Regular FR4 did spread out our 1988 GaAs-logic generated 1.25nsec wide pulses. We were using big boards.
The effect is probably less noticeable on the tiddly little high-spped boards that you post from time to time.
In practice, the obvious advantage of the exotic substrates was broader high impedance tracks.
We had a narrow track with a 150R characteristic impedance on the Blanking Output board, which a small single layer daughter board with an alumina-loaded Teflon substrate ....
-- Bill Sloman, Sydney
Sure. As in traveling wave tubes, too.
As I was well aware at the time
The patent that came out of the 1983 work explicitly mentions a traveling structure.
Now fit it into a regular electron microscope column at the second conjugate point. We thought about it from time, but the mechanical problems looked expensive to solve.
-- Bill Sloman, Sydney
True. But nobody who limits collector current and collector voltage to the absolute maximum ratings is going to need to take full advantage of the pul sed power advantage at any lower temperature.
Think about it.
True. But your peak pulsed power shouldn't be higher than 600mW (20V at 30m A) and you can dissipate 280mW continuously at up to 88C.
If they meant that you could safely drive more than 30mA through the device under pulse conditions, they'd say so explicitly. The figure on page 5 are all about power, not current.
No. That's much dumber. It just tells you the mass of the silicon you are h eating up when you temporarily exceed the maximum permissible sustained dis sipation.
I'm sure that it was, but it doesn't say anything about exceeding the maxim um permissible collector current. You are misunderstanding what it means to justify your evil over-driving habits.
That's not the job they did. The maximum allowable collector current is exp licitly specified on the first page of the data sheet. The graphs on page 5 are all about getting the die hot, and don't relax the front page limit in any way at all.
It isn't. It's in your self-serving misinterpretation of the data sheet.
Run the transistor in a 146C ambient. There's nothing complicated going on.
For a pulse that can get down to 0.5nsec wide? Maybe today, but not in 1990 .
rd.
800MHz was fairly fast in 1988-90. Back then, the board layouts were done b y printed circuit draughtsmen on a horribly user-unfriendly computer-aided design system. We got to comment on the tracking, and could insist that it was changed, but managing the Gerbers was their business at the time, not o urs.I got closer to the coal face in the Netherlands - I did the layout for our 200MHz ECLinPS board that replaced a very dodgy TTL-based board (where the parts had had to be selected to make it work, and tended to stop working a s required as they aged, and the propagation delays moved away from the swe et spot).
I was emotionally involved. The system design for the machine was pretty mu ch all mine. I'd put it together more to satirise the mad technical directo r than because I thought that it was the right degree of difficulty to aim at, but I quite liked what I'd thought up, even though I thought it pushed the sta te of the art a little too hard for comfort.
Not true above 146C.
Run the transistor in an environment at 146C \
Not if you haven't got hardware out ion the field that depends on understan ding it wrong.
You aren't paying attention.
It's trivial. Sustained power dissipation is limited to 280mW. Peak power d issipation is limited to 600mW(20V collector voltage and 30mA collector cur rent). At room temperature you need to worry about the die getting too hot if your duty cycle at maximum permissible power dissipation is higher than
46.6%.The die has got a definite thermal mass, and you can't sustain 600mW dissip ation for longer than about 100usec, if the transistor is an ambient below
88C.If the ambient is warmer, the die gets dangerously hot sooner. At an ambien t of 146C the core of the die can get too hot in as little as 100nsec after it starts dissipating 600mW.
-- Bill Sloman, Sydney
[...]
If I had my druthers and it would be a real mass product I'd even try it out on 2-layer phenolic :-)
No kidding, I've got some UHF stuff on phenolic in the garage and it works. With microstrip on there.
-- Regards, Joerg http://www.analogconsultants.com/
You don't need to limit to abs max DC values. But you seem not to understand that, so I'll give up.
[...]
So how does power come about without the requisite current? Jeannie's bottle?
[...]
It was easy even in the late 80's where I built pulsers and samplers. But if you are always so hesitant to use parts to their real limits you certainly could not get there. Even if you don't understand or doubt the datasheet all it would have taken is one call to an FAE. Back then they still came out to your site. Even today some do, like for Skyworks a couple weeks ago here.
[...]
Not at all. In ham radio people did several-GHZ designs back then. I wasn't much interested in it because of the dearth of radio traffic up there but helped others with their designs. Got a huge book about it here (Weiner "UHF-Unterlage") which AFAIK was unfortunately never published in English. It's a treasure trove of information on how to go about GHZ projects with a limited budget.
It was yours. Every set of Gerbers must be accompanied by a fab spec which, among other things, spells out the layer stack.
That sounds like an ugly job but I also have to do those on occasion.
[...]
You honestly believe someone in their right mind would run a transistor stage at 146C?
That's ridiculous. I take it you cannot demonstrate it.
[...]
So you honestly think Siemens/Infineon wrote page 5 for operation at 146C?
ROFL. Send that to Infineon, they'll get a nice chuckle out of it :-)
-- Regards, Joerg http://www.analogconsultants.com/
how much cheaper is phenolic?
do they even make 2 layer?, from what I understand vias are iffy in phenolic
-Lasse
Incredibly cheap. Especially single-layer punched hole/outline.
Only the vias that are made by silver ink crud.
