Which university produces good analog EEs?

[Joerg]

miso@sushi.com wrote:

[...]

Oh yeah, the lack of soldering skills. That would require the student
to have actually built something. These younguns just know how to
program. You've seen the posts where a pic uP is the solutions to any
task, not a state machine comprised of memory elements and
combinational logic.

For the vast majority of applications, a uC is the right solution,
certainly over the discrete implementation you suggest.

I certainly would not say "vast". Many times I have pondered the use of
a uC in a design only to come to the conclusion that 50c is still too
expensive. It's amazing how cheap "poor-man's logic" can be.

The problem with using a uP in such projects is if you are designing a
chip, you need to know how to do it with gates as often that is the
smallest and lowest power solution. The ability to hand craft logic is
disappearing rapidly, but is very much needed in mixed mode chips
which are not done on fine geometry processes.

It sure is disappearing, just like analog skills seem to be almost gone
in fresh grad.

I've ripped out many uC solutions for reliability and cost reasons. Same
for PALs/GALs because that's the era back in the late 80's when
mixed/discrete design skills began to tank. The topper was a system
where I ripped out so many (plus went to 100% AC terminations) that the
power supply kept tripping off and one of them blew. We had managed to
get underneath the minimum load. IIRC the PALs was guzzling 30-40mA just
sitting there. Each.

What's the old saying? If they only learn how to use a hammer every
problem will begin to look like a nail. Heck, I've seen one-shots being
done with a uC. That almost made me sick.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Vladimir Vassilevsky]

Joerg wrote:

Heck, I've seen one-shots being
done with a uC. That almost made me sick.

Did they use a RTOS ?

VLV

 

[Rich Grise]

On Wed, 03 Oct 2007 21:10:36 -0700, John Larkin wrote:

On Wed, 3 Oct 2007 22:51:41 -0400, krw <krw@att.bizzzz> wrote:

I'm hoping I don't have to go through that again. What a pain,
daytime TV sucks, and not cheap either.

From what JohnL has said, he likely did the guts of the MRI machine I
was inspecting the innards of, too. Good company here! ;-)

No, I only do the gradient drivers for small-bore stuff, like 4" and
8" machines for imaging lab animals, and even smaller micro-imaging
systems. Those can be dinky but fast high-precision, 100 amp, 170-volt
(17 KW per axis!) drivers. The full-body stuff is much higher power,
always switchers (Copley is pretty good) and slower and a lot noisier.

I had my head MRI'd earlier this year (spare me the predictable
witticisms, guys) and it was noisy and mostly boring. It's the
gradient coils that make the noise.

Oh, come on, John. Predictable witticisms are part of the reason some
of us hang out here! ;-)

Cheers!
Rich

 

[Joerg]

nimo wrote:

thank u very much u just raised my hope i thought that our
universities in egypt are so what too say primative in practical and
specially
in analog design
but for a change
i think analog/mixed design is reviving again it's the chance
i my self am designing a VGA for a communication channel AFE
well it may be a sand particale compared to the whole design but those
digital geeks using VHDL and denying the ability of human
to get new designs of his own can't do it never because it just need
the talent and someone loves Transistors
Analog era will never end it's the backbone of every thing even
digital they just need the finnest as they need less numbers these
days
any way if someone knows a college in US that will help an egyption
student like me to get his master at in analog/mixed design
(considering i can't be a penny)
i'll be glad of him

US universities tend to be quite expensive and I bet the cost of living
is also much higher here than in your country. Why can't you get the
degree in Egypt? And why do you really need a master? Ok, I've got that
degree myself but honestly nobody ever asked about it or actually wanted
to see the degree, except government agencies. All they wanted to know
is whether I could solve their problem and how long the redesign would take.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Joel Kolstad]

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:1c9Ni.5861$6p6.4832@newssvr25.news.prodigy.net...

miso@sushi.com wrote:
What's the old saying? If they only learn how to use a hammer every problem
will begin to look like a nail. Heck, I've seen one-shots being done with a
uC. That almost made me sick.

