Hams building equipment/ensuring compliance/due dilligence question

[cdev]

If you are a ham, and you build your own transceiver, using some conservative design (i.e. follow a well-trod path to your device) how can you make sure you're not emitting spurious signals?

Is it enough to, say, know what the fundamental output frequency is precisely (cheap now) and have a simple band-pass or low pass filter that you know has a roll-off steep enough to eliminate harmonics and view it via a cheap but relatively powerful SDR (compared to the equipment hams had just a decade or so ago) (also cheap), or do you have to test it with (expensive) equipment you don't have? Can somebody (some other ham near you) listen to your signal on the air to verify its not dirty?

[Towger]

A SDR is useful for as a poor man's spectrum analyser.

Does not need to be someone else, I have used a SDR Play with a couple of inches of wire for an antenna and the transmitter into a dummy load.

[babysitter]

There are more hams with boat anchor S/A or Rigols or Test receivers and BPF and stuff than you can imagine.

[medical-nerd]

Yup!!

I've got a Marconi 110Mhz TF2370 boat anchor spectrum analyser and a HP 8754A Network analyser. 


Cheers

[German_EE]

If you build an HF transceiver that has a correctly biased power amplifier and a well designed low-pass filter you should be OK. The amount of bias should be calculated from the P.A. device data sheets and there are a number of filter design utilities on the Internet (plus of course ELSIE)

However...............

If you have access to a spectrum analyzer then use it. Print out the screenshots of your output checks and paste them into your station log, not just checks for harmonics but checks for IMD as well. Above all else though, if someone tells you that you have a poor signal then it's ESSENTIAL that you go off-air and find out why. We need to set a good example.

[cdev]

If I build a simple SDR I think I can build a clean transmitter, one valid approach as far as harmonics is simply using filters. if I build my own HF radio, I can put quite-decent filters on it.

I have a $75 ebay Tektronix scope.. so I can visualize the envelope of the signal..

I have a frequency standard (and two low digit frequency counters - both of which don't have great resolution but both did come well calibrated..both didnt need any further tweaking, and resolve down to a few Khz, which is close enough for me to know that I am inside of bands.) 

I have a Soft Rock Rx and my GPSDO and using the GPSDO 10 Mhz (and a home made attenuator made with an L-pad) its much much much better than WWV for everything, it gives me a signal I am sure I can get the error on the softrock whittled away to almost nothing with. (Its just time consuming doing that with usbsoftrock, if anybody has any tips on that they would be welcome. The PE0FKO DLLs to tweak its Si570 dont work for me at all under wine.)

So, I think that now with my new frequency standard and one addition, a cheap multidigit counter (which one to get, though?) that can use it as a reference signal, there shouldn't be any need for any additional anything. I think I will be all set for measuring frequency for the foreseeable future.

Basically I just have to figure out how to make the most effective, economical use of that accurate 10 MHz.

[bd139]

The cheap solution is an absorption wavemeter. That's basically a dead simple tunable L/C based crystal set type receiver with a panel meter as an output device. As you tune above the fundamental transmit frequency of the receiver into the harmonics, you get a reading on the harmonics. It's not terribly scientific but it's a surprisingly good differential tool. A grid dip oscillator with the oscillator turned off also does the job.

Otherwise a spectrum analyser which is at least two orders of magnitude more expensive or old and manky

This all really only matters though if you're pumping some serious watts. The harmonic content with a canned design self built harmonic filter/LPF or one that has gone through simulation is not terrible.

Personally I've had some luck, at least on the 40m band with the FFT on my DS1054Z. It's good enough to see the harmonics but not measure them accurately.

[CJay]

Otherwise a spectrum analyser which is at least two orders of magnitude more expensive or old and manky

65p for an absorption wavemeter?

Seriously though, that's a handy tool and used to be mandatory in the UK as part of the licence I think.

I've used a variety of methods oin the past, some better than others but have always asked advice and had stuff checked by a local ham on occasion.