--sp
Even when I was a kid they did and you can still buy it:
It's not just Chinese shops that offer boards with it, also Western places like this one in Germany:
Remember riveted vias? From the days Lassie showed up on TV. But the real art is to make sure one needs no vias, where they all end up at some pin of a thru-hole part. Doing a really optimized layout on 2-layer phenolic is like solving a rubix cube or a 1000-piece puzzle.
-- Regards, Joerg http://www.analogconsultants.com/
By the same logic you wouldn't need to limit to the absolute maximum DC vol tage values either.
You've got away with ignoring the absolute maximum DC current limit, but it doesn't follow that it was a good idea.
The power limits depend on ambient temperature.
Not for the applications where I was working.
We certainly had that kind of relationship with Philips, and I installed ou r first "voltage contrast" electron microscope - the predecessor of the spe cialised electron beam tester - at the Siemens central research lab in Muni ch in 1984, having installed my Chinese copy of the original Siemens voltag e contrast research system at Thompson-CSF at Grenoble in 1983.
But I wouldn't have been silly enough to ask them if an absolute maximum cu rrent rating on a data sheet was in fact negotiable.
Not the same thing at all. You are talking about radio-frequency circuits u sing a high frequency carrier to transfer much lower bandwidth information modulated onto the several-GHz carrier frequency.
We built a system using synchronous logic locked to an 800MHz clock - much more broad-band.
Sure. That's where Kevin Jackson - our high-frequency consultant - came fro m, fresh from designing a cheap satellite receiver in Hong Kong.
Sure. But I'm talking about the way a UK company actually worked, not the w ay it should have worked. I would have been happy to take the responsibilit y, but it would have cut across the company power politics, so it was never going to happen.
And ignore absolute maximum current specifications.
No, but the pulsed power curves on figure 5 of the Infineon BFR92 data-shee t are a regular feature of transistor data sheets, generated by plugging in the right numbers and plotting out the results. The people who generated t hem weren't thinking about the fine detail of what they implied.
It may be ridiculous, but it's logical. The peak pulse power ratings are ge nerated by a process of more or less mindless calculation, taking into acco unt the thermal mass of the silicon around the collector and the thermal di ffusion coefficient of silicon. Reading the results as permission to exceed the absolute maximum collector current rating is perverse.
?No. I think they generated the page 5 curves with the same computer program that they used for other, bigger transistors, with higher maximum collecto r voltages and maximum collector currents, and didn't think about the impli cations.
-- Bill Sloman, Sydney
Hell, I'd ask them about everything! Back in ye olde tube dayes, the datasheets were "suggestions", not maximums! ;-)
(I'm in a nostalgic mood today as I'm wearing my tube t-shirt...
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com
As I said, it doesn't make sense to try to explain that to you any longer.
[...]-- Regards, Joerg http://www.analogconsultants.com/
But tubes were more or less hand-made. Modern semiconductors are produced b y vastly more heavily automated processes, and there's a lot less variation from device to device.
And "asking them everything" means hanging on the phone for hours - the peo ple whose job it is to answer the phone don't know much, and they are remar kably reluctant to connect you to somebody who knows enough to answer inter esting questions. The people who know that much have got more profitable th ings to do with their time than answer questions from people who think that absolute maximum ratings are negotiable, and might sue you if you tell the m something that they might misinterpret as justifying their daft design.
-- Bill Sloman, Sydney
Yes, it is like talking to a wall, isn't ?
Jamie
Den torsdag den 6. marts 2014 23.47.28 UTC+1 skrev Tim Williams:
-Lasse
You are kidding, right?
On the contrary. A typical semiconductor process see up to +/-30% variation and I've seen worse. Tubes can be less, their tolerances are largely based on mechanical dimensions of their innards.
Even the old Mullard plant was highly automated. The tube manufacturing machines they built were absolutely remarkable:
Nope. The people who know often even drive all the way up here and I am not located in a metropolis so that takes effort. Like with a Schottky diode where I needed a spec well beyond what's in the datasheet. The engineer drove up here on a Monday morning. Now I have my answer, backed up by another engineer at Skyworks headquarters, and can complete the design. This is how cutting edge engineering is done efficiently.
-- Regards, Joerg http://www.analogconsultants.com/
board-double-sided.html
Back in the 80s, when the cube was all the rage I could solve it on a 30 seconds average. My daugther got one for christmas and I'm back to under the minute. Not bad after so much time, and having forgotten most of the best tricks...
I wish I could P&R my very last board (8000+ components) at the same speed. (Un)fortunatly not :-)
-- Thanks, Fred.
It didn't make any sense to start trying to explain it to me in the first place. As far as I'm concerned you are trying to tell me that the sun rises in the west.
-- Bill Sloman, Sydney
DC
but > > > it doesn't follow that it was a good idea.
er.
When it comes to being persuaded that maximum current ratings can be ignore d, I'm pretty wall-like. I really don't like nonsense - even if it comes fr om Joerg who is almost always commendably correct (and he's allowed the occ asional idiosyncrasy, just like the other gurus around here).
You aren't allowed any idiosyncracies, because you are wrong much too often .
Regard me as a "sweet and lovely wall" as featured in Shakespear's "Midsumm er Night's Dream".
-- Bill Sloman, Sydney
I've got an almost identical one, except the tube is horizontal and it says "Tubes Rock".
You're useless, and stopped talking to yourself.
Jamie
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.