ephemt as diode

That could be just the bias conditions.

Cheers

Phil Hobbs

IME they're much stabler than SiGe bipolars (which are admittedly quite a bit faster). You can make stable source followers out of pHEMTs, even with a big capacitance on the gate ('big' meaning 40 pF or so). Of course with the Avago ones the voltage gain will be like 0.6 on account of the terrible Early voltage. (The Skyworks SKY65050 is somewhat better that way, but its 1/f corner is about 50 MHz.)

It would be interesting to look at pHEMTs for double-balanced mixers. The old Siliconix Si8901 MOSFET array was amazing. (Ed Oxner was always worth reading, too.)

MCL filters are usually a bit better than that, but their passbands are always wider than you'd expect from their names. The quality of their published data leaves everybody else in the dust, even though their specs don't.

On the other hand, you can get beautiful things like the GALI-74+, which is a 100-mW class InGaP Darlington HBT device with a +38 dBm IP3, for like $2.

If they didn't exist, they'd have to be invented. (I've been a fan for over 30 years now, since the SBL-1 days.)

Cheers

Phil Hobbs

You drink gin and tonic, too.

I draw them as zeners, but I am a confessed barbarian.

That ATF part measures asymmetric capacitance at zero bias, on my AADE meter. I guess I should try it on the Boonton to get better data.

We've collected a lot of data on several phemts, but only as needed, which is sort of disorganized.

The filter I used was a fat metal-tube SMA connectorized thing; I don't know what's inside.

The MLC filters look great. I wish someone could make lower frequency parts, for DDS synthesizers. An LC filter takes more room on a board than the DDS and opamp. There ought to be companion IC filters for the DDS chips.

We used a lot of ERA5's, in pulse amplifier applications, and one day they changed. The new ones were nothing like the old ones.

I've strung together a whole lot of experimental setups with those, including that fancy electronically-scanned laser microscope. You can make a fair BPF by cascading MCL lowpass and highpass filters of the same nominal cutoff.

I feel your pain. (Well, not really, but it worked for Bill, so why not for me?) ;)

That can happen with process changes in ICs too. For instance, I used to really like the MAX900 quad comparator, but the foundry went away and took the part with it. At least MCL's RF customers got something that met their specs (if not your measurements).

Cheers

Phil Hobbs

Yes, like everyone else who has a sentimental attachment to the British Empire. It being Saturday afternoon, I have one next to me right now. With a little green umbrella, because it's Christmas. ;)

Well, at least you're well adjusted. Do remember to paint yourself blue before storming the barricades.

Cheers

Phil Hobbs

Interesting. Measuring their capacitances under normal bias would need a bit of thought, because they're so fast. You'd probably need a network analyzer.

Cheers

Phil Hobbs

Or hang a capacitively (large C) coupled inductor of known value in parallel and use a dipmeter. That's how I did it as a kid when I wouldn't even dream of having access to a network analyzer.

For the folks under age 50: Those are them thar oscillators where a little meter shows a dip as you move the frequency dial, showing that energy is being sucked away at that particular frequency.

I've recently rhapsodized about the Measurements 59 in this very boutique--I'm a fan.

Kind of tough to measure 0.25 pF that way, though (S21 = 0.041 and 72deg @ 500 MHz).

Cheers

Phil Hobbs

Yikes. Talk about undefined behaviour.

Normally correct functioning is at more of a premium than board space!

Cheers

Phil Hobbs

Yes, something like that happen to me. I used a fault function of a chip as a feature. Newer chips didn't have that function anymore.

I took avantage of a latch up issue in a chip if input got a little high. At first this was a problem for us but I then figured out that I could use this to trip a current sense using a sensitive gate thyristor(SCR), which in turn started a series of events to signal a fault to a monitoring circuit and reset the system on the fly.

You see the problem was, we actually needed to know when the input got slightly into this range and it never damaged the ic but just created enough (I) in the (+) supply line to this chip for a useful reference.

Board space was at a premium, fixed later with smaller packages :)

Jamie

Yes true, but many if not most of those units are still in use.

Oh well. Must be the reason why some projects specify GENUINE PARTS :)

Jamie

I've got one of those Boonton 59 megacycle meters.

The Boonton goes to 400MHz. Or megacycles as they call it. My Heathkit HD-1250 only goes to 250MHz, and that only with a full battery and a wimpy Q so it has trouble showing a delta of 0.25pF.

In the old days I also had a home-made "broomstick oscillator", a circuit that oscillated with just about anything connected to it. That was very useful for small capacitance change measurements. However, only after I had enough money to buy a pre-scaler chip for my home-made counter. Those were well north of $10 back then which was hard to stomach for a student. So I sometimes borrowed a counter from on of the older ham radio guys in our club.

But ... WWV to the rescue. Run the oscillator really close to 15MHz so it would beat with WWV which could be heard clear across to Europe. If you kept the windows closed, held your breath and otherwise didn't move much you could note the beat frequency change when connecting an ever so small unknown capacitance. Fractions of a picofarad were no problem. Guys like me with next to nothing in musical talent also needed an audio oscillator to figure out the beat frequency.

Or one can get the UHF head, which goes up to 950 or something. It's more the problem of connecting to the 0.25 pF without destroying the accuracy.

Fun. I remember having to save my allowance for radio parts too.

Cheers

Phil Hobbs

That is a very useful thought! Are you collecting?

Well, in that case, it is pilot error that is fixable by use of a modicum of intelligence (hint to all except JL).

In your diagram it isn't clear if the gate is controlling the diode, or if the gate is acting as the diode. It looks like the gate is always the anode, while drain and source are cathodes. You could get the same behavior from back to back diodes. But I suppose you confirmed that the gate is not conducting.

You got an allowance? I saved my lunch money and went hungry for months at a time.