[bd139]

I made mine out of an old Hacker radio I got for 50p and a blown up micronta VOM so not far off. That was used for checking AM transmitters I made for around the house. Was used (classy I know) to pipe 90's rave tapes from my skip dived tape deck.

The vintage radio collectors would kill me for destroying that radio.

[TheSteve]

And now we know why the HF bands are all roughly harmonics of each other - So you mainly only mess up other hams if your transmitter is terrible.
If you're on the air with a radio that is splattering all over you will generally be told pretty quick.
Still some sort of verification before operating is recommended.

[cdev]

How is an "absorbtion wavemeter" different than a grid dip meter?

"absorbtion wavemeter" sounds like a fancy name for the same thing.

Couldn't you make one with a siggen and a calibrated tunable "tank circuit"? (with the current being measured as you tune it)

[bd139]

An absorption wavemeter is just a variable tuned circuit with a detector. It has no active components.

A grid-dip oscillator is a powered oscillator that is used to resonate the target circuit. At that resonant frequency the DUT will draw power from the oscillator causing the dip.

A grid-dip oscillator happens to usually be reusable as a wavemeter if you turn the oscillator off.

If you have an RF signal generator you have a power source only. You can mix the DUT's output with the output of the signal generator and pass it through a narrow filter then put a detector on the end of that and sweep the generator and it'll tell you where the harmonics are. However at that point you've built a spectrum analyser, albeit a bit of a crude one

[voltz]

By law, you do need to verify anything you build to transmit is within allowed spurious limits. Those limits vary on which country you live in, which document you read and what advice you may be given from another ham.. My current understanding (UK) - and please correct me if this is incorrect, but 50dB down is the minimum spec from the fundamental on HF. I believe VHF/UHF is around 50-60dB down. Some proper figures here would be useful. Its never that clear.
So using a spectrum analyser with a dummy load is an obvious choice as this directly reads in dB and you get to see any spikes they may be present along the way. I personally use an IFR1200 radio test set for this task. As mentioned already, correctly biasing the pre driver and final PA is essential to keep the transmitter clean. So follow a tried and tested design, or be very careful with your calculations if you design your own.
You can use much cheaper methods, like an SDR with dummy load or absorption wave meter, as long as they are calibrated in a way to show relative dB reduction. Fundamental, 2nd and 3rd harmonics are the points of main interest as this is where any nasties will normally occur.

[janoc]

Another thing that hasn't been mentioned is QRP.

If you are homebrewing, a small rig outputting few milliwatts is going to do much less damage on the band (and outside of it) than if you are running a large power amplifier. That doesn't mean you should be sloppy but you can get away with a lot more in a QRP radio than a 100+W station.

[bd139]

Good point with QRP.

Also TX amp is less likely to lose bits of power transistors in your face (I made an illegal FM TX once which blew up big time).

[cdev]

That actually is my plan, once I get my ticket, to get an SSB/CW rig working at the milliwatt range and then build a basic power amp to amplify it.

I don't plan on it being dirty, I think I can build a clean rig.

HF rigs are not so difficult to build now it seems due to all sorts of great modular parts becoming available but a power amplifier (the transistors) still are pricey-er than I think they should be given the falling price of semiconductor devices generally. .

Another thing that hasn't been mentioned is QRP.

If you are homebrewing, a small rig outputting few milliwatts is going to do much less damage on the band (and outside of it) than if you are running a large power amplifier. That doesn't mean you should be sloppy but you can get away with a lot more in a QRP radio than a 100+W station.

[radiogeek381]

For the most part, the fancy test equipment isn't at all necessary. 

The wavemeter approach has economy on its side. 

