Never put DC signal into a signal analyzer?

[Jim from Chicago]

I was watching this video and the guy says never to put DC into a signal analyzer? (link is timestamped to where he says this)

Is this true? I am trying to measure LDO noise performance which is obviously DC and in my case around 3V. My signal analyzer has a toggle which says "RF Input: DC or AC" and I select DC, and I thought everything was fine.

Should I put a blocking capacitor in like he says? What is the reasoning behind his claim? Thanks.

[RoGeorge]

Yes, the gold rule is to never put DC into a spectrum analyzer.  DC can damage the instrument.
Always use a DC blocker (pass-through adapter) attached to the input of the instrument.

However, there might be some models that support DC, but that's rather rare.  What instrument do you have?
Read carefully the user manual to find out if yours supports DC or not, and for what levels.

[basinstreetdesign]

author=Jim from Chicago
I was watching this video and the guy says never to put DC into a signal analyzer?

You bet!
I accidentally put 30V into the input of a 141T HP spectrum analyzer.  It blew the input filter in the RF section of the SA.  It is a 1.5 GHz LPF and was open from one end to the other.  I had to fix it to get the SA to work again so with dremel and fine point soldering iron I cut a hole in the very small tube encasing it and put a very small blob of solder to bridge the even smaller gap.  It now works but is now a 700 MHz LPF.

So don't try it just to see what will happen.  You won't like the results.

[rfclown]

Read the manual. All analyzers aren't the same. I have two analyzers in front of me right now. One says "0VDC MAX", the other says "26 VDC MAX". The one that says 26V max has an attenuator on the input inside the box.

[bdunham7]

Is this true? I am trying to measure LDO noise performance which is obviously DC and in my case around 3V. My signal analyzer has a toggle which says "RF Input: DC or AC" and I select DC, and I thought everything was fine.

Should I put a blocking capacitor in like he says? What is the reasoning behind his claim? Thanks.

RTFM, my friend!  What model analyzer?  Your AC/DC input selector may well be like that of an oscilloscope where it means DC coupled or AC coupled, and you probably want AC coupled if you are looking at noise on a DC voltage.  Sort of backwards if you don't realize what is going on.  The reason not to allow a DC bias on the input is to avoid overloading the input amplifier or burning something out like an input resistor.  It's possible, even likely, that your analyzer can tolerate 3V without damage and even without distorting the signal--but check the manual as I'm pretty sure they'll tell you.

[Berni]

There are ones that work down to DC, but they are usually called something different like a Dynamic Signal Analyzer and usually work in the digital domain.

That being said most spectrum analyzers will handle a little bit of DC on the input just fine. There is usually some sort of filter in the way that has DC blocking properties, but this being RF the components inside tend to be really tiny and delicate so they won't handle a whole lot of it. However there are some spectrum analyzers that WILL for sure get blown if you send DC into it. Some older high performance units have the mixers diodes sitting right on the input, since getting deep into microwave bands back then was hard. For this reason i have a DC block sitting on the input of my HP 8566B. But even a DC block has a certain voltage rating and so will blow up at some point, likely sending all the DC trough it. Even then the huge voltage step of touching a large DC voltage with a probe might produce a pulse large enough to damage the input.

If you do often need to probe DC voltage rails then you can still DIY yourself a probe for the task. Build the DC blocking and overload protection into your probe, test it to its limits on an oscilloscope so that you know it works, then plug that into your spectrum analyzer.

That being said the lower performance spectrum analyzers that only go into the low 100s of MHz are pretty resilient because much less RF optimization is required inside. They are also cheap on the used market and are easier to repair due to having less fancy special nude virgin manufactured RF black magic components inside.

[Vovk_Z]

To TS: When we say 'spectrum' that means we are talking about periodic signals, that is AC, but not DC.
Is there a DC blocker inside your SA or not - you should look at the manual.

[radiolistener]

Is this true?

Yes.

Should I put a blocking capacitor in like he says? What is the reasoning behind his claim? Thanks.

Yes, you're needs DC block capacitor. It needs to keep signal bias offset at at 0 V.

[David Hess]

There are ones that work down to DC, but they are usually called something different like a Dynamic Signal Analyzer and usually work in the digital domain.

If the first mixer of a spectrum analyzer is wired with the LO and RF ports swapped, then it can operate with inputs down to DC, but it also means a small excessive DC current will destroy the mixer.

Dynamic Signal Analyzers usually have a completely different input architecture than spectrum analyzers.

[SilverSolder]

I have a 1970s spectrum analyzer here (HP3571A) that I like to play with sometimes, just because it is so hard to use LOL.

It does seem to have unusually robust input circuitry....   its (DC coupled) input is rated as follows -



Maximum DC input voltage (All ranges):  +/- 100V DC

Continuous Overloads (All ranges):  10V RMS +/- 100V DC

Momentary Overloads (All ranges, <5 seconds):   100V RMS  +/- 100V DC



Wonder what's changed in spectrum analyzer design - is it the higher bandwidth of modern devices that require smaller and more fragile semiconductors?

[Berni]

There are ones that work down to DC, but they are usually called something different like a Dynamic Signal Analyzer and usually work in the digital domain.

If the first mixer of a spectrum analyzer is wired with the LO and RF ports swapped, then it can operate with inputs down to DC, but it also means a small excessive DC current will destroy the mixer.

Dynamic Signal Analyzers usually have a completely different input architecture than spectrum analyzers.

Yeah hence why they don't tend to do this as well as working with the DC down the rest of the signal chain can cause other problems. Still some go down to 100Hz or even less so that is pretty impressive.

In a HP 8566 it gets even worse. The mixer is just a diode directly looking at the input while a LO is fed back trough it to switch it. The switching creates higher harmonics that mix the RF input and come back out the same port the LO was fed in. At the same time a DC bias is fed into the LO to keep the diode on the perfect edge of switching in order to get the most IF output signal. Amazing all of this even works but that's how they got it to do 22GHz back in the late 1970s. It also means that you can really easily blow this delicate mixing diode with mere volts of DC. And they are very serious about you not putting any DC into it as they put a "0V DC MAX" label next to the input.

I have a 1970s spectrum analyzer here (HP3571A) that I like to play with sometimes, just because it is so hard to use LOL.
...
Wonder what's changed in spectrum analyzer design - is it the higher bandwidth of modern devices that require smaller and more fragile semiconductors?

That is quite the ancient dinosaur alright. Impressive that it nearly handles mains voltage directly on the input.

One thing about high frequency RF is that things inevitably have to get smaller to reduce parasitics even just capacitors, inductors, transformers get pretty special in the high GHz ranges. Similar goes for any semiconductors that needs to switch this quickly, a regular transistor just wont switch at 4GHz. Combined with some of these instruments pushing the boundaries of performance means that some of the protection features could degrade performance, so they are skipped.