Them pesky cheapo 433Mhz transmitters!

[Blueteeth]

Hi, long time no post! I realize this is RF circuitry really, but as I'm sure some here have used these devices, as well as it is more general 'electronics debugging' than specific to RF, though I would post here.

Well I had a recent project that required a simple remote (PIC micro job), using 433Mhz OOK/AM modules. I bought a bunch of super cheap ones that are everywhere these days:

**broken link removed**

Now, after much debugging, and problems I eventually went with better receivers (superheterodyne ) from Quaser.
http://www.maplin.co.uk/p/quasar-am-super-heterodyne-receiver-433mhz-n66de

The original cheapo ones worked! but where inconsistent, and I assumed it was the fact they weren't crystal/resonator controlled, so these new ones from maplin are much better. I have enough confidence in the frequency stability of these to use them as a 'benchmark' for testing the transmitters. (that is to say, that are bang on frequency, 433.92Mhz).

Now, I am having issues with the transmitters from the first link. They are deceptively simple, and despite using a 433Mhz resonator, I suspect their frequency isn't particularly accurate, nor is the output power controlled well. Essentially, one of my transmitters works (with the maplin Rx) very well. With a stubby antenna on the tx, a helical antenna on the Rx, I can get 60m+ range, without errors. However, the other transmitters don't seem to communicate at all with the receiver - that is until .....I touch one of the coils with my finger. Whilst I am not exactly an expert on RF, I know this will add small capacitance to ground

Here's a close up of the tx/rx pair. Right now I'm just concerned with the Tx.
http://danyk.cz/avr_btep03.jpg

A schematic found on Google of the tx - with different values, and C1 is not populated on my board, but this is more or less it.
http://www.hobbycomponents.com/images/forum/MX-05V.jpg

The only way this will transmit anything, is if I touch the larger coil, on the schem, that is L1, which is 7-8 turns, going from VCC to the oscillator.

My question is, although it is a rather crude oscillator, could it be that me touching the coil adds enough capacitance to 'detune' it back to 433Mhz? And as it *appears* that this coil is just feeding power, is it simply a filter, so decoupling the VCC to ground would provide more accurate frequency?

I am just unsure why one of these transmitters worked so beautifully well, but the other 4 I have all exhibit the above behavior, even though all the values appear to be the same. I don't have the facility to measure the frequency output of these, only real test I have available to me is using the stable Rx as an indication.

Any analogue guru's who knows pierce oscillators please chime in They are so cheap I should just buy 'better' ones, but curiosity got the better of me, I like to know 'why' things are inconsistent

BT

 

[MikeMl]

I have looked at these and others on my spectrum analyzer, and their operating frequency is all over the map. According to the regulations, the manufacturers are obligated to meet this spec: 433.050-434.790 MHz ISM LPD433 (70-centimeter band).

As a practical matter, their operating frequency doesn't matter much, because the matching receiver is a super-regenerative TRF reciver with only a single LC network to define its operating frequency, and it has the characteristic that it will lock onto the strongest signal it sees within +-5Mhz of its nominal frequency.

I am a licensed ham, and can legally transmit between 420MHz and 450MHz. Even a 2w walkie talkie transmitting 10Mhz from 433MHz will block the superregen receiver and prevent it from decoding...

Some of your problems are likely caused by unrelated signals that happen to near 433Mhz, especially if you are in a large metro area.

I have had much better luck with the 315MHz ISM band.

btw- putting a decent (1/2 wave dipole antenna) on both the transmitter and receiver will help a lot.

 

[4pyros]

Mike; Can you point to better RF sets, like at mid range pricing.

 

[Tony Stewart]

Finger capacitance to ground will lower Saw filter resonance, so an enclosure is essential for optimal performance with shielding and an access hole to the tuning cap using a plastic screw driver with tiny metal blade.
THis can be simply done with tin plated brass or copper, cut folded and soldered with short resistor wires< few mm to main board ground plane on bottom around the box. All seams need to be sealed with solder. Or buy an off the shelf can with lid and add tuning hole. Keep at least 2cm away from coil to avoid loading the coil and SRF issues. HInt... the two coils are orthogonal for a reason to avoid coupling and any reflections off inner wall of shielded enclosure, must respect these reflections as well.

The same principle applies to Xtal oscillators which are less sensitive to a few pF stray capacitance at lower frequency.

