Most interesting recent products / improvements?

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

Dirt cheap FPGAs.

Hurts the small guy but great if you can afford it.

Hurts the small guy...

Free resistors.

But no DACs to match. :-(

Boy howdy!

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

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

On a sunny day (Wed, 26 Aug 2009 20:05:23 -0700) it happened John Larkin wrote in :

Looks like 'AMP" and 'WIDTH' connectors are too close together to allow both inserted?

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

On a sunny day (Thu, 27 Aug 2009 07:17:35 -0700) it happened John Larkin wrote in :

inserted?

OIC. Hey, those are nice trimpost, at least you cannot push them over from the outside :-)

inserted?

outside :-)

Very nice ones, I think they are originally a Japanese design- Tokyo Cosmos.

inserted?

outside :-)

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

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.

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.

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 :-(

Free?

But nearly all LDOs and thus off limits in this here lab,

--
Regards, Joerg

http://www.analogconsultants.com/

Avoid THS3062. It's fatally messed up.

John

I am not much of a CFB fan anyhow but what did they botch on this one?

--
Regards, Joerg

http://www.analogconsultants.com/

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.

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

What's a d*pst*k?

I never really trusted those automatic bidirectional things.

John

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

An apparently useful part--what's wrong with it?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal

I just got a quote from Actel for $2.90 for a 750ish LUT/FF, 77 user I/O, flash-based FPGA.

You have a different definition of "pricey" than anyone else on the planet. ;-)

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.

Less than the cost of putting them there.

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.

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 ...

--
Regards, Joerg

http://www.analogconsultants.com/