Spectrum Analyzers -- Input Protection??

[olsenn]

I recently asked this in another thread, but I thought I'd start a new specific to this issue... are inputs into spectrum analyzers low impedance? If so, why? Also, what kind of RMS voltage (sinusoid) can be tolerated by the average bench spectrum analyzer?

Basically, I am looking for a replacement for the FFT function on my Rigol DS1102E... I wan't to be able to hook up any waveform or even a DC voltage to about 60V peak-to-peak. I just figured that a spectrum analyzer would do this and provide greater accuracy and resolution, not to mention bandwidth, over my cheap ass scope... but if the darned thing is just for connecting to an antenna and sensing microscopic waves in the air then I am not interested in it.

Can anyone help?

[jahonen]

Generally, spectrum analyzers don't like DC at all. There might be a tolerance for something like 30 VDC or so, but even that is not possible if input is configured as DC coupled.

And yes, inputs are low impedance, 50 ohm usually. Maximum steady state input power is around 30 dBm, or 1 watt. That is usually assuming certain minimum value of input attenuation. Consequence of too much input power is burn out of input attenuator and/or first mixer which can be very expensive to fix. Note that input power is total power, so even "out of screen" signals count.

I have found out that 1 k DIY probe is quite usable with a SA, it limits the power which can get into SA so that that 1k resistor will burn-out before the SA front-end is damaged. External DC-block is also useful safeguard to protect against connecting DC into SA.

Regards,
Janne

[alm]

Just like high-bandwidth scopes, spectrum analyzers are usually low impedance. It gets increasingly hard to develop high-impedance inputs at higher frequencies. At 1 GHz, even 1 pF of shunt capacitance will result in an impedance of about 160 ohm. Not much point putting a 1 Mohm resistor in parallel with this, is there? High-impedance probes also get difficult since impedance and reflections get hard to control without using transmission lines. Try to calculate what a 1 Mohm transmission line would look like on a common FR-4 layer stack-up or as coax with common dielectrics. Few if any probes driving high-impedance inputs go beyond 500 MHz, most spectrum analyzers go much higher. Even those expensive 500 MHz passive hi-Z probes don't work terribly well, loading at 500 MHz is much higher than passive Z₀ (50 ohm) probes, and similar to a plain 50 ohm input.

If you need less DC loading, use a $$$ active probe. If you just need less high-frequency loading, a Z₀ probe like Janne suggests should be able to get quite close in performance. In many cases, you may not need a probe at all. If all you need are relative measurements or you can deal with calibration yourself, just using a piece of coax with some of the shielding removed should make an acceptable antenna. I wouldn't expect excellent flatness across the bandwith of the specan with it, but it is very unlikely to destroy the front-end this way.

[olsenn]

Thanks for the responses guys; I'm starting to get a better pricture of hoe these things work. I will have to look into getting a high impedance probe if I do get the SA I'm looking to buy... I know I will accidentally connect it to a DC 12V line or high rms wave by mistake and I don't want to break a 2000 dollar device. I may even buy a 1GHz Rigol oscilloscope instead and use the fft feature when I need it, but those things cost a hell of a lot more than the SA

[alm]

That 1 GHz scope won't be 1 GHz either in Hi-Z mode.

[olsenn]

That 1 GHz scope won't be 1 GHz either in Hi-Z mode.

buck futter! So just out of curiosity, what kind of voltages can be input at 1GHz?

[alm]

The 50 ohm input can probably handle something like 5 Vrms, see datasheet for exact details. The 1 Mohm input will have to be derated with frequency because the input at 1 MHz is much lower than at DC. This is usually not the case for 50 ohm inputs, although I wouldn't push it. At least it doesn't mind a few volts of DC. Many scopes will also automatically switch to 1 Mohm if the voltage gets too high, no idea about this particular scope.

Any probe you attach will have its own limitations. For example, an attenuating probe (hi-Z or Z₀) will decrease the voltage the scope's input sees, but the probe will have its own limits. A Hi-Z probe will have a fairly low impedance at higher frequencies, so the max. voltage is again derated with frequency. There's often a chart in the datasheet of the probe. Z₀ will have finite power handling in the resistor, you can't exactly use a 5 W wire-wound resistor in that position. Active (FET) probes will usually have a dynamic range somewhere between 2.5V to 10V or so, and a somewhat higher non-destructive max. voltage. It depends what that probe was designed for, some are designed for particular logic families and their voltage levels. The general purpose ones should be able to handle 12 VDC without damage, although the signal may be distorted if it's outside its dynamic range.

[ejeffrey]

You can use a 10x or 20x low-Z probe with a spectrum analyzer, and you can build one yourself if you like.  That will protect you from too much RF power as well as avoiding loading the circuit too heavily.  Some spectrum analyzers have a moderately high DC voltage standoff capability (I have an old Tek that says 300 VDC on it).  If yours is DC coupled, has insufficient DC protection, or you are just justifiably paranoid you can get an in-line DC block that will protect your gear.