pick-and-place

Interestingly the pHEMT wasn't the problem at all--it was the SiGe. A

Cheers

Phil Hobbs

Sense or not, that's the way things are.

CPUs and FPGAs, things that pull amps of low-voltage core power and stress the power distribution system. It's trivial to power glue logic parts running at 3.3 volts.

Sometimes I think you know some stuff and other times I think you are a complete idiot. FPGAs do contain PLLs or DLLS, but they are only clocked by them when the user decides to use them.

Regardless, that has nothing to do with the topic at hand, not that you made a useful statement about using PLLs. Did you have a point?

I took a 150mm x 150mm board and measured the impedance, hope it's not a bogus measurement:

First resonance at 144MHz (no capacitance added)

This resonance may be due to the fixture. I only had 5 minutes for measurements......

Cheers

Klaus

I believe the subject was synchronicity of Icc spikes from various chips on the same board. Jasen correctly pointed out that the current spikes are unlikely to be simultaneous. The fact that big logic chips are often clocked from on-chip PLLs make things even less likely to be simultaneous.

The ARM chips that we use are always internally clocked by a PLL. Our FPGA designs are almost always clocked by internal PLLs, often with several internal clock domains. Big power-hoggy chunks, like serdes and PCIe, have GHz internal clock generators, and will have corresponding GHz power supply modulation.

What's idiotic about any of that?

Looks vaguely reasonable compared to JL's data--flattish and well below an ohm, it looks like.

Cheers

Phil Hobbs

Pretty reasonable, in fact.

Here's the TDR of a 6" slab of 45 mil thick FR4. It conveniently hangs on that SMA connector.

The initial flat part is about 1.2 ohms for maybe 2 ns. A thinner sheet would be a better model for a multilayer plane pair, and correspondingly lower impedance.

It looks a little different if I move the SMA to a corner, but then the board keeps falling off.

Made perfect sense to me--I knew exactly what he meant.

He's not an idiot, if that narrows it down for you. Quite the opposite.

It's weird what weird things people start to think they know about someone over usenet.

E.g. Bill thinks he's had to clean up after people like John. He's got that backwards, in spades.

On the bypass thing, a fairly smart guy like Fred speculates, then gripes when John actually measures, calls John a "mechanic." For being right. Bizarre.

Using caps wayyyy over SRF is standard practice. Above SRF a chip cap's an inductor, set mostly by package size, and not much else really matters.

Cheers, James Arthur

What blather.

The bypass caps furnish power to the chips in the mid-frequency range, where the power regulators are too slow and the copper planes can't store enough energy. They also lower the high frequency plane impedance a bit, if there are enough of them close to big fast loads.

I don't know why so many people get worked up over bypassing. It's easy and almost always works first try. I haven't had a bypassing problem on a digital board in a decade or three.

Wideband analog designs can have problems if you're not careful. Some opamps have PSRRs near unity, sometimes worse, so a little supply noise can cause big trouble in a high-gain, wideband amplifier circuit. That's about the only place where ferrite beads or resistors belong in power distribution systems.

What, me? Vain?

We use 0603's because we have them, they are cheap, and they work fine. Via inductance dilutes the value of 0306's. I've used a few 0306 types in superfast signal paths, and in bypassing where the cap was between layer 1 pours, without via inductance. Exotic stuff.

response you are on, C or L. Just need Xc to roll of against some R or L.

Right. I don't care what the SRF of my bypass caps might be.

the frequency is increased, just as the curves show. I look at some designs from AD and they have a stack of caps on every power node. What a waste. If it were not for ground planes, most designs would not operate.

Exactly. The plane dominates at interesting frequencies.

If experimenting and measuring make me a mechanic, all I want to be is a good mechanic. I don't need to publish pages of equations, I only need to build things that work, preferably work the first time.

In the real world, it's unavoidable. We use chips with multi-GHz internal clocks.

Above SRF a chip cap's

tress the power distribution system. It's trivial to power glue logic parts running at 3.3 volts.

complete idiot.

e over usenet.

hat backwards, in spades.

John's fault is that he doesn't know the theory all that well. He compensat es by doing a lot of observation. The people that I had to clean up after c learly didn't know the theory, and were not all that careful about understa nding the obviously limited observations they bothered to make.

I'd be very surprised if I could improve on many of John's designs. On thos e rare occasions where he uses a bought-in high-frequency transformer, I mi ght be able to design and wind something that did better, but getting it ma de would probably screw up his supply chain something horrible.

when John actually measures, calls John a "mechanic." For being right. Bi zarre.

John's weak on theory - always has been. People who aren't do feel superior to the more empirical "mechanics". Tesla versus Edison.

Give us a frigging break will ya! Bullshit may have once gotten you to the top, but guess what? If that ever did happen, you sure have been at the bottom for as long as I've know you here. You don't need to continuously display how ignorantly arrogant you are, we already know.

Hypocrite.

Jamie

I try to do that if a part has a dominant frequency or if I'm trying to filter a particular frequency. I learned that when I had problems with a TDMA ISM band transmitter on the board getting into everything (in particular, the PN junctions of a bipolar opamp). Decoupling with the appropriate caps (using SRF) at 2.4GHz and as filters on the opamps helped bunches.

If it happens within the risetime of the clock, it's synchronous. Absolutes aren't needed.

Please, please don't make me clean up his LC oscillator!

I learned about that trick from the ARRL VHF Handbook, circa 1970, but I've very rarely used it myself. For radio and other usually-narrowband applications, it can work great.

Cheers

Phil Hobbs

It's not nice to mess with the priest of the holy Baxandall and Blumlein. ;)

Cheers

Phil Hobbs

On Wednesday, 3 September 2014 10:34:49 UTC+10, Maynard A. Philbrook Jr. w rote:

rote:

nd stress the power distribution system. It's trivial to power glue logic p arts running at 3.3 volts.

re a complete idiot.

te.

meone over usenet.

ot that backwards, in spades.

nsates by doing a lot of observation. The people that I had to clean up aft er clearly didn't know the theory, and were not all that careful about unde rstanding the obviously limited observations they bothered to make.

those rare occasions where he uses a bought-in high-frequency transformer, I might be able to design and wind something that did better, but getting i t made would probably screw up his supply chain something horrible.

ipes when John actually measures, calls John a "mechanic." For being right . Bizarre.

rior to the more empirical "mechanics". Tesla versus Edison.

You certainly don't deserve one.

but guess what? If that ever did happen, you sure have been at the bottom for as long as I've know you here.

Your usual failure of judgement. I don't do bullshit, and never have. I don't like it any more than you claim to and have a long history of being rude about it.

Your problem is that you don't understand much, and write off stuff that's beyond your comprehension as "bullshit". It makes you look a bit of idiot, but then you are a bit an idiot, so it saves everybody else from having to take you seriously.

arrogant you are, we already know.

My arrogance, such as it is, is rather better-informed than you know - or could know.

Not true, but how would a half-wit like you know that?

It's like krw calling me ( and half the universe) a liar, because what we claim doesn't match what he thinks he knows.