Most interesting recent products / improvements?
Davej
the most interesting recent product developments? Apparently LED's
have become quite a bit more efficient. Thanks.
John Larkin
wrote:
>I've been away from board-level design for about seven years. What are
>the most interesting recent product developments? Apparently LED's
>have become quite a bit more efficient. Thanks.
FPGAs with enormous data crunching capability.
Really fast opamps.
Super-low drift chopper opamps.
More module-type things, like Linear's switchers
Silicon oscillators and delay lines
Cheap precision thinfilm resistors
Polymer aluminum capacitors
High-value ceramic caps, like 22uF and 100uF
Lots of cool RF stuff: plls, rf detectors, amps
Really fast ecl/cml logic, 6 GHz range.
Outrageous fast cheap microprocessors, like the ARM stuff.
Everything going to BGA and leadless/chip-scale packages
Everything getting smaller
Digital pots
Lots of transistors and fets that cost 3 cents
Gallium Nitride and Silicon Carbide and SiGe
Tiny Logic
Super-precision delta-sigma ADCs
Cheap SAR ADCs
Cheap video/pipeline ADCs
Too many voltage regulators
John
krw
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>wrote:
>
>>I've been away from board-level design for about seven years. What are
>>the most interesting recent product developments? Apparently LED's
>>have become quite a bit more efficient. Thanks.
>
>FPGAs with enormous data crunching capability.
Dirt cheap FPGAs.
>Really fast opamps.
>Everything going to BGA and leadless/chip-scale packages
Hurts the small guy but great if you can afford it.
>Everything getting smaller
Hurts the small guy...
>Lots of transistors and fets that cost 3 cents
Free resistors.
>Gallium Nitride and Silicon Carbide and SiGe
>
>Tiny Logic
>
>Super-precision delta-sigma ADCs
But no DACs to match. :-(
>Too many voltage regulators
Boy howdy!
Allan Herriman
>
> Too many voltage regulators
>
hahaha! Often a third of the area of my (ostensibly "digital") board
will be DC/DC converters or linear regs just to generate the various
rails needed for the various digital ICs.
Regards,
Allan
John Larkin
<allanh...@hotmail.com> wrote:
Or sometimes half. A simple waveform generator needs three voltages
for the FPGA, four for the analogs, one for miscellaneous logic.
We're using the LTC uModules a lot lately, everything in one small
package.
Two here, one +12 to +5, one +12 to -5, LDOs after that.
ftp://jjlarkin.lmi.net/J245_R5.gif
John
Jan Panteltje
<jjla...@highNOTlandTHIStechnologyPART.com> wrote in
<ujtb959ceppe7564b...@4ax.com>:
>Two here, one +12 to +5, one +12 to -5, LDOs after that.
>
>ftp://jjlarkin.lmi.net/J245_R5.gif
Looks like 'AMP" and 'WIDTH' connectors are too close together to allow both inserted?
John Larkin
Those are right-angle trimpots, as is the threshold pot on the other
side.
ftp://jjlarkin.lmi.net/J240_top.JPG
ftp://jjlarkin.lmi.net/J240_box.jpg
John
Jan Panteltje
<jjla...@highNOTlandTHIStechnologyPART.com> wrote in
<kl4d959tfm3okn4rh...@4ax.com>:
OIC.
Hey, those are nice trimpost, at least you cannot push them over from the outside :-)
Spehro Pefhany
<pNaonSt...@yahoo.com> wrote:
>On a sunny day (Thu, 27 Aug 2009 07:17:35 -0700) it happened John Larkin
><jjla...@highNOTlandTHIStechnologyPART.com> wrote in
><kl4d959tfm3okn4rh...@4ax.com>:
>
>>On Thu, 27 Aug 2009 11:20:12 GMT, Jan Panteltje
>><pNaonSt...@yahoo.com> wrote:
>>
>>>On a sunny day (Wed, 26 Aug 2009 20:05:23 -0700) it happened John Larkin
>>><jjla...@highNOTlandTHIStechnologyPART.com> wrote in
>>><ujtb959ceppe7564b...@4ax.com>:
>>>
>>>>Two here, one +12 to +5, one +12 to -5, LDOs after that.
>>>>
>>>>ftp://jjlarkin.lmi.net/J245_R5.gif
>>>
>>>Looks like 'AMP" and 'WIDTH' connectors are too close together to allow both inserted?
>>
>>Those are right-angle trimpots, as is the threshold pot on the other
>>side.
>>
>>ftp://jjlarkin.lmi.net/J240_top.JPG
>>
>>ftp://jjlarkin.lmi.net/J240_box.jpg
>>
>>John
>
>OIC.
>Hey, those are nice trimpost, at least you cannot push them over from the outside :-)
Very nice ones, I think they are originally a Japanese design- Tokyo
Cosmos.
John Larkin
Those are Bourns 3329X. Kind of expensive - about $1.90 - but
mechanically nice for stuff like this. They have a big slot that's
easy to locate and keep a screwdriver in. I hate fishing around inside
a tiny hole for a tiny screwdriver slot that I have to turn 10 or 15
times. Grrrrr.
The SMA edge launches are from Shining Star... cost less than the
trimpots!
John
Joerg
> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>
>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>> wrote:
>>
>>> I've been away from board-level design for about seven years. What are
>>> the most interesting recent product developments? Apparently LED's
>>> have become quite a bit more efficient. Thanks.
>> FPGAs with enormous data crunching capability.
>
> Dirt cheap FPGAs.
>
Where? Which ones? Back in the 80's I regularly beat other guys when
doing stuff like address decoders. Their PALs and GALs cost over a buck
and burned tens of milliamps while mine cost 20-30 Cents and needed a
few tens of microamps. Have seen much of a change :-)
Ok, this is more CPLD domain but still, it's pricey.
>> Really fast opamps.
>
Texas THS-series, blazingly fast +/-12V amps. Most of the others can
only do +5V, often not too practical like when driving huge FETs that
need 10V swings.
>> Everything going to BGA and leadless/chip-scale packages
>
> Hurts the small guy but great if you can afford it.
>
>> Everything getting smaller
>
> Hurts the small guy...
>
Oh yeah! You can't even find local techs who can do a prototype assembly
on a one-day basis. I am in that pickle right now, need someone in the
area between Sacramento and Lake Tahoe who can do TSSOP and 0603 and has
a Metcal and stuff. Even the local LTC guys didn't know anyone :-(
>> Lots of transistors and fets that cost 3 cents
>
> Free resistors.
>
Free?
>> Gallium Nitride and Silicon Carbide and SiGe
>>
>> Tiny Logic
>>
>> Super-precision delta-sigma ADCs
>
> But no DACs to match. :-(
>
>> Too many voltage regulators
>
> Boy howdy!
But nearly all LDOs and thus off limits in this here lab,
--
Regards, Joerg
http://www.analogconsultants.com/
"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
John Larkin
wrote:
>krw wrote:
>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>
>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>>> wrote:
>>>
>>>> I've been away from board-level design for about seven years. What are
>>>> the most interesting recent product developments? Apparently LED's
>>>> have become quite a bit more efficient. Thanks.
>>> FPGAs with enormous data crunching capability.
>>
>> Dirt cheap FPGAs.
>>
>
>Where? Which ones? Back in the 80's I regularly beat other guys when
>doing stuff like address decoders. Their PALs and GALs cost over a buck
>and burned tens of milliamps while mine cost 20-30 Cents and needed a
>few tens of microamps. Have seen much of a change :-)
>
>Ok, this is more CPLD domain but still, it's pricey.
>
>
>>> Really fast opamps.
>>
>
>Texas THS-series, blazingly fast +/-12V amps. Most of the others can
>only do +5V, often not too practical like when driving huge FETs that
>need 10V swings.
Avoid THS3062. It's fatally messed up.
John
Joerg
I am not much of a CFB fan anyhow but what did they botch on this one?
Spehro Pefhany
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>On Thu, 27 Aug 2009 12:51:51 -0700, Joerg <inv...@invalid.invalid>
>wrote:
>
>>krw wrote:
>>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>>> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>>
>>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>>>> wrote:
>>>>
>>>>> I've been away from board-level design for about seven years. What are
>>>>> the most interesting recent product developments? Apparently LED's
>>>>> have become quite a bit more efficient. Thanks.
>>>> FPGAs with enormous data crunching capability.
>>>
>>> Dirt cheap FPGAs.
>>>
>>
>>Where? Which ones? Back in the 80's I regularly beat other guys when
>>doing stuff like address decoders. Their PALs and GALs cost over a buck
>>and burned tens of milliamps while mine cost 20-30 Cents and needed a
>>few tens of microamps. Have seen much of a change :-)
>>
>>Ok, this is more CPLD domain but still, it's pricey.
>>
>>
>>>> Really fast opamps.
>>>
>>
>>Texas THS-series, blazingly fast +/-12V amps. Most of the others can
>>only do +5V, often not too practical like when driving huge FETs that
>>need 10V swings.
