Measuring power output of 500MHz oscillator?

[Chris Wilson]

I have built from a kit a small 500MHz oscillator circuit. It is on frequency and the sine wave looks fine. The equipment it's destined for is an HP8444A tracking generator. It's to mimic an option 059 that it currently doesn't have. The HP specs say the original option 059 oscillator should have an output of 4dBm plus or minus 0.5 dB I do not have a power meter, but do have a fairly good old Tektronix 7854 scope with a 7a19 plug in that is good for 500 MHz. Is this sufficient to view the output in mV's and check it from that? I know 4dBm is 2.5118 mW, but I need a sanity check how to go from there to mV. The 7A19 has a 50 ohm input impedance. I watched W2AEW 's latest You Tube video on attenuators the other night, couldn't have come at a more opportune moment if this thing is outputting too high. Thanks for any tips.

[lewis]

The voltage can be calculated from ohm's law. 4dBm would be 2.5mW into 50 ohms, so sqrt(50 * 2.5118mW) = 354mV rms, assuming the HP8444 also has a 50 ohm input.

[Chris Wilson]

Hmm, that's a good point. I am not sure what the input impedance is of the 8444A. The oscillator has a BNC output jack on the back of the 8444A, and when you need to use it you just loop the output to a "third LO" input BNC jack, with a short piece of co-ax. If I put a BNC T in between could I then measure the voltage with it actually feeding the impedance of the 8444A? And hope it's not so high as to do any harm? Here's 2 screen shots of the oscillator feeding the scope direct. Thanks for the fast reply!

[lewis]

It's almost certainly 50 ohms, but the voltage does seem a bit low for the tolerances quoted. The manuals are here: http://bama.edebris.com/manuals/hp/8444a/

I'm sure the RF guys on here will kill me, but teeing in shouldn't be a problem. I'd use a tee and measure with a 10x, 10M probe right at the tee - don't put any co-ax on the tee'd bit, just shove the scope probe right in the tee. This will minimise any reflections causing erroneous readings. If the voltage level is about the same as you're reading now with the oscillator connected directly to the scope, you can be relatively sure that the input impedance is 50 ohms.

[Rufus]

Tektronix 7854 scope with a 7a19 plug in that is good for 500 MHz.

Good for 500MHz doesn't mean good to measure accurately at 500MHz. Scope bandwidths are generally specified at the point where signals are no more than 3db down. And there is the bandwidth of the mainframe to consider. If you have detailed specifications for the plug in and mainframe you might be able to estimate.

[Chris Wilson]

Is a power meter the correct way to do this? Any other ways? I'll try the probe on X10 right at the T, but my probes are cheap and probably only 100 MHz max.  I also see an anomaly in that if I T in the scope on the back panel, where the co ax loop takes the output of my oscillator into the third lo input, I see 168 mV. But if I connect the scope, with the same coax lead, to the paralleled front pane third lo input BNC socket, I see only about 110 mV. I suspect my measurement techniques are poor.  EDIT, the probe won't see the waveform at all, evn on X10, probably because it's an El Cheapo Ebay one Thanks for the advice.

[PA0PBZ]

You have a tracking generator but no spectrum analyzer? Otherwise it would be very easy to measure the output of the oscillator...

[Chris Wilson]

I do have an SA, an HP8568B this is hopefully going to be run with, but I am literally terrified of blowing it up! Especially measuring stuff I have made that may have DC signals, or have an output I level I am unsure of (hence the need to measure it <LOL> I guess). But would using the SA make any differences to connector and cabling losses interfering with my measurements, or is it just down to the scope being at the end of its capability at 500 MHZ? Thanks.

[lewis]

...but I am literally terrified of blowing it up!

I know the feeling.... Some RF attenuators will come in handy. Kev (KJDS on here) sells some good stuff and is a pleasure to deal with.

100MHz probes are definitely no good for this, especially the ebay cheapies. I have some for throw-away use and you're lucky to get 50MHz out of them.

[PA0PBZ]

That SA can handle +30dBm (1 Watt) on the input, so if the input power of your final stage in the oscillator is less than that you should be safe. I'm assuming that the output is AC coupled so there should not be any DC present, otherwise just put an inductor over it.  If you are not sure, can you share the diagram?

[Rufus]

I do have an SA, an HP8568B this is hopefully going to be run with, but I am literally terrified of blowing it up!

From the specs I found both inputs are rated at up to +30dbm and input 2 is ac coupled with +/- 50vdc capability. 

Do you really think your oscillator is going to output more than 50vdc or 400 times more power than it is supposed to? Your scope already told you it isn't.

