usual fast preamp questions

[Mr.Jaick]

Hello
i have some questions regarding some of the previous preamp designs in our forum
off we go:
the example

if anyone would take a minute to answer them, I'll highly appreciate it.
I'll try to explain my thoughts in detail so you can tell where i'm wrong.
Questions and my current answers to them:
starting from AMP feedback caps

1_ 38pF and 23pF capacitors on the feedback path, is it for high frequency and noise cancellation or what else?
the cutoff (-3dB) frequency of first stage is about 748 khz
the second stage is about 844 khz
I calculated these with the feedback resistor(the large one of course) and feedback caps, is it right?
these numbers look a bit low for me since this particular two stage pre amp is capable of more than 1Mhz BW (based on calculations)


2_ another low pass filter in between the two stages, 1k and two 470pF in series, hummm
the cut off frequency of this guy is about 677 khz
beside the weird combination of the filters on cutoff frequency, " WHY to use two 470pF in series to form a 235pF cap? "
is it the value that needs to be precise or some other reason? this config ain't gonna give lower ESR or ESL, but maybe lower leakage possibly


3_ the RC low pass "or" high pass filter on the output of any preamp regardless of the preamp itself.
my way of understanding it:
Low Pass = Noise cancelling? although I'd worry about signal fidelity if i'd use one, gotta be careful with the cutoff frequency.(what should we aim for?)
High Pass = no need to adjust the DC offset, but noises are free to run like the wind blows...
IS IT CORRECT?
what if we add a good 2nd or 4th order sallen key bandpass? (from what Fmin to what Fmax i'm wondering)


4_ now let's get into the lovely bounding diodes.
first stage:
clips the output to one diode drop (+0.7V) when coil is charging i.e tx on (great)
but on the decay side it clips the decay curve to two diodes (-1.4V) which is very unnecessary in my humble opinion because the second stage
will amplify it by 37, so only a 0.2v is needed to absolutely saturate the second stage output! can't we just use two back to back schottky diodes?
schottky diodes will give it serious rise in dv/dt which is slew rate basically.

second stage:
when tx on it behaves the same as the first stage +0.7v
and when decay comes on, it suppose to limit to negative three diodes (-2.1V) , but does it?
because there are two diodes between output and ground so i think it will stop at -1.4V and make those three diodes useless, am i wrong on that?
BTW what's the purpose of the 270 ohm res.?

I apologize for this long romance, i couldn't help myself
it's been almost two weeks that i can't sleep fine with these questions hovering in my head
Op-Amp nightmares every night, you can imagine...

[6666]

Invalid Attachment

[Mr.Jaick]

[Carl-NC]

1. Numbers are correct. It's often more useful in PI systems to think of the preamp BW in terms of tau. Without the cap the first stage is limited by the opamp to 63MHz/25.5 = 2.5MHz, which has a tau of 64ns. Generally you don't need a preamp tau any higher than 5x the smallest tau you want to detect, which in this case would be 300ns. Are you wanting to detect a 300ns target? If not, then throttle back and cut some thermal noise in the process. If you only want to detect a 1us target then shoot for a 200ns tau. For a 5.6k resistor that would be a 36pF cap. Hey!

2. I would not use 2 caps in series. Probably ever. This extra filter provides very marginal noise reduction.

3. Both LP & HP can reduce noise. HP does decouple DC offset but can also reduce e.g. power line noise.

4a. I agree, not sure why the first stage has 2 diodes. 1 diode will do it.
4b. The 270R is to current-limit the opamp when the 2 output diodes turn on. The output can still continue to drop and if it drops another 0.6v or so the 3 feedback diodes will turn on.

[Mr.Jaick]

Thank you very much master
certainly learned a lot

i'm also wondering about this!


integrator in the PreAmp!
right now the only thing that i can think of is:" it will kill the signal "
it's gonna average out everything, how we're going to see those subtle changes produced by target metals?

[Carl-NC]

i'm also wondering about this!
[ATTACH]54562[/ATTACH]

Probably a typo. I deny responsibility.

[Mr.Jaick]

Understood.
Thanks again.

[Mr.Jaick]

This question is about Pre-Amp for a specific ADC:
let's say we have an ADC with 100kHz Sample rate

1) For a PI design, what should be the bandwidth of the Pre-Amp?

as per Nyquist theorem we need to sample 2 times the highest frequency, but that's definitely not enough for our applications
how about 5 times minimum and 10 times for medium work
so we need 20kHz (with 5X sample) or 10kHz (with 10X sample) pre-Amp bandwidths.

2) what exactly happens if we use a Pre-Amp that is as fast as ADC sample rate (100kHz in this case) or beyond ?
random numbers all over the place ?