>
>Avoid THS3062. It's fatally messed up.
>
>John
>
Speaking of fatally screwed up, I just had to deal with an ADG3301
that a certain d*pst*k spec'd. Avoid. The sad thing is that none of
them had to be bi-directional, it was just the latest and greatest I
guess.
Nemo
>the most interesting recent product developments? Apparently LED's
>have become quite a bit more efficient. Thanks.
In addition to that comprehensive component summary by John I'd add...
Lithium ion batteries. OLED displays. Blue LED's. Memories have
continued to get bigger of course, and nonvolatile ones are improving,
which is changing how folk write code.
POWER DISSIPATION: New (surface mount) packaging topologies, often
involving pads underneath the component (tricky to solder) allow small
packages to dissipate much more heat. Combined with much more efficient
switchmode IC's working at say, 2MHz instead of 100kHz, heatsinks are
getting smaller or disappearing altogether on many boards.
External universal-input mains power supplies are so cheap, it is
unusual to design your own mains supply these days.
EMC: has become a much more important area due to regulatory
impositions: you need to design stuff with good immunity and there are
plenty of new, small filters to help.
ROHS: pointless Euro-regulations called ROHS forced everyone to move to
lead-free solder which doesn't wet as well as leaded solder. Doesn't
help when soldering those new fine pitch components. Sorry, you said
interesting, not important!
PCB manufacturing has become better: even small suppliers can do finer
pitch multilayer stuff than was common 7 years ago. Many "pool"
prototype PCB's onto one biscuit, so you can get a handful of prototype
4 layer PCB's in say 3 days for $100 - $200.
Far Eastern component manufacturers have extended their distribution
into the West. Very cheap commodity linear components, though I tend to
steer clear of their IC's as the product life cycle is often only a
couple of years.
Lots of Chinese piracy driving us to build antipiracy features into new
products.
3D rapid prototyping ("printing") allows you to get accurate plastic
models of your proposed fancy moulded enclosure in a few days for a few
hundred $, see the component clashes, and iterate complex designs if
needed. Yeah, you could use CAD, but most people find a physical thing
in their hands is a much simpler way of discussing issues.
--
Nemo
John Larkin
What's a d*pst*k?
I never really trusted those automatic bidirectional things.
John
John Larkin
wrote:
If you drive it to output a healthy amplitude sine wave, say 20 volts
p-p at 8 MHz or so, way within specs, it latches up, draws huge supply
currents, goes nuts, and gets red hot. Burned a hole in me finger, it
did.
Amazingly, it recovers if you crank the signal amplitude down close to
zero. Me finger didn't.
After weeks of harassment, TI told us, essentially, "don't do that."
It is apparently suited to making ADSL waveforms and nothing else.
Their datasheet graphs all carefully avoid the voltage-frequency
danger zone; "large signal" is apparently 2 volts p-p, even when using
+-15 supplies.
The datasheet is seven years old. I wonder how many people have been
literally burned by this deliberate deception.
John
Phil Hobbs
An apparently useful part--what's wrong with it?
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
krw
wrote:
>krw wrote:
>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>
>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>>> wrote:
>>>
>>>> I've been away from board-level design for about seven years. What are
>>>> the most interesting recent product developments? Apparently LED's
>>>> have become quite a bit more efficient. Thanks.
>>> FPGAs with enormous data crunching capability.
>>
>> Dirt cheap FPGAs.
>>
>
>Where? Which ones? Back in the 80's I regularly beat other guys when
>doing stuff like address decoders. Their PALs and GALs cost over a buck
>and burned tens of milliamps while mine cost 20-30 Cents and needed a
>few tens of microamps. Have seen much of a change :-)
I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user
I/O, flash-based FPGA.
>Ok, this is more CPLD domain but still, it's pricey.
You have a different definition of "pricey" than anyone else on the
planet. ;-)
>>> Really fast opamps.
>>
>
>Texas THS-series, blazingly fast +/-12V amps. Most of the others can
>only do +5V, often not too practical like when driving huge FETs that
>need 10V swings.
>
>
>>> Everything going to BGA and leadless/chip-scale packages
>>
>> Hurts the small guy but great if you can afford it.
>>
>>> Everything getting smaller
>>
>> Hurts the small guy...
>>
>
>Oh yeah! You can't even find local techs who can do a prototype assembly
>on a one-day basis. I am in that pickle right now, need someone in the
>area between Sacramento and Lake Tahoe who can do TSSOP and 0603 and has
>a Metcal and stuff. Even the local LTC guys didn't know anyone :-(
I have to use the more expensive Actel parts because their cheap stuff
only comes in micro-BGA. We don't want to do anything smaller than
.8mm. They're .4mm and .5mm.
>
>>> Lots of transistors and fets that cost 3 cents
>>
>> Free resistors.
>>
>
>Free?
>
Less than the cost of putting them there.
>>> Gallium Nitride and Silicon Carbide and SiGe
>>>
>>> Tiny Logic
>>>
>>> Super-precision delta-sigma ADCs
>>
>> But no DACs to match. :-(
>>
>>> Too many voltage regulators
>>
>> Boy howdy!
>
>
>But nearly all LDOs and thus off limits in this here lab,
Doesn't matter what they are, there are too many. Lets see, our base
unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
(times three), +12V (times two), -12V, +19V, and various references
(and I'm sure I've forgotten at least one). That's something like 15
supplies totaling 12W.
Joerg
Oh, bad news. At least they should have fessed up to this and also
update the datasheet.
I had a similar comeuppance with a TPS-something linear regulator which
I'll never again use. When the input power came up too fast which it had
to for that client ... tsk ... phut ... *POP* ... nice li'l crater in it
and definitely not a good thing to test without goggles.
So I contacted tech support. After some waffling I wanted the SPICE
file. "Can't have that". So I asked them to throw my fairly simple
circuit onto their sim and try. "Nope, can't do that either". Hmm ...
Joerg
> On Thu, 27 Aug 2009 12:51:51 -0700, Joerg <inv...@invalid.invalid>
> wrote:
>
>> krw wrote:
>>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>>> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>>
>>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>>>> wrote:
>>>>
>>>>> I've been away from board-level design for about seven years. What are
>>>>> the most interesting recent product developments? Apparently LED's
>>>>> have become quite a bit more efficient. Thanks.
>>>> FPGAs with enormous data crunching capability.
>>> Dirt cheap FPGAs.
>>>
>> Where? Which ones? Back in the 80's I regularly beat other guys when
>> doing stuff like address decoders. Their PALs and GALs cost over a buck
>> and burned tens of milliamps while mine cost 20-30 Cents and needed a
>> few tens of microamps. Have seen much of a change :-)
>
> I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user
> I/O, flash-based FPGA.
>
If I'd put a $2.90 part on a design without a serious reason such as
"the world would stop spinning if we didn't", many of my client would
have my head examined :-)
>> Ok, this is more CPLD domain but still, it's pricey.
>
> You have a different definition of "pricey" than anyone else on the
> planet. ;-)
>
Sure looks like it ...
>>>> Really fast opamps.
>> Texas THS-series, blazingly fast +/-12V amps. Most of the others can
>> only do +5V, often not too practical like when driving huge FETs that
>> need 10V swings.
>>
>>
>>>> Everything going to BGA and leadless/chip-scale packages
>>> Hurts the small guy but great if you can afford it.
>>>
>>>> Everything getting smaller
>>> Hurts the small guy...
>>>
>> Oh yeah! You can't even find local techs who can do a prototype assembly
>> on a one-day basis. I am in that pickle right now, need someone in the
>> area between Sacramento and Lake Tahoe who can do TSSOP and 0603 and has
>> a Metcal and stuff. Even the local LTC guys didn't know anyone :-(
>
> I have to use the more expensive Actel parts because their cheap stuff
> only comes in micro-BGA. We don't want to do anything smaller than
> .8mm. They're .4mm and .5mm.
I wouldn't either, and I really don't like BGAs. If it doesn't come in
QFP or something then I typically won't design it in. Seen too much
grief with BGA.
>>>> Lots of transistors and fets that cost 3 cents
>>> Free resistors.
>>>
>> Free?
>>
> Less than the cost of putting them there.
>
I guess you assume domestic production. That's expensive, even worse in
Europe. But in China, very different thing. Many folks don't know, there
was even one guy here in the NG who just refused to believe that 5%
resistors would make sense anymore. They do, seriously.
>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>
>>>> Tiny Logic
>>>>
>>>> Super-precision delta-sigma ADCs
>>> But no DACs to match. :-(
>>>
>>>> Too many voltage regulators
>>> Boy howdy!
>>
>> But nearly all LDOs and thus off limits in this here lab,
>
> Doesn't matter what they are, there are too many. Lets see, our base
> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
> (times three), +12V (times two), -12V, +19V, and various references
> (and I'm sure I've forgotten at least one). That's something like 15
> supplies totaling 12W.