 

[MikeMl]

Finger capacitance to ground will lower Saw filter resonance, so an enclosure is essential for optimal performance with shielding and an access hole to the tuning cap using a plastic screw driver with tiny metal blade.

My point is that the actual transmit frequency is likely to be pulled by hand capacitance, battery voltage, temperature, etc. However, that will not cause the transmitter to get outside the receiver's capture range. The real underlying problem is that the receiver will lock to the strongest signal it sees, be it from your transmitter, or something else. If you live on a ranch in Arizona like I do, it works most of the time. In a busy metro area, the receiver is picking up all kinds of nearby stuff, which makes for low probability of a message being received properly.

 

[Blueteeth]

Hi guys, thanks for the replies!

Yes, I did do some reading up on types of receivers, superhet's, vs super-regen, that's what swayed me to go with the superheterodyne receivers, which btw, use this chip:
**broken link removed**

That receiver picks up a LOT of noise (as any receiver would) but it happily picks up transmissions from one of my tx's, and the packet decoder is happy. I realize the 433/434 band is crowded, almost every keyboard transmitter I've seen uses it, but I had reliable results with a sufficiently robust protocol. The cheapo super-regen ones had less noise on their output, but *do* work with the cheapo transmitters, they just required very careful tweaking, and with a good antenna's, didn't have the best range. Cheers mike for the info about supergens being less selective.

About antenna's: using a helical coil on the Rx, mounted in a plastic box, horizontally, kept at least 4cm from any wiring. The tx is mounted in a plastic handheld enclosure, (actually the tx board is glued to the back of my main PCB...) with a 3cm length of 50 ohm coax going to an end launch SMA, mounted on plain FR4 (core soldered directly to the SMA pin, rest is ground). All connections were checked, and double checked for shorts/open circuit.
Antenna used is this: http://uk.rs-online.com/web/p/telemetry-antennas/7934345/

But I am happy with the receivers I have - the problem now is the transmitters!
So, if Tony Stewart is correct, then these transmitters seem to operate at a slightly higher frequency - the receiver only picks up the packets I send if I touch the larger coil. Doesn't matter if I touch the other coil, ground, power, data etc.. only that larger coil, and if I use a bit of pressure. They are insulated copper coils, enameled, so it isn't an electrical connection, at first I thought that I was simply acting as another antenna, but then began thinking about the detuning thing, as the coils are clearly part of the resonant circuit. I guess the one decent transmitter I had just so happen to be spot on frequency !

I could mount these in a small metal can enclosure, with only power, and data in, and antenna coax coming out (with the can of course being ground), and will try that. Space is tight as the transmitter enclosures I'm using, plastic, are only 10mm high, 20mm by 110mm, so I don't think I can abide by the '20mm' distance from metal enclosure to coils... My thinking was that the boards being double sided ground plane would be fairly immune to proximity effects (few mm) from ones hand, as long as it I used a shielded coax straight to the SMA and kept power/data wires short ( <30mm). The 'working' transmitter - which I no longer have as it is being used by the customer right now - was identical in construction, perhaps it was a different batch?

So, although I will buy better transmitters (I didn't expect such cheap devices to work anyway, was more out of curiosity!) would be nice to see if I can mod these ones for better frequency stability. Perhaps attempting to add that unpopulated 0805 capacitor and see what different it makes, or measuring, then replacing the coils with an SMD inductors (mounted on FR4 to convert to through-hole). After Christmas I think I'll invest in a pre-scaler chip and knock up a reasonable frequency counter to measure the center frequency. In the meantime, I cannot find software to simulate the circuit, even when using LC equivalent circuits in LTspice/spice it fails to oscillate.

 

[Tony Stewart]

My point is that the actual transmit frequency is likely to be pulled by hand capacitance, battery voltage, temperature, etc. However, that will not cause the transmitter to get outside the receiver's capture range. The real underlying problem is that the receiver will lock to the strongest signal it sees, be it from your transmitter, or something else. If you live on a ranch in Arizona like I do, it works most of the time. In a busy metro area, the receiver is picking up all kinds of nearby stuff, which makes for low probability of a message being received properly.

