Post by ogrevorbis on Mar 17, 2015 19:26:35 GMT -6
So I am thinking about building my own FM transmitter. Most schematics I find online for FM stereo transmitters are way over my head. I'm not an electronics guru; I'm actually a programmer, so I'm having the most fun programming the PIC chip
Anyway, I am not really knowledgeable enough to build a full fledged stereo transmitter, so I began looking for transmitter ICs (better than the cheap BH1415F). Do you know of any FM stereo transmitter designs that are somewhere in between a BH chip and a fully integrated design (or maybe just a really high quality chip that I'm not aware of).
So far I've looked at the BH1415F and the KT0803L/K on a spectrum analyzer. The BH chip had much less spurious signals, but a pretty wide carrier and bad quality sound. The KT had pretty good sound, but the spurious signals were very bad (it was -25dB down across the entire band). There literally was a spur that covered the entire FM band.
I just recently discovered the RDA5820 and the QN8027. I looked at their datasheets and at least the RDA seems far superior. I've ordered both of them and will review them in a bit.
Just wondering if anyone has any input (suggestions for chips or a better idea)?
From my experience most of the stereo MPX all in one chips simply suck compared to using a well designed circuit that separates the audio switching IC from the 38kHz pilot generator and properly closes/opens the L/R lines individually as it is doing its high speed multiplexing. You really will not get the quality of a professional rig using the all in one ICs, but you can come close with proper audio/RF filtering. If you can find a premade MPX stereo generator board that is your best bet for audio quality. The problem with the all in one chips is the lack of oversampling and the internals being so small and close together that stereo separation rarely reaches above 35dB. With proper circuitry found in the more professional designed boards utilizing multiple ICs all working independently to get the job done you will get far better stereo separation and less hash from the switching above the regular audio range. As far as causing spurs and such, that should NEVER be an issue if the RF stages are designed properly. If you have spurs from the MPX IC getting high frequency crap into the modulation then there are bigger issues at hand. This is a common issue with Part 15 micro broadcaster transmitters. But at that low of power it usually is not an issue since no one is listening outside of a few hundred feet. Any proper MPX generator should filter out high frequency stuff above the audio range so that your RF doesn't become a mess. Band pass RF filtering will further reduce issues, though often with VHF gear we often only low pass filter the RF output.
TLDR: Save yourself the headache with these cheap all in one ICs and get a real MPX stereo generator board. If you absolutely have to use the all in one jobbies, read the specs and pick the one that has the least high frequency switching noise and the best stereo separation. Broadcast standards usually aim for 45dB+.
Post by ogrevorbis on Apr 17, 2015 19:03:09 GMT -6
Well, I finally got the chance to test all of those FM ICs. I didn't have to test all myself.
Updated. Maybe will soon try SN761634 (Texas Instruments chip) and BK1085.
EDITED: 8/2018
In order from best to worst:
1. QN8007
2. QN8066
3. BH1414K: >-60dBc across band, -50dBc +-200khz (decent sound quality, least spurious) 4. RDA5820: -45dBc across band, -60dBc +-200Khz (best sound quality, but ONLY WORKS FOR 5 MINS, THEN PLL RANDOMLY UNLOCKS) 5. BH1415F: -50dBc across band, -20dBc +-200Khz (but sound quality is worst) 6. Si4713: -30dBc, -25dBc +-200Khz - has a double carrier 7. AS6001/2: -25dBc (fills most, but not all of band) - also has a triple carrier -200 and -400khz 8. KT0803L/K: -25dBc (fills entire FM band with spurs) 9. QN8027: mostly -25dBc some -20dBc
I'm pretty sure that's all the chips on the market. Let me know of any others.
RDA5820 would probably be the winner if it didn't keep unlocking. Audio quality seems best on this one as well. Sounds exactly the same as the source signal to me, but in terms of spurious radiation the BH1414K is best.
The RDA5820 showed some nasty close-in spurs on the two boards I tested.
These ICs are NOT the way to do it. They're OK for their intended purpose - playing MP3s into a car radio - but they're crap for use in a proper transmitter.
My basic 10 Watt driver board shows no close-in spurs greater than -96dBc, no in-band noise at all, and second harmonic at -78dBc. It runs from a 12 - 15V supply and uses the 15W Mitsubishi FET for the output. It uses a simple 2-IC PLL for frequency control and fits into a standard tinplate box. There's an optional stereo coder that goes with the board, and RDS is easily added too. These are going to be made available at a reasonable price from the Netherlands in the next few months. There will be no point in wasting your money on those nasty Chinese spurious generators!
Most china made fm transmitters use one chip in the exciter to either produce the stereo or fm, they are garbage and best avoided no point wasting your money, either build a quality transmitter yourself there are plenty of decent circuits floating about on the net or buy good quailty modules leaving you just to box them up.
This 1 watt module on ebay seems ok as a starting point for anyone wanting to get going, both quality and price seem reasonable, you can then add on what ever amplifier units you wish or just box it up as is for a low power station, remember any good fm broadcast transmitter starts with a good exciter.
You're probably right about those all in one ICs. The RDA turned out not to be as good as I thought. The in-band spurs where -45dBc max, so not too bad, but not really professional. It also has a problem with the PLL randomly unlocking after about 5 mins of operation and I narrowed down the problem to the chip over a few months.
My basic 10 Watt driver board shows no close-in spurs greater than -96dBc, no in-band noise at all, and second harmonic at -78dBc. It runs from a 12 - 15V supply and uses the 15W Mitsubishi FET for the output. It uses a simple 2-IC PLL for frequency control and fits into a standard tinplate box. There's an optional stereo coder that goes with the board, and RDS is easily added too. These are going to be made available at a reasonable price from the Netherlands in the next few months. There will be no point in wasting your money on those nasty Chinese spurious generators!