>
And if you don't sequence them meticulously a $100 FPGA blows its lid?
Oh man, I am glad I am an analog dude :-)
Joerg
>> I've been away from board-level design for about seven years. What are
>> the most interesting recent product developments? Apparently LED's
>> have become quite a bit more efficient. Thanks.
>
> In addition to that comprehensive component summary by John I'd add...
>
> Lithium ion batteries. OLED displays. Blue LED's. Memories have
> continued to get bigger of course, and nonvolatile ones are improving,
> which is changing how folk write code.
>
Meanin' lots of sloppy coding and bloat :-(
> POWER DISSIPATION: New (surface mount) packaging topologies, often
> involving pads underneath the component (tricky to solder) allow small
> packages to dissipate much more heat. Combined with much more efficient
> switchmode IC's working at say, 2MHz instead of 100kHz, heatsinks are
> getting smaller or disappearing altogether on many boards.
>
> External universal-input mains power supplies are so cheap, it is
> unusual to design your own mains supply these days.
>
And some ultra-cheap ones lay on one spectacular pyrotechnic show when
the super-cheap electrolytics in there give up.
> EMC: has become a much more important area due to regulatory
> impositions: you need to design stuff with good immunity and there are
> plenty of new, small filters to help.
>
> ROHS: pointless Euro-regulations called ROHS forced everyone to move to
> lead-free solder which doesn't wet as well as leaded solder. Doesn't
> help when soldering those new fine pitch components. Sorry, you said
> interesting, not important!
>
Don't get me started. Now they banned lightbulbs, starting Tuesday. I
guess that'll save 0.0001% of carbon-dioxide over there. Pathetic.
> PCB manufacturing has become better: even small suppliers can do finer
> pitch multilayer stuff than was common 7 years ago. Many "pool"
> prototype PCB's onto one biscuit, so you can get a handful of prototype
> 4 layer PCB's in say 3 days for $100 - $200.
>
Where? My last quote was almost $2k, two (!).
> Far Eastern component manufacturers have extended their distribution
> into the West. Very cheap commodity linear components, though I tend to
> steer clear of their IC's as the product life cycle is often only a
> couple of years.
>
Rohm and NJR are pretty good though. The English in the datasheets can
be a bit rocky though, occasionally to the point where stuff becomes
ambiguous.
Frank Buss
> Nemo wrote:
>
>> PCB manufacturing has become better: even small suppliers can do finer
>> pitch multilayer stuff than was common 7 years ago. Many "pool"
>> prototype PCB's onto one biscuit, so you can get a handful of prototype
>> 4 layer PCB's in say 3 days for $100 - $200.
>>
>
> Where? My last quote was almost $2k, two (!).
I don't know, if he means $100-$200 per PCB, or maybe small PCBs?
pcb-pool.de: 5 eurocards (160x100), 4 layers, each one for 84 Euro, but in
8 days, 5 days is more than twice as expensive. But you get a free laser
stencil, too and the quality is always very good. They accept many formats,
without the need for you to convert it to gerber files. I've downloaded
their Eagle policy file for electrical rule check and there was nearly
never any questions. Only once, when I submitted the schematic file instead
of the board file at a sleepy night :-)
And using their reflow kit is nice for prototypes, because it is faster and
looks better than hand soldering.
--
Frank Buss, f...@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
Joerg
> Joerg wrote:
>
>> Nemo wrote:
>>
>>> PCB manufacturing has become better: even small suppliers can do finer
>>> pitch multilayer stuff than was common 7 years ago. Many "pool"
>>> prototype PCB's onto one biscuit, so you can get a handful of prototype
>>> 4 layer PCB's in say 3 days for $100 - $200.
>>>
>> Where? My last quote was almost $2k, two (!).
>
> I don't know, if he means $100-$200 per PCB, or maybe small PCBs?
> pcb-pool.de: 5 eurocards (160x100), 4 layers, each one for 84 Euro, but in
> 8 days, 5 days is more than twice as expensive. But you get a free laser
> stencil, too and the quality is always very good. They accept many formats,
> without the need for you to convert it to gerber files. I've downloaded
> their Eagle policy file for electrical rule check and there was nearly
> never any questions. Only once, when I submitted the schematic file instead
> of the board file at a sleepy night :-)
>
Oh, then they must have adjusted their prices. When I checked one or two
years ago they were expensive compared to the places over here. Mostly I
used this place:
> And using their reflow kit is nice for prototypes, because it is faster and
> looks better than hand soldering.
>
Interesting. Especially since I just finished a design and (so far) have
no idea who can solder it. It's mostly 0603 and some TSSOP pitch stuff.
In case anyone else wonder what this reflow kit is:
http://www.reflow-kit.de/rkuk/order_product_details.html?p=13
Hey, I guess when done we could bake a pizza for the guys in the oven :-)
John Larkin
wrote:
Is TI the new Maxim?
John
Jim Thompson
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
These threads concerning bad products and worse "support" should be
copied to the appropriate management.
Perhaps a new usenet group: sci.electronis.dud-products ?:-)
Nothing has come of it yet, but I am recently seeing inquiries of the
form, "Can you design a replacement for this part? But fix..."
...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.
Joerg
No, then you could not even buy a reel :-)
Other than that TI has been fairly good to me. And the THS4021 opamps
work like a champ. Two years and counting.
Joerg
Usually gets ignored :-(
> Perhaps a new usenet group: sci.electronis.dud-products ?:-)
>
> Nothing has come of it yet, but I am recently seeing inquiries of the
> form, "Can you design a replacement for this part? But fix..."
>
I receive the same requests but for circuit board designs. Roughly 50%
of my work consist of redesign of someone else's design. Keeps food on
the table so no complaints here :-)
Spehro Pefhany
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
That springy thing you use to check your oil level. Like a d**ch*
nozzle, but less useful.
>
>I never really trusted those automatic bidirectional things.
>
>John
All those unidirectional jobs with a 245 in the part number work and
are cheap. I guess that's a bad thing.
Spehro Pefhany
Hi, Phil:-
Check out page 16 of 20, top right paragraph:
http://www.analog.com/static/imported-files/Data_Sheets/ADG3301.pdf
John Larkin
Yikes! What a bag-o-worms.
John
Jon Kirwan
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
Interesting drive requirements. So was this really just some 50-75
ohm [coax] level converter for some prior client?
It does say on the front page that one of the applications is:
"Portable POS systems." And we know what POS may stand for.
Jon
Phil Hobbs
krw
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
Nah, TI has cost us a fortune. Maxim hasn't. ;-)
krw
wrote:
>krw wrote:
>> On Thu, 27 Aug 2009 12:51:51 -0700, Joerg <inv...@invalid.invalid>
>> wrote:
>>
>>> krw wrote:
>>>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>>>> <jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>>>>
>>>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>>>>> wrote:
>>>>>
>>>>>> I've been away from board-level design for about seven years. What are
>>>>>> the most interesting recent product developments? Apparently LED's
>>>>>> have become quite a bit more efficient. Thanks.
>>>>> FPGAs with enormous data crunching capability.
>>>> Dirt cheap FPGAs.
>>>>
>>> Where? Which ones? Back in the 80's I regularly beat other guys when
>>> doing stuff like address decoders. Their PALs and GALs cost over a buck
>>> and burned tens of milliamps while mine cost 20-30 Cents and needed a
>>> few tens of microamps. Have seen much of a change :-)
>>
>> I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user
>> I/O, flash-based FPGA.
>>
>
>If I'd put a $2.90 part on a design without a serious reason such as
>"the world would stop spinning if we didn't", many of my client would
>have my head examined :-)
Even if it replaced six parts worth $12?
>>> Ok, this is more CPLD domain but still, it's pricey.
>>
>> You have a different definition of "pricey" than anyone else on the
>> planet. ;-)
>>
>
>Sure looks like it ...
I know I won't get any argument about this here on SED. ;-)
>>>>> Really fast opamps.
>>> Texas THS-series, blazingly fast +/-12V amps. Most of the others can
>>> only do +5V, often not too practical like when driving huge FETs that
>>> need 10V swings.
>>>
>>>
>>>>> Everything going to BGA and leadless/chip-scale packages
>>>> Hurts the small guy but great if you can afford it.
>>>>
>>>>> Everything getting smaller
>>>> Hurts the small guy...
>>>>
>>> Oh yeah! You can't even find local techs who can do a prototype assembly
>>> on a one-day basis. I am in that pickle right now, need someone in the
>>> area between Sacramento and Lake Tahoe who can do TSSOP and 0603 and has
>>> a Metcal and stuff. Even the local LTC guys didn't know anyone :-(
>>
>> I have to use the more expensive Actel parts because their cheap stuff
>> only comes in micro-BGA. We don't want to do anything smaller than
>> .8mm. They're .4mm and .5mm.