I used to be opposed to using a tiny microcontroller to do something "simple"
like flash an LED in various patterns depending on, say, 2 or 3 inputs, but
these days I think it's often justifiable if only from the standpoint of
scheduling: Cranking out all of 20 lines of C code is a lot faster than doing
the "discrete" design. With tiny microcontrollers going for ~$0.25 these days
in 100-ish quantities, it's pretty hard to justify even a day spend on a
discrete design for low quantities (<1000). Of course, for the high-quantity
stuff you do, I realize it's a different game.

Right now I'm considering using one of those quarter microcontrollers just to
generate the output signals for a standard H-bridge creating a stock "modified
square wave" output (essentially 3 different states) based on an "enable"
signal and a check of whether or not the rail voltage is high enough (using
the cheesy built-in ADC on the micro). Any suggestions on a "better"
approach?

---Joel

 

[John Devereux]

Joerg <notthisjoergsch@removethispacbell.net> writes:

[...]

It sure is disappearing, just like analog skills seem to be almost
gone in fresh grad.

I've ripped out many uC solutions for reliability and cost
reasons. Same for PALs/GALs because that's the era back in the late
80's when mixed/discrete design skills began to tank. The topper was a
system where I ripped out so many (plus went to 100% AC terminations)
that the power supply kept tripping off and one of them blew. We had
managed to get underneath the minimum load. IIRC the PALs was guzzling
30-40mA just sitting there. Each.

What's the old saying? If they only learn how to use a hammer every
problem will begin to look like a nail. Heck, I've seen one-shots
being done with a uC. That almost made me sick.

I just took apart a transducer, they used two PICs as comparators...

--

John Devereux

 

[Joerg]

Joel Kolstad wrote:

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:1c9Ni.5861$6p6.4832@newssvr25.news.prodigy.net...

miso@sushi.com wrote:
What's the old saying? If they only learn how to use a hammer every problem
will begin to look like a nail. Heck, I've seen one-shots being done with a
uC. That almost made me sick.


I used to be opposed to using a tiny microcontroller to do something "simple"
like flash an LED in various patterns depending on, say, 2 or 3 inputs, but
these days I think it's often justifiable if only from the standpoint of
scheduling: Cranking out all of 20 lines of C code is a lot faster than doing
the "discrete" design. With tiny microcontrollers going for ~$0.25 these days
in 100-ish quantities, it's pretty hard to justify even a day spend on a
discrete design for low quantities (<1000). Of course, for the high-quantity
stuff you do, I realize it's a different game.

Right now I'm considering using one of those quarter microcontrollers just to
generate the output signals for a standard H-bridge creating a stock "modified
square wave" output (essentially 3 different states) based on an "enable"
signal and a check of whether or not the rail voltage is high enough (using
the cheesy built-in ADC on the micro). Any suggestions on a "better"
approach?

It probably could be done for very high volume where it is warranted to
put a lot of engineering hours into it. Some of my designs are like
that. You sit there agonizing over a BSS123 because it costs about two
pennies more than a BJT. Then after half an hour a light bulb goes off
and you figure you can really do it with a BJT plus a resistor, saving
15 milli-Dollars per unit.

Which uC were you thinking about? 25c with ADC on board sounds like a
good deal. But only if it doesn't need a crystal or resonator.

Of course, Jim would say that he just puts it all into Spice and then
has the chips made.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Joel Kolstad]

Hi Joerg,

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:TkdNi.3402$oA2.2153@nlpi068.nbdc.sbc.com...

Which uC were you thinking about? 25c with ADC on board sounds like a good
deal. But only if it doesn't need a crystal or resonator.

The Atmel ATtiny11. When I say "cheeseball ADC" I'm referring to its on-board
analog comparator. It does have an internal RC oscillator, although it isn't
too horribly accurate (for an "accurate" RC oscillator they'd like to sell you
an ATtiny12, which of course costs more).

DigiKey gets $0.31 in quantities of 100 or more, so I *think* it's fair to
call it a "quarter microcontroller" in large quantities.

---Joel

 

[Vladimir Vassilevsky]

Joerg wrote:

You sit there agonizing over a BSS123 because it costs about two
pennies more than a BJT. Then after half an hour a light bulb goes off
and you figure you can really do it with a BJT plus a resistor, saving
15 milli-Dollars per unit.

