Up/Downconverter issues

[bryan.mitchell]

Hi all,

I am having some issues with my up and downconverter. What I'm trying to do is to receive a 125 kHz signal, upconvert it to 433.125 MHz, transmit it, receive that signal, and downconvert it back to 125 kHz. I have 2 circuits which are shown in the attached images. I have verified that the mixers are working properly but there's something wrong going on and I can't seem to know what it is. Somewhere along the way the signal either isn't being transmitted or received properly, and I'm in the process of figuring out where. I would like to get some pointers and also to see if there is something wrong with the design of the circuit that I have missed.

Thank you

[edavid]

The FS1000A doesn't have the frequency accuracy or stability to be used this way.  I don't know if that's the only problem.  You can try using the same FS1000A as the LO for both mixers, and see if the rest of the circuit works.

[PA0PBZ]

"There is something going wrong" is not exactly a problem description we can work with.
What exactly are you trying to do? I can see RFID in the schematics, are you trying to extend a RFID reader? You know that you need a duplex link for that?
I can also see some problems with your mixers, for instance you are producing 2 signals, LO+IN and LO-IN without filtering, and filtering 100KHz @ 433 MHz is almost impossible.

[uncle_bob]

Hi

What are the values for the components used? For instance, are L1 and C1 taken from the Linear app note?

What sort of level are you getting out of the "RFID Loop". If it is -100 dbm, the system you have will not transmit much of a signal.

Bob

[lostengineer]

hello everyone, i'm working with the OP on this project and I have some more info to post. Thank you for any help! We are both novice engineering students and are pushing the limits of our RF knowledge with this project.

The goal of the project is to take a 125kHz signal (same format as RFID but NOT directly RFID) and upconvert it to a 433.125MHz signal. This means that the LO is 433MHz. The up-converted signal will be transmitted about a hundred meters and down converted back into the same 125kHz signal in order to excite a 125kHz receiver.

As I said, it is not directly RFID. What I mean is that we do not need to get a response back from the end receiver - so a duplex system is not needed. We chose 433 as the LO because it seemed to have good range and was easy to work with (I don't think we need to do too much impedance matching or special PCB traces at this frequency - but I may be wrong). Otherwise, the LO freq. choice was arbitrary.

We have loop antennas for 125kHz built and they appear to be working fine. We are trying to use the FS1000A simply as a signal generator / LO source because it was cheap and easy. It's just a SAW stabilized Colpitts oscillator I believe. After testing we came to the same conclusion as edavid did. Multiple tested FS1000A's all gave us different frequencies within about 1MHz as shown on a spectrum analyzer. With this knowledge we did connect one FS1000A to both the up and the down converter. This also did not work. Can anyone recommend a better circuit or pre-built module to generate a LO @ 433MHz?

I believe the up-converter circuit is working. When powered on we get a nice spike on the spectrum analyzer around 433. As soon as we move our 125kHz loop antenna next to an RFID reader module we see two new spikes pop up on the spectrum analyzer - one at 433.125 and one at 432.875. I will post a screenshot of this shortly. As PA0PBZ said, this is the correct response - a signal at LO+IN and LO-IN. Also, it is true that we are not filtering anything at the moment.

I was hoping the down converter would be able to receive the LO+IN signal while the LO-IN signal was just lost / unused power. We're not seeing the original 125kHz signal at the output of the down-converter mixer. This is our problem.

My best guess at the moment is that the down-converter mixer needs a more powerful input signal - possibly a LNA inline with the antenna instead of just a wire antenna at 433MHz?

My next best guess is that the op-amp circuit on the down-converter output is not working properly. The purpose is to make the down-converter mixer output from differential to ground referenced as well as amplify. This circuit was copied directly from the LT5560 datasheet under the low frequency application example. 99% of component values used in the design were directly from the LT5560 datasheet.

I'm working on getting real measurements (I will post scope & spectrum analyzer screenshots in a bit). For now, is there anything fundamentally wrong with our RF design?

Thanks again!

- lostengineer (KJ4AWM)

[PA0PBZ]

The path loss at 433MHz for a distance of 100M is about 65dB, so you will end up with virtually no signal at all.
Do you get anything when you connect the output of the up-converter directly to the input of the down-converter?

[lostengineer]

Oh darn. Well that path issue doesn't bode well for us. Would we be able to move to a different link frequency to increase range? I know from a paper that some researchers build a similar system but used 2.4GHz - I thought path loss would be greater at a higher frequency though? In reality, range is not yet an issue - we just need to prove that the up and down conversion works first before optimizing range.

