Agilent InfiniiMax 7GHz Active differential probe teardown

[Berni]

If any of you have been wondering whats inside a Agilent/Keysight 1134A 7GHz active differential probe then wonder no more.



EDIT: Corrected a silly typo in the title

[Berni]

Closeup of the ASIC wizardry

[Berni]

And finally the scope end

[plesa]

wow, thanks for sharing!! I did not noticed that these probes can be dissassembled without damage. It is damaged? Otherwise it is quite risky business ;-)

[Berni]

wow, thanks for sharing!! I did not noticed that these probes can be dissassembled without damage. It is damaged? Otherwise it is quite risky business ;-)

I got the probe with the amplifier end missing its plastic case, no idea why. So i don't know how easily that end can be opened up, but i would assume you have to take off the sticker and pry it apart until some sort of clips let go (Likely will leave marks). The scope end is fairly easily opened up by pushing on clearly visible clips with a screwdriver so it leaves no visible damage if done right.

For the hell of it i did try plugging it in and turns out it works just fine . I only had a peak inside because the shield on the amplifier end simply pops right off and while i was at it i noticed the scope end is not too difficult to get in to. I would have taken the scope end apart further if the probe was indeed dead.

I hope to get a spare plastic case for it to restore it to its original glory, but if not il have to make one using 3d printing or something.

[plesa]

Oo, I get it. you can check the parts on Keysight website for this probe. I suprissed that bond wires were not damaged.

[mos6502]

Jeez, so the PCB production got messed up, they accidentally produced a mirrored version of the layout, and instead of making new boards, they just mount the IC upside down? Some quality there ...

[HighVoltage]

Thanks for a great teardown.
It is good to know, that these probes can be easily taken apart and repaired, if needed.

 

[Theboel]

I am not sure the "up side down style" is a mistake I believe its on purpose for tuning purpose but may be I am totally wrong

[MadTux]

Quite boring, to be honest. One chip all there is. Old RF parts inside my spectrum analyzers were build from discrete componets, connected by wirebonding, repairable at least in theory.
Also good to know that if you kill it with too much voltage (more than 30V), it will be dead. Not much to fix there.

[Berni]

Yeah they can be taken apart but i don't think its possible to really fix anything since if anything blows it would most surely be that asic. I did expect to see inside wire bonded dies, but i was expecting there to be 3 or 4 separate ones rather than a single one.

Its not the only active probe i have (Sorry no teardown on others) and i do know how delicate these things are. Luckily the digital circuits these things are usually used on don't tend to contain voltages that high.

Turns out Keyseight does hold the plastic case for it:

And they also have the label that goes on it so it looks the part too:

[free_electron]

Jeez, so the PCB production got messed up, they accidentally produced a mirrored version of the layout, and instead of making new boards, they just mount the IC upside down? Some quality there ...

Its not a pcb but an ALOX hybrid module, and what you see is naked silicon that is bonded to the substrate.

[Someone]

Agilent had their own GaAs and Si fab for a lot of their microwave ICs, I assume Keysight continued it. You can be sure this isn't some die thats available on the open market.

[Fraser]

Many high performance RF IC's use open bond wire construction techniques. Nothing hobbyist about them  If you open those nice solid looking RF amplifier modules from Avantek or similar you will often be faced with the horror of a silicon wafer on Ceramic substrate and fine gold bond wires. Not something I could repair even if I could lay my hands on the failed chip. The bond wires are in an air dielectric, far better than passing through a plastic or ceramic shell. No need of such in modules that are sealed, and so protected. These chips are just basically a die and its associated bond wires minus the consumer friendly housing and pins/pads.

Thanks for sharing the images. As has been stated, do not buy one of these types of probes untested, they are not really a DIY repair proposition for the hobbyist. Physical damage to the case, repair of the scope end, yes.....repair of the probe head .......  nope.

People sometimes wonder why high performance microwave frequency modules cost so much. It is not just the physical component cost, it is often the equipment and technicians time to hand tune and test the module during production. Some of these modules need quite a lot of skilled 'tweaking' to ensure that they meet the specification. I knew a chap from Marconi who told me that they used Hewlett Packard test gear on their microwave test benches as it was the Best. They often found that their own microwave test gear barely met spec, yet the HP stuff far outperformed basic spec. Quality kit, no doubt about it. Sadly I can't afford such quality in my lab so I make do with Marconi, R&S, and Advantest RF test gear.

Aurora

[AF6LJ]

Nothing new in regard to the open bond wire tech, makes for more predictable preformance since air is the dielectric. Also someone mentioned ease of tuning in regard to the chip.

