Is it ok to use GBW as a lowpass?

I can buffer the ADCs with another opamp/follower, have 3 of them,

• 4 spare CT/SC blocks which can be used as PGAs. I'm trying to figure out how simple it can be (the metric is the number of external components, I don't care about the internal complexity of the programmable analog fabric). The knowledgeable folks here started growling at this oversimplification, so it must be wrong, but its metric is 2 (two external resistors).

Best regards, Piotr

Switched-capacitor filters are evil.

Digikey's unit price for the LTC1562 is $19!

Once opamp gain falls to not much, it's not really being an opamp. Its char acteristics cease to depend too much on the feedback network. The only way to know for certain if that can work for you is to try one & see.

I'm wondering how an internal opamp capacitor causes differing behaviour fr om one on the nfb loop, I don't think I can properly answer that.

Unless anyone here has actually tried to make what you're proposing work, I doubt folk will really know if such a kludge can be made adequate. How man y people would say yes, the LM386 can be used as a radio receiver and audio amp all in one? Well, it can, running it way way down on gain. So who know s. I don't see any definite answer but building it.

NT

Why? You can spare me two days of checking it myself, so please explain. :-) They are digitally tunable, and that is a major advantage.

Best regards, Piotr

Pricey, but it is excellent. A quiet 8th order linear bandpass filter occupying maybe 1.5cm^2. If you need that level of performance, there is IMHO nothing better. The problem is you rarely do.

Best regards, Piotr

1. They sample the input at the clock rate, so alias high-frequency signals and noise down into the passband. So they need an analog pre-filter.
2. They couple and alias power supply noise too. So they need very quiet power supplies.
3. They shoot nasty charge spikes out of both the input and output, and that can mess up opamps. The output is steps and spikes, so they need an analog output filter, too.
4. They are just plain noisy, probably because of the cmos process.

Other than that, they are great.

It's basically eight opamps and eight capacitors for $19.

A switched-capacitor filter is a sampled-data system, so you have to respect Nyquist there, and you do need analog anti-alias filters.

On the LF band, there is so much crud on the band thet you do need proper front-end filtering before attmpting to listen.

Something isn't clear to me here. How can it be follower mode if, as the first post says, gain is set to 20?

He's got a quad opamp. GH

Piotr is ignoring lots of real-world constraints: slew rate, gain inside of active filters, etc.

Must be young and inexperienced. ...Jim Thompson

This opamp is a part of the PSOC5LP "microcontroller". There four identical copies of this analog block (with direct connectivity to the external world) and four more burried inside, within the Switched Capacitor/Continuous Time blocks. In other words, you may assume there are two quad OpAmps.

Best regards, Piotr

Thanks a lot, John!

Best regards, Piotr

The receiver No. 1 is designed as follows. A loopstick antenna, then a JFET preamp (BF545A), then an 8th order bandpass Bessel filter (LTC1562-2) with BW=2.4kHz, the main amplifier (MCP6022) and the PSOC5LP chip. The PSOC contains a switched capacitor mixer to bring the signal to the baseband (homodyne), PGA and a sigma-delta ADC. Then the DSP part begins.

I've learnt that there is so much noise that the JFET preamplifier is mostly limiting my dynamic range and not helping very much. The gain should be placed after the filter. Its task was also to buffer the loopstick, but it worked too well: I want to base the reception on the phase modulation part of the DCF77 signal, so the Q should not be higher than 30. This happens for the parallel resistance value of ~60kOhm, and it is well within the capabilities of the filter chip, so the JFET preamp can be stripped off.

Now, the sampling frequency of the sigma-delta is another limiting factor (and it implies the existence of the analog frequency conversion frontend, which also works at 77.5kHz, coupling to the loopstick and results in a massive feedback). I would like to switch to the two other SAR ADCs, interleaved, and sample the input signal at 1.86Ms/s, that is 12x the minimal 155kHz sampling frequency. The rest (decimation by 6, a digital quadrature heterodyne/mixer, further decimation and filtering down to the baseband) will be done entirely digitally, so another part of the analog frontend could be removed.

The interleaved ADC will cover the band from DC to 930kHz and I am only interested in the part >=60kHz and

Low noise opamps are not that cheap either. But I agree with you, the chip is 4x too expensive.

Best regards, Piotr

The slew rate is 3 V/?s for 200-pF load, seems plenty. I would like to avoid active filters due to their precision components demands and the lack of tunability. The preferred direction is to move as much as possible to the digital domain, so ignoring the internals of an unwanted component is rather excused.

I wish I were young, but the second part is sort of accurate: I am just a hobbyist and weak signal reception in the presence of massive noise is not my area of expertise even by these standards. I prefer high power (that is,

I got burned on this when TI got rid of the OPA655 and I had to replace it with the OPA657. To TI the OP657 was the same as the OPA655 but better beca use it had more GBW. For me, I had more noise in the stop band because the GBW was my second pole. I had to add more filtering so the OPA657 wasn't a drop in replacement for the OPA655.