissions. They are also used primarily by a large set of different remote controls, such as the central locking/alarm key-fobs for cars, garage doors, wireless doorbells, wireless temperature senders and energy monitors to name a few. They also are used by early wireless headphones and IR extenders. Because the ASK/OOK modules are simple and not greatly frequency controlled, there is no concept of channelization, and the whole band is essentially “one channel”.
This necessarily means that interference is to be expected, and at best, half-duplex operation is the only mode available. Further to this, it’s also anticipated that such links cannot be utilized on a continuous basis, and instead, should only be used in very short and infrequent bursts to avoid jamming other transmitters. With these restrictions in mind, they can still be quite useful, however guidance on actually using them for a data link is outside the scope of this post.
Instead, this post will be looking at four different low-cost sub-AU$5 433.92Mhz ASK/OOK transmitter modules. In many cases, you may be interested in remote controlling devices – for example, a bunch of 433.92Mhz ASK/OOK remote controlled mains switches, where the receiver is already existent and fixed. A common problem with such devices is users complaining of insufficient range or unreliable transmission. To this end, many users seem to suggest changing transmitter module to a particular module they think is best and potentially increasing the voltage to the transmitter.
I wondered whether there was significant differences between modules in terms of their output and how the output scaled as a function of input voltage so as to make a decision whether the benefits of going with a multiple-voltage power supply (or adding a regulator) outweighed the additional complexity and cost.
Four modules were used in testing. Three of these modules were very recently purchased from eBay, directly from China, whereas one module I have kept in spare for many years and is likely no longer widely available for purchase and is used as a benchmark.
This module is the reference module, and is the unit I have had in spare for many years. This module, like many others, utilizes a surface acoustic wave (SAW) resonator and what appears to be a Pierce oscillator. The unit uses surface mount transistors and inductors on the rear for filtering and amplification. The unit claims a voltage range of 3-12v with an output of up to 16dBm and data rate up to 3kbps. The module has all of its connections along the bottom row, although instead of using nice golden header pins, it uses flat strips of thin metal which has reliability issues with connectors, so I soldered golden header pins to it instead. The unit is relatively small, about thumb sized.
This unit is a very popular one, and is pretty much the “anonymous” module you get if you purchase a $2 RF Link kit. This one also uses a SAW resonator, but in a larger TO-39 can. Data connections are made from the bottom side, with the antenna on the top left. Inductors appear to be hand wound, and the PCB is roughly twice the size of the Summitek module above. A datasheet for the module isn’t easily available, although generally, most webpages suggest it takes 3-12v, with an output of 14-16dBm and a claimed 10kbps data rate.
This rather anonymous looking module was a challenge to track down, but ultimately, it is a 2.1-5.5v module with a 16dBm output power at 5v, with a data rate up to 10kbps. Again, a SAW resonator is used. This module was incorrectly listed on eBay as a 3-12v module with a 500mW output which would have been a whopping 27dBm! As a result, I ended up incorrectly pushing the module beyond its intended design during testing, but it lived to tell the tale. The antenna and data layout is similar to the FS1000A above, being them on opposite sides.
Shenzhen LC Technology SYN115 F115
This was the smallest module of the bunch, and is different as it is a low-voltage optimized module that uses crystal synthesized frequency (32 multiplier) rather than a SAW resonator. The unit uses an intelligent IC from Synoxo, targeting battery operation. It delivers 10dBm, operating between 1.8-3.6v with a data rate up to 10kbps. This module has all the connections on the same side, with no header pins fitted, but for some reason, the antenna pin spacing is not a regular spacing.
All modules were prepared by soldering gold header pins to their connection points if not already provided. As testing looked at different voltages, the supply voltage was provided by a Manson HCS-3102 switching bench-top power supply. However, this was not sufficient, as the data pin input must not transition above Vcc supplied to the module AND must reach at least 50% of Vcc to register a logic 1. As a result, a MOSFET based circuit was used to switch Vcc to the data pin so as to satisfy these conditions for all range of tested voltages despite the data signal being generated by an Arduino Uno.
The MOSFET switching was tested and proved to be fast enough for the small currents involved, reaching the target level within 3us as tested with the PicoScope 2205A.
The test consisted of the module being keyed on for 1 second, then off for 9 seconds for a 1:10 duty cycle. The output was connected to DuPont jumper wire, which was clipped to the end of a BNC oscilloscope lead which was then connected to the Tektronix RSA306 real-time spectrum analyzer. Because the modules did not have proper RF connections, and I wasn’t going to solder leads to each and every module, this method of coupling is lossy and cannot provide the absolute power output level. The DuPont jumper wire was quite an efficient radiator, with the estimated loss being 24.8dBm by comparing the measured power values with datasheet values. Even without this correction, it still allows us to determine the relative power output between modules.
