Active filter with gain

Hah, Here's an easy poor man's lockin. Take a reference oscillator, something that will modulate your signal. (maybe wiggling a B-field, or blinking an LED) Send that to channel

2 of your DSO, and trigger on it. Put your noisy signal into channel 1. Hit the average button... maximum number... Bingo there's your signal! (You've probably already done something like this in the past.)

I hope you will buy the fancy LT part and report results.

(I think Phil is concerned with repeatable performance from matching filter parts, but if the C's are internal (and trimmed by LT) and you use 0.1% R's it might be fine.)

George H.

Reply to
George Herold
Loading thread data ...

Piotr,

If you're going to listen to DCF77, please make a decent superheterodyne receiver. It is not difficult to mix the signal down to say 20 kHz and sample it from there with quadrature A/D sampling. The rest is just code.

--

-TV
Reply to
Tauno Voipio

Yes, the PCB house has just started with my design, so it'll take some time.

The 0.1% parts are easily available, but FilterCAD proposes to use 1% resistors. Strange.

BTW, I am reading some papers on lock-in amplifiers and the basic theory behind them is very simple. But it is so just in the case of locally available exciting signal, then it could even be easy. BUT: if you plan to use it in a receiver, then how are you going to lock your LO to the input signal if it is way below the noise floor? A PLL with a a VERY narrow bandpass filter? A kind of chicken and egg problem. Enlighten me, please. ;-)

Best regards, Piotr

Reply to
Piotr Wyderski

Aha! See "WWVB-Schematic+Data.pdf" on the S.E.D/Schematics Page of my website for a design I did _43_years_ago_... basically a PLL doing synchronous AGC to detect the time signals. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

The key advantage there is that the narrow filter in a PLL is at baseband, so you use a simple active RC lowpass. One wrinkle is the frequency compensation--a VCO plus phase detector is an integrator (the input controls the frequency and the output measures the phase). You usually want a second-order loop, which is easy to do: an op amp with series RC feedback acts as an integrator at low frequency (for low static phase error), then at the RC corner frequency turns back into a proportional amplifier. You put that corner (which is a zero of the FB function) near where the integrator * VCO/phase detector would have crossed unity gain, then add a couple of RC lowpasses with corners well above there, to suppress the phase detector ripple.

At really low SNR and low bandwidth, pull-in is too slow and unreliable., so you'll need an acquisition aid. I usually use a trick I invented about 1982 (which somebody else had come up with a few years before) which turns the loop amp into a slow triangle oscillator when it loses lock. You just add a little positive feedback--when the loop loses lock, its negative feedback loop gain goes to zero, so the weak positive feedback dominates, so it sweeps back and forth till it reacquires lock.

You need to make sure that d^2 F/dt^2

Reply to
Phil Hobbs

Yeah from my point of view a lockin always has some reference oscillator. Synchronous detection.. maybe we need two names? If you've got a reference that is moving around some, I know they make PLLs that lock-on to the reference, but I've only used them and never built one. I'm afraid I have little enlightenment to offer.

George H.

Reply to
George Herold

I used a EG&G 124A back in grad school, it had some search function when it lost lock, but I'm not sure how it worked.

George H.

Reply to
George Herold

For anyone interested in low-level signal acquisition and tracking, the 124A makes interesting reading.

Specifications: https://testequipment.center/Product_Documents/EG-G-124A-Specifications- D9A48.pdf

An interesting point is they claim tracking over a 100:1 frequency range. This implies a multivibrator-type vco, not known for low jitter.

User manual with complete system schematics:

formatting link

Reply to
Steve Wilson

With good enough components it might work, but remember that the tolerance on GBW is nowhere near as tight as that.

In a 2-phase lock-in, you need the quadrature phase anyway, so you can easily use a phase-frequency detector for acquisition. The internal reference is locked to a high-SNR reference input rather than being derived from the weak signal input.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

It isn't necessarily done in a single range. And your average old timey lock-in has a minimum output time constant of 1 ms, i.e. 160 Hz bandwidth. So even a 555's jitter can generally be controlled pretty well by the loop. In any case, nobody expects better than 0.1% accuracy from a lock-in, and they're only that good in a narrow range around 1 kHz.

If they liked, they could use an octave-band LC VCO and divide it down in octave steps. A counter or F-V converter attached to the reference input would let it choose which step to be on fairly straightforwardly. You could get really wide range varactors for AM radios back then, which had just the right characteristics for the job.

I have an EG&G PAR 5208 two-phase lock-in that needs re-capping but is otherwise in fine shape.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Well I've only made active two poles filters...

But as long as you're far enough away from the GBW it doesn't matter so much.

And how 'good' is the GBW of an opamp? I look carefully at this one filter 100kHz (max) LP, OPA2134 (GBW=8MHz) Q=.707 (at low end) At 100kHz the Q is ~0.725. (don't quote me on the number I'd have to check if it mattered.) But there isn't much variation part to part. (I was thinking I should put a better opamp in this circuit.)

I'm going to have to buy Gardners' phase lock techniques book.

I've always wanted to make a two phase lockin. I made an analog quadrature oscillator, ~1% phase error in the past. DDS these days... maybe buy someone else's box?

George H.

Reply to
George Herold

?

You can make a pretty good 2-phase lock-in with two dual SPDT mux chips, a

nd Cs.

Making something as general as an EG&G box takes a few more steps, of cours e. ;)

Cheers

Phil Hobbs

Reply to
pcdhobbs

?

and Cs.

OK If I ever have a use for it I'll ask for details... I'm mostly thinking about using my rigol func gen. (dual outputs) and some comparators-> +/-1 oapmps.

rse. ;) SRS makes an updated version of the 124A, several $k. From talking to the guy who designed it there's still some market demand in the low temperature physics community.. maybe others.(?)

George H.

Reply to
George Herold

That works OK at low speed, but most op amps have somewhat asymmetrical slew, so the accuracy / frequency tradeoff isn't as good.

Cheers

Phil Hobbs

Reply to
pcdhobbs

What's the analog signal path for your circuit... through the mux to an RC? I hadn't thought about slew rate.... I haven't thought about making lockins that much.... The only thing we sell uses the AD630.. I could use two of those (at low freq.)

George H.

Reply to
George Herold

Each 2->1 mux switches between +1 and -1 copies of the input signal. One mux runs on the 0 degree phase of the clock and one on the 180 degree phase. Matched RC filters follow (just to knock the fast edges down to something the next stage can follow easily) and then you subtract.

Doing it that way gets rid of the charge injection to leading order, which helps reduce the offset.

There are other ways to do it, but that's the best one I know, mostly because there are excellent diff amps that do a great job making +1 and

-1 gain.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Syncronous detection, the PSoC way:

formatting link

Cheers

Klaus

Reply to
Klaus Kragelund

Cool, you can replace a, HC74 with a PSoC! What will they thin of next. ;)

Cheers

Phil Hobbs

Reply to
pcdhobbs

{>:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

I agree that it would only make sense if you have the PSoC in there anyway

But it's not just the HC,it's also the modulator, RCs and ADC, so a bit more

Also the PSoC does context switching, so you can use the synchronous detection blocks for something else just reprogramming the SoC

You can't do that with the HC

I once did a PSoC 1 design that used the same blocks 4 times over

Saved over 50 parts and squised the HW into a volume that would have been impossible otherwise. Hard to put a price tag on that

Cheers

Klaus

Cheers

Klaus

Reply to
Klaus Kragelund

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.