J
Joerg
Spehro said:
I think they would flog me if I'd just mentioned that
-
Categories
-
Platforms
-
Content
I think they would flog me if I'd just mentioned that
J
Joerg
Spehro said:
I think we can cover that part as well because one of the guys in our
group is quite experienced in the production of really small stuff. Not
an EE, but in the end an inductor is nothing else that a coil spring
from an ME point of view.
D
Dimitrij Klingbeil
Joerg said:
Ouch! Looks decidedly non-trivial.
Whatever you end up doing, beware of the temperature dependence of ferrites.
Unless the temperatures involved are practically constant, getting the
resonance of a ferrite-core inductor stable over temperature may prove a
considerable headache.
Recently I've been bitten by this less-than-well-documented property of some
no-name ferrite. Got a new, but rather inexpensive spectrum analyzer and
found out that it would not track the signal of it's own built-in tracking
generator. When turned on, set to 20 kHz RBW, and connected directly
output-to-input it would see no steady signal, just wavy lines on the
screen, resulting from slope demodulation of some PLL's loop ringing, the
receiver's center frequency being way off-center from the TG. As it warmed
up, ca. 20 minutes later, it would slowly drift into the frequency it's been
adjusted for at the factory and show a stable signal for a while. After 1-2
hours, as it heated up even more, it would drift out of tolerance and the
signal would slowly disappear into wavy lines again. At 400 kHz RBW however
all would look well all the time.
The manufacturer from Shenzhen seemed to hardly understand the issue, and
shipping the thing back looked like a lot of hassle and additional expenses,
so I ended up repairing it instead. I found out that the receiver's third IF
stage was done with LC filters, the Ls wound on some unknown ferrite core
(all 6 Ls in the rather symmetrical filer being equal). The Cs were NPO and
had sufficient stability, but the ferrites were really drifty, although all
equal to each other (which fortunately made the whole filter drift equally
in one direction and not spread).
In the end, I ended up replacing the local crystal oscillator that was
driving the mixer of this stage with a wide tuning range (350 ppm) VCXO
(plus a little amplification and filtering on the VCXO's output) and using
an LM34 temperature sensor plus an opamp to control the VCXO's frequency.
The LM34 was put into the drifty third stage module near one of the filters
so it could track the temperature. After setting this up, running the
spectrum analyzer throug some 5 or 6 temperature cycles and adjusting gain
and offset of the newly-added 'compensation' as it went, I finally got it
sufficiently stable. It ended up with a below 3-ish kHz drift over
temperature as opposed to the manufacturer's originally specified 10 kHz and
the actual measured 50 to 70 kHz drift when I first got it. A factor of 20,
more or less, and no more wavy lines.
When I finally had a look at the service manual that they decided to send me
after some waiting, and found there instructions to calibrate the center
frequency after at least half an hour of operation, somehow I was no longer
surprised. So much for ferrites and their more insidious propertied.
So, if you end up with a micro-miniature ferrite in your application, and
need to track it's resonance frequency, make sure to design in some local
temperature sensing, so you can correct out the nonlinearities later. If
that's not possible, make sure to select and test the ferrite very well.
Sure, given your well-known profession, if it's a medical device that can
take a more or less 36.6°C environment for granted, the above may not apply,
it also won't apply if you are only interested in the changes of the
frequency and can safely ignore the absolute value. Otherwise a ferrite-LC
may go well with some compensation.
Regards
Dimitrij Klingbeil
J
Joerg
John said:
Running another cable is completely impossible in this app.
We could do that. But doing it in the frequency domain is probably
easier and not so prone to nightmares at the EMC lab.
J
Joerg
Dimitrij said:
That's why there's engineers in the world
Luckily this application sees an almost constant temperature, and one
that is always the same. But slow drift would not hurt because we could
calculate that out. And we'll have to because just the mere
repositioning of the cable will cause drift. It is inevitavbly part of
the resonant circuitry, at least to some extent.
Yikes! That looks like a design from Uncle Chen's Backyard Works.
I probably would have returned the whole thing. But I understand the
urge, you are an engineer and it just has to be fixed. I am the same way
sometimes.
The analyzer I bought last year is a Signalhound. It was something
around $1600 including the tracking generator. It tracks beautifully and
has to in my cases. For example, when you must check out crystal filter
structures the track gen has to be in lockstep even when RBW is only a
few Hertz. That's impossible with anything purely LC-based.