My point is if you put on a Faraday Shield, fingers wont affect it at all and it will be more stable to stray interference. near field capacitance of the hand will lower the frequency as will the shield, so it is designed for some shielding. But the shield will block hand waving interference. Either way there is a tuning cap to optimize reception at resonance. IN the 70's our 400MHz transmitters were mounted in solid milled aluminum cases with a lid and access holes for all tuning parts. The coils are air core high Q and should not be moved .

 

[Blueteeth]

Hi,

Just wanted to point out the transmitters aren't tunable, no tuning caps, in fact, the only cap on it is the coupling to antenna, then its two inductors, SAW resonator, two resistors and a transistor!

 

[MikeMl]

If you believe in the SAW, then you will have a hard time pulling the transmitters very much. Since you are using superhets, then they have a much narrower bandwidth, and the transmitter must match the superhet receiver. Surely, the mfg. of the superhet sells a more accurate transmitter to go with it. You shot yourself in the foot by buying the cheapo transmitters and using them with an expensive receiver.

Get one of these to use to determine the actual transmit frequency of your transmitters, or do the repurposed TV-tuner SDR

 

[Blueteeth]

Hi again,

Yeah I think I probably did 'shoot myself in the foot' using an unstable (well, in terms of frequency) transmitter, along with a very selective receiver.

Well, now I have decided to go with a better transmitter for any more of these units my customer wants, I'm stuck with 4 x tx/rx pairs to play with! As an exercise in RF design/debugging, I'm trying to simulate the tx in LTspice - not the best software for that task, but seems to be sufficient for checking if it oscillates and what frequency. Using a combination of measuring, tracing, and yes.. guess work... I believe I have a half decent schem of the FS1000A transmitter:



This only includes parasitics of the SAW itself (the two 0.5p caps to ground). I suspect these modules actually rely on the parasitics on the PCB to get it to oscillate.... The output of this is anywhere from 800Mhz to 1.1Ghz, although with the above values, 883Mhz (close to 2nd Harmonic of 433?) seems to be the peak.

Took a lot to get it to oscillate, seems this crude incarnation in LTspice requires a load, and the output cap to be 1.5-4pF. I noticed on one device, if I put a 5pF ceramic cap in the 'unpopulated' footprint, the transmitter worked well with the superhet receiver - but *only* when no antenna was attached. As soon as I put a 50-ohm stubby on there, the receiver stopped getting anything. As the output is directly from the collector I'm guessing it is very easily pulled off frequency.

It is odd, most of my googling showed the most common version of a 433Mhz SAW transmitter is a colpitts oscillator, this appears to be a pierce, one that isn't well designed. I could just take the SAW resonators out of these and knock up my own 'better' transmitters, but would be nice if one could work out a matching network, or at least some tweaks to get these to have better frequency stability - be it replacing the inductors with SMD, adding shunt caps etc.. I'm out of my depth, but the first step is knowing how and more importantly *why* these work, then some mods to see if they can be improved - or at least tuned.

BT

 

[Blueteeth]

Righty,

Haven't forgotten about this post, just spent a while testing other 'cheap' transmitters/receivers, as well as PLL controlled ones, or those that do 'on chip antenna tuning' for output impedance.

Safe to say, a good combination of frequency stability, antenna matching, antenna placement, data protocol (in the presence of noise) are all required for any OOK/ASK system. In my search I found many many posts in other forums about people having trouble with these, so, when I finally get some test equipment delivered, should I post the results here? or make a sticky about how exactly to get the cheaper modules working? I realize Nigel already has an excellent tutorial on such devices/modules but if anyone is interested. Note some of the forum posts were from arduino fans, who may not know that much about RF/electronics, and included some dodgy info.

 

[upand_at_them]

I've used the HopeRF units in a metro environment with good results. They're also channelized. The RFM12/22B is popular. Or was; I think HopeRF has some newer/better ones out now.

 

[alec_t]

only that larger coil, and if I use a bit of pressure.

Perhaps applying pressure moved coil turns closer together and changed the coil inductance to tune it better? Have you tried deliberately squashing/stretching the coil a tad?

 

[granddad]

I started a PIC remote monitor project with simlar 433Mhz TX RX parts , could not get consistent results , eventually got two AUREL RTX MID 5V RF MOD, TXRX, ASK, 5V, 433.92MHZ and a couple of ceramic antenna from Rapid. a lot more £££ but result sucess...

