It is not clear if it can emit LOC and GS at the same time; most of the cheap boxes can’t, and you cannot test modern avionics with those because any EHSI presentation flags the GS if there is no LOC.
It has no attenuator, so only crude functional testing is possible.
Is testing this way not approved?
The Yaesu FTA750 does ILS but this is a receiver, not an emitter. That Sun box emits the signals so you can do a rough test on the aircraft systems.
The IFR4000 family is what all the better shops have, but is expensive – upwards of USD 6k depending on the version. I have been watching Ebay for years but everybody knows what they are worth 
Posts moved to an existing thread on the same topic.
I recently bought an SDR transceiver (LimeSDR) which covers 100kHz up to 3.5GHz. It has two independent transmitters and receivers, and adjustable output power (up to 10mW). One thing I’ve considered using it for is transponder testing, but I need to learn a huge amount on the subject before I can even begin to contemplate that subject (I’m just on the ‘barely understanding how to construct an FM radio in software’ stage).
The easy way to do a lot of this waveform generation stuff is to generate the waveform values (with some Basic prog on a PC, etc), program them into an EPROM, and drive the address lines of the EPROM from a binary counter 
A D-A converter driven off the data pins (8-bit will usually do) will then output the required waveform… This is a great way to generate e.g. a VOR signal on a specific radial. However I suspect you may need 16 bits of resolution of generating RF directly.
Generating RF from a digital source directly is more about sample rate (and stability of your timing source) than number of bits. Extra bits just give you more dynamic range (6dB per bit). If 48dB is enough dynamic range you can use 8 bit direct digital synthesis of the waveform (and a suitable low pass filter to get rid of the harmonics). However at VHF direct digital synthesis starts getting a bit harder (the sample rate must be at least twice the frequency you’re interested in – see Nyquist – not an awful lot of eproms or flash have access times low enough for direct digital syntehsis the way you’d suggest at VHF – instead you’d have to do the synthesis at some lower frequency and mix it up to the frequency of interest)
An actual SDR transceiver like the LimeSDR can be so much more flexible though and is already designed for the job (although you still need the low pass filter – I did some testing on the unamplified output of mine last weekend and the 2nd harmonic was only approximately 25dB down – a commercial transmitter’s 2nd harmonic is expected to be at least 50dB down). GNURadio for instance already has modules for decoding transponder replies and ADS-B (although probably not for transmitting the interrogations).
I think you know much more than I do about DDS but AFAIK to get any practical usability you need to go well above 2x sampling. Otherwise, the 2nd harmonic is too strong and the sort of filter that’s needed has to be steep, and that causes other problems. But I was always crap at maths I looked into all this when trying to develop the product I mentioned above. There is a variety of DDS chips out there.
I am not sure 8 bits is enough if you are trying to directly generate the RF, with the modulation around the carrier. For generating the analog waveforms in old aviation navigation (VOR etc) which are all in the audio band, you can just use a Z80 at 4MHz
As regards memory access time, most of these “arbitrary waveform” generator products use RAM (obviously), and you can achieve any desired access time (for sequential reading) by interleaving multiple chips. So by interleaving say 10 20ns SRAMs you can get an access time of 2ns which is enough for 500M samples/sec.
Transponder emissions carry digital data but the bandwidth required can’t be that great otherwise the thing would take up a lot of spectrum. So I can’t see why it could not be generated in the same way.
You can interleave memory, but you’ll start hitting issues with things like the rise times required (and PCB design issues) by the time you’re hitting VHF (especially if you’re looking at a sampling rate of >500Msps). It just becomes a lot easier to do all your modulation etc. at some lower IF, then mix it up to the final RF. You’ll note the price of DDS arbitrary wave form generators doesn’t scale linearly with maximum frequency :-) I think it’d be silly to even try DDS at 1030 MHz!
For a homebrew VOR tester you could even use a sound card on a laptop to generate the waveform and an SSB transmitter suitably modified so it’ll work on the VHF nav airband (with suitable attenuators, you don’t want to be dumping watts of RF power down the coax to your VOR receiver. It might even be possible to do that for a transponder, depending on the occupied bandwidth). But the nice thing about an software defined radio, just that one board will take care of making the RF and you can just use something like GNU Radio to build the transmitter or receiver you want. Something like the LimeSDR can transmit literally everything form morse code to 4K HDTV literally by running a different program on the PC it’s connected to, and will support modes that haven’t even been thought of yet.
Yes there are several solutions which use a sound card to generate the avionics audio-band signals. I have one here – details posted previously. It is written in java IIRC. You then connect the sound card output to the external modulation input of a suitable signal generator, as shown in one of the videos here here.