>
>
>I wouldn't either, and I really don't like BGAs. If it doesn't come in
>QFP or something then I typically won't design it in. Seen too much
>grief with BGA.
BGAs haven't caused us *too* much grief. QFNs are the ones that drive
us nuts. The larger Actel parts ($2.90, above) are in QFP-100s.
>>>>> Lots of transistors and fets that cost 3 cents
>>>> Free resistors.
>>>>
>>> Free?
>>>
>> Less than the cost of putting them there.
>>
>
>I guess you assume domestic production.
We do our own.
>That's expensive, even worse in
>Europe. But in China, very different thing. Many folks don't know, there
>was even one guy here in the NG who just refused to believe that 5%
>resistors would make sense anymore. They do, seriously.
They don't for us - inventory. A reel of 5,000 1% resistors is less
than $10 from DigiKey, of all places.
>>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>>
>>>>> Tiny Logic
>>>>>
>>>>> Super-precision delta-sigma ADCs
>>>> But no DACs to match. :-(
>>>>
>>>>> Too many voltage regulators
>>>> Boy howdy!
>>>
>>> But nearly all LDOs and thus off limits in this here lab,
>>
>> Doesn't matter what they are, there are too many. Lets see, our base
>> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
>> (times three), +12V (times two), -12V, +19V, and various references
>> (and I'm sure I've forgotten at least one). That's something like 15
>> supplies totaling 12W.
>>
>
>And if you don't sequence them meticulously a $100 FPGA blows its lid?
Not so much anymore. Actel (and ALtera, IIRC) parts are guaranteed to
not need sequencing. Actel even supports hot plugging.
>Oh man, I am glad I am an analog dude :-)
No FPGAs in the above. The analog circuits take about half of the
above supplies.
AtTheEndofMyRope
wrote:
>I wouldn't either, and I really don't like BGAs. If it doesn't come in
>QFP or something then I typically won't design it in. Seen too much
>grief with BGA.
You should work with or examine the work of those that succeed at it
then.
Frank Buss
> I wouldn't either, and I really don't like BGAs. If it doesn't come in
> QFP or something then I typically won't design it in. Seen too much
> grief with BGA.
What are the problems with BGA? I think there is no modern complex consumer
device without BGAs.
Hardcore technicians can desolder and solder BGAs even by hand, without
expensive equipment:
http://www.youtube.com/watch?v=JB1InDsWCjQ
http://www.youtube.com/watch?v=YdN53F5YQwk
Maybe it's more complicated, if you use 0.4 mm pitch BGAs, and
package-on-package PCBs:
http://focus.ti.com/lit/an/spraav1b/spraav1b.pdf
I wonder how many stacks they can do? Building a full computer with
separate CPU, flash and RAM chips with the size of a sugar cube would be
nice :-)
TheM
> BGAs haven't caused us *too* much grief. QFNs are the ones that drive
> us nuts. The larger Actel parts ($2.90, above) are in QFP-100s.
Any good advices on QFNs?
M
Jan Panteltje
<f...@frank-buss.de> wrote in <x05rkysg0u4i.epa86i80w1zz$.d...@40tude.net>:
>Joerg wrote:
>
>> I wouldn't either, and I really don't like BGAs. If it doesn't come in
>> QFP or something then I typically won't design it in. Seen too much
>> grief with BGA.
>
>What are the problems with BGA? I think there is no modern complex consumer
>device without BGAs.
>
>Hardcore technicians can desolder and solder BGAs even by hand, without
>expensive equipment:
>
>http://www.youtube.com/watch?v=JB1InDsWCjQ
Amazing!
krw
wrote:
Yeah, if at all possible, don't use 'em. Unfortunately, it's not
possible. :-(
Joerg
I can't remember the last time where six digital parts would have cost
$12 in any of my designs. More like 10c times six :-)
[...]
>>>>>> Lots of transistors and fets that cost 3 cents
>>>>> Free resistors.
>>>>>
>>>> Free?
>>>>
>>> Less than the cost of putting them there.
>>>
>> I guess you assume domestic production.
>
> We do our own.
>
Yep, that gets expensive. Maybe you are better off in Alabama than we
are in California.
>> That's expensive, even worse in
>> Europe. But in China, very different thing. Many folks don't know, there
>> was even one guy here in the NG who just refused to believe that 5%
>> resistors would make sense anymore. They do, seriously.
>
> They don't for us - inventory. A reel of 5,000 1% resistors is less
> than $10 from DigiKey, of all places.
>
We sometimes calculate them in millicents :-)
>>>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>>>
>>>>>> Tiny Logic
>>>>>>
>>>>>> Super-precision delta-sigma ADCs
>>>>> But no DACs to match. :-(
>>>>>
>>>>>> Too many voltage regulators
>>>>> Boy howdy!
>>>> But nearly all LDOs and thus off limits in this here lab,
>>> Doesn't matter what they are, there are too many. Lets see, our base
>>> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
>>> (times three), +12V (times two), -12V, +19V, and various references
>>> (and I'm sure I've forgotten at least one). That's something like 15
>>> supplies totaling 12W.
>>>
>> And if you don't sequence them meticulously a $100 FPGA blows its lid?
>
> Not so much anymore. Actel (and ALtera, IIRC) parts are guaranteed to
> not need sequencing. Actel even supports hot plugging.
>
>> Oh man, I am glad I am an analog dude :-)
>
> No FPGAs in the above. The analog circuits take about half of the
> above supplies.
Maybe someone should have a chat with the analog guys ;-)
Joerg
Some noble names among the ones I've seen. Now there are even places
specializing in BGA failure repair:
http://www.laptoprepairspecialist.com/specialist_equipment
Quote "We fix 80% of BGA failures, ..."
Joerg
> Joerg wrote:
>
>> I wouldn't either, and I really don't like BGAs. If it doesn't come in
>> QFP or something then I typically won't design it in. Seen too much
>> grief with BGA.
>
> What are the problems with BGA? I think there is no modern complex consumer
> device without BGAs.
>
Small ones can be ok but large BGA cannot easily stand the flexing of a
circuit board. All it takes is one fall. BGA don't have any compliance
to speak of while the leads of TSSOP and others can bend a little
without popping out of the solder.
> Hardcore technicians can desolder and solder BGAs even by hand, without
> expensive equipment:
>
> http://www.youtube.com/watch?v=JB1InDsWCjQ
> http://www.youtube.com/watch?v=YdN53F5YQwk
>
That is amazing. To be honest, I wouldn't attempt this job. I'd send the
board away to a pro-shop.
> Maybe it's more complicated, if you use 0.4 mm pitch BGAs, and
> package-on-package PCBs:
>
> http://focus.ti.com/lit/an/spraav1b/spraav1b.pdf
>
> I wonder how many stacks they can do? Building a full computer with
> separate CPU, flash and RAM chips with the size of a sugar cube would be
> nice :-)
>
The top ones should hold well but the problem is the circuit board.
krw
wrote:
No wonder you don't think digital stuff is fun. You're too cheap to
buy interesting parts. Other than passives and transistors, there is
not much we buy that costs less than a buck.
>>>>>>> Lots of transistors and fets that cost 3 cents
>>>>>> Free resistors.
>>>>>>
>>>>> Free?
>>>>>
>>>> Less than the cost of putting them there.
>>>>
>>> I guess you assume domestic production.
>>
>> We do our own.
>>
>
>Yep, that gets expensive.
It's sometimes necessary. Our main business is so cyclical that we
need the control in-house manufacturing gives. Certainly control over
production trumps cost.
>Maybe you are better off in Alabama than we are in California.
In more ways than one. ;-)
>>> That's expensive, even worse in
>>> Europe. But in China, very different thing. Many folks don't know, there
>>> was even one guy here in the NG who just refused to believe that 5%
>>> resistors would make sense anymore. They do, seriously.
>>
>> They don't for us - inventory. A reel of 5,000 1% resistors is less
>> than $10 from DigiKey, of all places.
>>
>
>We sometimes calculate them in millicents :-)
Our product markups are better than what Larkin claims (and our
quantities higher), so millicents aren't the main issue. I just
started a cost reduction design pass on our products but I'm certainly
not looking at millicents. Our fixed costs drown that out. Molds are
*expensive*.
>
>>>>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>>>>
>>>>>>> Tiny Logic
>>>>>>>
>>>>>>> Super-precision delta-sigma ADCs
>>>>>> But no DACs to match. :-(
>>>>>>
>>>>>>> Too many voltage regulators
>>>>>> Boy howdy!
>>>>> But nearly all LDOs and thus off limits in this here lab,
>>>> Doesn't matter what they are, there are too many. Lets see, our base
>>>> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
>>>> (times three), +12V (times two), -12V, +19V, and various references
>>>> (and I'm sure I've forgotten at least one). That's something like 15
>>>> supplies totaling 12W.
>>>>
>>> And if you don't sequence them meticulously a $100 FPGA blows its lid?
