TSP

In this episode Shahriar takes a close look at a defective Agilent E4440A spectrum analyzer. Initial analysis reveals that the startup errors are caused by a defective pre-amplifier module. During initial alignment and calibration, the defective microwave relays inside the pre-amplifier module fail insertion loss specifications and cause alignment failures.
Unfortunately, the relays are not repairable and must be replaced. Two new relays are purchased and replacement clears all alignment issues. The main motherboard battery has also failed; a replacement battery corrects any time and date keeping errors. The repaired instrument is cycled through several alignment procedures without any problems. The performance of the unit is verified by examining the DANL of the instrument with and without the pre-amplifier activated.
The Signal Path
www.TheSignalPath.com
/ thesignalpath
www.Patreon.com/TheSignalPath

[@WillArtie]

Lol. "It's not very high freq. Only 3GHz..". Tell that to my hacked 1054z! Been mainly on audio this year, so 50k seems high to me! Thanks for the vid Shahriar.

[@rfsniffer3175]

The repair videos are the most valuable ones especially when you couple them with the theories involved.

[@joes5669]

Ahhh the good old E4440. We still have several in our lab. There's something to be said for the OS in these devices compared to the new Windows boxes. Loved the switch tear down. Very valuable. Thank you.

[@douro20]

They run HP-UX, don't they?

[@TheRadioShop]

Very nice detail in construction of that relay. Thanks for sharing.

[@Factory400]

Surprised the relays are only $250.....that is a lot of work and takes a LOT of specialized skills, equipment, and fixtures to pull off. Excellent video....

[@Dyaxxis]

Great video, especially the detailed teardown of those relays. 17:31 Hopefully that tool wasn't too expensive. I tend to use the "cheap" hobby style metal picks and/or probers on destroyed and/or non-sensitive devices so my more expensive tools may be spared. The Homer Simpson "Doh!" reference was a bit humourous though!

[@sivalley]

Wow, for a machine from 1974 it's in VERY good condition! ;)

[@darkoharamina7229]

Hi, very good video, speed down video.... your usual speed is perfect.

[@askild_eide]

Thanx for continuing to deliver quality videos! Of all the "technical yuotube vids" I watch, yours is a separate league when it comes to WAF (Wife Acceptance Factor ((she's an art historian))), so I usually can watch yours whenever i discover a new one :) I think I have watched most of your videos now, but only started supporting you on patreon a few months ago... But I have a request, it might be difficult to make a long video, and you have already touched the subject, but if we look at instruments today and back to, lets say 80's or even 70's, the change have for spectrum analyzers have been from all analog to the digital part have moved forward in #the signal path#, and switches has gone from mechanical to solid state, and amplifiers has gone from discrete to IC or MMIC, but one ( or two) major parts has not changed.... The YIG oscillator and YIG tuned filters! I think the inventors and the technology really deserves more attention, not many inventions lasts so long with no other technology even being close.. (as far as my knowledge reach:) Would you consider this as a tribute to those that made all that we love possible?

[@rfsniffer3175]

A noise figure analyzer tear down would be great!

[@movax20h]

Nice video. Could you make a dedicated video about just RF switching, relays, solid state switches, and maybe directional copulers.

[@ruhnet]

Great stuff! Thanks for sharing!

[@roymercer9065]

That was interesting and very well done. Thank you sir.

[@pu1ypj]

Please! Simulate that relay in a software. I'd love to see how this is done.

[@maxsnts]

Hi. Nice video. Out of curiosity, any good 3d Simulator on the free/open side of software?

[@garybevis8691]

HI Shahriar, I had an idea about a way to remove the back cover on the relay, rather than using a ball end tool for Dremel. I think you could use a vertical mill (Bridgeport type) with end cutting tool. If you position the relay orient back cover up in a vice, perpendicular of the tool, carefully mill down the cover to the surface of the mono-block housing. Effectively removing the cover and epoxy. Subsequent, I think you could do the work on the switch contacts and glue a cover fashioned from same type material, such as aluminum. Good luck and best wishes for further project such as things like this.

[@AF6LJSue]

Good stuff, thanks for posting this.

[@cat637d]

Very interesting and FUN!

