I measured Zin and transconductance several ways; just posted the measurements. Zin is probably around 80 ohms. Transconductance is around 900 ohms. That's enough data to make me look for another part.
To bias it properly, for starters. And to evaluate nonlinearities.
They characterize them with a VNA using either time gating or an equivalent way to de-embed the test fixture. Should be the same as a TDR plot at the end of the day, but it's easy to misidentify the package entry point. Might be worth double-checking with a different fixture.
If your S11 measurements differ hugely from theirs, it would be interesting to drill down and find out why. Their customers live and die by the S- parameter sword, so they have to as well.
At DC it would pretty much have to. At RF (which is all they care about) the phase of S21 will vary with the stimulus frequency.
They don't use SPICE models because they assume linear operation is always going to be the goal (otherwise you probably would be using a discrete transistor rather than a class-A amplifier MMIC, right?) S parameters are sufficient for that regime. They aren't on the data sheet but if you back up a level on the MCL site you'll see links to both graphical and tabular copies of all four S parameters for pretty much every two-port device they sell.
For a while, HPAK was pushing something called "X parameters" that was supposed to be like S parameter measurements under nonlinear conditions, but I don't know if it ever caught on. The S parameters don't account for internal heating during pulsed operation, for instance.
-- john, KE5FX
Reminds me of "A Meeting with Medusa" by Arthur Clarke, where a remotely-piloted camera platform got its control and sensing feed routed via satellite by mistake. Not a pretty result, but a good story.
The real unit will use a 1-inch flex made by Molex. It's super slick--0.33 mm pitch, but the conductors fan out into three rows of 1-mm pads. That gets you high density and low thermal load, together with the misalignment tolerance of normal 1-mm ones. Brilliant.
The connectors are 5025982393. Check 'em out.
They make some of their products out of them, so I expect they will. They still make the SRA-1 mixer, which I used to use in 1981.
Cheers
Phil Hobbs
Sure, but they didn't know that!
MMICS are blindingly fast, stable, and cheap. It's hard to get multi-GHz gain from discretes.
We like MMICs as cheap, fast pulse amplifiers. I often pump a bit of DC current into the input pin, to offset the output bias point, when I want to make big positive or negative pulses. RF and telecom assume a symmetric waveform.
MMICs are terrible for DC, so I split the signal into separate AC and DC gain paths, and combine at the output.
I guess there is so little demand for pulse use that the MMIC people have no big incentive to generate Spice models.
We're working with distributed amps too, for pulses. They are similarly assumed to work with symmetric telecom waveforms. No Spice models!
Shouldn't be necessary. You ought to only have to provide a suitable power supply and then treat it like a black box with Zin and Zout both
50 ohm purely resistive. The whole idea of these things is so you can daisy-chain them together in a modular approach. The only thing you should have to worry about is preserving the characteristic impedance when 'piping' the signal between black boxes. And yet you're saying that's not what you're finding? Could it be the manufacturer you're sourcing these parts from is not exactly a top rank supplier? Get some other mmics from a decent, well-respected manufacturer and test those. I'll bet it's poor manufacturing/quality control that's the issue here.
That's one way to look at things. RF is seldom quantitatively precise.
MiniCircuits is generally considered to be a pretty good RF supplier.
I've measured a bunch of MMICS and only found one that was reasonably close to 50 ohms Zin, an obsolete Sirenza SiGe part. It could be tuned to exactly 50 by tweaking the biases. Most are lower, generally in the high 30s. That's what RF people call a good match. I was surprised that the GVA-63 is closer to 80 ohms in.
Except that they aren't that linear and they aren't frequency-independent. MMICs have much softer limiting than op amps, for instance. They also tend to slow down when you hit them hard.
Grandmothers and eggs come irresistibly to mind at this point. ;)
Based on what?
Cheers
Phil Hobbs
Grandmother have produced children who have produced grandchildren.
Cursitor Doom doesn't show any sign of every having produced anything - even the silly ideas he propagates are copied from the right-wing propaganda he is gullible enough to take seriously.
One more of Cursitor Doom's delusions of competence.
I used to work at MCL 20+ y.a. They used to do "specials" all the time for customers, like selecting parts per spec. Did you try talking to a Applications Engineer? (or maybe they fired them all since then... :( Cheers, Rich S.
They still listen. I think I'm single-handedly responsible for MCL selling barefoot pHEMTs. (JL may have done a bit of begging too, I forget.)
Cheers
Phil Hobbs
If you inspired the SAV series, you are an outright hero.
But on the other hand, hardly top-drawer, either.
I'd like to propose a test. Get together a range of MC modules from across a large frequency spectrum. If their performance is close to perfect at the lower end, then gets progressively worse as you go up, then perform the same tests against a similar range from say SELEX, GE, BA or whatever. See if the problem disappears with a more highly regarded manufacturer. Then you'll know MC are punching above their weight.
Well, whining at the rep till he took pity on me is hardly the stuff action heroes are made of. ;)
Cheers
Phil Hobbs
How did you find out that they even had the parts?
He told me that they were introducing pHEMT amps up to 3 GHz, so I asked him about just packaging the bare transistors, and whined about dirty old Broadcom being a bully. ;)
Cheers
Phil Hobbs
By RF standards, a 75 ohm load in a 50 ohm system is only a VSWR of
1.5, return loss -14 dB, a pretty good match... to some people.My customers expect a rising edge to be a boring Gaussian step without features.
Do it and let us know what you find.
My neighbor just across Arbor Street worked in the Avago phemt fab. I try not to glare too hard at him.
For a radio ham stringing up a dipole PERHAPS. But not for pro-RF. You should be within a whisker at most of 50 ohms zero reactance. The whole idea of those devices is all the matching is done for you- 'plug and play' as it were. Just look after the inter-stage feedline impedance, give 'em a stable power supply and you should be laughing.
Well, you jest, but I am actually in the process of assembling some test gear to do something very similar (although probably not at the sort of frequencies you need).
Not quite sure what that has to do with the subject under discussion here, Bill, but at least I'm *trying* to make a constructive contribution here.
Modular amplifiers are nearly always around 2:1 VSWR. YCLIU. Fancier ones with their own chassis and connectors are often somewhat better. What do you mean by "Pro RF"--broadcast? Wifi? marine radio? UWB radar?
Do you have actual data for a board-level MMIC amplifier with Zin "within a whisker of 50 ohms zero reactance" over a wide band? Could be a useful part, for sure, but I know of none.
Cheers
Phil Hobbs
Have something to add? Share your thoughts — no account required.
Ask the community — no account required