The power and centre frequency were measured by pausing the analyzer while the transmitter is keyed, setting the marker to the peak value, and reading off the relevant values. Resolution bandwidth was set to 1khz for the analysis, although as the output should be keyed on as a CW with zero bandwidth, this setting should not make a major difference to results.
Transmit Power vs Voltage
On all modules, increasing voltage caused an increase in power output in most cases. However, some interesting behaviour was observed. The Summitek module actually consistently decreased in power output above 11v. This appeared to be because of amplitude overload of the SAW resonator, resulting in increased energy in the spectral noise component and reduced power in the carrier. The filtering of the module may be a little inadequate, as spectral spurs about 300khz either side of the carrier can be seen in the output. As it turns out, the Summitek module I am most familiar with has very good power output by comparison over the majority of the tested voltage range.
Summitek ST-TX01-ASK at 5v
Summitek ST-TX01-ASK at 12v
In the case of the FS1000A, the transmitter was a bit cleaner with no spurs, with some raised noise near the carrier as expected. When testing, the transmitter became unstable above 12v, resulting in strange drop-outs in transmission. On the whole, even though it is the cheapest module, its performance is not too bad, as at 5v, it’s about 3dB (half the output power) of the Summitek, and at 11v, it’s about 6dB (a quarter of the power) of the Summitek. I suppose this may explain the difference in range experienced by some users.
XD-FST FS1000A at 5v
XD-FST FS1000A at 12v
The Wireless-Tag RF-T-ASK module, having not known its true voltage range, was observed to latch into continuous output above 10v, heating up the module significantly. However, the module had the best power output, at 5v, it was about 3dB ahead of the Summitek and 6dB ahead of the FS1000A. This would be a great module to choose for 3.3v or 5v Arduino operation as compared with the other tested modules. Its filtering is also quite decent.
Wireless-Tag RF-T-ASK at 5v
Of all modules, the SYN115 produced the lowest output power as expected from the datasheet, although it was relatively consistent over the voltage range. It seems that the output was even lower than expected when tested, falling 6dB behind the Summitek reference at 3v, and about 8dB behind the Wireless-Tag RF-T-ASK. The brown-out detection caused it to fail to transmit below 1.7v. Despite the datasheet claims of it being a very suitable device for battery operation, it seems that the output power lags behind resulting in range limitations. The synthesized nature does result in a different characteristic of the output, with wideband noise around the carrier, rather than shaped noise near the carrier, although the noise is still about 40dB below carrier.
LC Technology SYN115 at 3v
Other modules did continue transmitting down to 1.1v or even 0.8v, although many of the modules did have a particular voltage which causes the output to oscillate unstably (e.g. Summitek at 1.3v).
As a rule of thumb, increasing the voltage did increase the output power, however, the gains diminished as the voltage increased, and at a point, the Summitek even exhibited a reversing trend with the output power decreasing with additional further voltage increase. At 4-5v, the 1v change causes a 3dB output change on most modules. Above 5v, this increase is close to halved. About 7.5dB extra transmit power can be had from the FS1000A by increasing from 5v to 12v. About 10.5dB extra transmit power can be had from the Summitek from 5v to 11v. As a result, it can be quite worthwhile to increase the supply voltage if practicable.
Center Frequency vs Voltage
Frequency stability, while not of absolute importance due to the wide acceptance range of the receivers, can be seen to vary quite significantly with SAW based devices at the low end of their operational voltage. However, the stability of the synthesized unit wasn’t clearly better than the SAW units when tested in the air-conditioned environment.
It seems that the low-cost 433.92Mhz modules can vary quite a bit on output power even at similar cost. The Wireless-Tag RF-T-ASK seems to be a good choice for direct 5v/3.3v powering, and produced the highest output power as tested. The output at 5v matched the FS1000A at 10v, and the Summitek ST-TX01-ASK at 7v. Of the other transmitters, the Summitek and the FS1000A both benefited significantly from increasing voltage, as long as it was not overly increased to the point of transmitter instability and reducing output due to overloading the resonator. Of all the modules, the LC Technology SYN115 seemed to produce the lowest output power, and despite touting the battery friendly voltage range and synthesized frequency output, it seems that it was at a significant disadvantage for output.
Of course, output power is not everything, as the cleanliness of the keying may be of importance if dealing with higher data rate signals, however, all modules with the exception of the Summitek claim a 10kbps data rate.
One key way, as with all RF based devices, of increasing range is to use a proper antenna of the right length!