2N7002 would probably save two pennies plus the stuffing cost of the
resistor.

VLV

 

[Joerg]

Vladimir Vassilevsky wrote:

Joerg wrote:


You sit there agonizing over a BSS123 because it costs about two
pennies more than a BJT. Then after half an hour a light bulb goes off
and you figure you can really do it with a BJT plus a resistor, saving
15 milli-Dollars per unit.


2N7002 would probably save two pennies plus the stuffing cost of the
resistor.

Usually not, it's around 3.5c, plus Vds doesn't suffice for many apps.
Stuffing costs are extremely low (China) and I always calculate the total.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

Joel Kolstad wrote:

Hi Joerg,

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:TkdNi.3402$oA2.2153@nlpi068.nbdc.sbc.com...

Which uC were you thinking about? 25c with ADC on board sounds like a good
deal. But only if it doesn't need a crystal or resonator.


The Atmel ATtiny11. When I say "cheeseball ADC" I'm referring to its on-board
analog comparator. It does have an internal RC oscillator, although it isn't
too horribly accurate (for an "accurate" RC oscillator they'd like to sell you
an ATtiny12, which of course costs more).

DigiKey gets $0.31 in quantities of 100 or more, so I *think* it's fair to
call it a "quarter microcontroller" in large quantities.

Ok, that counts as a quarter-part. But your modified sine frequency
might be a bit drifty. The one thing I (and my client's purchasers)
would not be so fond of is the fact that it's single-sourced.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Joel Kolstad]

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:JueNi.390$Pv2.83@newssvr23.news.prodigy.net...

Ok, that counts as a quarter-part. But your modified sine frequency might be
a bit drifty.

In my case it's hard-wired to a particular load that, at least from my
testing, doesn't seem to care.

Certainly one couldn't get away with building a general purpose inverter
without adding the $$$ crystal (or similar), though.

Hmm... I wonder what those $20 100W 12V->120V inverters use for a timebase? I
would expect them to be at least +/-5% accurate...

Some years ago (wow, probably about 15 now!) I had one of those cheap
inverters self-destruct... smoke pouring out, the entire bit; I suspect it may
well have started on fire if I hadn't been there to unplug it from the
cigarette lighter.

The one thing I (and my client's purchasers) would not be so fond of is the
fact that it's single-sourced.

Yeah, understandable.

---Joel

 

[krw]

In article <vhp8g3h9vtg91quoml6einrhkjh6mtqac1@4ax.com>,
jjlarkin@highNOTlandTHIStechnologyPART.com says...

On Wed, 3 Oct 2007 22:51:41 -0400, krw <krw@att.bizzzz> wrote:


I'm hoping I don't have to go through that again. What a pain,
daytime TV sucks, and not cheap either.

From what JohnL has said, he likely did the guts of the MRI machine I
was inspecting the innards of, too. Good company here! ;-)

No, I only do the gradient drivers for small-bore stuff, like 4" and
8" machines for imaging lab animals, and even smaller micro-imaging
systems. Those can be dinky but fast high-precision, 100 amp, 170-volt
(17 KW per axis!) drivers. The full-body stuff is much higher power,
always switchers (Copley is pretty good) and slower and a lot noisier.

Ah, I misunderstood you incorrectly.

I had my head MRI'd earlier this year (spare me the predictable
witticisms, guys)

My mother had a head MRI many years ago to try to find the cause of
some balance problems. During the consultation after, the doctor
told her that they didn't find anything, thought for a second, and
started apologizing profusely after she burst out laughing.

and it was noisy and mostly boring. It's the
gradient coils that make the noise.

I was wondering. It sounded like something moving around in there.
They warned me of the noise and offered ear-plugs. It wasn't *that*
bad. It *was* close, though.

--
Keith

 

[krw]

In article <1c9Ni.5861$6p6.4832@newssvr25.news.prodigy.net>,
notthisjoergsch@removethispacbell.net says...

miso@sushi.com wrote:

[...]


Oh yeah, the lack of soldering skills. That would require the student
to have actually built something. These younguns just know how to
program. You've seen the posts where a pic uP is the solutions to any
task, not a state machine comprised of memory elements and
combinational logic.

