A look inside some Mini-Circuits modules

[D Straney]

Had the opportunity a while back to take the covers off a variety of Mini-Circuits modules in lab stock.  It's pretty standard stuff as far as RF is concerned, so this'll be old hat to most people here, but I'm only tangentially involved in the high-frequency world so I'm still easily impressed by things like special winding patterns on wideband RF chokes.

Many of the older modules contained just a PCB with the metal-can version of the equivalent Mini-Circuits part (which makes perfect sense from keeping design/stocking/production efficient), and one switch module would've been visible only if desoldering the connectors, but there's still some things to see in the rest of them:

ZX60-6013E-S+ wideband amp
Was surprised at the small number of components actually - just a single transistor and/or MMIC, DC filtering then a choke and series resistor to bias the output/collector/drain, and DC-blocking caps at the input and output.

Also looks like there's a small stub or capacitive bit on the output for matching (the frequency range does extend up to 6 Ghz).

ZFSC-2-1W-S+ splitter

Seems to use the common "center-tapped transformer with resistor across it + 2:1 auto-transformer for input impedance matching" scheme (described in more detail here: https://www.minicircuits.com/app/AN10-006.pdf, but with a little bit extra in the form of some planar spiral inductors (which must have values in the single nH range or below) in series with that resistor across the outputs, and an extra capacitor to ground.  If I remember correctly, I think that capacitor to ground partly compensates for the magnetizing inductance of one or both transformers, and the series inductors with the 100Ω resistor compensate for the output transformer's leakage inductance - but can't find a reference on that right now.

U2C-1SP2T-63VH SPDT RF switch


Looks like what's an SPDT switch at a high level actually uses 6 SPDT switch ICs inside.  The switch directly on the COM port seems to have been added so that it can be disconnected from both NO & NC and terminated to 50Ω.  One of the extra pairs of switches on the NO & NC ports is there to terminate the port to 50Ω when unused (you can see the termination resistors, larger on the ones closer to the port and smaller on the switches one level back from the port), and the second pair I think is to add extra isolation when unused.

Kind of funny in a digitally-controlled model like this (it takes USB, I2C, and SPI inputs) where the microcontroller and power supply just for interpreting inputs has 100x more transistors than are in the RF "business end" of things.

LZY-22+ RF PA

You can see the signal path clearly here, where from the left it goes through a crossed diode pair to protect against input over-voltages, and then 3 gain stages: first a small SOT-89 class-A stage, then a larger heatsink-mount class-A stage, and finally what looks like a class-AB dual-transistor stage to provide the full output power.  The 100Ω resistors next to the output transistors seem to be involved in providing some local feedback for the output stage.  There's a bunch of stuff around for biasing that I haven't mapped out, regulators etc.  The big TO-247 transistor screwed to the case seems to be a voltage regulator for the incoming power, and the "Airpak" device turns out to be a thermal switch for over-temperature protection.

ZFSWHA-1-20+ switch
Finally, an amusing example of the "standard Mini-Circuits PCB-mount part within a module":

I like the hack-ish shield hanging over the board, hand-soldered at the connectors and wire-bonded to the IC package.  Looks like someone realized too late that there wasn't enough isolation in the design... (or the enclosure created some kind of resonant mode: would think a square of damping material would be a simpler solution there, but who knows about the details)

[RoV]

Nice photos, thanks!
I add one that I found interesting: a ZFL-1000VH2, a so-called medium power amplifier (25 dBm out) 10-1000 MHz. It surely required a lot of hand work to be assembled!

[D Straney]

Also, that sure is a lot of glue...

[stephen_g]

Wow, nice! I use a lot of their surface mount parts in our designs at work, it figured that their modules would mostly look like this and use their own parts but I'd never had a chance to look inside any!

Did you take a picture of the other side of the drop-in wall on the U2C-1SP2T-63VH? It looks from the PCB like it had a form-in-place RF gasket?

[D Straney]

I didn't, I don't think that module had a back cover (looked like it was milled from a solid block), or if it did I didn't notice.

Here's a bonus photo from one that I actually have in person, a ZX73-2500-S+ voltage-variable attenuator that I kept and opened it after it got fried with some over-voltage on the power supply pin.  Inside is basically just a filter on the power input, and a Mini-Circuits RVA-2500 SMT module that I removed the cover from:

This one's super simple, as far as I can tell it relies on using the small-signal equivalent resistance of a diode, which changes depending on the forward bias, as a variable resistor.  Same principle as the diode ladder filters in old musical synthesizers.

You can pretty much read the schematic right off the board, and the RVA-2500's datasheet (https://www.minicircuits.com/pdfs/RVA-2500.pdf) shows the arrangements of the diodes minus the extra passives.  Only differences are:

• Control voltage goes to a shunt cap, and then through 332Ω in series, before going to the diodes.  This resistance together with the 562Ω described below on each diode makes the control act more like constant-current bias than constant-voltage bias for the diodes; putting a "voltage source" across the diodes as in the datasheet's simplified schematic wouldn't make any sense.  Setting the forward current instead keeps it from being wildly sensitive to individual diode characteristics and temperature.
• "Matching" on the input and output looks more like DC blocking than anything else, as it consists of a series cap, and 562Ω to ground
• Power comes in through a single shared series resistor, and then branches off to a series 1.5KΩ resistor and shunt cap for each diode pair, for filtering

It's also possible, since that shared input-power-series resistor looks like a different family than the others, that it's actually a low-value thermistor to do some rough temperature compensation with the bias current and keep the gain a bit more stable over temperature.