The Larkin front end

[xray]

I'll try not to insult Jim personally, but it
can be difficult, can't it

Very!

 

[Eeyore]

Not stupid, just stubborn, stubborn as ... a mule. And perhaps a
little overly resistant to real data.

I don't get the controversy over the beads--kill the real RF, let the
diff amp handle the rest. Seems simple enough.

I was concerned over the beads causing unbalance in the interfering signal which
then means that differential rejection will be compromised.

Doesn't explain where all that energy is going in my little
switcher--57% efficiency, and I can't find the loss. Dang it.
(inductor suspected)

Did you wind you own ? What's the material ?

Graham

 

[Eeyore]

Only from a nearby antenna. I have a set of the old tube GR unit RF
oscillators, from 10 MHz to about 2 GHz, and they output something
like a watt. So you can poke a banana lead in the connector as a
splash-everything antenna, or strip a bit of braid off a hank of coal
and make a more local source you can wave around. It's not
quantitative, but you can find resonances and do a rough before/after
comparison on things.

I've been meaning to buy a real EMI-type antenna, for crude
quantitative suceptability and emissions tests, but haven't got around
to that yet.

My own tests were targeted at the AM broadcast band.

We had a sig gen that worked up to 5MHz that had AM modulation.

One of the IEC standards gave a suitable coupling network so the signal went straight up the front end.

No surprise perhaps that FET input op-amps are much less bothered than bipolar.

Graham

 

[John Fields]

Considering you've never seen me !!!!

 

[Jim Thompson]

Very!

It can't be done ;-)

...Jim Thompson

 

[John Larkin]

I assume that your coal was actually coax. My coal never came with a
braid.

Spelling shouldn't count much here, but it threw me off when I read it.

I've been wavin' 'round a coal wand, all the live long day.

I never did learn to type. It makes me a better programmer, gives me
time to think about what I'm doing.

John

 

[Genome]

Considering you've never seen me !!!!

Can you still pull 20 somethings ?

Graham

I wouldn't be surprised if some of the people you are trying to show off to
have 20 somethings called daughters. I can't begin to imagine a situation
where they might feel the need to waste your time by taking the piss out of
you.

DNA

 

[Eeyore]

I wouldn't be surprised if some of the people you are trying to show off to

Who said anything about showing off ?

have 20 somethings called daughters. I can't begin to imagine a situation
where they might feel the need to waste your time by taking the piss out of
you.

Uh ? Oh I see. No, that doesn't apply here.

I do know a couple of Janeys ( mum and daughter ) and they're both fun to be
with. I think I'm about half way between them age wise.

Graham

 

[Jim Thompson]

I wouldn't be surprised if some of the people you are trying to show off to
have 20 somethings called daughters. I can't begin to imagine a situation
where they might feel the need to waste your time by taking the piss out of
you.

DNA

I have 40-somethings called daughters ;-)

As for "taking the piss out of" Eeyore, a 12-year-old could probably
handle that ;-)

...Jim Thompson

 

[John Larkin]

My own tests were targeted at the AM broadcast band.

We had a sig gen that worked up to 5MHz that had AM modulation.

One of the IEC standards gave a suitable coupling network so the signal went straight up the front end.

If one were to do that, inject RF directly into the connector pins, I
wonder what might be reasonable voltage levels to simulate typical
hazards, like broadcast, nearby cell phones, wi-fi nodes, cops with
walkie-talkies? There's so much RF around these days, some mobile
enough to get very close to the gear.

No surprise perhaps that FET input op-amps are much less bothered than bipolar.

Yeah, jfet and cmos opamps are much better. And there's a huge range
of sensitivities among different bipolar amps. I saw a web site or
something on this once, can't recall where... does anybody have a
link?


John

 

[John Larkin]

It can't be done ;-)

...Jim Thompson

Jim, you seriously upset a lot of people by refusing to be neurotic.

John

 

[John Larkin]

Not stupid, just stubborn, stubborn as ... a mule. And perhaps a
little overly resistant to real data.

I don't get the controversy over the beads--kill the real RF, let the
diff amp handle the rest. Seems simple enough.

Doesn't explain where all that energy is going in my little
switcher--57% efficiency, and I can't find the loss. Dang it.
(inductor suspected)

James Arthur

We recently bought a FLIR thermal imager. You can *see* where the heat
is going!

John

 

[[email protected]]

[email protected] wrote:
[snip]

I don't get the controversy over the beads--kill the real RF, let the
diff amp handle the rest. Seems simple enough.

I was concerned over the beads causing unbalance in the interfering signal which
then means that differential rejection will be compromised.