For the vast majority of applications, a uC is the right solution,
certainly over the discrete implementation you suggest.


I certainly would not say "vast". Many times I have pondered the use of
a uC in a design only to come to the conclusion that 50c is still too
expensive. It's amazing how cheap "poor-man's logic" can be.

You must have very small problems. I've never had one of those. ;-)
The project I'm working on now is in a pair of XC2V-6000s (my part)
and a couple of huge (don't know the model) Virtex-4s. I think I'm
only going to fill 5% of it (and fewer I/Os), but...

Seriously, I think you're only looking at a small niche. I do believe
uCs are the right solution for the vast majority. ...even the
birthday card I got has one in it. :-/

The problem with using a uP in such projects is if you are designing a
chip, you need to know how to do it with gates as often that is the
smallest and lowest power solution. The ability to hand craft logic is
disappearing rapidly, but is very much needed in mixed mode chips
which are not done on fine geometry processes.


It sure is disappearing, just like analog skills seem to be almost gone
in fresh grad.

I've ripped out many uC solutions for reliability and cost reasons. Same
for PALs/GALs because that's the era back in the late 80's when
mixed/discrete design skills began to tank. The topper was a system
where I ripped out so many (plus went to 100% AC terminations) that the
power supply kept tripping off and one of them blew. We had managed to
get underneath the minimum load. IIRC the PALs was guzzling 30-40mA just
sitting there. Each.

What's the old saying? If they only learn how to use a hammer every
problem will begin to look like a nail. Heck, I've seen one-shots being
done with a uC. That almost made me sick.

One shots make me sick too. ;-)/2

--
Keith

 

[Jim Thompson]

On Wed, 03 Oct 2007 21:10:36 -0700, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

[snip]

I had my head MRI'd earlier this year (spare me the predictable
witticisms, guys) and it was noisy and mostly boring. It's the
gradient coils that make the noise.

John

Actually I did also, about two years ago... trying to find a cause for
chronic sinus infections. Nothing found there ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

America: Land of the Free, Because of the Brave

 

[Don Bowey]

On 10/4/07 6:13 PM, in article ur3bg39pmoj8c8v7afspmestee4porjpho@4ax.com,
"Jim Thompson" <To-Email-Use-The-Envelope-Icon@My-Web-Site.com> wrote:

On Wed, 03 Oct 2007 21:10:36 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

[snip]

I had my head MRI'd earlier this year (spare me the predictable
witticisms, guys) and it was noisy and mostly boring. It's the
gradient coils that make the noise.

John

Actually I did also, about two years ago... trying to find a cause for
chronic sinus infections. Nothing found there ;-)

...Jim Thompson

Of course not.... People of your party have little substance.

 

On Oct 3, 11:59 pm, nimo <minagadelsa...@hotmail.com> wrote:

thank u very much u just raised my hope i thought that our
universities in egypt are so what too say primative in practical and
specially
in analog design
but for a change
i think analog/mixed design is reviving again it's the chance
i my self am designing a VGA for a communication channel AFE
well it may be a sand particale compared to the whole design but those
digital geeks using VHDL and denying the ability of human
to get new designs of his own can't do it never because it just need
the talent and someone loves Transistors
Analog era will never end it's the backbone of every thing even
digital they just need the finnest as they need less numbers these
days
any way if someone knows a college in US that will help an egyption
student like me to get his master at in analog/mixed design
(considering i can't be a penny)
i'll be glad of him

yours
gad elsayed,mina

Generally if you get a BS degree, you can get a job at a university
while you get your MS. Of course, it takes longer to graduate that
way, but the schools like the slave labor. I'm having a hard time
thinking back to my undergraduate years of a TA that wasn't from a
foreign country. You have to realize that the time you spend getting a
MSEE is time you weren't earning real money, and I'm not entirely sure
the advanced degree buys you much. I got my MSEE going part time and
on my employers dime.

 

[Joerg]

Joel Kolstad wrote:

"Joerg" <notthisjoergsch@removethispacbell.net> wrote in message
news:JueNi.390$Pv2.83@newssvr23.news.prodigy.net...

Ok, that counts as a quarter-part. But your modified sine frequency might be
a bit drifty.


