Topics

DC voltage on a DC coupled spectrum analyzer

DW
 

I am well aware that a DC coupled spectrum analyzer might take up to +30dB peak but no more than 0V DC. The slightest DC voltage can destroy a DC coupled spectrum analyzer.

Curious if anyone has any thoughts about what goes on inside a DC coupled spectrum analyzer where if a DC voltage is applied why damage occurs. Thanks.

DW
 

https://www.edaboard.com/showthread.php?196119-Why-DC-be-not-given-input-to-a-spectrum-analyser

Here is what I found if anyone is interested. It seems when you apply a DC voltage the mixer consists of diodes and when you apply enough forward bias to it, it will turn on and supposedly current increases exponentially damaging the mixer.

Ed Breya
 

DC is just 0 Hz - there's no need to worry about the DC level if it's within the power input rating. If an SA is rated for operation to 0 Hz (DC-coupled), then it is typically for a wide bandwidth including 0 Hz, or at least to very low frequencies, and no coupling capacitor (or transformer) is used, which would limit the low-end response. Also remember, this assumes a 50 ohm or whatever source resistance - it's the current that can blow the input mixer if too much voltage is applied from too little source resistance.

The main reasons for specifying that no DC is allowed, are because DC offset on the input may upset the mixer balance, causing errors in the conversion process, and that excessive DC may be inadvertently applied due to operator error or measurement failure. SAs are typically for looking at AC/RF signals - that's where the information is - not particular DC levels, unless the information is at very low frequencies. Is 1 Hz approximately DC? How about 1 mHz? It depends on the application.

A common risk to SAs is in probing around an RF circuit, say an amplifier, which has DC levels present, for operation. An operator error like hooking up to an RF node that has significant DC available, without assuring that it's either AC-coupled, or that the source resistance is high enough to limit fault currents, can damage the SA. The safest approach and rule of thumb and specification is to not allow any DC at the input. If you don't heed this advice, then it's your own fault. In reality, some DC is no sweat, but it's best to just say zero.

Ed

DW
 

Great answer, thanks

radioconnection@...
 

Unless you have a need to go into the LF region, a DC blocking accessory is cheap insurance.

Chuck Harris
 

Often, a DC block isn't insurance at all.

Every DC block I have met is simply a series capacitor, usually
about 0.1uf... those for lower frequency SA's, can be several uf,
so as to not degrade the low frequency performance too much.

If you suddenly apply a high DC voltage, from a low impedance
source to the DC block, it will apply all of the charging current
impulse directly to the SA's front end. This impulse can quite
easily destroy the mixer, if it is directly exposed without an
attenuator.

Best is to be cognizant of what you are doing, and discharge long
coax runs, and antennas, before connecting, and simply don't connect
any DC charged source to the SA without using using a high
impedance probe... preferably with a series capacitor.... 10M scope
probes are often a very good choice for probing in dodgy situations.

-Chuck Harris

radioconnection@... wrote:

Unless you have a need to go into the LF region, a DC blocking accessory is cheap insurance.



DW
 

Here is a thought, for example a HP 8566 which can go down to 100 Hz. I assume they choose that number as that is the lowest it can go for certified performance, but I wonder if they choose that number could also be the minimum safe frequency range. For example if you go below the 100 Hz minimum to say 60 or 30 Hz, or maybe even 10 Hz, I wonder if that presents a hazard to the mixer as there is a longer but brief duration of positive and negative DC without a DC blocking capacitor that could cause eventual damage.

Ed Breya
 

You also have to worry about plain old component failure. RF amplifiers are typically capacitor-coupled, and relatively safe DC-wise - unless the output cap shorts out, for instance. Then the DC from inside can be applied to the SA. For higher frequencies, the cap is usually small and reliable, but for very broad band and low frequency coverage, it may need to be quite large, including getting into tantalum territory, with its reliability risks.

One particularly scary situation I have is in one of my broadband noise generator projects (I think I discussed it somewhat here or in the HP-agilent group a while back). The noise generator uses a breakdown diode as the source, which is amplified through about ten stages - all coupled with rather large solid Ta caps, in order to have a very low cutoff frequency. These are off-shelf modules, but I can open them up to modify them, and I can add external protection features. A large part of the project effort is in making it "safe" for SAs and such, without compromising the RF and high crest factor capability. It is still unfinished, partly because of this issue. Not only is there risk of a Ta cap shorting anywhere in the chain, but also just powering it up and down can cause a large surge at the output while the caps all get charged or discharged to their new quiescent levels.

Ed

Jim Ford
 

Good points, Chuck.It's a good habit to discharge coax cables that have been lying or hanging around for a while before connecting them to ESD sensitive equipment.  Obviously if you have just been using a cable there's no point in discharging it, but cables are capacitors and will charge up.  Easy enough to forget that.Jim Ford Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: Chuck Harris <cfharris@...> Date: 5/28/19 9:54 AM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] DC voltage on a DC coupled spectrum analyzer Often, a DC block isn't insurance at all.Every DC block I have met is simply a series capacitor, usuallyabout 0.1uf... those for lower frequency SA's, can be several uf,so as to not degrade the low frequency performance too much.If you suddenly apply a high DC voltage, from a low impedancesource to the DC block, it will apply all of the charging currentimpulse directly to the SA's front end.  This impulse can quiteeasily destroy the mixer, if it is directly exposed without anattenuator.Best is to be cognizant of what you are doing, and discharge longcoax runs, and antennas, before connecting, and simply don't connectany DC charged source to the SA without using using a highimpedance probe... preferably with a series capacitor.... 10M scopeprobes are often a very good choice for probing in dodgy situations.-Chuck Harrisradioconnection@... wrote:> Unless you have a need to go into the LF region, a DC blocking accessory is cheap insurance.> > > >

