Spectrum analyzers DC blocking devices connected to input..

[jpdesroc]

I have a HP 8594A Spectrum Analyzer that I would like to protect from any incoming DC voltage to its 50 ohms input.
I saw some DC blocking BNC to BNC devices around and I was wondering what is the internal circuitry of such devices.
Maybe only a series capacitor around 0.1uf between both input and output
Any clues ??

[MrBungle]

That's basically it. There are two types though: inside and outside DC blocks. The inside blocks DC voltage from the conductor of the transmission line and the outside block blocks DC voltage on the shield. I never worked out which one to use when and only ever used the inside block myself as both halves of the system were connected to RF ground.

They are high pass in nature and tend to have a min frequency as well so you have to watch out for that.

[jpdesroc]

I'm interested from 1Mhz and above on 50 ohms. So 0.1uf gives 1..3ohms at that frequency and less above it. That should do.

Suddenly applied DC to a spectrum analyser given enough voltage will kill the input, best use the smallest capacitor you can get away with, even 10nF would be 15 ohms approx at 1MHz and hardly affect measurements. 0.1u could give enough energy to pop the input quite easily if applied to 100V.

[jpdesroc]

Good advices.. 0.1uf is a little high.
I could use 2 x .022uf or even 3 x .033uf in series to give a final 0.01uf.

I had chosen 0.1uf because I will very often play
in the 500khz-1600khz AM band and wanted a flat
response in this range but with DC protection..
0.01uf @ 500khz gives 32 ohms (61% of signal at the 50 ohms connector)
0.01uf @ 1600khz gives 10 ohms (83.3% of signal at the 50 ohms connector)
Not a flat response on my future readings..
Hmmmm..

[cmjones01]

Originally Posted by MrBungle

That's basically it. There are two types though: inside and outside DC blocks. The inside blocks DC voltage from the conductor of the transmission line and the outside block blocks DC voltage on the shield. I never worked out which one to use when and only ever used the inside block myself as both halves of the system were connected to RF ground.

My Tek 7L12 spectrum analyser has a DC block on the output of its first local oscillator. I wondered why it was there, and why it was encased in an annoying lump of plastic that gets in the way. It turns out that the whole YIG oscillator is sitting at a significant DC offset, so both the shield and inner need DC blocking.

Outside blocks are useful when you've got two systems at different, or uncertain, DC levels. Inside blocks are much more common, though, and probably the right thing for the input to a spectrum analyser.

Chris

[MrBungle]

Thanks for clarification. Much appreciated

[Argus25]

If you know what the max voltages are you can expect to see at the input, you can use back to back diodes in series with zeners above that voltage, which have little capacitance and won't affect the measurements in a 50R environment, then if a large voltage spike comes along they help clamp it and protect the input circuitry.

[jpdesroc]

Yes. Back to back diodes in parallel with the DC blocker's final output to the SA 50R input.
Since my SA has a max of +30dbm input tolerance (10vpp)
I'm thinking of using enough back to back schotky diodes to do a +/- 5v clipping protection.. with a 0.01uf cap in series between both BNC center
conductors.
Again I'm wondering the freq flatness around 500kHz to 1600khz
using a 0.01uf @ 50R input imp. as a DC blocker..??
What do you think ?

[MrBungle]

Just a point with the diodes: they will have capacitance even if they’re not conducting and therefore side effects. The Zener across the transistor in my PA caused me no end of trouble.

[jpdesroc]

Just a point with the diodes: they will have capacitance even if they’re not conducting and therefore side effects....

Even schotky diodes ? very low capacitance ?

[MrBungle]

Yep. Not only that they also change capacitance with voltage

[John_BS]

0.01uf @ 500khz gives 32 ohms (61% of signal at the 50 ohms connector)

The voltage appearing across the C is of course in quadrature to the voltage developed across the resistive input of the analyser. If you assume your source is also 50 ohms, the loss due to the 0.01uF at 500kHz (relative to the C being removed) is only about 0.5dB (0.42dB theoretical I think). For a zero source impedance the worst case loss is 1.5dB.

I'd use 0.022u and make sure the input attenuator is always set to something > 0dB when first connecting the DUT!
John

[G0HZU_JMR]

I use a commercial 100kHz to 18GHz (n-type) DC block with my old HP8566B spectrum analyser and I think the internal capacitor is a nominal 40nF.

Even with a decent DC block inline, it's never a good idea to have a low resistance path (via a bias tee?) to a meaty PSU. I always try and have a decent external attenuator inline and additional DC blocking 'in circuit'.

So the 18GHz DC block is the last line of defence. I've also got a suite of DC blocks and limiters and fused limiters that I use with my analysers here at home.

I measured the 18GHz DC block a couple of years ago on my VNA and it has low insertion loss up to 8.5GHz as you can see below. So I try and keep it inline as much as possible. It's rare that I need to remove it from the HP8566B

[Argus25]

Originally Posted by jpdesroc

Even schotky diodes ? very low capacitance ?

If you build your diode network into a male-female in line BNC housing or similar, you can then plug it in and out to see if it is having any practical effects on the measurements, obviously any capacitive effects increase with increasing frequency.

[Craig Sawyers]

Or, if you are prepared to pay £50, you can buy one http://www.leobodnar.com/shop/index....roducts_id=275

No personal interest other than having bought a 30ps rise time pulse generator from the guy.

[G0HZU_JMR]

For low frequency and through VHF it's possible to make an effective limiter using 4 x 1N4148 diodes arranged in two back to back pairs. This can be arranged to have very low loss up through VHF and will provide good limiting against accidental RF overload for signals up to several watts across this frequency range. A bare limiter with just the two pairs of back to back 1N4148 diodes should have << 0.1dB loss and low VSWR by 500MHz if built correctly. But it won't limit reliably up at UHF and may cease to be a limiter up at these frequencies so best used where the threat of overload is at HF or VHF. It won't offer much protection against DC unless the current is limited to maybe 200mA and a more elaborate version would include a series DC block cap at the input and also an inline RF fuse. But this will add some loss. Maybe 0.5dB loss by 500MHz for the whole circuit made from the capacitor + RF fuse + 4x1N4148 ?

For operation from high VHF through UHF a simple PIN diode limiter could be used like the circuit below. Big RF signals in the VHF/UHF range will cause the diode to conduct and it will behave like a low value clamping/limiting resistor that will reflect energy back to the source. But it will have a high pass response due to the shunt inductor and the series cap and the diode won't work well at low frequencies. So no good for lower frequencies.

Both limiters will introduce distortion if any wanted signals get up much beyond 1 milliwatt so care must be taken to manage the wanted signal levels into the limiter.