[@WaltonPete]

Interesting to see inside the relay. You can see why they were so expensive, what with the machined housing and other precision parts. I wonder if they could be manufactured additively (using laser sintering or some other 3D process) to make them cheaper? It's the sort of thing 3D printing excels at, making small numbers of highly specialised components.

[@movax20h]

The machining isn't that expensive, it also produces nice smooth edges and surfaces. 3d printing isn't necassiarly as good (the cavity where switches are might be tolerance critical), and it might be significantly slower if producing them one by one. It is also possible they do have some precasting first to get close to shape, and then just machine cavities. I think it is the case because, the external surfaces, and the coil cavity are not that smooth, and because they used some weird alloy of aluminium, magnezium or zinc, as seen from the cristaline structure of chips. But you could probably manufacture additively in bigger machine and manufacture lets say 100 or 150 at the same time. The problem with sinteing is that you also need to control dimensional changes during sintering process very precisely. The main cost is really in the fact that these are relatively specialized and low volume devices, and main cost is actually in design and testing, not manufacturing.

[@philippoMC]

Could you make a video on simulating RF cavities? That would be very interesting.

[@electronic7979]

Useful video

[@DeeegerD]

$250 each! I'd have told Agilent (or whatever they call themselves this week) that they were out of their minds! Glad it worked. $500 is my budget for an analyzer ;) (maybe used?)

[@MarkFunderburk]

although I don't think it was too bad considering one of those units used cost as much as my car did new.

[@morto360]

Just came to leave a like, ill watch the video later with a cold beer!

[@movax20h]

18:25 minor mistake. The pins at the right are not powering coil, these are outputs.

[@timun4493]

those slots on the controller board look like GSC, i couldn't find a release date for the instrument but from the way the board looks i would expect a PA-RISC 7100LC under the heatsink

[@timun4493]

so probably an hp-ux variant rather than linux

[@liviococcia]

I like watching all these great teardown and repair videos, even though I have a very very basic rudimentary knowledge of electronics, so thank you for creating them. I am always trying to learn, and pick up basic principles of how electronic components work, and how a peice of equipment functions. I have a question about the battery testing which I have no understanding of though please, when you tested the bios battery cell you placed the positive and negative lead onto the negative side of the battery, why is this? (I always assumed that the DVM positive lead should go on the positive side, and the negative lead on the negative side to give the cells Volts DC value), thanks again

[@atta30]

If you reverse Polarity on a dvm you just get a negative voltage you could measure 3 volts or -3 volts by swapping the polarity but either way it tells you 3 volts

[@ebb2421]

interesting experiment to implement some solid state rf switches in place of the relays...

[@michaelnobibux2886]

Relays that switch low level ac/rf can fail like this.A solution that is sometimes used is called 'dc wetting' ,whereby the signal is superimposed on a few milliamps of dc.

[@filipamator]

We had in our lab one of those PSA analysers with upper freq. limit at 6.7 GHz. One day I discovered that we can go up to 26 GHz.. the battery was dead and it seem that the instrument forgot about its configuration!

[@glasslinger]

I would like to see that analyzer looking at a signal at the low end of the range, say 20 hz or even 10 hz. This seems impossible to me! Magic!

[@btouw8558]

Always Interesting, bit short

[@galileo_rs]

Common trick in HAM radio gear with intermittent relay issues like this was to inject some DC.

[@kissingfrogs]

and a similar trick used to bring old telephone lines back to life

[@galileo_rs]

Didn't know that, nice :)

[@sefarkas0]

Since the other two are bad, would it be possible to try drilling a tint hole through the back and injecting cleaner through it, it may take a long time to dry or maybe two holes.

[@michaelnobibux2886]

Steve Farkas This might not be a good idea,as some debris may fall inside. Also rf relays are sometimes sealed and pressurized with dry nitrogen!!!

[@mattnottingham3663]

Are you going to take a look at the Siglent SVA1015X in the near future? Its an interesting price point.

[@movax20h]

It is pretty expensive with all options in. I still think it is overpriced for what it is, but the truth is, it is priced better than competition. So hopefully prices will lower in few years, market would grow too, which would allow price to lower further.

[@mattnottingham3663]

With its current frequency range it's of no use to me. I just hope its the start into the (low-end) VNA market which they build on.