[Carl-NC]

Are you talking about direct-sampling the preamp (that is, no demodulators)? Regardless of this, you normally set the BW of the preamp based on the desired targets you want to detect. For example, if you want to detect a 10us target then the preamp tau should be no more than 2us, which means the BW should be no less than 80kHz.

In a direct-sampling application you will normally sample at the pulse rate, or maybe 2x or 4x the pulse rate. If the pulse rate is 1kHz and you need a target sample plus a late EFE sample, then you would run the ADC at 2kHz. In direct sampling, such a low sample rate with such a wide BW results in multiple Nyquist zones of noise getting folded into the sample, which decreases SNR. You can improve SNR by averaging a large number of samples, say, 64x. The more the better SNR, but the worse the response latency. If you run the ADC way faster than the pulse rate then you end up sampling the decay curve all over the place which is not helpful unless you intend to do some really intense DSP.

[Mr.Jaick]

yes i meant to say direct sampling

not sure how 80khz came along the way, 2us is 500khz and 10us is 100khz
are we being generous about that ?

other than that, well explained and well understood thank you.

and i'm thinking about the previous questions and answers.
1) you said "throttle back and cut some thermal noise in the process"
how do we reduce thermal noise ?, by lowering the Amp speed with lowpass filter?
because i think the op amp will not operate slower internally but it will shove that fast current to feedback cap so the output is slow

2) I am hesitant to use capacitors in the feedback of the preamp because i think they will greatly reduce Slew Rate of the Amp
any other way to limit the bandwidth without compromising?

3) about the three bounding diodes in the second amp feedback and those two extras on the output
couldn't we just use either one of them, or is it necessary to go with both? maybe a leftover from the original scheme?

[waltr]

you said "throttle back and cut some thermal noise in the process"
how do we reduce thermal noise ?

Resistors are a source of thermal noise known as Johnson noise.
https://en.wikipedia.org/wiki/Johnso...3Nyquist_noise

Reducing the resistor values can reduce this noise.

[Mr.Jaick]

Resistors are a source of thermal noise known as Johnson noise.
https://en.wikipedia.org/wiki/Johnso...3Nyquist_noise

Reducing the resistor values can reduce this noise.

yeah exactly
but what i'm wondering about is what Carl meant about lowering Johnson noise with limiting the bandwidth
maybe I'm getting it wrong

[Carl-NC]

not sure how 80khz came along the way, 2us is 500khz and 10us is 100khz
are we being generous about that ?

That's for omega. Bandwidth (f) is 1/(2*PI*tau).

1) you said "throttle back and cut some thermal noise in the process"
how do we reduce thermal noise ?, by lowering the Amp speed with lowpass filter?
because i think the op amp will not operate slower internally but it will shove that fast current to feedback cap so the output is slow

Yes, bandwidth-limit the opamp. It will also bandwidth-limit the noise.

2) I am hesitant to use capacitors in the feedback of the preamp because i think they will greatly reduce Slew Rate of the Amp
any other way to limit the bandwidth without compromising?

Opamp BW affects the linear (non-saturated) operation. In non-linear (slew-limited) operation the feedback cap has little effect.

3) about the three bounding diodes in the second amp feedback and those two extras on the output
couldn't we just use either one of them, or is it necessary to go with both? maybe a leftover from the original scheme?

The 3 diodes limit the large-signal gain of the opamp but don't necessarily limit the output voltage. The 2 diodes limit the output voltage.

[Mr.Jaick]

the 3 diodes prevent the amp from hitting the rails(for better slew rate all that) but why should we limit the output voltage that goes into switch/integrator?

I am currently thinking about differential preamp and I have several ideas but not certain
you might've seen some preAmps before, but get ready my friend

but before that i need to know this:
is the noise coming from the coil common mode noise or not? I guess not

A) if it was common mode we could pre amplify then give it to a precision single to differential ended converter and we can use those two signals for our dual ADC
or just using a differential amp after that and then to demodulators.

I got familiar with current transformers just today
i think we can use them
B) let's say we measure the R2 current(the second damping resistor under protection diodes) to just get the under 0.7v current density not the whole flyback
that's our primary winding, and we give the secondary to a shunt resistor and measure that with a current amp, or alternatively to super matched BJTs
this is good for a simple approach
now is that differential? 90% not
but i think it will reduce noise because of transformer

the Alternative to this is to measure the damping resistor current so we get the flyback shape but "Current" material and we can analyze it better
now that is differential I believe

thank you at the end I appreciate your time and sorry if i'm being annoying sometimes.

[Vadim18]

Please correct me if I am wrong about the time constant of the opamp.
If i want to reject targets such us medium size nails with TC say ~80uS, my amplifier minimum TC should be 80/5=16 uS.
And finaly In classical one stage amplifier for TC 16 uS i should place ~15 pF capacitor and 1000k feadback resistor.
Thanks.