Menu
What's new
Find's Treasure Forums

Welcome to Find's Treasure Forums, Guests!

You are viewing this forums as a guest which limits you to read only status.

Only registered members may post stories, questions, classifieds, reply to other posts, contact other members using built in messaging and use many other features found on these forums.

Why not register and join us today? It's free! (We don't share your email addresses with anyone.) We keep email addresses of our users to protect them and others from bad people posting things they shouldn't.

Click here to register!

Need Support Help?

Cannot log in?, click here to have new password emailed to you

NE5534 v AD8031 v AD8055

Hi Carl,
I guess if have a differential sampler that has a gain of +1 in one state and -1 in the other state. Then sampling 1 volt dc would go +1,-1,+1,-1 at the sampling frequency. Integrated this would be zero volts. Sampling zero volts dc would yield zero volts and no sampling frequency.
Not sure if Robert needs to worry too much about this getting through the integrator (some small amount always will) since it would go back and forth a little at the sample frequency but not really cause a dc shift representing a target.
JC
 
JC
Exactly! A DC input of +V produces an output sequence from the differential sampler of +V, -V, +V, -V . . . at a frequency of half the sampling rate. My point was that this is the frequency that you do not have to worry about. This frequency coming out of the sampler is so far out of the pass band of the integrator that the output of the integrator will effectively be 0.
The input frequencies you do have to worry about are those near the pulse rate and all its odd harmonics. Those are the input frequencies that will give an output from the sampler that is near 0 Hz. These low frequencies coming out of the sampler will not be attenuated by the integrator, they will be amplified.
So for a 5 kHz pulse rate it is input noise that is near 5 kHz, 15 kHz, 25 kHz, and so on that is going to get through the integrator and show up as noise at the output. Whereas input noise that is near 0 Hz will be transformed to a frequency that is very effectively blocked by the integrator.
I have been describing this as if the circuit were split into a differential sampler followed by a simple integrator because I think it is easier to describe this way, and that is how it is implemented when using an ADC to do the sampling. Carl's circuit is actually a two channel sampler followed by a differential integrator. The result is the same but my description may be hard to follow for someone who is looking at Carl's circuit.
Robert
 
Nick
Ok, if the charge pump is not synchronous with the pulse rate then it could be a source of noise. Also the audio chopper could be another source, and there might be others. But I do not know how to estimate the amount of noise that could be picked up from these sources. The only way I would be able to do it is build the circuit and test it. I would be interested to hear what you find.
Robert
 
Both the Aquastar and Goldquest use synchronous charge pumps, so there is no noise from that source. The audio chopper is not driven from a synchronous oscillator, but I have not seen any evidence of interference. It is well away from the front end and doesn't have a ground return common with the receiver.
Eric.
 
Robert,
The effect of frequencies both near the sampling rate and at the odd harmonics is well taken. As a point of interest, most new designs take two different samples. The first sample being the main sample with the second sample subtracted from the first. The second sample is taken after the receive signal has decayed leaving the earths field component and noise common to both samples.
With a difference between the two samples at say 100uS a 60Hz power line hum signal advances only 2.16 degrees between the samples. Using a circuit with a high CMRR one can see that there is not much noise left after subtracting the second sample from the first.
I believe that much of the noise which gets through many designs is due to poor differential integrator CMRR. I use a precision difference amplifier instead of a differential integrator. My lab TV set sits just eight feet from the searchcoil and I get almost no interference from the TV at all.
As is so often the case with detector circuits, dual sampling was invented by Eric Foster back in the 60's before Whites reinvented it! Hats off to Eric.
 
Yes you are right. Seeing it graphically makes it much easier. Some people can look at the formulas and see the graphical representation. I wish I could!! <IMG SRC="/forums/images/smile.gif" BORDER=0 ALT=":)">
HH
Beachcomber
 
I was actually thinking of the period between pulses as being the problem - when the analog switches are turned off, the integrator inputs are "floating" and presumably the charge pump is operating. I just did a quick test with my detector running "quiet" ie. I reprogrammed the microcontroller to produce no pulses and keep the switches off, disconnected the coil and looked at the noise at the output of the integrator on my oscilloscope, connecting a 10k resistor between the input of the integrator and the ground seemed to slightly reduce the noise but not by much. This test was pretty "quick and dirty" though, if I find anything more definite I'll post more.
 
Here are some chart recorder plots of the noise from the front end amplifiers under discussion. The board used was as used in the Goldquest SS running at 10kHz pulse rate. The search coil was not plugged in, so the amplifier input was just to the damping resistor network, which totals about 600 ohms. The chart recorder was connected to the output of the high pass filter/amplifier that follows the integrator. This stage has a gain of 48x, so dividing the results by this figure will give the noise at the integrator output. The chart recorder was run at 10mm/min, so each plot represents 2.5 minutes running time. The horizontal scale for all the amplifiers is the same, at 200mV per 10mm. At a chart speed of 10mm/min the noise spikes are compressed together, but the difference in noise levels is quite clear and would be representative of the smoothness, or otherwise, of the audio threshold. For comparison, a 709 was tried and came out as the quietest of the bunch. It also settles fast enough to sample at 10uS with the coil connected.
To look at the noise contribution of the sampling gates and integrator stage, the front end amplifier was removed and the sources of the two jfets grounded via a 470 ohm resistor. This represents the impedance seen by the sources, back to the amplifier output. The chart recorder sensitivity had to be turned up to 2mV/10mm and this plot is marked as REF. Hence the pk-pk noise of the sampling and integrating circuits are is about 100 times less than when a 5534 is inserted.
Eric.
 
Eric
I think you have talked me out of the 8031, but you have not talked me into a 709 yet. I am pretty sure I have a couple of 709's in metal cans sitting a drawer somewhere, but I think that is where they are going to stay.
Your results are very interesting though.
Robert
 
Eric,
Good stuff, but like Robert I think my 709s will stay in hiding for a while...
What about the 797?
Nicko
 
Hi Nicko,
Hopefully, I should have some AD797 samples arriving shortly. Don't worry, I'm not going back to the 709, although National still make them. Interesting to compare a 1960's IC though.
Eric.
 
You must log in or register to reply here.
Top