I got that, but here the remaining imbalance is but a fraction of an
already small error, hence unimportant.

Did you wind you own ?
Yes.

What's the material ?

Dunno. After using Genome's handy skin-resistance calculator, I
re-wound a commercial SMPS inductor bi-filar to chop its resistance to
1/4th. The rewound inductor gets warm, but it's so intimate with the
other tiny components it's hard to be sure who's doing the heating.
The original was rated to take my peak current, and now it has half the
turns on it; I measure saturation at >>2x my peak current. Only 1-2%
improvement from the re-wind, which rules out copper losses.

I'm driving the switch from a signal gen, *hard*, to rule out
switching losses. Time to mung in a megainductor and see what's what.

Best,
James Arthur

 

[Jim Thompson]

Jim, you seriously upset a lot of people by refusing to be neurotic.

John

I having an evil amount of fun tweaking chins ;-)

...Jim Thompson

 

[Genome]

I have 40-somethings called daughters ;-)

As for "taking the piss out of" Eeyore, a 12-year-old could probably
handle that ;-)

...Jim Thompson

I had the humanity of your daughters in mind when I wrote....

"I can't begin to imagine a situation where they might feel the need to
waste your time by taking the piss out of you."

DNA

 

[Eeyore]

[email protected] wrote:
[snip]

I don't get the controversy over the beads--kill the real RF, let the
diff amp handle the rest. Seems simple enough.

I was concerned over the beads causing unbalance in the interfering signal which
then means that differential rejection will be compromised.

I got that, but here the remaining imbalance is but a fraction of an
already small error, hence unimportant.

That isn't what I've seen.

Is that simply an assumption you're making ?

Dunno. After using Genome's handy skin-resistance calculator, I
re-wound a commercial SMPS inductor bi-filar to chop its resistance to
1/4th. The rewound inductor gets warm, but it's so intimate with the
other tiny components it's hard to be sure who's doing the heating.
The original was rated to take my peak current, and now it has half the
turns on it; I measure saturation at >>2x my peak current. Only 1-2%
improvement from the re-wind, which rules out copper losses.

I'm driving the switch from a signal gen, *hard*, to rule out
switching losses. Time to mung in a megainductor and see what's what.

Epcos's Magnetic Designer application is very handy for inductor and transformer
calculations btw. You will need to know the core material of course.

Graham

 

[Rich Grise, Plainclothes Hippie]

I never did learn to type. It makes me a better programmer, gives me
time to think about what I'm doing.

I want to nominate this for the "rationalization of the year" prize. ;-)

Cheers!
Rich

 

[John Larkin]

I want to nominate this for the "rationalization of the year" prize. ;-)

Cheers!
Rich

No, seriously. There's a new school of programming called XP, Extreme
Programming, and one of the methods is to have two people sit in front
of one computer and discuss every line of code before one of them
types it in. The astounding loss of productivity (two people working
maybe 1/10 as fast) is more than made up by the savings in debug time
and the cost of bugs. I manage 300+ lines of tested, bug-free code per
day, even though I can't type and waste a lot of time in newsgroups.

Slowing down is the best way to get programming done fast.

John

 

[Eeyore]

No, seriously. There's a new school of programming called XP, Extreme
Programming, and one of the methods is to have two people sit in front
of one computer and discuss every line of code before one of them
types it in. The astounding loss of productivity (two people working
maybe 1/10 as fast) is more than made up by the savings in debug time
and the cost of bugs. I manage 300+ lines of tested, bug-free code per
day, even though I can't type and waste a lot of time in newsgroups.

Slowing down is the best way to get programming done fast.

The best way of all is to carefully analyse what needs doing and then code it.

Graham

 

[John Larkin]

The best way of all is to carefully analyse what needs doing and then code it.

Maybe so, but I don't do that. Often I don't know exactly what I want
to do when I start. So I write some code, read it and think it over,
and rewrite it until it seems right, and code mostly bottom-up. In a
realtime hardware/software system, you can't really understand all the
interactions until you're deep into the details. And if I don't like
the hardware interface, I change it. That's especially easy when most
of the logic is an an FPGA.

I just finished a serial communications thing with a command parser.
As I coded it, I found that I didn't like the command language syntax,
so I changed it to make the parser code cleaner and more robust
against pathological input strings. That's sort of working up and down
the abstraction stack as needed.

I think the most important thing, at least for a lone programmer, is
to comment carefully and *read* the code many times, and tune it until
it looks perfect, before you ever run it. Hardware designers tend to
be careful in this manner because it's superficially a lot more
expensive to fix hardware blunders than to fix software mistakes.

John