Here are some screenshots & measurements:

1: The output of the up-converter with no 125kHz signal near the loop antenna. The spectrum analyzer antenna is about 1ft away from the up-converter 433MHz antenna.

2: The output of the up-converter with the 125kHz loop antenna directly next to a RFID reader module generating 125kHz. The spectrum analyzer is now about 6ft away from the up-converter.

3: The oscilloscope reading directly on the output of the loop antenna while it is next to a RFID reader module (probe is 10x).

[PA0PBZ]

Oh darn. Well that path issue doesn't bode well for us. Would we be able to move to a different link frequency to increase range? I know from a paper that some researchers build a similar system but used 2.4GHz - I thought path loss would be greater at a higher frequency though? In reality, range is not yet an issue - we just need to prove that the up and down conversion works first before optimizing range.

No, a different frequency will not help, you will need to amplify the output of your up-converter and add some RF stages to your down- converter or in other words: build a proper transmitter and receiver    But first try what I suggested, couple the modules together and see if you get any output.

[uncle_bob]

Oh darn. Well that path issue doesn't bode well for us. Would we be able to move to a different link frequency to increase range? I know from a paper that some researchers build a similar system but used 2.4GHz - I thought path loss would be greater at a higher frequency though? In reality, range is not yet an issue - we just need to prove that the up and down conversion works first before optimizing range.

No, a different frequency will not help, you will need to amplify the output of your up-converter and add some RF stages to your down- converter or in other words: build a proper transmitter and receiver    But first try what I suggested, couple the modules together and see if you get any output.

Hi

Since you are trying to do a direct conversion process:

The output of the "transmitter" is 125 KHz + local oscillator one

The output of the "receiver" is  125 KHz + local oscillator one - local oscillator two

If LO1 and LO2 are each good to 1%, then they will each be off by about 4 MHz. The difference could be 8 MHz. Your output will be something between DC an 8 MHz.

If you design and build a custom LO chain for each end that is based on a TCXO, you might get them to 1 ppm. Then they both will be within 433 Hz. Your output will be at 125 KHz +/- 1KHz.

How close to frequency are you expecting the signal to be?

Bob

[qn1993]

Hi,

As many people point out, there are lots of problem with your circuit. I will try to clarify them as much as I can:

- Your local oscillator is of questionable quality. I search for an image of FS1000A and so far I can't see any crystal on it, which means troubles. I guess the method which they use to generate the signal if from an ceramic resonator (best case) or from an R-C circuit (bad). These two methods could have frequency deviation up to a few percent of the nominal frequency, even if people pay attention into design details.

- sub - 400MHz definitely needs impedance matching and so far I can't see it in your design. As a rule of thumb, any trace that run longer than one tenth of a wavelength needs impedance matching. A tenth of a wavelength with 400MHz running on FR4 PCB is approximately 19 mm. Is your trace from the FS1000A longer than that??

- I see in your design lots of passive components and I am not sure the functions of each of them. Let's assume that they do something, have you checked their resonance frequencies? Consumer-grade components, even SMT, sometimes can only run up to 100 MHz, after that their ESR and ESL dominate and they act in a totally different way.

- Seeing the images from your spectrum analyser, I assume that you do have something out of the mixer. Have you try connect the up-converter directly to the down converter without any antenna. If they work, it means that there is something wrong with the transmitting part. Have you try to get a 433.125MHz signal from a well-designed source to make sure that your down converter works? Break up your circuit into smaller parts, test and verify each of them (divide-and-conquer tactic) and you should be able to identify where the problem is? Don't give up

- 100m surely will attenuate your signal a lot. The problem is one part associated with frequency, another because of how you actually transmit it. Let's assume that you get your signal to the antenna 100%, perfect matching. After it is aired, as it gets further away from the antenna, it gets weaker because signal strength has to be distributed over a larger area. Thinking it's like this, you pump 1W into dipole antenna over 1s. The signal is out on a doughnut shape. As the distance increases, the doughnut become bigger but the amount of energy of the signal is still the same, no wonder why it gets weaker.

And yes, you are true. As frequency gets higher, it attenuates faster but many people still want to push for higher frequency since it means they can make antenna smaller -> less cost, circuit become more compact. This also means that even for low cost, narrow band application you still can get more bandwidth.

- You mentions that you take this design from a low frequency datasheet and thought that it will work on the frequency you want. As I mention earlier, high frequency (>1 MHz)design differs greatly from low frequency one. In such circuit, a simple wire ain't a wire anymore. They are a mixture of parasitic resistor, inductor and capacitor distributed over the wire's distance and a simple bend may deteriorate the signal greatly. Pay attention to impedance matching and available bandwidth, that should be the first thing you do when approach a high speed design.