I'm a big fan of using things you have around the house.  Around my house, I've got receivers that can tune the HF bands. They're more sensitive than the spectrum analyzer on my bench, and the mechanism between the input port and the output port is simple.  Fire up the transmitter into a load, and tune the receiver for the fundamental (does it sound pleasant?  If you put in enough attenuation on the RX can you get a pleasing tone?  Once you get to that point, are there "images?"  Is the carrier suppression adequate? Splatter?  (If you don't hear splatter when the RX front end attenuator is off, then you spent way more on your receiver than I could afford   The point here is that not every spur you hear in the RX came that way out of the transmitter.  When overloaded by very strong signals, most receivers will generate images, birdies, and all manner of other things.)

Next tune to harmonics -- 2x, 3x, as far as you can go.  You'll probably hear something, but is it sufficiently far down (can the S-meter help?  possibly.  But remember, this ain't Rhode and Schwartz we're dealing with.)  I know people who measure relative amplitude with a voltmeter on the audio out.  But remember to turn off the AGC.

Good luck in your new adventure!

73...

[bd139]

Good point with sensitivity. Also some of the radio is actually in your head. Your audio processing system is pretty good at isolating signals compared to an SA

[cdev]

For the most part, the fancy test equipment isn't at all necessary. 

The wavemeter approach has economy on its side. 

I'm a big fan of using things you have around the house.  Around my house, I've got receivers that can tune the HF bands. They're more sensitive than the spectrum analyzer on my bench, and the mechanism between the input port and the output port is simple.  Fire up the transmitter into a load, and tune the receiver for the fundamental (does it sound pleasant?  If you put in enough attenuation on the RX can you get a pleasing tone?  Once you get to that point, are there "images?"  Is the carrier suppression adequate? Splatter?  (If you don't hear splatter when the RX front end attenuator is off, then you spent way more on your receiver than I could afford   The point here is that not every spur you hear in the RX came that way out of the transmitter.  When overloaded by very strong signals, most receivers will generate images, birdies, and all manner of other things.)

Next tune to harmonics -- 2x, 3x, as far as you can go.  You'll probably hear something, but is it sufficiently far down (can the S-meter help?  possibly.  But remember, this ain't Rhode and Schwartz we're dealing with.)  I know people who measure relative amplitude with a voltmeter on the audio out.  But remember to turn off the AGC.

Good luck in your new adventure!

73...

Here are some simple ways to measure harmonics that might work.. Xmitter connected to dummy load.. As you say, something like an RTLSDR with a strong variable attenuator connected to its front end so it receives almost no signal.. (and isnt overloaded) Tune to fundamental, hear weak signal, but able to bring it 60 db above noise..  then tune to 1st, 2nd, 3rd, 4th, harmonics.. should hear nothing.  Then bring it to 70 db above noise.. can you hear anything on harmonics? and so on. Repeat with other receivers. If you have a simple, well characterized multiple pole low pass filter of sufficient ability, you can turn even a pure square wave into a sine wave without much difficulty, (for one or maybe two bands per filter)

So a switchable bunch of bandpass or low pass filters could filter all bands adequately.  I already have a bunch of band pass filters I made (in dip plugs with tiny capacitors so they are definitely only adequate for QRP, if that) but the concept would work, switching filters in by hand even if I were to use something like WSPR-berriPi - i.e. Raspberry Pi GPIOs for my transmitter (which produces square waves..  )

I got the DIP plug resident BPF idea from "Zao" - a maker of SDRs in Japan.. he uses plug in band pass filters.

Making filters is a good reason to get a DER EE5000, I think.. (*sigh*)

[medical-nerd]

Hiya

Are the DER EE 5000 available from a supplier in the UK or Europe?

Cheers

[LaserSteve]

You can also verify the output filters during construction with a clean signal generator and just a impedance matched detector, provided your rig is well shielded.   Most problems will show up with sweeping with a signal generator or proper sweeper, well before you hear them with the on the air test.

Note: I'm  Not so sure I would call a cheap EBAY DDS RF board clean, but it would help in many cases.