[AndyC_772]

I have built from a kit a small 500MHz oscillator circuit. It is on frequency and the sine wave looks fine.

Are you sure? A 500 MHz scope will show you the fundamental, but it certainly won't show you any higher harmonics. Any wave at that frequency will look like a sine wave, unless you have a very much faster scope (think 5 GHz +) and a probe to match. Alternatively you could use the spectrum analyser to show if any higher harmonics exist in the oscillator's output.

[Chris Wilson]

I have taken my brave pills and have two scans from the HP8568B.

The first is the direct output of my oscillator, straight into the SA. the second is T 'ing off between the oscillator and the HP8444A third LO input.

I'm going for a lie down, as you can tell, my confidence in using this new beast is pretty low

Thanks guys.

[PA0PBZ]

Well, obviously you are a few dB's short, the output is around -5 dBm and you need +4. When using the T you are putting another 50 Ohm in parallel so your output will drop as expected.

[PA4TIM]

3dB point of mainframe is 400Mhz, plugin is 600 MHz http://w140.com/tekwiki/wiki/7A19. The mainframe is the limmiting factor and how fast it drops is rather unpredictable. You need a spectrum analyser, (and if you are building a TG I presume you have one  )

Or build something. A good and easy way is an 8317. i made two powermeters with the latter. Very easy. (if you have good eyes)
Otherway is to use an stepattenuator and start low enough and then increase it just until it works fine.

[Chris Wilson]

3dB point of mainframe is 400Mhz, plugin is 600 MHz http://w140.com/tekwiki/wiki/7A19. The mainframe is the limmiting factor and how fast it drops is rather unpredictable. You need a spectrum analyser, (and if you are building a TG I presume you have one  )

Or build something. A good and easy way is an 8317. i made two powermeters with the latter. Very easy. (if you have good eyes)
Otherway is to use an stepattenuator and start low enough and then increase it just until it works fine.

If I had more time, and fewer uncompleted projects, I would have a go at building this:

https://sites.google.com/site/lofturj/intelligent_attenuator

Sadly I suspect it would be months before I finished it, realistically I need a commercial solution that keeps the SA safe from my current incompetence.  I can do most things, but measuring unknown RF power levels without coming out in a sweat over the 8568B is something I need to address. Thanks!

[Rufus]

If I had more time, and fewer uncompleted projects, I would have a go at building this:

https://sites.google.com/site/lofturj/intelligent_attenuator

Which has the same maximum input rating as your SA so you would just damage it instead.

Stick a 30db attenuator on the front of your SA and you will need more than a kW to damage it, you will need a 1kW attenuator of course.

[PA4TIM]

One thing you need for sure with a SA is a set of attenuators. I have a bunch of  bnc style attenuators (3,6,10,20dB), GR attenutors ( 10 and 20 dB)  and 3 adjustable ones (two of them 0-80 dB in 1 dB and 10 dB steps)

Then you need some directional couplers. You connect that to the generator or tranmitter you want to test and a dummyload on the other side. And most have a 40 dB attenuation to the testport ( use  the forward port in this case)

I made a few couplers myself. Just inuctive coupling. On port with a single turn coil, the other two connected through a second loop. I made a list of the atternuation / freuquency.

You may need a DC blocker.

I also made some E and H-field probes and have a small wideband antenna. That is a nice way to look at unknown signals. Even if they are a few hundered Watt.

( i use this stuff for my (5) SA's and (3) VNAs )

If you can read German, Funkamateur magazin sells 3 nice books from Hans Nussbaum about using scopes and spectrum analysers. Part 2 and 3 are nice for you. Really great stuff.

Tektronix has a given free the copyrights on the concepts range. One ( or two) are about spectrum analysers. Free to download on many places. ( very good books, I have about 15 original books of this series and a few on pdf) Tek and Agilent have more pdf books about this subject. Read them if you want to keep your SA working ;-)

[KJDS]

Looks like you need a small buffer amp on the output of the oscillator. I can't remember the accuracy of that spec ana but at best it will be +/-1dB so if you can borrow a power meter once you've got the buffer amp sorted that would help.

One of these will do the buffering job

http://217.34.103.131/pdfs/MAR-3SM+.pdf

http://www.ebay.co.uk/itm/MAR-3SM-Manu-MINI-Encapsulation-SMT-86-MONOLITHIC-AMPLIFIERS-50-Ohm-/110886772085?pt=LH_DefaultDomain_0&hash=item19d15ddd75


You'll struggle to blow up the input stage of a spec ana until you start deliberately generating hundreds of mW unless you accidently feed DC in when you shouldn't. It's worth making up a little DC block board. Buying a high power attenuator once you need one will be a lot cheaper than replacing the front end mixer.