Phase Matrix EIP 28B 26.5GHz Microwave Frequency Counter mini teardown

[DaJMasta]

Got my hands on an interesting counter for a pretty cheap price, so I took some pics in my initial cleanup/testing in case someone else was interested.



The EIP 28B is a CW microwave counter in a fairly small (but very long and heavy) form factor that's good to 26.5GHz, the 25B model is the same except it tops out at 20GHz.  It had three input bands, one low frequency, one mid (to 1GHz) and one up to the stratosphere (relatively speaking) which also supports power measurement, and while there are some AM/FM measurement and other measurement functions, it's basically what you expect for basic counter functionality.  It's simple enough that the sticker on the top runs through all the measurement functions and auxiliary testing functions (but be careful when cleaning, the ink on the sticker comes off with IPA.)  My unit will measure down below advertised spec of levels (I could get to -33dBm or so at 20GHz, but didn't have a source with a low enough variable output to really check (though my cruddy cables patched together managed to get me an extra -14dB or so!), but while the maximum operating level is only rated to +10dBm, the damage level is +45dBm, so the input seems pretty durable and about the same sensitivity of most microwave counters.



It defaults to a one second gate time and will give you 11 digits of measurement at 10GHz+, showing 1Hz as its least significant digit, though I haven't tested whether that LSD is mostly noise or what.  If you turn the power meter on you lose 5 LSD on the display, but it may still count it for GPIB users and such.  Worth mentioning that the single fan on the back starts at normal audible test equipment level and then steadily ramps up for takeoff.... I will be replacing it, though it seems from the internal thermal design (or lack thereof) it's only for the linear power supply.



Inside, the instrument is built to be durable... it weighs a bit over 20lbs on its own (a little over 9kg) and almost all of it seems to be the seemingly die cast chassis and the large transformer.  Construction looks late 80s/early 90s but my unit is from 2001 (a la SRS), and the YIG filter on the input of band 3 is flying the jolly roger... though I don't remember seeing the same marking on other YIG devices, I wonder if it's for that or if there's beryllium grease in it or something.  The cards are all through hole but the tabs are helpfully labeled, I haven't looked into it with any depth, but I don't think schematics are available.




Haven't spent much time with it yet, but it seems to be a very nice straightforward counter.  They can be had a little cheaper than some of the other standard options and there are full-width variants with pulse counting and other features too, so they could be worth considering if you're looking for a microwave counter.

Mine came missing a foot and the feet are attached by self tapping screws through the bottom plate, so you have to pull it off to mess with the feet.  I'll be designing and 3d printing some replacements soon, but there's a bit of a line for my printer's time at the moment, will share when available.

Edit:
Replacement feet: https://www.thingiverse.com/thing:3817945
Image of the back panel:


The.... operations manual:

[0culus]

+45 dBm damage level is actually really good. I have a 5350B microwave counter and it's rated damage level is only +25 dBm. Neat instrument, thanks for sharing. 

[edpalmer42]

I have an older EIP 545 counter that goes to 18 GHz.  I've always thought that the best feature is the ability to focus in on a particular frequency range and ignore other signals.  i.e.  "Show me the frequency/power of the highest power signal between 2.6 and 3.7 GHz".  You can narrow the range down to a span of 0.1 GHz and a resolution of 0.01 GHz.  This works from 1 GHz to 18 GHz.  Looks like your counter has the same feature.

Ed

[PA0PBZ]

Seeing it has 1Hz resolution at 20GHz makes me wonder what the reference accuracy is.

[DaJMasta]

I was lazy and didn't run the test since I didn't have space on my bench near my reference or other counter.... but I set it up under the bench to check the accuracy since I was curious too.

Vs. a Fluke PM6690 (CNT-90 rebadge) locked to the same external reference, one least significant digit variation, but usually matched at 2GHz (the 6690 can't do double-digit GHz to see the finest detail, but it's otherwise an accurate counter).

Running from the internal reference after some warm up time (though it was the same initially), I see an error of 1.5x10^-6 (about 3kHz high at 2GHz), with the normal reference and no recent calibration.  May tweak it, but when I get a spot on it for my bench, it will probably be on the lab reference anyways.  I think the base spec for the standard reference is something like 10^-7 ballpark after a year.

[DaJMasta]

Edited the first post to show the back face of the thing and the sticker on top that constitutes the field manual.

Also designed and some printed some replacement feet, they work well and are likely compatible with all EIP counters from the pictures of the full-width counters I've seen.
https://www.thingiverse.com/thing:3817945