Digital RF Capacitors

If you're using porcelain or sapphire caps, you probably have not yet met with the cost accountants and bean counters. Reading between the lines, this sounds like a commercialized military contrivance.

Hopefully, tx and rx are both on the same frequency. If not, doing a binary search through the capacitor banks for best noise matching (hopefully with no stray signals at the antenna) can take forever. The usual method is to tune for best VSWR on transmit, and take what you can get on receive. That works fine for most HF systems until you run into a really high Q antenna, such as a magnetic loop with a loaded Q of about 100. Unless you do something disgusting, like a resistive attenuator in front of the receiver, you can never get the tx output impedance and rx input impedance exactly the same. Therefore, the magnetic loop will tune at slight different tx/rx frequencies, creating a mismatch for one of them.

The easy way out of this is to transmit on the receive frequency, autotune for best match, save the relay settings, and then do the same for the tx frequency. When switching between tx and rx, just use the saved settings. If something changes in tx, you can detect a change in VSWR. If something changes in rx, you're screwed.

Incidentally, nobody does a binary search over the entire frequency range every time you need a match. If it did that, you would soon exceed the cycle count on the relays. The usual method is to do it once, and record the relay settings every few hundred kHz. The antenna tuner has a built in frequency counter that measures the frequency. When it needs something between two recorded frequencies, it interpolated the relay settings and starts a narrow sweep over a much smaller frequency range. However, doing that might be a problem with a very narrow pulse with. You'll need a low level CW signal to run the frequency counter and to keep voltages low enough not to blow any capacitors.

I think that's what I used to say just before someone in a suit arrived to add his "special" requirements and make my life "difficult".

I can't tell from here what you're doing, but the entire purpose of the automagic antenna tuner is to retune when something changes to the antenna or the radio changes frequency. You can detect these changes with a directional coupler and frequency meter. If you do it too often, it will spend all its time tuning, and leave little time to actually operate the device. The trick is not a high speed calibration or tuning, but rather a high speed method of determining whether a calibration or tuning is needed. The relay setting for a change in frequency will be reproducible. The settings for a change in antenna position or configuration might as well be random.

Give them the tools and the knowledge and they're sure to break it.

That's the easy part. Just wait until they change the specs just before you think you're done.

Photofets are better, for low voltage.

RTDs and thermistors both suffer from self-heating.

Fairchild H11F1M. Not all that cheap, but still available.

What, no multiplying D/A converters?

They do look good - no CE mark but we've got other pieces and a system to certify so that's not a show stopper. The DPDT arrangement would allow switching two signal paths simultaneously to maintain common mode balance. Those parts are small enough to be annoyance free up to a few hundred MHz. It'll be interesting to see what you come up with above that.

I won't have boards for another week or so, probably. I'm not a sine-wave guy, so I'll TDR test them.

Environmentally hardened scientific apparatus. Doesn't need shake and bake for combat helicopter with gunfire or full function at 60k feet but needs to survive and work properly after a boondocks ride in a Land Rover or the equivalent.

The suits know what's real - we are them. We just don't wear ties or dress fancy because ties don't look good with solder spatters and dress shirts don't stay dressy with cutting oil on them.

A DC is already part of the system. So is a sensitive receiver with quadrature phase detectors. Plus, also, a sweepable frequency source. Software and tuning devices are what's in the works now.

Mostly, they just get confused and call for assistance. Support is expensive. If the interaction is "POST says it's broke", "a new one is on the way" and "return the old one for recycling" - things get simpler. You put your life cycle support into applications development and support - open new markets or enhance existing ones.

We work directly with the people who would discover the need for different specs. We sit in the same room and eat lunch together and run in and out of the labs together and gab a lot. Then stuff gets written into a development plan. Then the CEO and CFO and COO get filled in and stuff happens.

Joe Fourier solved that problem years ago. TDR --> FFT --> Frequencies!

Could you PWM between a low value R and high value R to interpolate?

My reasoning being the kind of things using thermistors are probably quite slow acting and inherently low-pass?

piglet

Thought about that, never came up with anything practical. We wanted to be a bipolar/AC resistor with good response time, so scanned acquisition systems will see a pretty realistic RTD. Accuracy needs to be in the 0.1% range over a wide range of excitation.

There was once a benchtop DVM (Digtek?) that had a 10-turn pot, a motor, a chopper, and a mechanical Veeder-Root type numeric display. It was pretty cool.

Den mandag den 8. august 2016 kl. 20.41.09 UTC+2 skrev John Larkin:

ell,

ta

ill

ame

I can't see that done without active circuitry on the isolated side which would be tough with AC

even with relays 0.1% isn't much compared to a PT100

-Lasse

It works, with a fair amount of isolated electronics per channel. As I said, it's difficult.

An RTD is about 4000 ppm/K, so 1000 ppm is usually acceptable.