Sounds very good. What IC are you using? Probably the TSA5511? Are you the dutchrfshop guy?
Are there any clear schematics of only the TSA, VCO and pic chip? I would really appreciate something like this because my knowledge of PLL + VCO design is limited.
I already have several amp stages that I'm going to use in my design. I actually finished my entire design based on the RDA and later found out the chips are defective, so now I'm looking for a better chip.
I have just made a test board with a QN8007 and power amplifier, and the output seems very clean; I can't see any spurs within a few MHz higher than -60dBc, looking at it with a pluto SDR. I'll check for far out spurs later and post the results. The power amplifier is a saturating type, so if there was any magnitude noise it would have been obscured, but phase-jitter wise it seems clean.
Note: controlling the chip properly through the i2c interface was a complete pain; the datasheet register listing is incomplete and misleading (there are lots of undocumented registers that are not even listed on the datasheet), but if you look up "chumbradiod github" and look at qndriver.c, you can get it working by adding your platform-specific i2c accessing code.
I have just made a test board with a QN8007 and power amplifier, and the output seems very clean; I can't see any spurs within a few MHz higher than -60dBc, looking at it with a pluto SDR. I'll check for far out spurs later and post the results. The power amplifier is a saturating type, so if there was any magnitude noise it would have been obscured, but phase-jitter wise it seems clean.
Note: controlling the chip properly through the i2c interface was a complete pain; the datasheet register listing is incomplete and misleading (there are lots of undocumented registers that are not even listed on the datasheet), but if you look up "chumbradiod github" and look at qndriver.c, you can get it working by adding your platform-specific i2c accessing code.
You've found it mate.
After I made that list I came to discover that the QN8007 is the best chip of all. The QN8066 comes in close behind. It has the advantage of being easier to solder however.
I made some PIC18 C source code to control it. It works fine. The programming was the hardest part of the whole process though. If you want a test board, I have one equiped with the QN8007 along with an MMIC producing 10mW and I have another with 100mW. I have a 15W, but I'm still working on it. I could sell you my test board if you want it. It is on professional PCB material. You need a 12V supply and a 3.3V I2C control using whatever MCU you want.
it is a small fm transmitter with mp3 player and class F amplifier, powered from USB; it outputs between 2 and 4 watts of power depending on an inductor value, and includes the lowpass filter as part of the output matching. The board is 4 layers, and was combined onto a 10pcs production panel that had some free space left over. QFN soldering isn't really a problem for me, and most of the passives on the board were pre-soldered by the pcb/pcba fab, so I could produce these boards very fast. I think I'll hand a few of these out for free locally to try to get people into radio transmission
I lost the picture of mine. The low power one is very similar except that it is controlled via an LCD instead of the USB port that it looks like you're using. If you're not already, I'd suggest using an FTDI (usb serial chip) and then using PC software to control the transmitter. If not, it would be pretty easy to expose the RX TX pins on the MCU and just get a separate USB to serial cable.
Post by ogrevorbis on Sept 4, 2018 19:36:25 GMT -6
RQA0009. We're using the same transistor. I was never able to get more than 500mW out of it even though I know it can do up to 4W. How did you figure out that matching?
The board is controlled using the 4 buttons and has pads for a spi oled module (currently not soldered). The UART pins of the micro is already used for the communications with the dfplayer, so it couldn't be used for control. The easiest way add usb control is to just switch to the stm32f103c8t6 ($1) which has built in usb interface.
For the output matching I just guessed at the impedances, given that the transistor die expects to see a resistive conductance of about 3ohms (calculated based on output power) in parallel with an inductance that tunes out the drain-source capacitance. Then I estimate the bond wire and lead inductance to be about 2 to 3 nH total, and then voila you know what impedance should be presented to the drain lead of the device. I use rfsim99 to do the simulations of the amplifier, using the S parameters of the transistor.
The board is controlled using the 4 buttons and has pads for a spi oled module (currently not soldered). The UART pins of the micro is already used for the communications with the dfplayer, so it couldn't be used for control. The easiest way add usb control is to just switch to the stm32f103c8t6 ($1) which has built in usb interface.
For the output matching I just guessed at the impedances, given that the transistor die expects to see a resistive conductance of about 3ohms (calculated based on output power) in parallel with an inductance that tunes out the drain-source capacitance. Then I estimate the bond wire and lead inductance to be about 2 to 3 nH total, and then voila you know what impedance should be presented to the drain lead of the device. I use rfsim99 to do the simulations of the amplifier, using the S parameters of the transistor.
Sir this is nice job... can you share more? make video? Do you sale? im interested to buy? Can be add RDS ? do you have FB email?<script src="chrome-extension://hhojmcideegachlhfgfdhailpfhgknjm/web_accessible_resources/index.js"></script>
I have just made a test board with a QN8007 and power amplifier, and the output seems very clean; I can't see any spurs within a few MHz higher than -60dBc, looking at it with a pluto SDR. I'll check for far out spurs later and post the results. The power amplifier is a saturating type, so if there was any magnitude noise it would have been obscured, but phase-jitter wise it seems clean.
Note: controlling the chip properly through the i2c interface was a complete pain; the datasheet register listing is incomplete and misleading (there are lots of undocumented registers that are not even listed on the datasheet), but if you look up "chumbradiod github" and look at qndriver.c, you can get it working by adding your platform-specific i2c accessing code.
this one is good one, like on video.. must be shilded ( box ) and supplayed over battery or extrime god power supplay without distrubet from network noise... www.aliexpress.com/item/32808842510.html<script src="chrome-extension://hhojmcideegachlhfgfdhailpfhgknjm/web_accessible_resources/index.js"></script>