HP 8593A 22 GHz Spectrum Analyzer - LO riding on TG output

[John_ITIC]

Hello, I recently bought an 8593A SA. When testing the tracking generator, I found that the LO is riding on top of the tracking generator output. For instance, when setting frequency to zero-span and 100 MHz, the TG outputs the 100 MHz and a 3.9214 GHz + 100 MHz = 4.0214 GHz signal. When setting zero-span frequency to 200 MHz, I get 4.1214 GHz and 100 MHz out, etc.

I took the tracking generator apart and all the little micro-modules looked ok. The signal path goes mixer -> passive filter -> output amplifier. All look okay. The filter is passive so cannot be broken. I verified that with the 600 MHz disconnected from the A3A15 module, the 100 MHz signal disappeared but the 4.0214 GHz signal remained with unchanged amplitude. So, it seems the A3A15 lets through the LO, 20 dBm lower (-30dBm) than the TG output (-10dBm).

I could confirm that the tracking generator (in full span) is level within 2dB across the full 2.9 GHz span. So it seems otherwise be fine.

My question is:  is it normal to have the LO riding on the 8593A TG output? When measuring filters and feeding back the filter output to the 8593A input, the LO will not be visible as the maximum TG frequency is 2.9GHz. Perhaps this is by design? I find it odd, however, as my 8590B outputs a clean 100 MHz for the same test scenario.

Select pictures attached.

Thanks,
/John.

[John_ITIC]

More pictures

[John_ITIC]

Some internal TG pictures

[Berni]

It is quiet common to have some extra junk in a tracking generator output, this is the reason why they should not be used as a signal generator on zero span. The point of the tracking generator is to provide a consistent large amount of RF power at the frequency the spectrum analyzer is looking at, so that it provides a lot of signal for the SA to see.

The spectrum analyzer is designed to ignore frequencies outside of what it is tuned for, so it doesn't affect the mesurement. They just do a decent attempt at reducing the other junk in order to avoid the junk from overloading or mixing, perhaps on any RF amplifiers you might be characterizing as a DUT.

How much junk is normal for your specific HP i don't know tho, some other forum member can probably answer that. Might be worth looking for a service and calibration manual for it.

[MarkL]

I can confirm the LO appears in the TG output on a 8595E.  Using your same setup, the LO shows up at 4.0214GHz and down -28dB from the 100MHz signal.

On the E series, the Calibration Guide specifies the Tracking Generator LO Feedthrough to be <=-16dBm from 3.9217GHz to 6.8214GHz @ -1dBm output.  There's no adjustment if it's out of spec.  I didn't see a Calibration Guide for the 8593A, but it's likely in the same ballpark.

[John_ITIC]

Berni, MarkL: Thank you for the confirmations. It all makes sense now. What confused me was that my 8590B has a 1.8 GHz LP filter on its TG output (A24 FL1) which filters out the LO effectively. But the 8593A apparently doesn't filter out its LO much (-20dBc on my system). Anyways, I'm confident that my unit is now working properly!

I actually bought a 2.9 GHz LP filter on Ebay (https://www.ebay.com/sch/i.html?_from=R40&_nkw=RLC+Electronic+L-1495A&_sacat=0&_sop=15) to see how much it will attenuate the signal in the pass band. If not too much, then I can still use the 8593A's TG as a signal generator in zero-span mode.

The same filter (HP A3A3 / RLC Electronic L-1495A) is used on the input side on the 8593A, when the SA is on the low-band <= 2.9 GHz.
 

Thanks,
/John.

[Berni]

Yeah the reason for this LO noise is that the tracking signal is usually generated by mixing the LO with a certain tone in a mixer. This mixer likely has quite a bit of LO feedtrough to its output and they probably need a pretty steep low pass filter to clean all of it out, and since its not critical they don't bother. This is also why there is no adjustment for turning this. It just depends on how good the mixer and filter is.

With a good filter this can definitely be cleaned up to make it more usable as a signal generator. It does have one feature that is very difficult to find in RF signal generators, continuous gap free frequency sweep over a large range.