Re: looking for class D amplifier

miso@xxxxxxxxx wrote:
On Sep 14, 9:50 am, Joerg <inva...@xxxxxxxxxxxxxxx> wrote:
m...@xxxxxxxxx wrote:
On Sep 12, 2:38 pm, Joerg <inva...@xxxxxxxxxxxxxxx> wrote:
Jamie wrote:
Joerg wrote:
Jamie wrote:
Bullwinkle wrote:
Hi,
i'm looking for a class D amplifier with these specifications:
1. mono
2. max3W
3. gain regulation (varying input pin's logical value and not input
resistor)
Can you help me?
Thanks in advance
Something like the
MAX9768 ?
http://datasheets.maxim-ic.com/en/ds/MAX9768.pdf
Which then goes unobtanium on you ... ;-)
Well, that's the way the market goes which is why you buy extra's
for your customers.
I prefer to design in parts from manufacturers where it's less likely
that things vaporize :-)
--
Regards, Joerg
http://www.analogconsultants.com/
"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
All I can say with the exception of one chip that developed issues
over time due to process shifts, all my stuff shipped routinely. Even
the problem chip (which was a big ass problem since it was in some
aircraft) eventually got fixed.
Got some examples? Not the problem chip, don't reveal any internal issues.

Examples of chips that sold without hitches? That would be the
majority of what I saw. [I had access to the WIP, which was maintained
on a VAX of all things. Design could telnet into the machine and there
were no limits to what I could look at. Yeah, idiots. Damn good thing
we were making enough money from the stock that nobody sold that
information on the street.] For all intents and purposes, there was no
product engineering at Maxim in my era (all of the 90s and some of the
80s). Yeah, there were a few product engineers, but it was not a big
effort like you would see at the big semis. So if something didn't
work, a design engineer would investigate it. To be clear here, the
designer would see chips that do not yield at the target. Most likely,
you would see it at the wafer level. Generally for QA, a company is
very reluctant to sell product from poorly yielding wafer, even if
there are sufficient dice to make it worth your while to package them.
This is not universal thinking. Some companies would sell floor
sweepings to make sales and supply orders. But generally if something
smells funny, most companies don't sell product from that wafer run.


What I meant was successful chips that you've designed _and_ that are now in stock (in serious qties) at distributors. Because that's what the typical problem is with Maxim. You can get samples but then often no large production quantities.


Now what you would not see as a designer are production screw ups. For
instance, some idiot running packaging backlaps the wafers to the
wrong thickness, and thus the product can't be packaged. So this
person wasted an entire lot of wafers. That causes delay in shipping.
The person running packaging might not order leadframes for the
product. This is common with power devices that use etched leadframes
rather than something they stamp out. [Etched leadframes are
expensive, but so is the tooling. So if you need a custom leadframe,
you start etched then go to stamped when volume is sufficient.]


If this happens a lot then they've got some training issues.


Also what will not be seen by the designer are test issues. If the
product reaches high volume, it will go offshore for testing. It is no
secret that the people off-shore really don't know a whole lot.
Nothing racist here, but the talent behind the product is in the
states, so these people off-shore are in a vacuum. When it is decided
that the situation for test off-shore is hopeless, the product is
shipped back to the states. [IIRC, all the wafer test was done in the
states, and only packaged parts were tested off-shore.] Now this crap
comes back to the US and somebody has to make time to figure out the
problem. Here again, the lack of product engineers means a test
engineer and perhaps the designer has to figure out the problem, even
though they have other duties. Since analog chips have critical layout
and need external components, often a dutboard failure is enough to
make the product fail off-shore but work here. Or that contactor
itself needs components attached to it, and the person off-shore did
not use the dedicated contactor. [The test program is suppose to check
that the right external components are there, but often the board
can't be 100% tested.]

So your product can be delayed for packaging and test screw ups. Now
if you knew who did what and watched the retirements, you would say
things are approving.


That is what we always had under network access. Without leaving my desk I could see WIP, scrap, throughput at every single production step. Ok, that was catheters but logistically it's the same thing.


Now if you had a lot of product in stock, the customer would not see
these problems. However, Maxim has way too many designs to keep all
that inventory. Most of the business managers are clueless, but they
are making the key decisions regarding what products to make. They
come up with way too many products that are very similar. To some
degree, this has been part of the strategy of the company, i.e.
fracture the market to make it harder for second sources to figure out
which chips to copy. But these business managers took fracturing the
market way too seriously. [The path would be make the Swiss the army
knife product, see what the customers do with the chip, then make two
or three chips that incorporate a specific set of the functions of the
Swiss army knife version.]


That is a huge mistake for a company. Either they are willing and able to produce all the stuff they announce or pare that selection down, big time. In poker that's called hold'em or fold'em :-)


The times you were able to build devices by getting family members to
sample chips probably seemed odd, but it made perfect sense to me.
There is a warehouse in the states that holds sample stock. They have
rails of chips to give away, plus the eval boards.


Yeah, but they won't likely stash as much as needed for 50k or 100k units/year and that's what some of my designs run at in production. Plus my clients usually don't have so many uncles and aunts to order samples :-)


A good operations person is probably worth the money you pay them. Too
bad Maxim never had anyone in operations that was good. Oh, stuff got
made, but it was not like clockwork. ...


Well, when is any of the executives going to wake up and put that in place? It ain't that hard. I've done it.


... Also I have no idea how things
ran for the bipolar parts, which might be what you were buying. The
MOS end of the business had a few glitches, mostly with that fab in
Oregon. The bipolar was mostly foundry as far as I know. It is hard to
run an analog bipolar business using foundries.


Haven't bought any of the bipolar parts either, and won't. I have only designed them out so far. Mostly video stuff and PWM.


A cog in the system at Maxim recently retired. Supply should improve.
Still a few more executives need to be shown the door though, but in
the last two years some real stinkers have left.
As I've said here plenty of times, the best thing that could ever
happen to Maxim is a company that knows what they are doing (say TI),
buys the company and dumps the management. ...
Sorry, but before that happens I won't design in Maxim parts.

The board is not very independent. Really, the board is tits on a boar
hog. If they didn't clean house when all the illegal stock repricing
activity was going on, there is no way they will vote to sell the
company now that it is listed again. Seems to me if you let the
company go to pink sheets, you are proven incompetent, and any decent
board will show the president the door. I always suspected the delays
in cleaning up the paperwork was how to do all this and stay out of
jail. Look at Steve Jobs, who pulled the same *** regarding stock
pricing. He has dragged the lawsuit out for years.


Oh-oh ...

Anyhow, then why don't they hire someone who knows operations, pay him well and let him run the place?

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
.

Quantcast