Hope this helps.

Quang

[vk6zgo]

Oh darn. Well that path issue doesn't bode well for us. Would we be able to move to a different link frequency to increase range? I know from a paper that some researchers build a similar system but used 2.4GHz - I thought path loss would be greater at a higher frequency though? In reality, range is not yet an issue - we just need to prove that the up and down conversion works first before optimizing range.

Here are some screenshots & measurements:

1: The output of the up-converter with no 125kHz signal near the loop antenna. The spectrum analyzer antenna is about 1ft away from the up-converter 433MHz antenna.

2: The output of the up-converter with the 125kHz loop antenna directly next to a RFID reader module generating 125kHz. The spectrum analyzer is now about 6ft away from the up-converter.

3: The oscilloscope reading directly on the output of the loop antenna while it is next to a RFID reader module (probe is 10x).

An "upconverter",a "mixer" & a "modulator" are really the same device----the latter term is more appropriate for what you are doing,as the two frequencies are so far apart.

I assumed from your earlier comments,that your "up-converter" suppressed the original UHF carrier (LO signal),producing a "Double Sideband Suppressed Carrier" ( DSBSC) signal.
(2) However,reveals that the carrier is still present,so what you have is good old-fashioned "Full Carrier Amplitude Modulation"(AM).
There are a number of disadvantages to AM--power wastage (the carrier does not "carry" any nformation),susceptibility
to noise,etc.,but in this context,the main downside of AM is that it doesn't play well with direct conversion reception.

If the Rx LO is even a bit off frequency,you will get heterodynes ("beats") between the LO & the incoming carrier.
These will fall within the spectra of the desired baseband signal,causing interference.
If you can frequency & phase lock the LO to the incoming carrier,all will be well,but that entails extra circuit complexity.

The upside to AM is that you don't even have to provide an LO at the Rx---you can just use a so-called "Envelope Detector",as the received carrier performs the function normally provided by an LO.

The output from such a detector will be your 125kHz signal.

As others have pointed out,you need some gain in front of your Receive Mixer ("downconverter"),as it won't work with the miniscule signals you are presenting to it.

RF is not "black magic",but neither is it "plug 'n play",either---Engineers & others  have addressed all these problems over many years,& came to the conclusion that simple circuitry is neither effective nor reliable at UHF.

[lostengineer]

ok wow. first off thanks for the help everyone! A bunch of stuff is starting to click.

So, I believe most of you understand what we're trying to do, but I'll clarify again so no assumptions are made. Please let me know if anything isn't clear. We're not trying to build a whole radio or receive and decode a signal at all - but maybe we must to make this work. We already have a system that we are trying to extend which generates a 125kHz data signal much like an RFID reader would. This system comes with a good receiver for this 125kHz signal which works within a few feet of the reader/transmitter as it is a magnetically coupled signal - not yet RF I think. No response to the 125kHz signal is produced by the end receiver....at least at 125kHz.

We also do not know or care what the modulation or data is in the originating 125kHz signal. This is a design goal. We want to transmit it as is and have it appear on an identical 125kHz antenna at the receiving side - simply increasing range of the signal. This receiving side antenna - the 125kHz loop antenna on the output of the down converter - will then excite the true receiver which came with the system.

Tomorrow is the next time I'll be working with the hardware so further testing will have to wait until then. We did try to connect the output of the up-converter to the input of the down-converter and the whole system appears to be working (albeit barely)! We even detected a proper output signal with the up and down converters about a foot away from each other with only antennas! I'll post screenshots of the spectrum analyzer tomorrow.

As for improvements:

1: We know the LO is a huge issue. We will be switching to a TXCO based design as recommended by uncle_bob. I don't think we need ridiculous precision but i'd like to be within a few kHz maybe?  Can you guys recommend a TXCO circuit which would work well with our mixers? I'm doing research on this now...

2: Power - bryan did not post our entire circuit design, but I figured transmit power would need a boost on the up-converter 433 side. We have already designed a RF power amplifier for this based on a reference design for the RF2126 amplifier IC. I have a bandpass filter on the output of this amplifier. This has not been prototyped yet. I'm posting a more thorough schematic.

   On the down-converter side, we'll add a matching band pass filter on the input followed by a LNA before the input to the mixer.

3: Impedance Matching - Quang says we need it. I think we can make all signal traces on our PCB less than 19mm as he recommends. The whole PCB is less than a business card in size and we have not finished their design. I mean this stuff is small... the LT5560 mixers are 3mmx3mm DFN's - that was a bi**h to solder. 

