RF is horrrible

Excellent, put up the results!

--
 Thanks, 
    - Win

Am 01.01.20 um 19:22 schrieb snipped-for-privacy@highlandsniptechnology.com:

The output side matters. These things live from low-impedance resistive feedback, internally.

So what. AC voltages may be Vpp, Vrms, dBv, dBm in 600 Ohms.

You can usually get design kits for the Advanced Design System or for Microwave Office. The drawback is it takes easily $100K for the software.

A free simulator that is still quite OK is QUCSstudio. Actually quite nice. Handles S-parameters, Touchstone files, can import Spice models etc. It connects a lot of the open source / free software together. The user interface is not clumsy, but different. Takes a day to get used to. Seems inspired by ADS.

< > < >

Why can't MCL be available at Mouser's or DK???

BTW, there is a new data sheet of the Interfet IF3602. It confirms what I suspected for a long time. Or Fred Bertoli here, 10 years ago. They choose not to clean up their process but their documentation. At ?65 per pop the IF3602 is now completely uninteresting. :-(

cheers, Gerhard

OK, but it will be pretty messy whiteboard pics. I want to know Zin and transimpedance gain, with my output load. In a darlington MMIC, the output load affects the input impedance.

I want to use a MMIC as a photodiode TIA, and I want a low input impedance to keep the input time constant down. The pd capacitance is

0.7 pF, call it 1 pF with strays, which is a time constant of 50 ps into an ohmic 50 ohm MMIC. So I'd like a part with real 50 ohm Zin, preferentially less. Maybe I can do some peaking after the MMIC.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

We have a lot to say in AoE 3, about the IF3602, but it was all based on actual measurements we've made. Interfet has serious customers, such as in Agilent's 34420A, 7.5-digit nano-voltmeter, see Figure 5.61. A tradeoff for large die size vs. capacitance, etc.

--
 Thanks, 
    - Win

MMIC Zin is usually around 30 ohms at lowish frequency, IIRC.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

I measured GAV-63+ at about 90! And that goes up if I AC short the output.

Sounds fishy. Maybe I should measure it different ways.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Relaxing the spec from 300 pVtyp to 500 pVtyp after maybe

20 years is not really fine practice. I got usually 380 pV from mine, but not from 1 @ 5 mA. It took both halves in parallel and 35 mA total. :-(

The 300 pV were the only selling point. For 500 pV I would better go to 4 of the sot-23 jfets from ON.

And they now show the noise molehill between 1 and 10 KHz that made me search the noise generator in my biasing circuit like crazy. The typ. spec is at 100 Hz, below the molehill. You would expect 1/f and flat, according to the old data sheet. Right?

The important sentence is "Custom Part and Binning Options Available" which means the good ones went to Agilent and us Mouser customers get the negative selection that's left over.

Oh, these 8 pairs of IF3602 are all precisely the same: <

> Pure gambling.

BTW I'm just building a box for your noise test circuit. I decided to remove the low pass cap after the 220K resistor in the bias circuit and rely on the 200mF at the base only. Seems to improve stability. And the wiper of the pot needs a current limiting resistor or it could be eol-ed when rotated fast to the GND side with the 200mF charged.

Cheers, Gerhard

They probably only made one wafer box worth, ever.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

You can do 300 pV with five 20-cent CPH3910s at 60-75 mA total.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Or maybe he needs to refine his measurement techniques.

He'd be the last person to realise that he might not be reading the data sheets or doing the measurements entirely correctly.

--
Bill Sloman, Sydney

Given the choice between believing a self-confessed RF neophyte's and MiniCircuits, my money is on Minicircuits.

The Minicircuits datasheets are excessively brief; charitably, they presume you know about using RF components.

The VSWR of your average MMIC amp is almost always around 2:1.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

I'm measuring the MMIC in time domain, with pulses, which preserve the sign of all the voltages and currents.

And the application is time domain, not RF. The RF guys sort of slosh buckets of input into an amp and expect some different buckets to come out.

I once asked MiniCircuits if one of their darlington MMICS inverted the signal or not. They didn't know.

I asked about Spice models too. They don't have any.

I guess an RF expert wouldn't ask questions like that.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Apparently in a random direction, more often below 50 ohms on the input side. Numbers ballpark 40 ohms or less seem to be common. For a TIA, I want low.

But the little darlington mmics are super fast, wideband, quiet, stable, and cheap.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Here are some rough measurements on GVA-63+. Enough to suggest that I want to use something else.

The older darlingtons were just that. Most newer ones have some sort of internal bias circuit that allows a constant-voltage power supply to be connected through just an inductor. In my wideband application, I really prefer the older ones that need an external current-set resistor.

Zin really looks high. I was eyeballing fuzzy scope traces, so the measurements are a bit scattered.

I'll get a few "classic" MMICS and measure them.

6 GHz, voltage gain 22, is 132 GHz gain-bandwidth, for 97 cents at q100.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Yup. They mismatch it like that to improve the noise and maybe the flatness.

BTW I'm using a SAV-551+ pHEMT (12 GHz f_max) to bootstrap a 400-pF MPPC array over a 100-mm flat flex cable. Seems OK with a 10-ohm BLM15BB bead in the gate.

It's really tough to make those things oscillate!

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Am 02.01.20 um 17:53 schrieb snipped-for-privacy@highlandsniptechnology.com:

The phase angle of S21 tells you how many degrees it "inverts".

cheers, Gerhard

That sounds very wrong. Are you sure you're testing methodology is sound?

Why would anyone need one??

Go nuts. Let us know how you get on.

--

"The BEST Deal is NO DEAL"

Bootstrapping a photodiode that's at the end of a transmission line gets interesting.

SAV551 and 541 are fabulous parts. I hope MiniCircuits keeps selling them.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com