Did an Aliexpress DC block kill my 27GHz SA?

[alexturner]

My SA is dead, and I'm very sad. I have an older W&G SNA-23, which has just been lovely to use, and I'd been using it a week and a bit ago attached to a LN antenna capturing anecdotal EMI data from an SMPS.

I'd powered it up yesterday to perform some measurements on a signal source I've been repairing and, prior to that, had installed an N connector DC block as a just in case. I noticed when performing my measurements through the DC block, there was nothing, just the noise floor and the same when I removed it. I hooked the SA up to a known signal source (an R&S SG) and again, nothing. At one point I saw an unusual square wave-type pattern run across the screen whilst the SG was sweeping but since then nothing at all.

At no point have I exposed this unit to DC or the possibility of a DC source. The issue persists across what I can test in the bands and throughout all attenuators. The only thing I've connected to this unit was a low-cost DC Block (DC-6GHz) from Aliexpress. Could this have killed it?

Next, where to start debugging? Given it seems to be uniform across the entire band should I start looking at reference sources and the YIG? Would appreciate all advice, ideas and condolences at this point.

[ArdWar]

Well, DC block on itself surely shouldn't be able to damage anything? Especially if you don't actually ever put DC signal through it before.

Can you fed your SNA from the IF input? If the DC block was indeed the culprit it hopefully kills something upstream.

[joeqsmith]

My SA is dead, and I'm very sad. I have an older W&G SNA-23, which has just been lovely to use, and I'd been using it a week and a bit ago attached to a LN antenna capturing anecdotal EMI data from an SMPS.

I'd powered it up yesterday to perform some measurements on a signal source I've been repairing and, prior to that, had installed an N connector DC block as a just in case. I noticed when performing my measurements through the DC block, there was nothing, just the noise floor and the same when I removed it. I hooked the SA up to a known signal source (an R&S SG) and again, nothing. At one point I saw an unusual square wave-type pattern run across the screen whilst the SG was sweeping but since then nothing at all.

At no point have I exposed this unit to DC or the possibility of a DC source. The issue persists across what I can test in the bands and throughout all attenuators. The only thing I've connected to this unit was a low-cost DC Block (DC-6GHz) from Aliexpress. Could this have killed it?

Next, where to start debugging? Given it seems to be uniform across the entire band should I start looking at reference sources and the YIG? Would appreciate all advice, ideas and condolences at this point.

I have had people write me suggesting a DC block is some sort of transient protection device.   Of course, you know that a DC block is a capacitor and while a capacitor can be considered an open to DC, it can be considered a short for AC (not really, just over simplifying).   Imagine the input to your SA is 50ohms to ground.  You place a capacitor in series with it.  Now you connect the other side of that cap to a 12V source.  Thinking you saved your front end from that DC.  Instead, you just supplied a short transient to the front end that ended it's life.  Think about it, you just applied a step function with very high frequency.   Of course, this depends on the SA front end design, the value of the blocking cap, the voltage applied..... 

I made a few videos showing where I used one of those lost cost VNAs to measure the impedance of a power distribution network.  Of course, you do this powered up.  So I use blocks for this.  They need to have a fairly large capacitance to get the low frequency cutoff I wanted.  I warned viewers how easy it would be to damage the VNA doing this. 

Along the same lines I have had people tell me how attenuators are also used as transient protection devices.  I made a video once that worked some of the basics on protecting a front end.  Like any design, its a give and take. 

[TERRA Operative]

In addition to what joeqsmoth said, is it possible that the capacitor in the DC block could have been holding a charge (ESD from packaging, general charge from the aether, etc) and that discharged into your instrument when you connected it?

[alm]

Could it be that the N connector on the DC block was out of tolerance and bent the center contact fingers on the spectrum analyzer? Have you inspected them closely?

[joeqsmith]

Could it be that the N connector on the DC block was out of tolerance and bent the center contact fingers on the spectrum analyzer? Have you inspected them closely?

Good point.  Imagine someone getting a good deal on some 75 ohm N connectors and thinking, they will be fine for their application.  Then they proceed to jam them into their 50 ohm N connectors.   

I'm guessing at 27GHz, those are some special connectors.  Not something most would plug a 90 cent connector into.

[joeqsmith]

In addition to what joeqsmoth said, is it possible that the capacitor in the DC block could have been holding a charge (ESD from packaging, general charge from the aether, etc) and that discharged into your instrument when you connected it?

Another very good point.  I mentioned that when making the PDN measurements with the DC blocks, there are two concerns.  Powering up, but also just as important powering down.  You can fry it both ways for the same reason you mention.  The DC block is now charged and turning off the DUT may inject a transient in the opposite polarity. 

[Gerhard_dk4xp]

On my SNA-33, the input plug was a 2 piece construction by
Rosenberger specially made for W&G. The first part was the
semi-rigid transition to something unusual; the 2nd part was
a kind of port saver from "unusual" to PC3.5 or whatever.
The port saver was missing on mine when I bought it in 2005.
I then called W&G; it turned out that none were left-over. They
could have asked Rosenberger to make some more but it would
have cost an arm and a leg.

The way out was to use a PC3.5 connector from a blown
HP54750 sampler plug in. It took some filing to make the front plate
hole bigger and then to make the semi rigid line fit.  It was quite
a lot of work but it can be done. It did its job for 15 years.
My SNA-33 is still in the state of the Zebra-thread here on
eevblog, i.e. defunct.

But I don't think it's the input connector. A signal generator output
should be visible even when the cable end is only in close vicinity.