Where I use to work we bought S-band power transistors from TRW they were hand tuned before the cover was epoxied to the device. I have fond memories of the days when we use to tune 40W and above S-band amplifiers by adding solder to the substrates in the amplifier. Now you just go to Mouser and buy a module off the shelf.

[daqq]

Thanks for the teardown! Especially for the ceramic board with the ASIC - it's not everyday I see stuff like that! Still, I'm always amazed by the prices these things cost.

There seems to be a fair amount of solder flux residue where the cables are soldered to the board. The non-obscenely-high-frequency connections I'm sure are OK, but I would have assumed some cleaning for the main 7GHz connection. EDIT: Also, there's some residue from the resistor trimming process. I'm sure they know what they are doing. Just surprised me.

[joeqsmith]

Closeup of the ASIC wizardry

Very interesting.  Thanks.

[Berni]

Yes at first it looked to me like something blew up on that resistor but after looking at it more closely i can see a lines cut in to it so that shows it was laser trimmed and it left a bit of residue.

I also would have expected to see the flux cleaned away, but i am guessing they probably didn't want to risk disturbing the gold bond wires with the cleaning process.

Another interesting thing is that the pins on the ceramic capacitors look like them have been gold plated or something. Not your average joe $5 per reel ceramic cap from china.

[Rupunzell]

Very typical microwave thin film (alumina substrate) construction that has been used in the microwave industry for decades.

Trimming resistor values by slicing at the resistive element is typical practice, some times chip capacitors are also modified to slightly alter it's value. The gold bond wires are length of a controlled length and they are some times moved around to affect impedance match between connections.

RMA solder is non-corrosive, trying to clean will likely result in spreading the flux all over the tim film circuit and causing serious problems. Thus, leaving the RMA flux on is wiser than it initially appears.

Hand tweaking is very common in microwave hybrids, this is one of the reasons why microwave bits that are any good cost what they do..

Question is, at 7 Ghz, what is the common mode rejection, input match and... like and how does one set up to measure that given all the various factors that will affect the measurement results. At these frequencies, typical probing practice mostly goes out the window making each and every connection requiring controlled impedance to make meaningful measurements (typical microwave systems are normalized to 50 ohms).


Bernice

[Fraser]

Regarding the comment on SMD capacitor gold plated end caps..... I used some in a high frequency RF project and they cost GBP80 each    Some bright spark then tried to reduce the cost by replacing them in the production run. The result was a badly compromised upper frequency response.  The sayings "quality costs" and "you get what you pay for" applies in some RF designs.

The good news is that hobbyists can often get away without such "perfection" in less demanding situations outside Industry. 7GHz is no longer a major challenge in terms of COTs RF building blocks like MMIC's and HEMT's, but the frequency response would need to be mapped and accounted for when taking measurements.

Aurora

[Rupunzell]

Microwave chip capacitors are made by folks like Johanson Technologies. Link to one of their chip capacitor pages:
http://www.johansontechnology.com/microwave-single-layer

These are not like typical SMD ceramic capacitors, They are often single layer with gold plating and made of dielectrics designed to operate at Ghz. This is an old technology as microwave chip capacitors have been around for decades.

The microwave folks have their specific group of parts and ways of making stuff refined over decades of trial and error and painstaking learning how to make it work and they keep these foundational lessons in place for the here and now and for the future. One of the things that makes microwave-RF stuff different from software-firmware-crippleware-digital-computing that is driven for consumer oriented mass market items.


Bernice

Regarding the comment on SMD capacitor gold plated end caps..... I some in a high frequency RF project and they cost GBP80 each    Some bright spark then tried to reduce the cost by replacing them in the production run. The result was a badly compromised upper frequency response.  The sayings "quality costs" and "you get what you pay for" applies in some RF designs.

Aurora

[Someone]

Question is, at 7 Ghz, what is the common mode rejection, input match and... like and how does one set up to measure that given all the various factors that will affect the measurement results. At these frequencies, typical probing practice mostly goes out the window making each and every connection requiring controlled impedance to make meaningful measurements (typical microwave systems are normalized to 50 ohms).

The user guides are available if you're interested in how to use the probes, but the heads attached to them put a pair of damping resistors right at the point of measurement (which is often soldered in) shown in pictures in the user guides. CMRR peaks at 40dB but can be very poor in worst case.

[marshallh]

I dug up this pic I have of (I believe another Infiniimax) from some promo material.

[Berni]

Oh that one looks quite fancy alright. I like the embedded coax running off to the output.

The circuit looks quite interesting too. The input part of it looks like its AC coupled but then it has some sort of amplifier in there for the DC component or something?