HP 5334A option 030 redesign

[ekoloski]

Hi All,

I apologize if this is should maybe go in the RF forum, as it seems to overlap.

Here goes. The HP5334A counter was available with an option 030 1.3GHz Channel C input module. It's next to impossible to find a spare board assembly (05334-60005) anywhere, unless it's ripped from a counter that was originally built with it.

At a high level this board does not seem overly complicated to recreate with modern components. I see that LA2YUA has actually attempted something similar for their HP 5335A https://longview.be/html/5335ao030.htm, sadly they only posted preliminary details with the promise to post back if a second 'rev b' attempt was successful. This was back in 2015.

My experience in designing anything over a few hundred MHz is non-existent, so I believe this could be a fun academic exercise and a useful update to my trusty 100MHz counter if successful. The first step is to reach out for peer review and feedback on the feasibility of my initial high level approach. I would not be attempting any fancy ALC or squelch circuitry, just an adjustable PIN diode attenuator for level control.

Essentially this is an RF frontend with a divide by 20 output that feeds back into the MFC block in the counter. Sounds simple, right? Maybe for an experienced RF engineer. Not so much for me, but that's what makes it fun.

I have found some generally available parts that I believe would be suitable. Namely, a MC100EL33DG divide by 4 prescaler, a programmable MC100EP139 prescaler configured for divide-by-5 operation, and a BGA2851 MMIC amplifier (I plan to use two of these at about 23dB gain each). The rough idea would be as follows:


Before going too far down the rabbit hole of the design process I was looking for some feedback on the validity of this approach. Additionally, I am comfortable designing PCBs for lower frequency applications but this would be my first attempt at anything approaching and exceeding 1GHz. From a materials standpoint the board (once designed) would likely be ordered through OshPark. Based on their capabilities a 2 or 4 layer FR4 is possible. Would this be up to the challenge or do I need to consider something more exotic? I am also unsure if I would need to go to a 4 layer board to make things like controlled impedance traces easier and help with isolation, or if a 2 layer would be 'good enough'.

[ekoloski]

Updating with a rough draft of the schematic (and I mean rough, it's ugly), if any better experienced engineers out there wouldn't mind tearing it apart and critiquing, please do! 

[ekoloski]

Not as much interest in this as I'd thought there may be. Perhaps I should just acquire a counter that has the factory option, but where's the fun in that?

Preliminary layout for the board is attached. There is a lot of wasted space, due to the board being the same physical dimensions of the HP one, made with all THT components. I am not sure I'm doing it right either, but this is a learning experience.

After some tweaking of the design I'll order a batch of boards and assemble it, and report back after seeing if it works.

[lowimpedance]

While I cannot offer any suggestions,  I am following with interest. I have a 5335a that could  benefit from a bandwidth  upgrade  !
Look forward to seeing the  results  .

[croma641]

Same for a friend that need this module !

[ekoloski]

Thank you lowimpedance and croma641, it's good to know that I'm not alone in my quest to modernize an older but reliable piece of test equipment.

I have ordered a batch of boards and have the parts on hand. Once they arrive and I've had some time to assemble one and test it out I'll report back with my results, whether it's success, a miserable failure, or somewhere in between 

If this does work out I'll gladly publish the design files and BOM, with the caveat that my go-to distributors are low on stock of the dividers and peak detector. They do have some inventory, but not the huge numbers I'd prefer to see when designing a project. The hardest part to source is the programmable divider (MC100EP139DTG) which allows for a divide-by-5. It has limited availability at Digikey and several are on order at Mouser.

HP used a divide by 20 design, which from a signals perspective makes perfect sense. They used some custom divide-by-10 IC to handle the business, which is apparently not a very common ratio in dividers these days. From what I have read the 5335A uses this same divide by 20 schema. The schematic for the board will be compatible (if it works), however the form factor of the board is different from the 5334A. A new PCB would need to be designed.

Lastly, some eye candy. Attached for comparison is a poor quality photo of the original HP part.

[RichardM]

Following with interest 😀

[SCSKITS]

Did a board with a couple of ERA2SM amplifiers followed by an HMC432 then two HMC434s in series with the output conditioned by an ADCMP600.
I needed a divide by 128 for my 53131A counter. With the HMC parts its top end was over 3Ghz.

