Re: gEDA-user: C++ HDL

[Joerg <joergsch@xxxxxxxxxxxxxxxxxxxxx>]

John Doty wrote:
> On May 15, 2009, at 3:10 PM, Joerg wrote:
> 
>> John Doty wrote:
>>> On May 15, 2009, at 2:06 PM, Joerg wrote:
>>>
>>>> John Doty wrote:
>> [...]
>>
>>>> I think the BF862 will be around for a long time,
>>> So will the 6L6, but few engineers ever have reason to design with  
>>> one.
>>>
>> However, those who do make oodles of money. Audio freaks seem to have
>> pockets of infinite depths.
> 
> Sure, but so what? There's no need to train the average EE to do this  
> work.
> 

Until they must design a large transmitter or HV gen. Right here, right 
now, I am working with a tube that is fed almost 100kV. Always one hand 
in pocket ...


>> There's one 6L6 on duty in our living room, inside a Hammond organ.  
>> The
>> sound is amazing although it could be done class-D these days. But not
>> back in 1961.
> 
> My bedside radio has a 42 for its output stage, so I've got you beat  
> there ;-). 1934 Stromberg-Carlson, just the right sound for a ball game.
> 

Cool. But I still got you beat, have a radio with one of these from the 
mid 20's. At the bottom of the link you can see that this multi-section 
tube is actually the very first integrated circuit:

http://www.radiomuseum-bocket.de/museum/int./english/loewe-3-nf-tube.php

So maybe you are right, the first signs that the world might be going 
towards ICs started in the roaring 20's :-)


>> If you need a new 6L6 you can still buy them fresh off the
>> belt from places like Sovtek.
> 
> Yeah, but try to find a new 42. Fortunately, tubes that survive their  
> first few years and aren't stressed too much tend to last a long time.
> 

But don't push your luck. Get a spare while you still can. Else you have 
to do the unthinkable like I had to with my old Astor BPJ, had to adapt 
the set so it can take a slightly different IF stage tube.

>>
>>>> else Digikey wouldn't
>>>> keep >10k of them in stock most of the time. The six bump BGA (I  
>>>> hate
>>>> BGAs...) won't ever happen at reasonable cost
>>> Doesn't have to be reasonable cost. The Amptek thing everybody uses
>>> is $250 apiece, and is too bulky and power hungry for something like
>>> an x-ray APD array. If NASA and DARPA ever quit arguing over who
>>> should fund XNAV, they have the need, and they have the money. But
>>> they're both as bad as Windows users: "We have a common interest so
>>> here's what I need you to develop for me for free...".
>>>
>> Looks like a case of Windows-people phobia :-)
> 
> Not phobia, just an observation. gEDA was created by *nix people. It  
> is maintained by *nix people. It is extended by *nix people. But  
> Windows people who wander in here never come to help, to roll up  
> their sleeves and contribute. They would be welcome if they did.
> 

They can still provide ideas. Some of the features found in 
Windows-based CAD are pretty cool.


>> Cost may not matter in your business but it does when it comes to  
>> telco
>> and other markets.
> 
> That's one of the reasons IC's are taking over. In the high volume  
> markets they're cheaper than bunches of discretes.
> 

Sure, but in the high power circuits. Apex and similar amps set you back 
hundreds of Dollars. A pop. The discrete solutions were never above $10.


>> Many people tend to immediately default to "ready to
>> go" modules such as Apex amps where gusto is required. Part of my  
>> income
>> is related to replacing such designs with discretes, with cost savings
>> that regularly stun people. Especially controllers and CFOs ...
> 
> As IC's continue to increase in power, they'll get there. Also, many  
> power devices these days appear to be discretes, but are IC's inside.  
> You can't model a "power MOSFET" as a simple discrete, need a  
> subcircuit.
> 

Some, yes.

>>
>>>> unless there is a huge
>>>> consumer app that needs it. Some laser measurement device or  
>>>> whatever.
>>>>
>>>> As for smaller geometry JFETs, those unfortunately do tend to
>>>> vanish and
>>>> someday we might even read a eulogy on ye olde 2SK3372 :-(
>>> The handwriting is on the wall.
>>>
>> Not yet. The -G version is in mass production and Digikey still has  
>> tons
>> in stock. If it ever goes lalaland it'll become hard to make  
>> micropower
>> oscillators that can live off fractions of a volt. Chips don't do that
>> well down there.
> 
> They keep getting better. Especially for micropower low voltage  
> cases. And they've been used in such applications in watches for years.
> 

Same for hearing aids. The Japanese were always good in those domains. 
But when Vbat get down to 600-700mV they usually quit. A fuel cell 
doesn't have more than that and you've got to start the electronics somehow.


> I remember this same argument over tubes. And while the haven't  
> exactly gone away (the reheated cup of tea in my hand is proof!) they  
> have been sufficiently eclipsed that for the purposes of a university  
> professor designing an EE curriculum they may safely be ignored.  
> Discrete semiconductors have the same future, but there is less here  
> than you think. Future designers will still have to design with  
> individual transistors, but they'll mostly fiddle with w, l, and m  
> (or whatever equivalent their HDL uses) to get them. Anybody who can  
> analyze the consequences of those choices can easily learn to use  
> discretes in the rare cases where they will be necessary. But they'll  
> be annoyed by the limited choices. ASIC design has spoiled me enough,  
> I'm already there.
> 

Students that only learn device geometries and such can be great 
engineers but I've regularly seen them crash through the ice when they 
signed off on a layout and the EMC lab session ended with a thick 
forest. That's when my phone rings :-)

-- 
Regards, Joerg

http://www.analogconsultants.com/



_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user