Of course your supposed to measure into a dummy load and then on the air, but a good sweep during construction works wonders on checking your filters, especially with home made inductors.

I personally use a PTS frequency Synthesizer in 1 Hz steps, a 1970s HP sweeper for my microwave addiction, a tunable YIG filter, a wide variety of detector diodes, a home made diode based harmonic generator, an ancient CATV spectrum analyzer with a CRT, a few directional couplers, and a RF bridge or two.

 The quickie box I use,  is a signal level  meter / monitor  receiver from 450 KHz to 950 Mhz by B&K used for CATV servicing. With a built in demodulator for FM and a crude RF level meter.. The B&K 430 is a kickin fun little tool.  If you buy one and its dead, remove the Ni-Cad pack and it usually will come back to life with an external PSU. It finds harmonics nicely.

see:

http://www.ebay.com/itm/BK-Precision-Dynascan-430-Signal-Level-Meter-/272862248918?hash=item3f87db7fd6:g:9YoAAOSw7GRZIPbz

or

http://www.ebay.com/itm/BK-Precision-430-Signal-Level-Meter-/282558425280?hash=item41c9cb68c0:g:OwAAAOSw8GtZXOeM

Internally its a adapted TV tuner, a calibrated IF board, a AM/FM demodulator,  and NiCad pack with a battery charger.

For a receiver side  sanity check I keep the external oscillator version of this around:

http://www.hsmicrowave.com/CMG.html
Steve

[janoc]

The eBay DDS boards (the AD9850/51 kind) are fairly clean. If it is not good enough, you can always low-pass filter it. It is plenty sufficient for sweeping a filter.

A good tool using these is the Simple Scalar Network analyzer:
http://rheslip.blogspot.fr/2015/08/the-simple-scalar-network-analyzer.html

Basically nothing more than a swept  RF generator (the AD9850 module + an Arduino to drive it) and a sensitive power detector using AD8307. I have built it for my playing around with filters and such and it makes the testing a lot easier.

Otherwise you can do the same by hand - a generator (one decent enough can be had for ~$50 on eBay - no need to calibrate the output levels as you only need relative measurements) and simple rf voltmeter (or scope if you have one).  Then sweep by hand and record the values into a spreadsheet. Subtract from a sweep without the DUT (generator directly connected to the detector) which gives you a reference level (output level of the generator often changes with frequency) and you have your filter response. 

If you have a scope, then the output of sweep of the generator can be used to trigger a scope and you get the filter response on the scope directly. Etc. There are many ways to skin the cat.

[bd139]

Damn it I’m going to have to build one of those scalar network analysers now! Similar to a sweeperino.

[janoc]

Damn it I’m going to have to build one of those scalar network analysers now! Similar to a sweeperino.

It is pretty much the same thing - the idea has been around for years and many variants exist.

However, if you want to work on HF HAM bands, then I find the AD9850 DDS better because of the wider frequency coverage at the low end. Also the Si570 generates a square wave, not sine, so you will get a ton of harmonics in the signal. That could make for quite a bit of spurious signal fun.

Don't forget to put the detector in an RF-tight metal box - those things are suuper sensitive and will pick up local radio stations and what not easily. Shielding is a must - on mine it made a difference of ~20dB noise floor level ...

Another good idea is to buy two 6-10dB coaxial attenuators and use those for connecting the DUT. Otherwise the reflected signals due to impedance mismatches will cause quite a bit of error in the measurement (the displayed filter pass-band shape will get distorted).

On the other hand, it is a very hands-on method of learning about impedance matching, return loss and similar - if you add a directional coupler you can even measure it. A super useful instrument for few bucks.

[bd139]

I saw a sneaky solution that someone came up with using a si570 at a frequency of 32-62MHz and mixing it with a 32MHz crystal clocked NE612. You use the LSB as the fundamental and stick a 30MHz LPF on the output. Bit more complicated and probably more expensive but if you don’t have an AD9850 lying around...