 

[MikeMl]

OTOH, I am using the el-cheapo EBat 315Mhz key-fob transmitters and the matching receiver and with an outside dipole antenna on the receiver to control a drive-through gate and my garage doors. I am consistently getting a range of > 250m (from inside a car). I use the key-fob transmitters and the receiver with the built-in encoder and decoder ship set. I use the four-button remotes and the four output receiver...

 

[Blueteeth]

The fact I got a few replies, tells me quite a few folk use various types, including the basic analogue cheapo ones... which proves my point that it maybe helpful to collate some info about them, and getting them working!

upand_at_them: Ahh yes I've seen a fair bit of info on them. I'd consider those another class of radio modules, intelligent, with decent antenna matching (analogue support circuitry) and single chip functionality. I have no doubt they are a joy to work with as the user interface is all digital. Other examples are the old 'EasyRadio' types which are modules containing a PIC, and tx/rx/trx - they all take the sting out of it, allowing one to pretty much communicate without touching 'rf' or the protocols they use. NRF24L01+ is in the same vein for 2.4GHz comms. All are a joy to work with, giving very good range, even with poo antennas.

@alec_t : When my frequency counter arrives, I'll do that test and post results! Given the transmitter in my original post was so basic (I don't think they could make one with less components) it's frequency is bound to be extremely sensitive to any change in those coils, parasitics, supply voltage etc.. I originally thought that rather than being poor antenna matching (possible) it was simply that they had awful frequency tolerance/stability, and as MikeMI pointed out - using super hetero receivers with narrow selectivity meant it was likely that anything that pulled the transmitters off-frequency would pretty much kill the reception. As that was my only available test (the Rx's I have have no RSSI output...) it was awkward to test that as I could only get a 'yes/no' from the receiver.

@granddad : Yes, I too have used some from rapid, with similar results to the original. The module themselves are probably quite high quality, with the Tx again being a colpitts oscillator. However, that again means that the antenna impedance can pull the tx off-frequency if not well matched, and the receiver spitting out random noise when no transmitter is on. That second problem can be fixed with half-decent radio protocol (preamble, a good sync pattern, careful selection of on/off periods) but it just makes like difficult in 'trying' to get things to work. I checked out the ones you recommended:

https://www.farnell.com/datasheets/332708.pdf

Noticed it has a 'buffer' between the oscillator output and antenna matching. That's very much a good thing as it means that whatever antenna you wack on the output, will only affect power output (range) but shouldn't pull the oscillator off-frequency. Its a colpitts oscillator again, you can tell by the layout of the pic on the datasheet I'll add those to growing list I'm hoping to 'review'.

@MikeMI : It seems that the chinese manufacturers fully test the fully-made keyfobs. Also, I'm guessing they use small printed loop antennas. With the antenna built in, and the fairly good tolerance of PCB track width, to me that said that they should be pretty damn consistent - shouldn't require tuning as the antenna forms part of the oscillator, so with better manufacturing tolerance > better frequency matching > more consistent. With modules that don't have a built-in antenna and just a pin... whatever one connects to that, can pull off the frequency significantly, so even super-regens don't get a good signal. I will however keep those in mind for future projects that are single/dual channel. Cheers!

Apologies for a long reply, just thought given how many people are using the various types of modules, of varying cost, and my own battle with them without having RF equipment, someone needs to sort the crap from the good! And now I have at least 8 of these cheap ones, I'm sure I can get them to work - just requires tweaks and some caveats to what baseband encoding one uses.

BT

 

[alec_t]

I've recently had problems with a 433MHz wireless door-bell.
Originally, I had the Tx mounted on the internal face of our house door, but operated by an external button via a push-rod passing through a hole drilled in the door. The idea was to protect the Tx from the weather. That worked fine for a couple of years, but last year we had a new door fitted and she-who-must-be-obeyed has forbidden the drilling of holes therein! So, the Tx was duly mounted on the house exterior.
All well and good for a couple of months, then the Tx stopped working. New battery bought and fitted. No joy. I brought the Tx indoors and dismantled it, cleaned off some flux residue and .... it worked! Remounted it outside. Failed to work again after 1 day.
Thinks: it's sulking after being banished from its cosy environment, but is that due to cold or humidity? Bring it back indoors. After a couple of hours it works again! Ok, pop it in the fridge for a few hours to see if it's unhappy with cold conditions. Still working at 5C. Conclusion: problem is due to humidity/moisture. I suspect flux residues or moisture-absorbing pcb is causing a leakage current which upsets the Tx circuit.
Remedy: admitted defeat and today bought a new wireless door-bell. Time will tell if the fault re-occurs .