All the passive parts & their values were taken from reference designs for 433MHz or 125kHz respectively from the LT5560 datasheet / RF2126 datasheet. As you can see from the newly posted schematic, I added a bunch of extra attenuators, solder bridges, and connectors for aiding in testing. I'm not sure we need all of that, but it doesn't cost anything extra to add pads to the PCBs so why not?

I'll update tomorrow with more testing screenshots. Thanks!

[lostengineer]

Here are two additional pictures of the up-converter prototype

[edavid]

Maybe you should digitize the 125kHz signal and transmit it by WiFi 

[rfeecs]

You should check on L1, C5.  This is a series resonant LC.  Your values of 22nH, 1nF resonates at 34MHz.  It is going to block a lot of your LO signal.

Your ceramic caps and other components with long leads might be a problem.  A 1cm lead has about 6nH of inductance.  This could seriously screw up any matching you might be doing.  Better to use surface mount parts.

As has been pointed out, you really need to revisit your whole approach here.  You are trying to do a direct conversion transmitter / receiver.  Look up the problems associated with that.  Your approach looks like it won't work at all.

Also pointed out, you are not really doing an up converter.  You are doing an AM modulator.  You are converting your 125kHz signal to three signals.  The carrier, the upper side band and lower side band.

On your receiver when you mix these back down you are going to get a horrible mess.  If your oscillators differ by 10kHz you are going to end up with three signals in the base band that are shifted by 10kHz.  So your carrier gets converted to 10kHz.  Your upper sideband gets converted to 135kHz.  Your lower sideband gets converted to 115kHz.  It just plain doesn't work.

If your object is not to design a radio link, but just get something that works, then you are better off buying some pre-made transmit and receive modules to get the job done.  Otherwise you need to rethink your whole approach.

[uncle_bob]

Hi

Ok, how to design a radio:

1) Take a look at the basic requirements for board layout and component selection at the frequency you have decided to use. Price out the parts and re-evaluate if this is the right band for you.

2) Evaluate the frequency references needed in the design. What frequencies do you need? How stable do they need to be? What power levels do you need? What kind of phase noise do you require? What sort of tuning range do you need? How will they be tuned?

3) Look at the link budget for the radio. You have a modulation approach, at minimum modulation, you have a signal. You put that signal out of your power amplifier at some level (10 watts maybe). That goes into your feed line and you loose something (2 db maybe). It then goes into your antenna and you do or do not get gain depending on the antenna design. It flows through space and you have loss. On the other end you have an antenna with or without gain, a feed line and an amplifier. That amplifier has a noise figure. Your receiver may or may not be limited by the noise figure of the input amp. If it is, the level of the signal at that amp gives you a signal to noise ratio (yes it's a bit more complex than that). That SNR should be something you can work with (say 40 db).  After you add up all the various losses and modulation issues --- do you get 40 db SNR (your goal) or not?

4) Look at your modulation scheme. There are a *lot* of ways to modulate a signal. Each has it's benefits and drawbacks. Some are easy to transmit and hard to receive. Others are hard to transmit, but easy to receive. Some are quite complex on both ends. Does the modulation you are using make sense for the application?

5) Given the design parameters so far, take a look at the power available for the radio. If it is going to run on batteries, will it run long enough with a practical battery pack? If it plugs into the wall, can the wall socket supply enough power to do the job?

6) Take a look at the possible antennas for the radio. Are they of a size that makes sense? If they are directional (directivity = gain) how will they be steered? Can they be designed with the tools you have? Will you use an off the shelf antenna?

Once you have that stuff worked out, take a look at the parts you can use. Do a trial design and see if it meets your cost goals. If not, look some more or change the goals. Design and evaluate each section of the radio. Compare it against the system requirements. Once you have a fully compliant design, look into the rules and process for RF or microwave pc board layout and design. PC boards are about the only practical way to get the job done with modern parts at the frequencies you are considering.

You are now out of the basic design phase and into the "build and test" phase. You likely will loop back through the design phase a few times before the whole system is up and running.

(Yes, the above is far from complete, it does hit enough of the points that you can get an idea of what is involved).

Bob

[vk6zgo]

ok wow. first off thanks for the help everyone! A bunch of stuff is starting to click.

