Hey everyone,

I'm looking to generate a ~500MHz -15GHz digital signal. It really just needs to be alternating HIGHs and LOWs. I don't need any control over the signal itself as I'll only be using it to measure insertion/return loss in a system. But I will need to be able to scale it at will between 500MHz and 15GHz

What's the best way of accomplishing this? The lower end of the spectrum isn't looking challenging. But up in the 15GHz range, my options seem very limited. Do any of you have suggestions? I am willing to settle with a max of 10GHz if that makes things any easier.

Thanks!

 

Here is one in EBay for just $27360.00

20GHz signal generator

 

I'm no expert, but I don't think anything that actually works reliably and accurately at those frequencies is going to be ... economical. What's your ultimate goal? You're looking for 15GHz but will settle for 10. It sounds like your project might be missing a specification. Specs can help you figure out exactly what you need, so perhaps in this case it's worth the time to put together a spec.

 

This will give you 2.5GHz, fixed frequency but hey, its only 50 bucks!

 

I'm looking to generate a ~500MHz -15GHz digital signal.

Getting a digital signal at that frequency is quite difficult.
It you are measuring return loss, then I don't see why the signal would be digital.

Likely you best bet for generating that frequency range is to look for a commercial signal generator, but it may be a challenge to find one you can afford unless you have deep pockets.
If this is short term, you might consider renting one.;

 

I'm looking to generate a ~500MHz -15GHz digital signal. It really just needs to be alternating HIGHs and LOWs. I don't need any control over the signal itself as I'll only be using it to measure insertion/return loss in a system.

How does that work? I'd think that a square wave digital signal would have lots of 3rd harmonics (and higher) so
the frequency being measured isn't a single value. Perhaps a comb or noise source and selective receiver?

 

You may want to do a Google of "YIG Tuned Oscillators" and you may find some which have been gutted from spectrum analyzers. That may get you some in ranges you are looking for.

Ron

 

Perhaps one can use several 1 GHz pulse generators. If we had ten of them and we set up a system where they were all locked slightly out of phase, then we could devise a system where we see 10 pulses per period, mixing these into a single output would then appear to be a 10 GHz pulse generator...

 

Thanks, everyone. I've been trudging through this and discovering as I go so I really appreciate all the help and suggestions.
To clarify a few points, specs call for 15GHz but the reality is that in practice I don't expect anything above the 8-10GHz range. If 15GHz costs me $27360.00, you can bet I'll settle for 10 and let companies worry about the upper end of the spectrum.

I've been digging around since I initially posted this and I've found the DS125DF1610/DS150DF1610 that should be capable of around the 6-8GHz range for PRBS test sequences (ish). Almost there, and I could find a use for the EOM, PRBS Checker, CDR, and DFE. So there's that.

The part that is throwing me for a loop is that there are multiple specs out there for high-speed data transfers that should, in theory, be around the same scale as what I'm looking for. I'm not all that knowledgeable about the 128b/132b encoding USB 4 Gen 3 uses, but USB4 Gen 3 supports 20Gbps per channel, which should be around the range of what I'm looking for. Although judging from the PAM-3 encoding of Gen 4, it's probably a decent guess to say that higher frequencies aren't tenable at consumer prices.
Part of me was expecting to find plenty of parts from the USB/Thunderbolt world, but now I'm wondering how common 20Gbps is in practice. (Or how much I'm misunderstanding something)

I'm not that knowledgeable when it comes to high-speed data transfer mechanisms so if you have any insight, please do share.

 

Things might be simpler if you can just use a vanilla CW pulse stream. It appears that you are not going to be sending data.

 

This will give you 2.5GHz, fixed frequency but hey, its only 50 bucks!

View attachment 292701

Am I mistaken that this would give 2.45-ish GHz. It seems part of the gray area is how precise the clock needs to be for this project.