Sallen-Key input Z

Sometimes you can have fun tweaking them under the chin.

A major player is this military-industrial complex had a mailbox inside the grounds. So you could simply walk down to the cafeteria and mail your phone-bill, etc.

When I pointed out that I could also mail myself floppies (dating the contract :-) and even circuit boards, they went into apoplexy.

It was moved to outside the guard shack the next day ;-) ...Jim Thompson

We've been talking about S-K filters in general. The OP was talking about the input impedance, so if your and miso's opining on component sensitivity and clipping is on-topic, so are the DC specs.

Where I'm coming from on this is that (a) the DC gain spec is a legitimate engineering consideration, frequently a very important one, (b) unity-gain S-K filters are the undisputed champs at DC, (c) in instrument design they are frequently more important than a couple of dB of noise peaking that can very readily be taken into account in the system design.

You chip folks were maintaining that there was no legitimate reason for using a S-K, which just isn't so.

It's true that S-Ks have serious drawbacks if you have to use high Qs, e.g. that the tuning uncertainty is something like 3*Q times the component tolerances, as the charts that miso posted point out. But that's not death and destruction, necessarily--it's just another one of the many things you have to keep in mind when designing gizmos.

Nowadays I don't use a lot of high-order analogue filters of any sort--in most of my work, the job of the analogue filters is to get the right SNR and adjacent-channel selectivity and avoid aliasing, not to set the final passband shape.

To get accurate and repeatable AC specs, IME the easiest thing to do is to oversample as much as I can afford, and do the fancy stuff digitally. At that point, any major sins of the analogue filters are way out in the stopband someplace, where I generally don't have to worry about them. Naturally, folks designing 10 Gs/s scopes probably have very different issues.

Cheers

Phil Hobbs

times

output,

by

Learn to read. The DC accurate output is after the resistor. It served Fluke well in its discrete implementation.

It is pointless to compare a switched capacitor design to a continuous time filter made from discrete bipolar op amps. Why would anyone who understand electronics think that is a legitimate comparison?

I have a feeling you wouldn't understand anyway, but thanks for playing.

of

Cool, thanks, I assume the cover holds 'em in place?

Someday, I'll have to do a thermocouple circuit. I've never used a chopper amp, but it seems like the right choice.

'course it'll need a filter too.

George H.

.highlandtechnology.com  jlarkin at highlandtechnology dot com

This is like arguing over Ohms law. There is the right way, i.e. dynamic range adjustment, and there is the wrong way (everything else).

Do me a favor and go tell Dr. Gabor Temes he is full of s*it, then let us know how that goes.

Do you realize that some customer of your company or a potential employee might read your nonsense?

Good engineering practice is just that: good engineering practice. Do you use parts beyond their rated voltage? Presumably the answer is no, since it is good engineering practice to use parts appropriately rated. But in the case of Highland Technology, I'm not so sure that is the case.

For a topology where all node gains can be independently adjusted, IT COSTS NOTHING TO DO SO. No additional parts are required.

Note the LT1062 is representative as a topology. As I pointed out once before, you can (OK, maybe only I can) design arbitrary filters with this topology provided you have a root solver and some knowledge of root locus analysis.

Grin, yeah, miso, I love ya, but I don't know if you're responding to me or someone else?

George H.

.highlandtechnology.com  jlarkin at highlandtechnology dot com

h nodes

e, as is

sign.

nt

to

but

ide quoted text -

Ouch, Back on topic???? Jim my original f-up has long been dead..

Nice thread though! All filter topologies have a use somewhere. (maybe?)

George H.

SK

t -

Hmm, some inherent DC accurate inverter? If you had a circuit that flipped something twice, (?) (using the same resistors??)

George H.

Larkin, and his sycophants, specialize in obfuscation >:-} ...Jim Thompson

Those who complain about your snipping are correct... did you READ what I posted? ...Jim Thompson

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

irrational

Give Larkin 10 extra points in the "village" designation challenge run-off >:-} ...Jim Thompson

e 5

on

Pity about the performance of that part of the kid that was still in school. I know you made the Dean's list on the stuff that you recognised as relevant, but the bits that you didn't recognise as relevant seem to have got the bare minimum of attention needed to pass the course, and were forgotten the day after the exam.

It shows.

times

output,

by

The "output" pin will still load the 28K resistor. Switched-cap filters don't have infinite input impedances; things need to be charged up every clock. Funny how a "DC accurate" parts has no specs for DC accuracy.

You cited this part, not me. It's really crappy.

irrational

Circular definition, pure prejudice.

Do

Sometimes. It's one way to get extreme performance. If a PHEMT is rated at Vd of

An 0603 resistor can dissipate half a watt reliably, if you use it right.

Presumably the answer is no,

Our stuff is very reliable in the field: multiples, often large multiples, of MTBFs calculated from Bellcore or MIL-HBK-217.

Yup, little strip of fairly dense foam.

Or use any old opamp and auto-zero it now and then.

Yup. TC circuits are especially easily fooled by RF and spikes. So you need some front-end passive filtering before you hit any opamps. Jfet opamps are much less sensitive to RF than bipolars.

We had 12 modified SK filters in the Microdyne 700 series Telemetry receivers. Selected by analog mux at the input & output. A buffer before the first mux. Nice clean filters, but some elect in test with