I thought it was clear: All of them. Nobody has a clue - everybody does it by the way that works for them, and there are no hard-and-fast rules, becaue reality doesn't work that way, at least on digital/analog grounds.
Thanks, Rich
Yeah, mid-terms, ergo nobody really cares, albeit there are rumblings in the media that the repugnocons are getting tossed out this year.
Thanks, Rich
Are you implying that there is such of a thing as a "sane" grounding scheme? ;-)
Thanks! Rich
What an odd reaction - I would laugh. ;-)
Cheers! Rich
Sure. Use a metal box. Ground it through the 3rd pin on the power plug, and also bolt it into a grounded relay rack if you can. All the pc boards inside should be multilayers and have a single solid internal ground plane. Boards should be mounted to the chassis with lots of metal spacers, hard-grounded to both the chassis metal and the pcb planes; PEMs are nice. Any connector shells that penetrate the chassis should be grounded, too. It's very simple, actually.
John
Hello John,
As a consultant I rarely get a peek at the financial hit a client took. Sometimes they mention it later over a beer. Many times they won't exactly know. It is hard to gauge how much a company lost because of the fact that they missed "the" yearly trade show. You can only project your sales versus the competition well when you are actually selling. In one case it has caused around a half year of delay and that has definitely amounted to a few million. Might sound arrogant but had they called a year earlier (IOW well before layout time) this could have been avoided.
The most sad story was a company that turned me down and I am very certain I could have helped them out of their pickle. It shut down because they couldn't really make their flagship product work. Then the usual, layoffs, loan write-offs etc.
-- Regards, Joerg http://www.analogconsultants.com
Hello Dimiter,
Oh, it can happen and I have seen it. One case was a "three-high" VME style board that had a slot through the ground plane at about 1/3rd from the front. There was a low impedance connection (big piece of wire) in the center of that slot. Then a 1kW switcher decided it was time to blow up. It wasn't the power supply for this unit but another. The ADCs in the center of the board survived but everything on the outside that was straddling the slot was toast.
Sometimes we have to. Mostly in medical gear that is connected to a patient. It has to withstand a full defibrillator jolt into its input. This requires at least 5kV of isolation and as little as possible in capacitive coupling.
-- Regards, Joerg http://www.analogconsultants.com
I've had several occasions where clients considered my quote too high and went with someone else. Then they come back a year or more later in a panic, wanting it done immediately... +50% ;-)
I think I told of this before...
Stopped in at a client's place, just because I was in the neighborhood.
Found them struggling with a strain gauge conditioner on one of their exercise machines.
Two days before a Las Vegas show.
I said, "Sure, I know how to solve the problem."
"But it's taken me years to develop the technique."
Told them $4K to show them what to do.
Haggle. Haggle.
They agreed to write me a check for $2K on the spot. I'm to draw the schematic. If it works like I claim they'll write another check for the additional $2K.
15 minutes later the technician had it working and they wrote the second check.They were a success in Vegas.
But they never called me again :-(
I'm on my 4th successor to an RFID tag company ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Hi Joerg,
...
Do you know the reason for the massive frying was the ground split? The ADC in the middle survived, you say; how come the rest died? If there have been parts connecting to both sides of the split, I can understand that - but it is just nonsense to design things this way. Other than that, if across the bridge between the planes the voltage has been low enough, it will have been lower over the planes themselves, the reason must have been something other than the split.
Well if you count 5 kV isolated planes ground, yes. To me, ground is something with no DC discontinuities. Having said that, your comment is actually correct, there are exceptions. For example in my HV modules I did two separate ground planes (conected somewhere), but I don't remember now why and how I did it, nor am I sure whether it was a must or I just went the easier/faster way, it was over 5 years ago last time I designed a HV source (these are very low noise - a few mV p-p - 100V to 5 kV things for biasing detectors).
Dimiter
------------------------------------------------------ Dimiter Popoff Transgalactic Instruments
------------------------------------------------------
Joerg wrote:
Probably because my circuits are so "prosaic" ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
ok, you're a connect everything to everything else everywhere possible type of guy,eh? :)
ooo thats a pretty nifty spec on that chip, Im also modulating a laser at high frequency but not quite that fast, atm I use a bfg591 in common emitter wich just switches the laser on and off it works at 1ghz but I suspect its running out of steamm at 2ghz, originaly was only designed for 10mhz+ I want to go as fast as I can but I think 20ghz might be a bit too dificult however I was thinking of using an MMMIC such as a phemt as the above seems to be, but I have a couple of questions ... what is the best way to attatch it to the laser at such high frequencies ... ac coupled with a bias supply ? or use the device as a switch ? and what about impedance with the high laser capacitance and inductance of the leads ? it also has 2 extra gate voltages wich is interesting, is this like cascaded stages or like a dual gate device ?
Colin =^.^=
"Didi" a écrit dans le message de news: snipped-for-privacy@p79g2000cwp.googlegroups.com...
You don't need to defibrillate your equipment to send it to hell.
I remember wery well, a payment terminal with a small separate VGA monitor where the video chip kept frying latched-up. Not at a bad rate, about 30% of the 2K produced were killed over about 6 months.
Guess what? The designer used a split plane to isolate the video IC part from the evil uP ground currents, as was told in the IC app note (SED1355 IIRC) with a nice 5x5mm track between both GND areas.
Not even one was destroyed since I had them making the GND plane full. Of course the video quality didn't suffer in any way.
-- Thanks, Fred.
nt
tor
end
Last year:
--=20 Keith
=20
Why don't you re-post as a v3 or v4 PDF. I can't read it.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
A SOT-89 mmic should make a pretty good modulator, up to maybe 150 mA p-p or so. Beyond that, a discrete mesfet or phemt, driven from a mmic, should be fine at 2 GHz. W-J (formerly EIC) makes some cheap high-current mmics. The CLY seroes of mesfets (CLY2 etc, formerly Infineon, now made by TriQuint) are nice, and NEC makes some high-current phemts. The phemts only need around a volt p-p of gate drive, so are easy to use.
With a mmic as driver, I'd ac couple into the laser and add a dc bias, through a small inductor maybe. With a fet, you can put the laser directly in the drain, and maybe servo the gate bias or, more crudely, adjust the drain supply voltage to tweak laser current. If you have a monitor photodiode, use it to servo fet bias.
Just bypass (or ground) the laser cold side, and keep everything zero length!
I just did a laser driver (1310 nm singlemode vcsel) that uses an MC10EL89 as a dc-coupled driver, but it's only good for about 70 mA and 1 GHz.
John
y
want
at
ten
nput
10 or dthe
ou,
.D and
s,
e , !PDFs suck, eh? It opens in 6.0 but gives a warning. Apparently it=20 really wants 7.0. Even PDFCreator barfs on it so can't reformat it=20 downward. =20
I don't have the source (though wrote/edited parts of it) and don't=20 own the site, so have no control over what is there. Sorry.
--=20 Keith
[snip]
Can you print it to a PostScript file, zip it up, and E-mail it to me?
Then I can run it thru Distiller to get a v4 or v5.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Hello Fred,
30% in six months? Ouch.
Same here. After a re-layout with a full ground plane the failures didn't happen anymore. Some glitches that folks were blaming on software also miraculously vanished :-)
-- Regards, Joerg http://www.analogconsultants.com
Yup. Preferably with big bolts.
John
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.