So, I believe most of you understand what we're trying to do, but I'll clarify again so no assumptions are made. Please let me know if anything isn't clear. We're not trying to build a whole radio or receive and decode a signal at all - but maybe we must to make this work. We already have a system that we are trying to extend which generates a 125kHz data signal much like an RFID reader would. This system comes with a good receiver for this 125kHz signal which works within a few feet of the reader/transmitter as it is a magnetically coupled signal - not yet RF I think. No response to the 125kHz signal is produced by the end receiver....at least at 125kHz.

We also do not know or care what the modulation or data is in the originating 125kHz signal. This is a design goal. We want to transmit it as is and have it appear on an identical 125kHz antenna at the receiving side - simply increasing range of the signal. This receiving side antenna - the 125kHz loop antenna on the output of the down converter - will then excite the true receiver which came with the system.

Tomorrow is the next time I'll be working with the hardware so further testing will have to wait until then. We did try to connect the output of the up-converter to the input of the down-converter and the whole system appears to be working (albeit barely)! We even detected a proper output signal with the up and down converters about a foot away from each other with only antennas! I'll post screenshots of the spectrum analyzer tomorrow.

As for improvements:

1: We know the LO is a huge issue. We will be switching to a TXCO based design as recommended by uncle_bob. I don't think we need ridiculous precision but i'd like to be within a few kHz maybe?  Can you guys recommend a TXCO circuit which would work well with our mixers? I'm doing research on this now...

2: Power - bryan did not post our entire circuit design, but I figured transmit power would need a boost on the up-converter 433 side. We have already designed a RF power amplifier for this based on a reference design for the RF2126 amplifier IC. I have a bandpass filter on the output of this amplifier. This has not been prototyped yet. I'm posting a more thorough schematic.

   On the down-converter side, we'll add a matching band pass filter on the input followed by a LNA before the input to the mixer.

3: Impedance Matching - Quang says we need it. I think we can make all signal traces on our PCB less than 19mm as he recommends. The whole PCB is less than a business card in size and we have not finished their design. I mean this stuff is small... the LT5560 mixers are 3mmx3mm DFN's - that was a bi**h to solder. 

All the passive parts & their values were taken from reference designs for 433MHz or 125kHz respectively from the LT5560 datasheet / RF2126 datasheet. As you can see from the newly posted schematic, I added a bunch of extra attenuators, solder bridges, and connectors for aiding in testing. I'm not sure we need all of that, but it doesn't cost anything extra to add pads to the PCBs so why not?

I'll update tomorrow with more testing screenshots. Thanks!

You are reading,but not absorbing !

Firstly, even though you insist you are making "up" & " down" conversions,that is really not what you are doing.
You are taking a "baseband" signal (your 125kHz) & modulating a 433MHz carrier,amplifying the resultant AM signal & transmitting it 100m to a Receiver,where it is demodulated,so you get your 125kHz back.

In concept,this is the same as "up" & "down" conversion,but in practice,it presents different problems.

Normal "up converters" mix a signal which differs in frequency to the LO,but not by such a huge factor.
For instance,a PAL TV transmitter vision IF signal will have the Vision Carrier (Vc) at 33.4MHz.

In order to present a "onair"  Vision Carrier at,say  189.0MHz,the LO will be at 222.4MHz (189.0+33.4MHz)

OOPS!,I messed that up in my attempt to keep it simple!

The IF carrier at  33.4MHz is,in fact the Sound  Carrier ( fs),the vision carrier (fv) is at 38.9MHz,so for an "on air" Vision Carrier at 189MHz,the LO will be at 227.9 MHz.

I stuffed up the next bit,too,but the overall argument is still correct.

The LO & the other  conversion product (189.0-33.4MHz=144.6MHz) are far enough away in frequency that they may be easily filtered out with normal LC tuned circuits.
The other product from this conversion will,in fact,be 266.8MHz.

A conversion product (LSB)at 432.875MHz,or (USB)at 433.125MHz cannot be simply filtered,if at all.
For this reason,when the baseband signal is very much lower in frequency than the LO,the process is regarded as that of modulation,rather than frequency conversion.

By the way,if you decide to stick with AM,your LO stability requirement can be less strict .
The Receiver can be a simple diode detector.
Using such a detector with an AM signal no Receiver LO is needed,as the difference frequencies between the sidebands & the carrier are at the original baseband frequencies.
If your Transmit LO drifts,the sidebands drift with it,maintaining this relationship.

Please do some real reading on Radio Transmission & Reception----App Notes simply do not hack it!
Your Uni will have a Library,& in that Library will be books written by Radio Engineers,& probably some by Radio "Hams"
 too--very often the same people.
If you can find an ARRL Handbook or an RSGB Manual,you will pick up a lot of useful concepts that a few paragraphs on this thread cannot convey.