Acutally, mutual phase-noise that is randomly distributed, and only 1 degree or so is perfectly tolerable for radio astronomy purposes in an interferometer. But when the phase noise magnitudes start b

In VLBI, each station has a *very* precise clock--usually a hydogen maser--that clock drives not only the time-stamping, but all of the relevant electronics. The clocks in a system are aligned with on

I've been doing radio astronomy both in the small-scale "amateur" end of things, and as an occasional technical consultant to the "big boys" since the mid 1980s. I will observe that you have a very ve

What is the Yagi -- commercial? Do you have a pointer to it online? Image synthesis at the kinds of resolution you'll get with a bunch of very-coarse single-pixel radiometers is also, I think, going t

I think that the reality is that if all your *other* problems with producing high-quality measurements have been solved, and you've arrived at the "how linear is the ADC" problem, you've achieved some

Indeed, at larger scales, of course, it isn't flat. For example, the synchrotron falls off fairly quickly above a few hundred MHz, so at the notiional "edge" of the synchrotron region, you might get s

Frequently switching to a load at a known temperature (sometimes at several Hz) is a technique that has long been used to factor out gain changes over the entire receiving chain. Robert Dicke invented

For us, "Binocular" is just a catchy name. We dont' require the two sides to be coherent. With the exception of spectral lines, the radio sky is, over "smallish" bandwidths, dead-flat. It consists of

Our dual-channel ("binocular") 21cm spectrometer has as low-loss RF relay in the front-end, in front of the LNA, which switches the LNA between the sky and a terminator at about 50C. The entire receiv

1GiGe has plenty of bandwidth for modest applications up to 25Msps or so with complex samples on the wire, 16-bit samples. USB2.0 has about half of that USB3.0 has about 5Gbit. But, as I've observed s

#poll

I ran a small hobby/research aerospace company in the 1990s/early 2000s that provide propulsion systems. These systems pretty-much *had* to be built to the kind of quality you'd expect in aerospace, o

#poll

I have found that 1GiGe is more predictable and reliable than any of the USB standards. It has historically been the case that it is more expensive to implement than USB for embedded systems.

#poll

User-land software going all-crossed eyed when you pull its connection to hardware is one thing, but causing the whole system to crash as well? That's just bad ballet right there. It should never be p

The temptation when you go to a much-higher-rate 'pooter interface is to offer a number of sample-rate options to the end user (where number is probably significantly more than two). That immediately

I think the problem is that USB3.0 "data pump" chips are still substantially more expensive than their USB2.0 sisters. That could be changing--for example the FTDI601 that Lime-Mini uses was chosen fo

On 04/24/2018 05:54 AM, prog wrote: On Tue, Apr 24, 2018 at 02:50 am, jdow wrote: But some residual error is inevitable. I leave that to the application engineers who design the final coherent systems

The origin of this thread, I think we have different perspectives on, which is where this arises. It was about, I thought "dBFS" in FFTs vs "dBm". And getting dBm, in lieu of calibration, is tricky--i

Sure, completely agree. Although you still end up with corner cases where your app may be using a library that produces a filter that isn't perfectly flat across the pass-band, and as an app developer

Suggest to me, then, what "clever algorithm" can, in the absence of hardware knowledge and calibration data (and/or a calibration *procedure*, using calibration sources) tell you what the dBm power le

Granted, "modern" hardware generally has batch-to-batch performance variability that is much better than it used to be, and temperature effects are of a smaller magnitude. But the key phrase here is "