Also did a 1GHz prescaler for the TEK DC502 counter using a couple MAR6SM amps followed by a MC12080 and an ADCMP600 to condition the output.
Made the DC502 similar to the DC508.

No AGC on either board, so that was a limitation.

ed

[W6EL]

Hey this is really cool! I have a 5328A, which is different but I often think of adding things like this.

My only input so far is that if you have a component which is not directly in the RF chain, to use a larger size like 0805 or 1206. It'll just make it less likely to blow off the board and disappear should you sneeze while soldering.

I will be watching this thread to see how your design goes!

--E
de W6EL

[Swainster]

Another interested party here - I've got a 5335A and I'd definitely be interested in a working circuit design for channel C. Currently my fastest counter is a venerable HP 5245L nixie job, and it sounds like a tractor unless you oil the fan every hour

[Brumby]

Hey this is really cool! I have a 5328A,

So do I.  (With an annoyingly blank panel in the middle )

which is different but I often think of adding things like this.

So do I.

My only input so far is that if you have a component which is not directly in the RF chain, to use a larger size like 0805 or 1206. It'll just make it less likely to blow off the board and disappear should you sneeze while soldering.

Sounds reasonable - but, like the vast majority of tinkerers, I have no real RF design experience.  While I can usually follow a conversation to a degree, it's mostly voodoo.

I will be watching this thread to see how your design goes!

As will I.

To be honest, though, I think you will get a lot more feedback if you put this into the RF section.

[DimitriP]

I can't find where I saved this from, but it exists...somewhere:


Ok, it's from here :  https://www.eevblog.com/forum/testgear/hp-5335a-timer-counter-anything-i-should-know/?action=dlattach;attach=1212020;image   

[croma641]

No, it's :64


"Ok, it's from here :  https://www.eevblog.com/forum/testgear/hp-5335a-timer-counter-anything-i-should-know/?action=dlattach;attach=1212020;image  "

[babysitter]

interested 5334 owner here. Who exchanged the cell for super cap and installed a STP OCXO yet.

BR
Hendrik

[LazyJack]

No, it's :64


"Ok, it's from here :  https://www.eevblog.com/forum/testgear/hp-5335a-timer-counter-anything-i-should-know/?action=dlattach;attach=1212020;image  "

It is either 20 or 64. Newer units with v1.1 firmware are 64. You need to match the prescaler to the firmware version. There are various varsions, the combinations are described in the latest service manual and quoted in the above forum topic.

[croma641]

Yes, I know. My replay was to  note that the 5334A doesn't have any ROM that accepts :64 prescalers.

[ekoloski]

Quick update for the community on this project. As with any first run of a board, there are some changes and some bodges. All things aside though, this looks promising. I have only tested from 60-990MHz, further testing to determine the upper limit will commence once I've cleared some bench space!

Some quick notes:

• I will need to look into improving the sensitivity. I believe the original HP version worked down to -23dBm without a preamp, I am seeing reliable input down to about -17, so some work needs to be done here. Perhaps another gain stage or an LNA on the frontend
• The pin diode attenuator is aggressive, only the top 1/4 of the range seems to work, with anything lower cutting the signal below detection. I'll adjust this
• The biasing on both clock lines between the two divider stages has been altered, I bodged in a voltage divider to provide about -2V and used that instead
• The biasing on the input to the first divider is now 50/150ohms

Update: After hooking up an SDR to test it, it appears the upper limit of stable readings is 2.1GHz. Not bad at all, considering this counter maxes out at 100MHz and this is a divide-by-20 frontend! I've also got it reliably working down past -23dBm on the input.

[ekoloski]

As promised, attached is a revised schematic reflecting the changes I had bodged into the board. I will also update the PCB and upload the gerbers, in case this can help anyone else out there.

Regarding other counters like the 5328A. The general design could easily be adapted to fit other models. For those with a 64:1 requirement, different prescaler configurations could easily be fit as well. I believe the real magic here is in the signal conditioning section and the peak detection to prevent false readings due to background noise.

I have also passed the point of no return and have drilled some holes in my nice counter. The level control and a type-N connector now happily reside on the front panel of the instrument. Next, I'll need to find a suitable knob and figure out a way to label the panel. Beyond that I may add a connector to bring power out for a LNA/preamp. I'm eyeing up TSS-53LNB+ to fit the purpose. I had briefly considered using a bias-tee arrangement, but the thought of injecting DC into anything being measured if I forget to switch it off when used is unsettling.

