Pre Amp gain

[golfnut]

What is the highest preamp gain possible using the 5543... 50dB , 55dB , 60 ?

The IDX sits about 46dB, would 10dB ectra help.


S

[Tepco]

Hardly can help, more likely will just produce more trouble. 5534 gain is limited by it's open loop response, that is like 80dB at IDX freq. but increasing it too much will also increase coil null signal and interferences picked up. Some moderate increase may be useful, add 10dB, but most of the gain is achieved in post-demodulator stages, better try to increase that.

[Davor]

Yes, it's only function is to improve S/N, while most gain is done by post-demodulator gain blocks. When saying "noise", it is both thermal noise voltage at front end and also noise in gain blocks. While NE5534 operates at about 4nV/sqrt(Hz), the gain blocks run well within the 1/f frequency range, and effectively at about 100nV/sqrt(Hz) or so. Hence to overcome the gain blocks noise you need over 25 gain (28dB). Upper safe limit is set by the residual unbalanced signal and the opamp distortion. THD is not giving the full spec, but may give you a sense of proportion. Short version: if your null is over 100mV - you are screwed.

[Tepco]

Yes, low noise front end, but on full BW, then downconverted, to whatever noise performance filters, with like 10Hz BW. Just wondering, if ever someone tried double conversion, first to upconvert and filter at some higher freq. above TX, and then downconvert to zero IF?

[Tepco]

And just one, unrelated observation. Time ago, I played with Russian design KROT detector, MCU based all digital. Limited with 10bit ADC on Atmega 8 chip, so obviously limited dynamic range. With very high transformer ratio coil, high gain amp, RX\TX separated only about 10%, I get like 1V null signal at chip input, and it worked just fine, all the way until ground signal became too strong to saturate ADC input. This time high gain worked, even with 1V residual voltage, actual detection threshold is less than mV imposed on it.

[golfnut]

Original gain is 45dB.


Its not a big mod, Ill try this week.


Is it better to go for smaller than the 1k
OR
Larger than the 180k

Ill aim for 470R I think for 51.6dB

I am aware that the extra gain will give more noise at the comparator, more compression problems from medium/larger targets, potential oscillation problems.

But I NEED that extra 1" on legitimate small interesting targets.

S

ps I was looking at the DC level on the comparator threshold for quiet in my home. It is about 1/2 travel is +1mV.
disc ah level sits about -1mV
Geb about 0v

Signals on both Channels getting about +3mV trips the tone - - - Knowing 3mV at the output is valid detect, the input signal is then about 10-20uV range.

Fairly impressive for older design cheapish kit.

S

[Davor]

Just wondering, if ever someone tried double conversion, first to upconvert and filter at some higher freq. above TX, and then downconvert to zero IF?

I actually thought of that for my future LF build, but due to a moment of epiphany, I realised that a much better option lurks from the fact that the synchronous switches are in fact choppers in disguise. So instead of low-pass filtering right after the switches it will be opportune to amplify the already chopped signal, configure the gain block as a "Correlated Double Sampling" (or other appropriate chopper amplifier) and low pass it later on. Provided the switchers are in QSD configuration, and switching is performed in quadrature T/4 fashion. It would alleviate 1/f problem once and for good, so a true slow motion or a true non motion detectors will become reality. There are a few bonuses as well.

The "option B" approach would be "autozero" correction amplifier applied to the offset circuitry to reduce 1/f, but it makes sense only with non-motion builds. Otherwise the synchronous detector/chopper is a more economical approach.

As for preamps, I'm inclined to (finally) make a front end with discrete BJTs. They are still performing much better than op amps at low source resistance. There are a few exciting configurations I'm playing in simulation occasionally, waiting for "a proof of a pudding" stage.

[golfnut]

first to upconvert and filter at some higher freq. above TX, and then downconvert to zero IF?

Upconvert will add noise, downmix reduces back to start noise, If noiseless - which do not exist.


DSP has to be the way to go, with good A/D 120dB dynamic range, keep Rx fully differential, full real time AGC to keep Rx signals in within full span of A to D for max S/N, compression and blocking tolerance.

S

[Davor]

Is it better to go for smaller than the 1k
OR
Larger than the 180k

Ill aim for 470R I think for 51.6dB

Don't reduce 1k as it affects coil loading, and phase shift in a process. An inverting configuration input presents itself as a 1k resistance (- input is equal to ground). 1k is also a good match for noise performance of this op amp.