In my case it's hard-wired to a particular load that, at least from my
testing, doesn't seem to care.

That's the kind of load I like best. Why is it that the loads I am
confronted with are always fussy?

Certainly one couldn't get away with building a general purpose inverter
without adding the $$$ crystal (or similar), though.

Hmm... I wonder what those $20 100W 12V->120V inverters use for a timebase? I
would expect them to be at least +/-5% accurate...

Most likely it's ye olde PWM chip and RC. At $20 retail I wouldn't even
expect a ceramic resonator in there. Heck, they usually don't even have
a choke after the rectifier which could possibly explain your fiery
situation below.

Some years ago (wow, probably about 15 now!) I had one of those cheap
inverters self-destruct... smoke pouring out, the entire bit; I suspect it may
well have started on fire if I hadn't been there to unplug it from the
cigarette lighter.


The one thing I (and my client's purchasers) would not be so fond of is the
fact that it's single-sourced.


Yeah, understandable.

---Joel

--
Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

krw wrote:

In article <1c9Ni.5861$6p6.4832@newssvr25.news.prodigy.net>,
notthisjoergsch@removethispacbell.net says...

miso@sushi.com wrote:

[...]


Oh yeah, the lack of soldering skills. That would require the student
to have actually built something. These younguns just know how to
program. You've seen the posts where a pic uP is the solutions to any
task, not a state machine comprised of memory elements and
combinational logic.

For the vast majority of applications, a uC is the right solution,
certainly over the discrete implementation you suggest.


I certainly would not say "vast". Many times I have pondered the use of
a uC in a design only to come to the conclusion that 50c is still too
expensive. It's amazing how cheap "poor-man's logic" can be.


You must have very small problems. I've never had one of those. ;-)
The project I'm working on now is in a pair of XC2V-6000s (my part)
and a couple of huge (don't know the model) Virtex-4s. I think I'm
only going to fill 5% of it (and fewer I/Os), but...

Well, if the sky is the limit in terms of cost ...

Seriously, I think you're only looking at a small niche. I do believe
uCs are the right solution for the vast majority. ...even the
birthday card I got has one in it. :-/

Ok, Chinese 4-bit micros exempted. Those can be had under 10c. But the
programmers should master Mandarin.

The problem with using a uP in such projects is if you are designing a
chip, you need to know how to do it with gates as often that is the
smallest and lowest power solution. The ability to hand craft logic is
disappearing rapidly, but is very much needed in mixed mode chips
which are not done on fine geometry processes.


It sure is disappearing, just like analog skills seem to be almost gone
in fresh grad.

I've ripped out many uC solutions for reliability and cost reasons. Same
for PALs/GALs because that's the era back in the late 80's when
mixed/discrete design skills began to tank. The topper was a system
where I ripped out so many (plus went to 100% AC terminations) that the
power supply kept tripping off and one of them blew. We had managed to
get underneath the minimum load. IIRC the PALs was guzzling 30-40mA just
sitting there. Each.

What's the old saying? If they only learn how to use a hammer every
problem will begin to look like a nail. Heck, I've seen one-shots being
done with a uC. That almost made me sick.


One shots make me sick too. ;-)/2

I don't use them, I use 74HC14 or CD40106 sections. Even as PWM chips.

--
Regards, Joerg

http://www.analogconsultants.com

 

[Jim Thompson]

On Thu, 04 Oct 2007 21:15:11 -0700, Don Bowey <dbowey@comcast.net>
wrote:

On 10/4/07 6:13 PM, in article ur3bg39pmoj8c8v7afspmestee4porjpho@4ax.com,
"Jim Thompson" <To-Email-Use-The-Envelope-Icon@My-Web-Site.com> wrote:

On Wed, 03 Oct 2007 21:10:36 -0700, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

[snip]

I had my head MRI'd earlier this year (spare me the predictable
witticisms, guys) and it was noisy and mostly boring. It's the
gradient coils that make the noise.

John

Actually I did also, about two years ago... trying to find a cause for
chronic sinus infections. Nothing found there ;-)

...Jim Thompson

Of course not.... People of your party have little substance.

Why are leftist weenies such jerks?

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

America: Land of the Free, Because of the Brave