I appreciate all of the feedback. To address some of the points raised:

Yes, I think I would have benefited from posting in the RF section as well. This has been a learning experience in the ways of 'magic plumbing' for signals greater than a few hundred MHz.

The components on the board are pretty small. The RF is mostly 402 sized smd, simply due to RF considerations. The remaining are 603, which is small but still workable for hobbyist assembly and repair. This was assembled using nothing more exotic than a solderpaste stencil, some fine tweezers, a hot plate, hot air station, and a cheap microscope to verify that I hadn't misplaced anything.

[Swainster]

What input fuse did you use? I notice that you went with something on the board, rather than something built into the BNC jack - makes sense as the BNC type seem to be unobtanium. My HP 5335A is the older version, so should be compatible with this prescaler design... I shall have to put it on my future projects list, though it might get bumped up if another 5335A enthusiast provides a compatible layout.

[ekoloski]

You're spot on about the fused connectors, I had to go with something on the board out of necessity.

Its probably a faux pas, but I went with a 1206 sized polyfuse, a 1206L012WR by Littlefuse specifically. It did add about 6 ohms of series resistance, and I am considering replacing it with a standard fuse instead. The intent with the polyfuse was to avoid ever having to replace it but now I am questioning my decision.

[LazyJack]

What input fuse did you use? I notice that you went with something on the board, rather than something built into the BNC jack - makes sense as the BNC type seem to be unobtanium.

Look for AMP 3284-3201-10 on eBay, it is available from various sellers. It's an obsolete part, I don't see it anywhere else.

[Swainster]

Yep, that's it - should have guessed that it would be SMC on the inner end. I'd never seen an SMC in the wild until I took apart an HP RF sig gen.

[RichardM]

I would be interested in trying to build this for my 5334a

Were the gerbers uploaded somewhere ? The small sized components may be a challenge for my hobyists skills.

Cheers

[ekoloski]

Hi Richard,

I haven't uploaded the Gerber's yet. My plan was to make a V2 with some upgrades, like an inline LNA. In reality though I've actually been pretty happy with the design as is. Maybe one day, but for now it meets my needs.

I will make the one or two revisions that were bodged in and upload the files by the end of the week. Glad to see this project is helpful to the community!

As for soldering the small parts, a solderpaste stencil and some tweezers made quick work of it. If you aren't comfortable with 0402 though, there are assembly services out there. I was able to do it with a hotplate and patience. To be honest my soldering experience exceeds my design experience for this sort of thing, so I may be biased in encouraging you to go for it.

[RichardM]

Thanks very much. No hurry.

I guess I could upsize some components by editing the Gerber for my needs and stock avaiability ? Is it single sided ?

Cheers

Richard

[ekoloski]

Hi Richard,

Apologies on the delay. I've updated the board to reflect the small changes that were made on the biasing of the differential lines between both divider stages. The revised schematic, gerbers, and BOM are attached.

The board is actually a 4 layer, simply to keep the controlled impedance traces to a manageable size. The components are a single side load. Modifying it for larger footprints is certainly a possibility.

[RichardM]

Thanks very much.

Wow, that ibom is pretty cool. How does that work

[ekoloski]

Glad to help!

If you feel up to attempting some 0402 assembly and performing a bodge, I do have a spare board that I never populated. If you want it, its yours for the cost of shipping. This is the original board, I never built a replacement since it works well enough.

The ibom is a handy little plugin for KiCad, it automatically generates the HTML file, which I use during hand assembly. Its available on [github][/https://github.com/openscopeproject/InteractiveHtmlBom]

[ekoloski]

One quick alteration. I have updated the gerbers once more, the center pin hole for the SMC/SMA connector has been enlarged. It worked fine for a SMC, however if you wanted to use a SMA the pin may not have fit, my experience is that it's a little bit larger. No other alterations have been made.

[RichardM]

Thanks for the offer. I would be interested in the spare PCB board. A new set of 5 boards is rather costly, >$100 Australian. Not sure what postage to Australia is. Maybe let me know via PM ?

I may be biting off more than I can chew with such small components. I still stick with through hole but feel I need to make the move some time and this could be my first test to see what new gear I may need 😀