>>
>> Not so much anymore. Actel (and ALtera, IIRC) parts are guaranteed to
>> not need sequencing. Actel even supports hot plugging.
>>
>>> Oh man, I am glad I am an analog dude :-)
>>
>> No FPGAs in the above. The analog circuits take about half of the
>> above supplies.
>
>
>Maybe someone should have a chat with the analog guys ;-)
Start with TI's. ;-) I'd even add a couple more to get rid of the
single-ended analog. We also have isolation requirements so some
supplies show up on both sides of the barrier. They are paranoid
about running the analogs and digital stuff from the same supplies
(and grounds!) too. ...even though neither are being taxed.
Joerg
> On Sat, 29 Aug 2009 15:48:21 -0700, Joerg <inv...@invalid.invalid>
> wrote:
>
>> krw wrote:
>>> On Fri, 28 Aug 2009 06:39:12 -0700, Joerg <inv...@invalid.invalid>
>>> wrote:
>>>
>>>> krw wrote:
[...]
>>>>> I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user
>>>>> I/O, flash-based FPGA.
>>>>>
>>>> If I'd put a $2.90 part on a design without a serious reason such as
>>>> "the world would stop spinning if we didn't", many of my client would
>>>> have my head examined :-)
>>> Even if it replaced six parts worth $12?
>>>
>> I can't remember the last time where six digital parts would have cost
>> $12 in any of my designs. More like 10c times six :-)
>>
>> [...]
>
> No wonder you don't think digital stuff is fun. You're too cheap to
> buy interesting parts. ...
I do occasionally get involved with uC but it's not much fun for me.
Usually I leave that to teh experts, just like DSP.
> ... Other than passives and transistors, there is
> not much we buy that costs less than a buck.
>
The you may be leaving money on the table :-)
>>>>>>>> Lots of transistors and fets that cost 3 cents
>>>>>>> Free resistors.
>>>>>>>
>>>>>> Free?
>>>>>>
>>>>> Less than the cost of putting them there.
>>>>>
>>>> I guess you assume domestic production.
>>> We do our own.
>>>
>> Yep, that gets expensive.
>
> It's sometimes necessary. Our main business is so cyclical that we
> need the control in-house manufacturing gives. Certainly control over
> production trumps cost.
>
>> Maybe you are better off in Alabama than we are in California.
>
> In more ways than one. ;-)
>
Yeah, maybe some day we'll join you.
>>>> That's expensive, even worse in
>>>> Europe. But in China, very different thing. Many folks don't know, there
>>>> was even one guy here in the NG who just refused to believe that 5%
>>>> resistors would make sense anymore. They do, seriously.
>>> They don't for us - inventory. A reel of 5,000 1% resistors is less
>>> than $10 from DigiKey, of all places.
>>>
>> We sometimes calculate them in millicents :-)
>
> Our product markups are better than what Larkin claims (and our
> quantities higher), so millicents aren't the main issue. I just
> started a cost reduction design pass on our products but I'm certainly
> not looking at millicents. Our fixed costs drown that out. Molds are
> *expensive*.
In the end it boils down to whether the lower cost is or was worth the NRE.
>>>>>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>>>>>
>>>>>>>> Tiny Logic
>>>>>>>>
>>>>>>>> Super-precision delta-sigma ADCs
>>>>>>> But no DACs to match. :-(
>>>>>>>
>>>>>>>> Too many voltage regulators
>>>>>>> Boy howdy!
>>>>>> But nearly all LDOs and thus off limits in this here lab,
>>>>> Doesn't matter what they are, there are too many. Lets see, our base
>>>>> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
>>>>> (times three), +12V (times two), -12V, +19V, and various references
>>>>> (and I'm sure I've forgotten at least one). That's something like 15
>>>>> supplies totaling 12W.
>>>>>
>>>> And if you don't sequence them meticulously a $100 FPGA blows its lid?
>>> Not so much anymore. Actel (and ALtera, IIRC) parts are guaranteed to
>>> not need sequencing. Actel even supports hot plugging.
>>>
>>>> Oh man, I am glad I am an analog dude :-)
>>> No FPGAs in the above. The analog circuits take about half of the
>>> above supplies.
>>
>> Maybe someone should have a chat with the analog guys ;-)
>
> Start with TI's. ;-)
Don't use the fancy parts :-)
> ... I'd even add a couple more to get rid of the
> single-ended analog. We also have isolation requirements so some
> supplies show up on both sides of the barrier. They are paranoid
> about running the analogs and digital stuff from the same supplies
> (and grounds!) too. ...even though neither are being taxed.
I am usually pretty brazen in that respect. Single-supply analog as much
as possible, I normally don't object sharing the supplies with the
digital guys as long as they share their Christmas cookies, and iso
supplies are home made using little transformers and some sort of bridge
driver (as long as not over a buck ...).
krw
wrote:
>krw wrote:
>> On Sat, 29 Aug 2009 15:48:21 -0700, Joerg <inv...@invalid.invalid>
>> wrote:
>>
>>> krw wrote:
>>>> On Fri, 28 Aug 2009 06:39:12 -0700, Joerg <inv...@invalid.invalid>
>>>> wrote:
>>>>
>>>>> krw wrote:
>
>[...]
>
>>>>>> I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user
>>>>>> I/O, flash-based FPGA.
>>>>>>
>>>>> If I'd put a $2.90 part on a design without a serious reason such as
>>>>> "the world would stop spinning if we didn't", many of my client would
>>>>> have my head examined :-)
>>>> Even if it replaced six parts worth $12?
>>>>
>>> I can't remember the last time where six digital parts would have cost
>>> $12 in any of my designs. More like 10c times six :-)
>>>
>>> [...]
>>
>> No wonder you don't think digital stuff is fun. You're too cheap to
>> buy interesting parts. ...
>
>
>I do occasionally get involved with uC but it's not much fun for me.
>Usually I leave that to teh experts, just like DSP.
UCs are all the same anymore. We have another group of engineers to
screw them up. ;-)
>> ... Other than passives and transistors, there is
>> not much we buy that costs less than a buck.
>>
>
>The you may be leaving money on the table :-)
My time is worth money too, as is another board spin.
>>>>>>>>> Lots of transistors and fets that cost 3 cents
>>>>>>>> Free resistors.
>>>>>>>>
>>>>>>> Free?
>>>>>>>
>>>>>> Less than the cost of putting them there.
>>>>>>
>>>>> I guess you assume domestic production.
>>>> We do our own.
>>>>
>>> Yep, that gets expensive.
>>
>> It's sometimes necessary. Our main business is so cyclical that we
>> need the control in-house manufacturing gives. Certainly control over
>> production trumps cost.
>>
>>> Maybe you are better off in Alabama than we are in California.
>>
>> In more ways than one. ;-)
>>
>
>Yeah, maybe some day we'll join you.
It's a pretty place (no mountains or snow, though) but rather quiet.
This is a fairly big "small town". If I moved it would likely be
100NE to Atlanta. Atlanta is a *really* big small town. ;-)
>>>>> That's expensive, even worse in
>>>>> Europe. But in China, very different thing. Many folks don't know, there
>>>>> was even one guy here in the NG who just refused to believe that 5%
>>>>> resistors would make sense anymore. They do, seriously.
>>>> They don't for us - inventory. A reel of 5,000 1% resistors is less
>>>> than $10 from DigiKey, of all places.
>>>>
>>> We sometimes calculate them in millicents :-)
>>
>> Our product markups are better than what Larkin claims (and our
>> quantities higher), so millicents aren't the main issue. I just
>> started a cost reduction design pass on our products but I'm certainly
>> not looking at millicents. Our fixed costs drown that out. Molds are
>> *expensive*.
>
>
>In the end it boils down to whether the lower cost is or was worth the NRE.
Yep. There were some decisions made that are going to be hard to
counter. The real savings (like 2x $15 for pots) are going to break
the mold.
>>>>>>>>> Gallium Nitride and Silicon Carbide and SiGe
>>>>>>>>>
>>>>>>>>> Tiny Logic
>>>>>>>>>
>>>>>>>>> Super-precision delta-sigma ADCs
>>>>>>>> But no DACs to match. :-(
>>>>>>>>
>>>>>>>>> Too many voltage regulators
>>>>>>>> Boy howdy!
>>>>>>> But nearly all LDOs and thus off limits in this here lab,
>>>>>> Doesn't matter what they are, there are too many. Lets see, our base
>>>>>> unit has 1.2V, 1.5V, 1.8V (times two), 3.3V (times two), 5.6V, 5V
>>>>>> (times three), +12V (times two), -12V, +19V, and various references
>>>>>> (and I'm sure I've forgotten at least one). That's something like 15
>>>>>> supplies totaling 12W.
>>>>>>
>>>>> And if you don't sequence them meticulously a $100 FPGA blows its lid?
>>>> Not so much anymore. Actel (and ALtera, IIRC) parts are guaranteed to
>>>> not need sequencing. Actel even supports hot plugging.
