I think of this as a two port black box. If the exact same thing goes in and comes out that’s what you have. Like the two components in series back in school in the box. Feedback is feedback use the right tool for the job.
But, comparing the same number of devices (2) in differential pair to 2 devices in parallel (complementary), it's 6dB with the same number of devices (2).
My earlier questions were aimed at maximizing efficient use of non-obtainium. How would one best weight the power supply interactions of folded cascode stages vs. its (idealized) 6dB Johnson/Nyquist advantage over differential stages? As simplified to the case of (2) non-obtainium devices and true-enough complementarity.
John has made a strong case for parallel (complementary) but at the expense of Herculean power supplies. Is the current state of art in power supplies good enough to prove him prescient for DIY guys (and gals)?
Much thanks, as always,
Chris
Last edited:
At Large Hadron Collider at Harwell they are using a new power suply for the most advanced nuclear detectors (transimpedance) which is actually a small SMPS with 10uV of total output measurable noise with 10% variation measured over 10 days.We need to steal that schematic 🙂
Look at it this way: the power supply is an integral part of your signal processing chain, and deserve as much TLC as the amp itself. The Constellation schematic JC posted illustrates this very well.
Chronologically, the power supply is:
- An afterthough, just make it as cheap as you can get away with
- A monster, as dictated by marketing, with Mega-Joules as the preferred units
- Maybe now it's time for a good balance?
Actually i wanted to say radiation detectors, not nuclear detectors...sorry, but i think that they actually use them for exciting the millions of coils taking care of the x-ray travel path.I just saw them physically...i don't know other details.
Last edited:
It is all about the good balance of PSRR of your circuit VS the ripple of its power supply, the current that your circuit is able to provide at each frequency (IE P.S. impedance) VS the current that the load requests.
Poor circuits request good P.S., good circuits can afford poor P.S. for the same result.
But the bear of Judge Roy Bean told me that the law of the west always applies: "A circuit is as good as its power supply."
On my opinion, it is not a question of "marketing", but the results someone consider as satisfying or good enough.
10uV noise (RMS, I presume) tells nothing about the noise performance, without specifying the bandwidth. It’s not amazing anyway, I would think: if the BW is 1MHz then the noise is 10nV/rtHz. The TI and Linear ICs quoted above are significantly better.
Diamond: Britain's answer to the Large Hadron Collider | Science | The Guardian
They have billion pounds laboratory equipment at Harwell...and they use intensively the most expensive electronic measuring equipment units i've ever seen in one place.
Given the fact that they deal with an x-ray fascicle focus and all radiation spectra needed to be measured in billion pounds research projects placed there...they might probably know what they made for their own use.
As a chief engineer said: "We are not financially constrained in prototyping anything, we are given anything we need to do out job, but we have to deliver anything we are asked to by the whole world research teams."
I personally never heard such a statement elsewhere...
You need to do a bit more research. Start with LLNL where Richard used to work. Plenty of places doing science with budgets for billet unobtanium.
Ok - I accept that you are correct on this.
I went back and did the sims, and over a narrow range and with bad compensation, you get peaking which looks like the independent gain bandwidth behavior of a CFA, but of course it’s not.
I stand corrected on this point. Mea Culpa
🙂
Last edited:
I don't know about Harwell, maybe this is true, or maybe this is seen from a distance.
Once upon a time, I worked a the CERN, and the budget constrain in the electronics dpt. was so bad the labs were cross-charged down to a per-resistor level.
I left shortly after I started, not because of the $, but the internal politics were totally uncool.
That was not aimed at you, if that author was just repeating something you said he didn't do all his homework. There are plenty of data sheets out there by now with a more complete picture. At frequency extremes both VFA and CFA deviate from their basic behaviors.
My current project is designing microphones for security use. I have done two designs. One is fairly basic a phantom powered line level output microphone that mounts in the space of a standard electrical outlet box.
The other is designed to feed the audio inputs of two security cameras. The cameras them connect to the rest of the system using internet protocols.
Unfortunately many of them use ITU G.711 standards. 8 bits of logish encoding at 8 K sampling rate. Some of these cameras expect an input signal level maximum of 0.1 VRMS. Others more.
So the design includes a limiter, selectable maximum level, a phantom power ability and dual outputs as one microphone may feed two cameras.
So I should have prototypes in hardware in a week. While waiting for the circuit boards and the UPS lost parts shipment, I am building an EMI test chamber.
I have a signal generator and 1 watt amplifier. I suspect I will build a 33cm metal cube as the test chamber. The launch antenna seems to be the sticky bit. My amplifier really doesn't want to see an SWR above 2.
So my first thoughts are use an attenuator to protect the amplifier.
Bill? Etc?
The other is designed to feed the audio inputs of two security cameras. The cameras them connect to the rest of the system using internet protocols.
Unfortunately many of them use ITU G.711 standards. 8 bits of logish encoding at 8 K sampling rate. Some of these cameras expect an input signal level maximum of 0.1 VRMS. Others more.
So the design includes a limiter, selectable maximum level, a phantom power ability and dual outputs as one microphone may feed two cameras.
So I should have prototypes in hardware in a week. While waiting for the circuit boards and the UPS lost parts shipment, I am building an EMI test chamber.
I have a signal generator and 1 watt amplifier. I suspect I will build a 33cm metal cube as the test chamber. The launch antenna seems to be the sticky bit. My amplifier really doesn't want to see an SWR above 2.
So my first thoughts are use an attenuator to protect the amplifier.
Bill? Etc?
Ed, depending on bandwidth you are testing over I might be tempted to just bang a circulator on the amp output. Depends quite how much worse that 2 your vswr is 🙂
33 cm? Per side? I must not be understanding what you are saying.
If your intent is to measure emitted RF from your DUT, you need a larger chamber. Not to meet some standard, but just to get the walls away from the near field of your DUT. Even then, you’ll need to use some types of RF anechoic material on all the surfaces to minimize reflections. These reflections will dramatically affect your antenna pattern.
If you want to see how your DUT is affected by external RF generators, the same constraints apply.
But, if just want a general idea of RFI susceptibility, you don’t really even need a chamber. Your RFI test source will likely be much stronger than any random external source, unless your lab is at a radio station transmitter site. (An actual possibility.)
Anyway, to answer your question about an attenuator: Yes that works.
VSWR to Return Loss Conversion Chart | Amphenol RF
The above shows that you need a pad of about 5 dB. Or, higher. That way, no matter what load you put at the output the VSWR “seen” by your amplifier will be better than 2:1.
Also, it's likely that your test antenna does not have a perfect nor flat impedance match across the band you are interested in. Unless your connecting cable to the antenna is very short, it probably will benefit you to place the pad at the antenna if you can. That way, increased loss in the cable due to mismatches at various frequencies won't be as severe. This can add to the non-flatness of the overall system. Just a detail, I know...
Yours truly,
Etc.
Last edited:
- Status
- Not open for further replies.