[joeqsmith]

Agree.  I suspect the clue was in their initial post:

and I'd been using it a week and a bit ago attached to a LN antenna capturing anecdotal EMI data from an SMPS.

Where I can envision someone touching the blocks input to the SMPS output....

I did have some hams posting about my video explaining the transient protection that the OP may find humorous.  I would have liked to have had a discussion with them about it but the persons running the group kept censoring my posts making it impossible, so I dropped it and unsubscribed.   

https://nanorfe.com/forum/ESD-safer-circuit.html

[alexturner]

Thanks for your input all.

Where I can envision someone touching the blocks input to the SMPS output....

I wish it were so simple, at least I've have my own stupidity to blame however the LN antenna was on the other side of the room debugging EMI from a GaN SEPIC converter which has proven very effective at generating airband RF! I digress...

Could it be that the N connector on the DC block was out of tolerance and bent the center contact fingers on the spectrum analyzer? Have you inspected them closely?

I'm but a budget operator and I thought my $60 DC should prove sufficient, there's every chance it was charged up in shipment and it's killed the unit. Moving forward, it's become a mission now-

I've begun validating the RF frontend, which has ruled out front panel connector issues and attenuator issues; keep in mind, this unit has a diplexer separating two functional bands, 0-3.2Ghz and >3.2Ghz in which the lower frequency is fed through an LP filter and the higher band through a YIG filter and on to the fundamental mixer. I only have a 1 GHz signal source, so I'm limited to the lower frequencies for now, but I'm hoping I can shed some light on something either way. Attaching my LN antenna to the input and tuning to the 5Ghz WiFi band I see no difference between the antenna being disconnected or connected, so I'm assuming the issue lies beyond the input section.

Everything in the signal path seems OK up until the lower band frequency conversion unit, where I'm able to confirm a stable 400MHz input from the TXCO; I'm not seeing any expected output onward to the IF change-over switch or fundamental mixer.

Measuring the 1st LO output on the back of the unit yields a very weak sinusoidal pattern with beats at around 100 MHz. However, the only measurement tool I have is a 300Mhz DSO (which I've checked with a 50ohm BNC pigtail and a probe with a BNC tip adaptor); I'm not sure if I should at least see some voltage or spurious emission on the scope here (esp at +10dBm).

I've attached some block diagrams of the measurement and conversion sections, would love any input on what a possible culprit could be (I know, it could be anything) but just hoping for a nudge in a direction in particular at present.

I'm fairly amateur in this realm and my experience and expertise is limited, especially when it comes to RF so please mind my ignorance which I promise accompanies a willingness to learn.

[joeqsmith]

, prior to that, had installed an N connector DC block as a just in case.

It makes me question what was the just in case you were concerned with?   The only time I would consider adding a block (or any other component)  is if the measurement required it. 

[alexturner]

Some measurements around the mixer, may be helpful, however, I'm not sure of the accuracy of the measurements given I can only capture them on my scope. There is nothing particularly promising here, with the exception that I was expecting to see the 422MHz at TP6 however, I query whether the level is too low for the noise floor of my MSO5000

[alexturner]

It makes me question what was the just in case you were concerned with?   The only time I would consider adding a block (or any other component)  is if the measurement required it. 

Honestly, I thought it was just a safe practice to have it there in permanence whilst working on potentially failed equipment.

[mawyatt]

, prior to that, had installed an N connector DC block as a just in case.

It makes me question what was the just in case you were concerned with?   The only time I would consider adding a block (or any other component)  is if the measurement required it.

We always leave a Mini-Circuits 10dB type N Pad connected to the SA input of our bench SA/VNA. This forces some thinking to remove when making measurements, hopefully identifying any potential hazardous condition before directly connecting to the input. Also immediately reattach 10dB Pad after any measurement, been forcing ourselves to religiously do this over decades so it's almost become muscle memory.

Don't ask why we starting doing this 

Best, 

[mr ed]

1) I wont connect any SA to any antenna longer than a rubber duck directly. Long wires pick up static just from air movement. DC blocks are only 50 volts or so, designed to strip out 12 volts etc bias.

2) I do spend money on good dc blocks and connectors to expensive equipment. And then i am real careful. The dc  block for me (hobbyist) is insurance only. For instance, i will meter check a signal gen output, and then still use a dc block. Never fried anything expensive yet.

3) my cheapo vna has 5db pads to try to provide some  protection. Again, never on long outside antennas.

[coppercone2]

I think its safe to hook up an antenna to a SA so long you don't have the DC block on it because it should be discharged by 50 ohms to ground. static should not build up that quickly unless you are in a tornado

 i mean unless there is some kinda activity around. but i can't see anyone getting anything done without hooking an antenna up to a SA. I guess you should short out the cable before connecting it though. and i would be worried about the rubber duck antenna doing something strange because it might have cheap rubber on it that is not ESD safe.

[jonpaul]

OP:

"I'd powered it up yesterday to perform some measurements on a signal source"

Sorry seems you blew the SA input/mixer.

Suggest to go to the RF group on that model or look in a W&G group or groups .io.


The input attn and mixer of any SA are the weak points for overload, DC or transients.

Any measurement of a "signal source" (what was it?) are dangerous.

Might the source have: 

Turn on transients?
Defects , setup wrong.  output is way higher than set?
Checked output of sig source on a scope?


We use a Mini Circuits 20 db inline attenuator in such cases.

Transient protectors for SA exist but need great connectors   and very low capacitance.

Finally all millimeter wave and microwave connectors are precision hardware, beware of any used/low quality/Chinese  mating connectors.

Bon Chance


Jon