continuously variable Bessel filter

Uh, pardon the observation, but you're insane.

And I bet you can't tune filters either.

John

John Larkin schrieb:

Hello,

you wrote: The range of interest is about 100KHz to maybe 600 KHz.

There are some values for the bandwith in the data sheet, 300/150/35 kHz. Sorry, the bandwith of the dpot of the link is much to low for your application.

Bye

at

There's a DC accuracy spec? OK, then redesign so the amplified bits are capacitor-coupled. As for linearity, yeah, there you're hosed. Linear, large dynamic range, low noise: pick any two. Trimpots are the norm, of course; laser-trimmed resistors and state-variable filters will also have plenty of trims and more than=20 zero DC offset to deal with, but you maybe could just buy two or three from the module manufacturers... whoever THEY are this week.

Whereas there may be theoretical and practical limits to how well you can do, there are no limits on how badly. In other words, some circuits manage the tradeoff better than others.

John

that

. We

hem,

st

s to

ts

ut.

load

rt

mp

ou

ol.

rse

re

ne.

n.

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0... When Tek was

to

er

on

ot

:-)

ow

One thing Tek did use was "split-path" amplifiers, separating a low frequency (dc accurate) path from a (screaming bandwidth) high frequency path. You might look into some of those techniques. The trick, of course, is how the paths get added back together without uglies in the transition band.

Just

despite

Aye. Neat stuff. Plus fun in getting the delays to match as well.

The 7000 series scopes used fast custom linear ICs that tended to have a lot of DC drift and thermal hooks. The fix was called "feed-beside", a parallel gain path done with slow opamps and such. Merging the signals was easy since the feedbeside stuff was pretty slow. See Addis' piece in Jim Williams' first "Analog Circuit Design" book, the

John

Maxim (I'll duck) has an 8-pole filter - the MAX274 - using the state- variable filter form. It's a bunch of op amps and smallish but nicely tweaked capacitors. It's most definitely not programmable, unless you consider changing resistors as programmable. Much less noisy than a switched-capacitor filter, but the capacitors are too small for it to be quiet, offset voltages can be obnoxious, and the bandwidth is limited.

OTOH I'd rather not have to need loads of high precision capacitors to make a high-order filter feasible.

I just got in a Krohn-Hite 202R, circa 1970, for $80. Dual channel, HP/LP, Besel or Butterworth. It was a bit flaky when I got it, but a bit of DeOxit fixed it right up.

It has 8 pots inside, in two banks of four driven by gears from the front panel knob. Men were real men in those days.(*)

I just missed getting a Krohn-Hite 3905b for $450. Bummer.

Cheers

Phil Hobbs

(*)And women were real women, and small furry creatures from Alpha Centauri were real small furry creatures from Alpha Centauri. (Douglas Adams, "The Hitchhiker's Guide to the Galaxy")