>>>>
>>>>> Oh man, I am glad I am an analog dude :-)
>>>> No FPGAs in the above. The analog circuits take about half of the
>>>> above supplies.
>>>
>>> Maybe someone should have a chat with the analog guys ;-)
>>
>> Start with TI's. ;-)
>
>
>Don't use the fancy parts :-)
Don't use TI.
>> ... I'd even add a couple more to get rid of the
>> single-ended analog. We also have isolation requirements so some
>> supplies show up on both sides of the barrier. They are paranoid
>> about running the analogs and digital stuff from the same supplies
>> (and grounds!) too. ...even though neither are being taxed.
>
>
>I am usually pretty brazen in that respect. Single-supply analog as much
>as possible, I normally don't object sharing the supplies with the
>digital guys as long as they share their Christmas cookies, and iso
>supplies are home made using little transformers and some sort of bridge
>driver (as long as not over a buck ...).
I don't like throwing away the "free" PSRR. I'm planning on combining
supplies to save space more than the cost of the components.
Interboard connectors are expensive and a PITA. The isos we're using
are bricks (=expensive).
Frank Buss
> Small ones can be ok but large BGA cannot easily stand the flexing of a
> circuit board. All it takes is one fall. BGA don't have any compliance
> to speak of while the leads of TSSOP and others can bend a little
> without popping out of the solder.
I think this depends on the application. A one mm thick mobile phone size
FR4 board should not flex very much and maybe for bigger boards you could
use more screws, but you might be right for the baking tray size boards you
do, used in Ariane rockets and similar things :-)
Joerg
> On Sat, 29 Aug 2009 16:52:02 -0700, Joerg <inv...@invalid.invalid>
> wrote:
>
>> krw wrote:
[...]
>>> ... I'd even add a couple more to get rid of the
>>> single-ended analog. We also have isolation requirements so some
>>> supplies show up on both sides of the barrier. They are paranoid
>>> about running the analogs and digital stuff from the same supplies
>>> (and grounds!) too. ...even though neither are being taxed.
>>
>> I am usually pretty brazen in that respect. Single-supply analog as much
>> as possible, I normally don't object sharing the supplies with the
>> digital guys as long as they share their Christmas cookies, and iso
>> supplies are home made using little transformers and some sort of bridge
>> driver (as long as not over a buck ...).
>
> I don't like throwing away the "free" PSRR. I'm planning on combining
> supplies to save space more than the cost of the components.
> Interboard connectors are expensive and a PITA. The isos we're using
> are bricks (=expensive).
I remember a Burr-Brown sales rep trying to wet my appetite for an iso
amp in the late 80's. I was interested a little, until he told me what
it cost. 80 (!) bux. I can't remember ever having used iso power
converter bricks or iso amp bricks/chips. Not even on expensive gear
where it would have mattered less. Besides cost you'll typically be
married to one single source and when that manufacturer can't deliver
things turn ugly real fast.
Joerg
> Joerg wrote:
>
>> Small ones can be ok but large BGA cannot easily stand the flexing of a
>> circuit board. All it takes is one fall. BGA don't have any compliance
>> to speak of while the leads of TSSOP and others can bend a little
>> without popping out of the solder.
>
> I think this depends on the application. A one mm thick mobile phone size
> FR4 board should not flex very much and maybe for bigger boards you could
> use more screws, but you might be right for the baking tray size boards you
> do, used in Ariane rockets and similar things :-)
>
For commercial aircraft it often has to be rated for 20G, rated meaning
it must not fail at all. On spacecraft it's several times that rating. A
target lifetime of 30 years is also quite normal. But even regular small
size boards are vulnerable. I don't remember the product but it was an
expensive computer play station for kids that had a noticeable BGA
failure rate (XBox?). Not sure if that's fixed by now.
BTW, since changing news servers I am having trouble with the German NG.
Often it disappears and all threads must be re-loaded. Other times like
this morning it doesn't download any new posts. It's only that NG, all
the others always work. Sometimes it comes back after an hour. Very
strange.
Frank Buss
> Often it disappears and all threads must be re-loaded. Other times like
> this morning it doesn't download any new posts. It's only that NG, all
> the others always work. Sometimes it comes back after an hour. Very
> strange.
You should contact your usenet provider, sounds like they have some peering
problems. But I don't see any new posting in de.sci.electronics since today
1:33 am GMT either, maybe it is a sleepy sunday :-)
Joerg
>> BTW, since changing news servers I am having trouble with the German NG.
>> Often it disappears and all threads must be re-loaded. Other times like
>> this morning it doesn't download any new posts. It's only that NG, all
>> the others always work. Sometimes it comes back after an hour. Very
>> strange.
>
> You should contact your usenet provider, sounds like they have some peering
> problems. But I don't see any new posting in de.sci.electronics since today
> 1:33 am GMT either, maybe it is a sleepy sunday :-)
>
Oh, then it's probably not a server problem. My usenet provider is now
the university server in Berlin, after our phone company cut all newsgroups.
I hope that group doesn't follow others and become a dinosaur ;-)
Do you think that having to reload all threads could also be a peering
problem? Sometimes in the morning all NGs load with new posts except
d.s.e. Then when I click on d.s.e. it asks as if I had just newly
subscribed, "Download all 56782 headers?". Of course then I only
download 1000 or so, otherwise this takes forever. Sometimes all of them
will be marked as un-read, sometimes not. I can't imagine this being a
Thunderbird problem because it never happens to any other NG.
Boris Mohar
There is a chip for that ;)
--
Boris Mohar
Jan Panteltje
<inv...@invalid.invalid> wrote in <7fvch9F...@mid.individual.net>:
News articles are indexed by number.
If the server maintainer edits the list, and changes numbering to start
with a lower number, then you have this effect.
He should not do that.
Change servers.
Try www.datemas.de or www.albasani.net to see if they still accept new users.
I am using both, you need to get an account.
As you are no spammer your changes are good you will get [and keep] one.
No binaries though.
Both are in Europe, and should carry most German groups.
If that fails use nntp.aioe.org, now located in the Netherlands IIRC.
Joerg
If it persists I'll first write to them. It's news.individual.net and
they are very responsive.
> Try www.datemas.de or www.albasani.net to see if they still accept new users.
> I am using both, you need to get an account.
> As you are no spammer your changes are good you will get [and keep] one.
> No binaries though.
> Both are in Europe, and should carry most German groups.
> If that fails use nntp.aioe.org, now located in the Netherlands IIRC.
>
Well, I just changed :-)
krw
wrote:
The bricks go for about a buck-a-watt. The bricks we're using are
standardized (we've switched a few times already). The real problem
with any of this is compliance. It costs a lot of money to change a
switcher.
The thing that appalled me was what we're paying for isolated RS-422
transceivers! We're using them for essentially DC, too. It'll be at
least another month before I get to that part of the system.
Joerg
EMC can be nasty. I have seen bricks fail at the lab while their
manufacturers were staunchly saying that they'd pass with flying colors.
Thing is, my time is expensive, as is yours. But spending 8h on a
switcher design is better than pulling your hair out after a day at the
EMC lab and having to call the boss asking him to tell the marketeers
that we won't make the trade show this time.
> The thing that appalled me was what we're paying for isolated RS-422
> transceivers! We're using them for essentially DC, too. It'll be at
> least another month before I get to that part of the system.
But be careful if they connect system parts at different locations. I've
seen a guy ask his colleague "Hey, why does this little data cable get
hot? How is that possible?"
Lightning and spikes are another concern.
krw
wrote:
We had a wall wart, that was already certified, fail in compliance
testing. Turns out that the manufacturers only test with a resistive
load at full ratings.
>Thing is, my time is expensive, as is yours. But spending 8h on a
>switcher design is better than pulling your hair out after a day at the
>EMC lab and having to call the boss asking him to tell the marketeers
>that we won't make the trade show this time.
EMC testing for a trade show? ;-)
>> The thing that appalled me was what we're paying for isolated RS-422
>> transceivers! We're using them for essentially DC, too. It'll be at
>> least another month before I get to that part of the system.
>
>
>But be careful if they connect system parts at different locations. I've
> seen a guy ask his colleague "Hey, why does this little data cable get
>hot? How is that possible?"
Sure, but there are cheaper ways of isolating "DC" signals.
>Lightning and spikes are another concern.
We're thinking about going to fiber for just this reason. The problem
is that it forces the customer to supply power. ...and we already
pulled the remote power option from the remote transceiver (see wall
wart issue above).
Joerg
That does not surprise me. One reason why I like to roll my own.
>> Thing is, my time is expensive, as is yours. But spending 8h on a
>> switcher design is better than pulling your hair out after a day at the
>> EMC lab and having to call the boss asking him to tell the marketeers
>> that we won't make the trade show this time.