[@douro20]

@@mattnottingham3663 Still can't think of a better VNA for the price than what Deepace makes. They also make very good training kits for microwave network analysis.

[@mattnottingham3663]

@@douro20 The Deepace one isn't made anymore. It'll be interesting to see what Harmon Instruments come up with.

[@douro20]

@@mattnottingham3663 www.deepace.net/shop/kc901v/

[@Henning_Rech]

This preamp module looks a bit overengineered for its 3 GHz bandwidth. Maybe part of it - around the relays - is identical to the full span preamp.

[@vidasvv]

Thanks for another GREAT video ! 73 N8AUM

[@armandoleantesanabria8355]

Hello nice job I have just question, i have advantest r3465 which didnt retain date or config after boot, so i guess battery is dead, however is any risk on erase the calibration or it just storage config? Thanks for your content

[@WisdomVendor1]

The vid somehow seems incomplete without a Pooch appearance :)

[@valakatz3967]

Great video thank you for your share info. Need to change the battery in my E4406A, any advised would be appreciated, thanks

[@twoody2148]

now that you know how its constructed and its spectrum profile would it be possible to use a solid state relay as a replacement or even a quality transistor ?

[@kirkpennock2997]

Drill a tiny hole in switch cover, squirt some deoxit in, cycle a lot?

[@kirkpennock2997]

But will the cleaner affect the freq response? Beyond me. It would be an interesting thing to test.

[@PlasmaHH]

In my experience the failure mode is more often the deformation and thus wrong contact pressure rather than oxidizations, especially since those things are generally gold coated and sealed very well

[@JlerchTampa]

Drill a hole, use the new hole to pressurize the cavity to remove the lid, fill the hole, clean the things, re-assemble?

[@jfbeam]

Except that's not why it failed. The springs wear out, and the contact area deforms.

[@movax20h]

The problem is small metal particles will get into switch, and possibly short contacts. :/ Also I do not think it is oxidation. All touching parts are gold coated, and it appears to be hermetically sealed. Would be better if filled with nitrogen, but even without that, there is not much oxidation that can there. Most likely the metalic strips deformed, making it not make enough pressure to maintain good contact.

[@movax20h]

How long it took manufacture to make them for you from the time of first contact?

[@douro20]

What was the lead time on those RF switches?

[@gacherumburu9958]

👍👍

[@OneBiOzZ]

"why spend so much on a single relay and LNA" several people have been saying ... its all in the calibration

[@dtiydr]

3Hz.. how do they even do it.

[@movax20h]

"Fairly low frequency, it is just 3GHz"

[@LevonAvagyan]

And i thought audiophile stuff is overpriced! 250$ for one relay!

[@allesklarklaus147]

The same relay for audiophile applications probably cost you 3 grand

[@dtiydr]

250$ for a relay with as little loss as possible and custom made to spec.. I would have expected way more.

[@allesklarklaus147]

dtiydr Yeah but it's old and probably not used anymore. However, with prices of 2000$ for a shielded "audio grade" powercable... The relay is way too cheap.

[@dtiydr]

Perhaps, but just the fact that they could still be made by request say me they are not necessary discontinued. Yea I have seen those cables had to be pure gold in those, way overpriced so say the least.

[@allesklarklaus147]

dtiydr Yeah it's probably just recycled chinese copper but it's shielded so awesomely! Because, powercable

[@JLSoftware]

Those mechanical relays go at 3 gHz? Somebody clarify this.

[@Thesignalpath]

Yes, those are 3GHz relays.

[@theondono]

Almost 170$/GHz 🤣

[@paulpillau5858]

That does not mean that they are switching at 3 GHz though. If that is the question...

[@risfutile]

JL Software the relay carries signals up to 3GHz. it's not switching the contacts at 3GHz.

[@ciano5475]

Correct me if I'm wrong, but what I understand is that the amplifier has 3GHz of bandwidth, but in bypass, the relay has the full signal and the full bandwidth of the instrument (26.5GHz).

[@SLOWOLITUS]

Reed contact would we better

[@michaelnobibux2886]

SLOWOLITUS Reed contacts have high RFskin resistance!!!