In this case I really don't have to worry about thermal drift. Only
about size, plus the fact that mankind has to be able to produce this.
T
Tim Williams
Spehro Pefhany said:
Wescott, I presume?
There's an awful lot of Tims (or Jims) and Williamses in this business... go
figure!
Tim (not the filter guy) Williams
S
Spehro Pefhany
<snip>
Care to mention the maker? Atten?
Best regards,
Spehro Pefhany
M
Martin Riddle
T
tm
Jeff Liebermann said:
I wonder if there is a recording head that might be that small? Maybe a
video head?
tm
S
Spehro Pefhany
Mmmm.. looks like it would be good with smoked porkchops (Kasseler)
from our local Euro meats place.
Do you recommend with or w/o horseradish? Generally I'd probably
prefer the latter.
Best regards,
Spehro Pefhany
J
Jasen Betts
can you stick an oscilator in there? A 555 die is not going to fit, but
perhaps there's something else that would.
M
Martin Riddle
Spehro Pefhany said:
I prefer with horseradish. I don't think I ever tried it without.
It has a nice Zing from the horseradish, but not over whelming.
Cheers
A
axolotl
Getting ferrite down to .004 may be a bridge too far. You may want to
look at shaving off a sliver of .004 amorphous steel. For
experimentation, you can find a piece in the anti theft tags of your
next big box store purchase.
Kevin Gallimore
D
Dimitrij Klingbeil
Spehro Pefhany said:
Yes. AT6011 of ca. 2010 make. Also rebranded SA6011T by MCP Shanghai. It
looks like a newer version of the Hameg HM5011 with a wider sweep span
range, and an LCD for the frequency and settings readout. Actually, given
the HM5011 horrible datasheet performance (100 kHz frequency offset, and
150 kHz drift per hour), the Atten looks like a noticeable improvement,
even when it's third stage is not compensated. I wonder, how the TG on the
Hameg's "original" could even have worked at all. Maybe it never did or
they've redefined "work" for 400 kHz RBW only. At least Atten used a TCXO
and a DDS IC followed by a PLL for upscaling where Hameg originally had used
a VCO with analog tuning and (as it seems) free-running without a frequency
feedback at all, so at least Atten's first two IF stages and TG are stable.
The AT6011's third IF stage is the only one not controlled by the central
TCXO, so while everything else is in lockstep, this stage relies on LCs.
In a pinch, it is fixable by way of a SI550BG32M1250DG (VCXO), LM35 (temp
sensor) and some other minor supporting circuitry, if you happen to have
one and need an idea how to make it behave properly.
That was 2010 however and they could have changed suppliers in the mean
time. No telling, what ferrite is in there now. No idea if the correction
would still work the same on newer models or if it would even be needed.
Given that re-use is popular in many places, chances are that this third IF
stage module with its 32.125MHz quartz LO and string of filters switchable
between 2 and 6 resonant LCs is also used in other models. The AT6011 is
very modular after all, with a minimum of interconnects between the modules,
so its design easily "invites" copying.
Regards
Dimitrij Klingbeil
J
Joerg
Jasen said:
Nope, can't :-(
J
Joerg
Jeff said:
Might work. I want to get the most L-increase possible, whichever
concoction of materials gives me that.
J
Joerg
axolotl said:
I disected one of those. The store left them on, just deactivated them
at the cash register (which I thought was environmentally rather
wasteful). There was only a very thin film in it.
J
Joerg
Martin said:
Considering the family name probably a Bavarian recipe. Maybe from
great-great-great-grandpa back in Europe.
W
whit3rd
All metals, with metallic bonds, are conductive; metglas just has higher
resistivity than most, it isn't quite an insulator.
It takes some work to shape (because mechanical stress and heat
will destroy the glass structure), but chemical etching is effective.
Some of the ferromagnetic metglas materials are VERY FAMILIAR
to most folk: for a decade or more, tuned-length metglas strips
have been used in security labels (check out your DVD collection
for little plastic 'security device' gizmos).
S
Spehro Pefhany
http://www.ediblecommunities.com/queens/fall-2010/notable-edibles-fall-2010.htm
According to this, it's a one-man show. And they use the GOOD mustard
seed.
I've ordered a case of the horseradish type ($25 shipped), I'm sure
it's at least comparable to good European mustards.
Best regards,
Spehro Pefhany