>
> EMC testing for a trade show? ;-)
>
Yes. At medical trade shows you must often be able to deliver next day,
right to the show location. In Europe it is customary that interested
hospitals send people with trucks, to land and execute a deal right then
and there. Others want it delivered to their hospital next week Monday
when they are back. IOW you better have a substantial amount of
merchandise on the shelf, ready to ship the millisecond a sales guy
calls in a deal.
Imagine you find "the" digital scope at a show. Drooling all over the
place. You'd probably walk away from it if the vendors says "Well, we
have a three month leadtime on this one".
>>> The thing that appalled me was what we're paying for isolated RS-422
>>> transceivers! We're using them for essentially DC, too. It'll be at
>>> least another month before I get to that part of the system.
>>
>> But be careful if they connect system parts at different locations. I've
>> seen a guy ask his colleague "Hey, why does this little data cable get
>> hot? How is that possible?"
>
> Sure, but there are cheaper ways of isolating "DC" signals.
>
Yes, for DC all need is a 20c optocoupler.
>> Lightning and spikes are another concern.
>
> We're thinking about going to fiber for just this reason. The problem
> is that it forces the customer to supply power. ...and we already
> pulled the remote power option from the remote transceiver (see wall
> wart issue above).
Or go with a transformer wall wart.
krw
wrote:
The safety certifications are easier with a purchased solution,
though.
>>> Thing is, my time is expensive, as is yours. But spending 8h on a
>>> switcher design is better than pulling your hair out after a day at the
>>> EMC lab and having to call the boss asking him to tell the marketeers
>>> that we won't make the trade show this time.
>>
>> EMC testing for a trade show? ;-)
>>
>
>Yes. At medical trade shows you must often be able to deliver next day,
>right to the show location. In Europe it is customary that interested
>hospitals send people with trucks, to land and execute a deal right then
>and there. Others want it delivered to their hospital next week Monday
>when they are back. IOW you better have a substantial amount of
>merchandise on the shelf, ready to ship the millisecond a sales guy
>calls in a deal.
By the time we have a saleable system there is no need for trade
shows. Different industry norms, I guess.
>Imagine you find "the" digital scope at a show. Drooling all over the
>place. You'd probably walk away from it if the vendors says "Well, we
>have a three month leadtime on this one".
Three months isn't an issue. It would take that long to convince
myself that I really wanted that model (and get the cash loose to buy
it). ;-)
>
>>>> The thing that appalled me was what we're paying for isolated RS-422
>>>> transceivers! We're using them for essentially DC, too. It'll be at
>>>> least another month before I get to that part of the system.
>>>
>>> But be careful if they connect system parts at different locations. I've
>>> seen a guy ask his colleague "Hey, why does this little data cable get
>>> hot? How is that possible?"
>>
>> Sure, but there are cheaper ways of isolating "DC" signals.
>>
>
>Yes, for DC all need is a 20c optocoupler.
Exactly where I was going. Since there is already an isolated supply
for this interface and opto-isolator is pretty cheap.
>
>>> Lightning and spikes are another concern.
>>
>> We're thinking about going to fiber for just this reason. The problem
>> is that it forces the customer to supply power. ...and we already
>> pulled the remote power option from the remote transceiver (see wall
>> wart issue above).
>
>
>Or go with a transformer wall wart.
I suggested that last week for a different application. I got funny
looks but no reasons it couldn't be done.
Joerg
> On Sun, 30 Aug 2009 09:34:46 -0700, Joerg <inv...@invalid.invalid>
> wrote:
>
>> krw wrote:
>>> On Sun, 30 Aug 2009 09:01:02 -0700, Joerg <inv...@invalid.invalid>
>>> wrote:
>>>
>>>> krw wrote:
[...]
>>>>> The bricks go for about a buck-a-watt. The bricks we're using are
>>>>> standardized (we've switched a few times already). The real problem
>>>>> with any of this is compliance. It costs a lot of money to change a
>>>>> switcher.
>>>>>
>>>> EMC can be nasty. I have seen bricks fail at the lab while their
>>>> manufacturers were staunchly saying that they'd pass with flying colors.
>>> We had a wall wart, that was already certified, fail in compliance
>>> testing. Turns out that the manufacturers only test with a resistive
>>> load at full ratings.
>>>
>> That does not surprise me. One reason why I like to roll my own.
>
> The safety certifications are easier with a purchased solution,
> though.
>
If they come from a manufacturer you can trust. The worst I've seen:
Fully "certified" open frame supply from one of the traditional PS
manufacturers. Submitted unit to the TUEV, they did the usual overload
tests ... *PHUT*. Could have been a fluke so they tested a 2nd system
... *PHUT*.
I read the folks at that vendor the riot act and had purchasing cancel
the whole order. Their sales guy was almost in tears.
>>>> Thing is, my time is expensive, as is yours. But spending 8h on a
>>>> switcher design is better than pulling your hair out after a day at the
>>>> EMC lab and having to call the boss asking him to tell the marketeers
>>>> that we won't make the trade show this time.
>>> EMC testing for a trade show? ;-)
>>>
>> Yes. At medical trade shows you must often be able to deliver next day,
>> right to the show location. In Europe it is customary that interested
>> hospitals send people with trucks, to land and execute a deal right then
>> and there. Others want it delivered to their hospital next week Monday
>> when they are back. IOW you better have a substantial amount of
>> merchandise on the shelf, ready to ship the millisecond a sales guy
>> calls in a deal.
>
> By the time we have a saleable system there is no need for trade
> shows. Different industry norms, I guess.
>
>> Imagine you find "the" digital scope at a show. Drooling all over the
>> place. You'd probably walk away from it if the vendors says "Well, we
>> have a three month leadtime on this one".
>
> Three months isn't an issue. It would take that long to convince
> myself that I really wanted that model (and get the cash loose to buy
> it). ;-)
Ok, for personal use, yes. Hospitals and much of the industry must time
their purchases into a tax year. Government-run places must often blow
out the remainder of the budget by this and this date or they'll have
next year's budget cut by that amount. That's sick, but this is
unfortunately how it is.
[...]
>>>> Lightning and spikes are another concern.
>>> We're thinking about going to fiber for just this reason. The problem
>>> is that it forces the customer to supply power. ...and we already
>>> pulled the remote power option from the remote transceiver (see wall
>>> wart issue above).
>>
>> Or go with a transformer wall wart.
>
> I suggested that last week for a different application. I got funny
> looks but no reasons it couldn't be done.
Pull out the bill for the last failed EMC test, along with a good faith
estimate of company engineering hours put into it, travel expenses, the
works. That ought to convince them :-)
Of course if your gear must ship internationally you don't have much of
a choice, then you nowadays need a 90-260VAC universal supply.
krw
wrote:
Not surprising. I'm not impressed by the compliance folks. OTOH, I
wouldn't do that job.
I meant for business use. I always had to plan my purchases at least
a year in advance. Sometimes the plans were only placeholders, but
our planning cycle was three and five years.
>[...]
>
>
>>>>> Lightning and spikes are another concern.
>>>> We're thinking about going to fiber for just this reason. The problem
>>>> is that it forces the customer to supply power. ...and we already
>>>> pulled the remote power option from the remote transceiver (see wall
>>>> wart issue above).
>>>
>>> Or go with a transformer wall wart.
>>
>> I suggested that last week for a different application. I got funny
>> looks but no reasons it couldn't be done.
>
>
>Pull out the bill for the last failed EMC test, along with a good faith
>estimate of company engineering hours put into it, travel expenses, the
>works. That ought to convince them :-)
>Of course if your gear must ship internationally you don't have much of
>a choice, then you nowadays need a 90-260VAC universal supply.
It does, and yes, it does. How about a 12/24V transformer? ;-)
Joerg
Oh, they did their job alright. The manufacturer had cut some corner and
that did them in.
Sure, but even there tax laws, accounting rules and such apply.
Sometimes a purchase must be made after a certain date or before a
certain date.
>> [...]
>>
>>
>>>>>> Lightning and spikes are another concern.
>>>>> We're thinking about going to fiber for just this reason. The problem
>>>>> is that it forces the customer to supply power. ...and we already
>>>>> pulled the remote power option from the remote transceiver (see wall
>>>>> wart issue above).
>>>> Or go with a transformer wall wart.
>>> I suggested that last week for a different application. I got funny
>>> looks but no reasons it couldn't be done.
>>
>> Pull out the bill for the last failed EMC test, along with a good faith
>> estimate of company engineering hours put into it, travel expenses, the
>> works. That ought to convince them :-)
>
>> Of course if your gear must ship internationally you don't have much of
>> a choice, then you nowadays need a 90-260VAC universal supply.
>
> It does, and yes, it does. How about a 12/24V transformer? ;-)
Seriously, 240V transformers can work here but only if the power
consumption is miniscule.
JosephKK
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:
>On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <gal...@hotmail.com>
>wrote:
>
>>I've been away from board-level design for about seven years. What are
>>the most interesting recent product developments? Apparently LED's
>>have become quite a bit more efficient. Thanks.