Jim Ford
 

Not sure why the 100 Hz low end, but I  did take my 8566A down that low a couple months ago just to see what it would do.   Lots of 1/f noise!  I would use my Stanford Research Systems SR780 FFT analyzer if I were looking at signals in the audio range.Jim Ford Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: DW <wilson2115@...> Date: 5/28/19 10:14 AM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] DC voltage on a DC coupled spectrum analyzer Here is a thought, for example a HP 8566 which can go down to 100 Hz. I assume they choose that number as that is the lowest it can go for certified performance, but I wonder if they choose that number could also be the minimum safe frequency range. For example if you go below the 100 Hz minimum to say 60 or 30 Hz, or maybe even 10 Hz, I wonder if that presents a hazard to the mixer as there is a longer but brief duration of positive and negative DC without a DC blocking capacitor that could cause eventual damage.

Rajesh VS
 

Thanks Chuck and Jim,
These are very valuable points, specially around DC block and co-ax
cables.
Any comments /suggestion about using fast acting lowest current pico fuses
(sub miniature ) as an added protection along the probe? about their
impact on input loading/parasitic ?

//** I almost burned the front end of my hp 8568B by accidentally plugging
it in to DC rail instead of RF out in a tracking generator board.
I was saved by the built in pico fuse **//

On Tue, May 28, 2019 at 11:07 AM Jim Ford <james.ford@...> wrote:

Good points, Chuck.It's a good habit to discharge coax cables that have
been lying or hanging around for a while before connecting them to ESD
sensitive equipment. Obviously if you have just been using a cable there's
no point in discharging it, but cables are capacitors and will charge up.
Easy enough to forget that.Jim Ford Sent from my Verizon, Samsung Galaxy
smartphone
-------- Original message --------From: Chuck Harris <cfharris@...>
Date: 5/28/19 9:54 AM (GMT-08:00) To: TekScopes@groups.io Subject: Re:
[TekScopes] DC voltage on a DC coupled spectrum analyzer Often, a DC block
isn't insurance at all.Every DC block I have met is simply a series
capacitor, usuallyabout 0.1uf... those for lower frequency SA's, can be
several uf,so as to not degrade the low frequency performance too much.If
you suddenly apply a high DC voltage, from a low impedancesource to the DC
block, it will apply all of the charging currentimpulse directly to the
SA's front end. This impulse can quiteeasily destroy the mixer, if it is
directly exposed without anattenuator.Best is to be cognizant of what you
are doing, and discharge longcoax runs, and antennas, before connecting,
and simply don't connectany DC charged source to the SA without using using
a highimpedance probe... preferably with a series capacitor.... 10M
scopeprobes are often a very good choice for probing in dodgy
situations.-Chuck Harrisradioconnection@... wrote:> Unless you have a
need to go into the LF region, a DC blocking accessory is cheap insurance.>

--
/Rajesh

DW
 

Disregard my post above as it has become obvious to me from Ed's post about exceeding the lower limits of the spectrum analyzer that source impedance is important when running low frequencies especially DC which I want to avoid anyways

I must add at one time I used the output of an iPhone and an iPad turned down low generating a sine wave before, it worked but now it has me thinking was that a bad idea?

Thanks for the excellent replies

Ed Breya
 

Jim wrote:

"Not sure why the 100 Hz low end, but I did take my 8566A down that low a couple months ago just to see what it would do. Lots of 1/f noise! "

That's not 1/f noise - it's just the DC impulse or zero frequency (or IF) response, viewed through the narrowest resolution BW filter (10 Hz). The 100 Hz bottom spec is just where it's still usable, since you can view a 100 Hz signal close to the noise floor, just past where the RBW filter response falls reasonably far down. Below 100 Hz, the signal becomes swamped and hidden under the filter skirt. You can see that it's a very narrow bandpass filter by viewing the negative frequency range. Set CF=0, SPAN=1 kHz, RBW=10 Hz, for example.

Ed

Jim Ford
 

Thanks, Ed.I'll have to try that sometime.  Maybe this coming weekend.Learn something new every day!JimSent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------From: "Ed Breya via Groups.Io" <edbreya=yahoo.com@groups.io> Date: 5/28/19 1:08 PM (GMT-08:00) To: TekScopes@groups.io Subject: Re: [TekScopes] DC voltage on a DC coupled spectrum analyzer Jim wrote:"Not sure why the 100 Hz low end, but I  did take my 8566A down that low a couple months ago just to see what it would do.   Lots of 1/f noise! "That's not 1/f noise - it's just the DC impulse or zero frequency (or IF) response, viewed through the narrowest resolution BW filter (10 Hz). The 100 Hz bottom spec is just where it's still usable, since you can view a 100 Hz signal close to the noise floor, just past where the RBW filter response falls reasonably far down. Below 100 Hz, the signal becomes swamped and hidden under the filter skirt. You can see that it's a very narrow bandpass filter by viewing the negative frequency range. Set CF=0, SPAN=1 kHz, RBW=10 Hz, for example.Ed