HP 8566B tracking gen build - what are the considerations

[siggi]

Hey y'all,

I'm thinking it'd be a fun (steep) learning project to build a tracking gen for my HP 8566B. I figure a sane goal is to cover the first two bands, 0-2.5GHz and 2GHz-5.8GHz.
Now I'm a total RF n00b, so please bear with me, and mind that the point here is to learn, more so than to get the best quality tracking gen at the lowest cost - whether in time or money. So, I'm thinking a "discrete" design, with a PCB and modern RF components is the way to go (for me), rather than plumbing together existing RF modules.

The old HP 8566B exposes the LO, which is offset by 3.6214GHz for band 0, and by 321MHz for band 1. I think I understand what the block diagram of a tracking gen ought to look like reasonably well.

• First you want to isolate the HP's LO a bit by attenuation or other means,
• then mix it up/down with an appropriate offset.
• From there it may need amplification
• and filtering,
• and along the way ideally there'll be some leveling.
• The leveling wants a directional coupler, as otherwise the SWR of the load will affect the output level.

I started looking into combined wide-band PLL/VCO chips like e.g. ADF4355 to provide the offset frequencies, but I'm not sure that's the best way forward. Specifically, the specs look lackluster in phase noise and harmonics when division is employed.

So questions I have include:

• What are the measurement effects related to phase noise in a TG? Is there some kind of relationship between a TGs and SAs phase noise?
• ADF4355 is specified to only -12dBc for third harmonic content when the output's divided. Will this not be a problem in a TG?
• What are some better ways to get precisely the above two synthesized LO frequencies for a TG?
• What else am I missing ?

Siggi

[KE5FX]

Hey y'all,

I'm thinking it'd be a fun (steep) learning project to build a tracking gen for my HP 8566B. I figure a sane goal is to cover the first two bands, 0-2.5GHz and 2GHz-5.8GHz.
Now I'm a total RF n00b, so please bear with me, and mind that the point here is to learn, more so than to get the best quality tracking gen at the lowest cost - whether in time or money. So, I'm thinking a "discrete" design, with a PCB and modern RF components is the way to go (for me), rather than plumbing together existing RF modules.

This approach will get you up and running on the 0-2.5 GHz band.  Also see Dan Doberstein's notes here.

The high band requires a completely different conversion scheme.  There will be few if any opportunities to reuse the same hardware that you use for the 0-2.5 GHz band, so you really have to treat them as two different designs. 

I started looking into combined wide-band PLL/VCO chips like e.g. ADF4355 to provide the offset frequencies, but I'm not sure that's the best way forward. Specifically, the specs look lackluster in phase noise and harmonics when division is employed.

The ADF4355 is fine, along with any number of similar parts.  Hint: look for evaluation boards on eBay.  Valon Technology sells some economical synths that would work well as LOs. 

Remember that the signal will ultimately be picked up by a tuned receiver (meaning the 8566 itself).  Isolation is really the only thing to worry about, but a good ALC circuit is also helpful.  You can make do with trace subtraction on the analyzer to 'calibrate' the measurement, although load VSWR won't be accounted for. 

You don't care much about harmonics, unless you're testing antennas and you want to be nice to other users of the spectrum.  Phase noise is also not important.  In fact, many people actually use broadband noise sources instead of TGs.  The measurement takes longer than it would if you had a real TG, and dynamic range is worse, but it gets the job done.  Or you can use a sweep generator with the analyzer set to MAX HOLD. 

Before I had a VNA, I used my homebrew TG for 0-2.5 GHz work, swapping in an HP 86290B plugin for 2-22 GHz measurements in MAX HOLD mode.  HP 8620C mainframes and plugins are so cheap nowadays that it's a no-brainer to use them as the basis for a tracking generator, even more so than when I built mine.

[hendorog]

ADF4355 is quite expensive (as are all of the newer Analog VCO/PLL's), another option is the MAX2871 which is quite a bit cheaper.

[KE5FX]

Yep, I like the MAX2871 too.  It offers good performance for the money and also takes less power than most other parts in its class.

That being said, all of the integrated PLL/VCO synths come in QFN packages that are difficult to use without a PC board.  So, for one-off projects with non-critical requirements, I always suggest shopping for cheap demo boards instead of specific ICs. 

[hendorog]

Yep, I like the MAX2871 too.  It offers good performance for the money and also takes less power than most other parts in its class.

That being said, all of the integrated PLL/VCO synths come in QFN packages that are difficult to use without a PC board.  So, for one-off projects with non-critical requirements, I always suggest shopping for cheap demo boards instead of specific ICs. 

Copy that, good advice - although I did find that 6GHz eval boards (or anything above 4.4GHz) were rare on ebay.

I've built up a couple of MAX2871 boards using a cheap hot air gun and solder paste, and while it took a couple of attempts to fix bridges etc, it wasn't _too_ difficult. I have some spare boards if the OP is feeling brave! Might have one or two MAX2870 chips to spare as well.

[KE5FX]

Another option would be the ADF4350.  That was one of the first of the integrated PLL/VCO parts, not a great performer but more than good enough for this job.  The ADF4355 is actually one of the better parts available currently (I've gotten them mixed up myself.)

ADF4350s have lots of options for prototyping, using boards like eBay 401169578378 or 231878104845.  You'd want to drive it with a VCXO to support a fine tracking adjustment control.   

[Skimask]

Good timing on this one.
Just picked up an HP 8590L (no options) for dirt cheap on fleabay.  Listing said it was working, dim display, out of focus, etc.  See what happens when it gets here.  Don't have much experience with the equipment.  Don't know if a guy can buy a TG unit that fits these models and just bolt it in or what.

[siggi]

Another option would be the ADF4350.  That was one of the first of the integrated PLL/VCO parts, not a great performer but more than good enough for this job.  The ADF4355 is actually one of the better parts available currently (I've gotten them mixed up myself.)

ADF4350s have lots of options for prototyping, using boards like eBay 401169578378 or 231878104845.  You'd want to drive it with a VCXO to support a fine tracking adjustment control.

Thanks - I ordered a cheapo-ADF4350 demo board. It'll be fun to see what comes out of it for starters...

[steve_w]

Check this out http://www.scottyspectrumanalyzer.com/  the slim concept may give you a start on a modular approach.


regards

SW