>
>FPGAs with enormous data crunching capability.
>
>Really fast opamps.
>
>Super-low drift chopper opamps.
>
>More module-type things, like Linear's switchers
>
>Silicon oscillators and delay lines
>
>Cheap precision thinfilm resistors
>
>Polymer aluminum capacitors
>
>High-value ceramic caps, like 22uF and 100uF
>
>Lots of cool RF stuff: plls, rf detectors, amps
>
>Really fast ecl/cml logic, 6 GHz range.
>
>Outrageous fast cheap microprocessors, like the ARM stuff.
>
>Everything going to BGA and leadless/chip-scale packages
>
>Everything getting smaller
>
>Digital pots
>
>Lots of transistors and fets that cost 3 cents
>
>Gallium Nitride and Silicon Carbide and SiGe
>
>Tiny Logic
>
>Super-precision delta-sigma ADCs
>
>Cheap SAR ADCs
>
>Cheap video/pipeline ADCs
>
>Too many voltage regulators
>
>
>
>John
Power supply voltages crowding 1 V at many A.
Much stronger concern for power dissipation.
JosephKK
wrote:
>Frank Buss wrote:
>> Joerg wrote:
>>
>>> Nemo wrote:
>>>
>>>> PCB manufacturing has become better: even small suppliers can do finer
>>>> pitch multilayer stuff than was common 7 years ago. Many "pool"
>>>> prototype PCB's onto one biscuit, so you can get a handful of prototype
>>>> 4 layer PCB's in say 3 days for $100 - $200.
>>>>
>>> Where? My last quote was almost $2k, two (!).
>>
>> I don't know, if he means $100-$200 per PCB, or maybe small PCBs?
>> pcb-pool.de: 5 eurocards (160x100), 4 layers, each one for 84 Euro, but in
>> 8 days, 5 days is more than twice as expensive. But you get a free laser
>> stencil, too and the quality is always very good. They accept many formats,
>> without the need for you to convert it to gerber files. I've downloaded
>> their Eagle policy file for electrical rule check and there was nearly
>> never any questions. Only once, when I submitted the schematic file instead
>> of the board file at a sleepy night :-)
>>
>
>Oh, then they must have adjusted their prices. When I checked one or two
>years ago they were expensive compared to the places over here. Mostly I
>used this place:
>
>http://www.4pcb.com/
>
>
>> And using their reflow kit is nice for prototypes, because it is faster and
>> looks better than hand soldering.
>>
>
>Interesting. Especially since I just finished a design and (so far) have
>no idea who can solder it. It's mostly 0603 and some TSSOP pitch stuff.
>
>In case anyone else wonder what this reflow kit is:
>
>http://www.reflow-kit.de/rkuk/order_product_details.html?p=13
>
>Hey, I guess when done we could bake a pizza for the guys in the oven :-)
The market for that seems to be far better than i had guessed. Perhaps
this is the beginning (well at least rising slope) of hobbyists round
4.
Joerg
Concern? Why?
The synchronous buck has been invented ...
John Larkin
wrote:
>>> Too many voltage regulators
>>>
>>>
>>>
>>> John
>>
>> Power supply voltages crowding 1 V at many A.
>>
>> Much stronger concern for power dissipation.
>
>
>Concern? Why?
>
>The synchronous buck has been invented ...
And how!
ftp://jjlarkin.lmi.net/SwitcherRise.JPG
LM3102. Beautiful, classic step-recovery diode sequence. I suspect the
risetime was actually scope+probe limited.
This freaks out opamps clear on the other side of the board. I don't
even want to think about the EMI situation.
John
Joerg
> On Thu, 03 Sep 2009 08:23:42 -0700, Joerg <inv...@invalid.invalid>
> wrote:
>
>
>>>> Too many voltage regulators
>>>>
>>>>
>>>>
>>>> John
>>> Power supply voltages crowding 1 V at many A.
>>>
>>> Much stronger concern for power dissipation.
>>
>> Concern? Why?
>>
>> The synchronous buck has been invented ...
>
> And how!
>
> ftp://jjlarkin.lmi.net/SwitcherRise.JPG
>
> LM3102. Beautiful, classic step-recovery diode sequence. I suspect the
> risetime was actually scope+probe limited.
>
That is beautiful.
> This freaks out opamps clear on the other side of the board. I don't
> even want to think about the EMI situation.
>
With Sutro Tower being that close I guess nobody would even notice ;-)
Joerg
Yeah, but I chickened out and ours will go to a contract assembler :-)
For hobbyists this is fine but in business it costs far too much time.
JosephKK
wrote:
Perzactly.
--
Transmitted with recycled bits.
Damnly my frank, I don't give a dear
----------
JosephKK
wrote:
>JosephKK wrote:
For a prototype it might be useful, i wouldn't dare use it for real
production. But this product is targeted at hobbyists. Which in a
sense was really the point, cheap reflow for hobbyists.
Klaus Kragelund
> John Larkin wrote:
> > On Fri, 28 Aug 2009 06:30:13 -0700, Joerg <inva...@invalid.invalid>
> > wrote:
>
> >> John Larkin wrote:
> >>> wrote:
>
> >>>> John Larkin wrote:
> >>>>> wrote:
>
> >>>>>> krw wrote:
> >>>>>>> On Wed, 26 Aug 2009 14:45:12 -0700, John Larkin
>
> >>>>>>>> On Wed, 26 Aug 2009 13:49:23 -0700 (PDT), Davej <galt...@hotmail.com>
> >>>>>>>> wrote:
>
> >>>>>>>>> I've been away from board-level design for about seven years. What are
> >>>>>>>>> the most interesting recent product developments? Apparently LED's
> >>>>>>>>> have become quite a bit more efficient. Thanks.
> >>>>>>>> FPGAs with enormous data crunching capability.
>
> >>>>>> Where? Which ones? Back in the 80's I regularly beat other guys when
> >>>>>> doing stuff like address decoders. Their PALs and GALs cost over abuck
> >>>>>> and burned tens of milliamps while mine cost 20-30 Cents and needed a
> >>>>>> few tens of microamps. Have seen much of a change :-)
>
> >>>>>> Ok, this is more CPLD domain but still, it's pricey.
>
> >>>>>>>> Really fast opamps.
> >>>>>> Texas THS-series, blazingly fast +/-12V amps. Most of the others can
> >>>>>> only do +5V, often not too practical like when driving huge FETs that
> >>>>>> need 10V swings.
> >>>>> Avoid THS3062. It's fatally messed up.
>
> >>> If you drive it to output a healthy amplitude sine wave, say 20 volts
> >>> p-p at 8 MHz or so, way within specs, it latches up, draws huge supply
> >>> currents, goes nuts, and gets red hot. Burned a hole in me finger, it
> >>> did.
>
> >>> Amazingly, it recovers if you crank the signal amplitude down close to
> >>> zero. Me finger didn't.
>
> >>> After weeks of harassment, TI told us, essentially, "don't do that."
> >>> It is apparently suited to making ADSL waveforms and nothing else.
>
> >>> Their datasheet graphs all carefully avoid the voltage-frequency
> >>> danger zone; "large signal" is apparently 2 volts p-p, even when using
> >>> +-15 supplies.
>
> >>> The datasheet is seven years old. I wonder how many people have been
> >>> literally burned by this deliberate deception.
>
> >> Oh, bad news. At least they should have fessed up to this and also
> >> update the datasheet.
>
> >> I had a similar comeuppance with a TPS-something linear regulator which
> >> I'll never again use. When the input power came up too fast which it had
> >> to for that client ... tsk ... phut ... *POP* ... nice li'l crater in it
> >> and definitely not a good thing to test without goggles.
>
> >> So I contacted tech support. After some waffling I wanted the SPICE
> >> file. "Can't have that". So I asked them to throw my fairly simple
> >> circuit onto their sim and try. "Nope, can't do that either". Hmm ...
>
> > Is TI the new Maxim?
>
> > John
>
> No, then you could not even buy a reel :-)
>
Supposively Maxim has changed their policy about 1 year ago. Will now
use subdealers which makes it possible for the small buyers to get
their parts and much more aware of their clients. But, is this true.
Any one with fresh experiences?
Regards
Klaus
Joerg
One of my litmus tests is Digikey. I just did it again, selected the
category "Voltage Regulators - Switching DC DC Converters & Controllers"
and then only Maxim. PWM chips is most of the Maxim stuff I have
designed out for clients.
Without clicking the "in stock" box: 1199 hits.
After clicking the "in stock" box: Poof ... down to 165.
They have promised so many times that it gets better and then it didn't,
so I really don't believe them anymore. I think it takes more than a
policy change, it takes upper management changes at Maxim.
In the end Digikey is similar to Walmart. If they don't have someting in
stock then, for the vast masses, de facto it doesn't exist :-)