 

[MikeMl]

...
@MikeMI : It seems that the chinese manufacturers fully test the fully-made keyfobs. Also, I'm guessing they use small printed loop antennas. With the antenna built in, and the fairly good tolerance of PCB track width, to me that said that they should be pretty damn consistent - shouldn't require tuning as the antenna forms part of the oscillator, so with better manufacturing tolerance > better frequency matching > more consistent. With modules that don't have a built-in antenna and just a pin... whatever one connects to that, can pull off the frequency significantly, so even super-regens don't get a good signal. I will however keep those in mind for future projects that are single/dual channel. Cheers!...

The transmitters I use have a pull-up whip antenna about 5cm long. I get slightly better range with the whip pulled-up compared to collapsed.

I am using 5 transmitters going to just one receiver. It seems all five of the Tx are within the capture range of the single super-regen Rx, because they all seem to work reliably over ~ the same range. I live in a rural area, where the likelyhood of the super-regen listening to non-related signals is unlikely. If there are a lot of other RF emitters nearby, especially if using 433MHz instead of 315MHz, then the range may suffer.

The encoder/decoder chipset is the kind with four encoded inputs/four decoded outputs. It has a strap-able address code (eight tri-nary inputs 8^3 = 512) plus four unique commands.

 

[Blueteeth]

alec_t : I'm not surprised dude, whilst the self contained units like keyfobs and doorbells are more stable because of the printed antenna, they are still essentially SAW stabilized oscillators, which can still be pulled off frequency by parasitics, contaminants etcc. As I mentioned before, at first I thought it was a radiated power issue, but even you shield most of a unit like that, enough RF should 'escape' for half decent range, so I suspect its the carrier frequency. Do you have a frequency counter? Although its not designed to 'receive RF' you might be able to hook up a 50ohm cable to the collector of the transmitters transistor to peak at its frequency when activated (the 50ohm load might pull it way of frequency though..).

The easiest/cheapest debugging solution I thought of was purchasing an intelligent 433Mhz FM module, that has RSSI output, be it digital, or analogue. Then one-by one, tuning it to a channel, and simply looking at the RSSI - effectively making a crude spectrum analyzer. That would mean you can see its operating frequency, and output power - as well as just using a module already designed to pick up/amplify RF signals with a decent antenna.

I hear you on 'buying a new one' Ultimately, geeky curiosity aside... it becomes a question of '20 hours of head scratching, simulations, tinkering, and buying some equipment' or ... 'Spend $10 on a new one'. Sadly, I almost always fall for the first option.

@MikeMI : I would be interested to see the frequency output of those transmitters. As you're using super-regen, they could cover a fairly wide band and still be reliably picked up eh. Those chipsets tend to be the venerable PT2262/PT2272 pair. In a recent project where a customer wanted an 8-channel remote system for fireworks, because of time constraints (< 7 days) I ended up having to buy premade transmitters used for controlling power sockets - which used those exact chips PT2262. Had to hack them for different channels, and better address codes (010101010101 is not a good address code if you're trying to pick that up in a noisy area..). Safe to say the basic encoding they use is inefficient, but *ideal* for these OOK radio systems. I have the softwre for encoder/decoder somewhere if you wish to roll-your-own on a PIC or AVR. There is similar software others have done, but it tends to be very unreliable and has strict timing constraints ( = crap for sending the signals over RF where pulse widths are strentched/shorterned)

 

[granddad]

(preamble, a good sync pattern, careful selection of on/off periods)

I eventually found this out ! preamble ,I start a TX (8,n,1) with H'00 ( as inverted this turns on carrier ) next 3 bytes H'AA 55 and XX xx is Receiver ID ( has to match ), then 20 odd bytes data , includes : Time stamp, period (mins) till next TX,( allows receiver power down mode ) sum check and message number . was in .asm but now 'C'

I am using possibly older ASK version.
https://uk.farnell.com/aurel/rtx-mi...+5v|match|b|plid|&CMP=KNC-GUK-FUK-GEN-SKU-MDC