The Tayloe Mixer: A Low Noise Solution

[simonbaker]

I'd be interested in anyone describing what "noise" problem is related to these switches and MD circuits and how one chip might be different from another.

Are the switches simply adding a noise signal while they are switched on, acting like a noisy wire? Or amplitude modulating the signal while they are on because the resistance doesn't stay constant? Is it just semiconductor noise?

Regards,

-SB

[moodz]

noise stuff ....

I am not sure about this upgrading of old detectors when the maybe the whole design should be changed ... however I have some experience with RF mixers ....and front end of ADC sample and holds ..... you do want the switch to be fast and have low on resistance AND the voltage source you are sampling to be able to "drive" the sampler because it has to charge the sample capacitor when the switch is on.

Think of a tap and bucket .... if the water pressure ( input voltage ) is low the tap ( switch ) will have to be on for a longer time to get a measureable sample ( bucket ).
Likewise if the tap only turns on and off slowly ( switching time ) then again it is harder to ensure an accurate sample....is the tap half on or half off. Furthermore lets say the tap is partially blocked ( on resistance ) .... for a certain pressure ( voltage ) it is easier to measure one cup of water that flows out than 1 drop for the blocked condition.

thats my 2 cents.

moodz

[mario]

Hi guys, I'm still trying to learn the Tayloe circuit before actually building myself a board. I I found the accompanying schematic and am currently studying the mixer/receiver end.

simonbaker

I'd be interested in anyone describing what "noise" problem is related to these switches and MD circuits and how one chip might be different from another.

Are the switches simply adding a noise signal while they are switched on, acting like a noisy wire? Or amplitude modulating the signal while they are on because the resistance doesn't stay constant? Is it just semiconductor noise?

The only reason why I can attribute that chirping noise to the integrators is because the noise was much more prominent before changing them to newer parts. The newer parts did have better "lower on-resistance, less charge-injection and better switch-to-switch matching" like Skippy mentioned. Had I not changed those old parts, I would have never attributed that noise to the integrators. Simonbaker, there are few Minelab owners who can answer your question. And unfortunately, I don't have enough experience with op amp harmonics to be one of them, sorry. However, If I had to guess it would be " amplitude modulating the signal while they are on because the resistance doesn't stay constant" or "semiconductor noise". It's definetly not noise like "noisy wire." And the noise definitely does not appear when the switches are "turned on."

Like I mention earlier, very little noise remains now and it's not persistent. It appears only after sampling, and it is some what more audible when not ground balanced properly. Having said that, It's not normal ground balancing noise because improper ground balancing noise is different -gb noises are not "chirps" like you hear in the video. It's hard to explain. You need to be familiar with the language of the detector.

Mick mentioned something about the coil picking up emi -I honestly don't know. I tried his suggestion of "disconnecting the front end receive wire and hooking it up to the coil ground." Although this method might be excellent at diagnosing internal noise, the source of this "chirping" noise is different. When I follow his method, I hear no chirping, just one consistent, long "wehhheeeeeeeeeeee." In other words, this makes it difficult to diagnose because the coil has to be connected as seen on the video. And when the coil is connected, because I'm testing at home, the coil picks up all kinds of additional emi as witnessed in the video. These Minelabs, as sensitive as they are, don't handle emi very well (Incidentally, I have eliminated all emi from planes flying overhead when using frequencies higher than 1.6Mhz ).

If you hear the audio in the 3 minute length video, you only hear that "chirp" at that a particular location, which is interesting. The remainder of the 3 minutes is free of that "chirp". However, my mind is made up and I would like to completely eliminate that noise by replacing those integrators with a better circuit.

[WM6]

and here mono coil solution:

[Mechanic]

Hi Mario,

Ok if you did the test and the result was a steady threshold at all frequencies, except when changing operating frequency, then it is either something not right in the tx cct(but this will be noticeable all of the time when away from emi) or it is just emi. If the earth field trimmers are not set properly, then this can cause issues for some emi too. It could be that there is a strong am radio station nearby.

Now for integrator noise. The 4066, just after it is turned on has a small amount of ringing, which could lead to some signal degradation and cause some noise if the ringing is not exactly the same every time, but this is only very low level noise, nothing like in the video. The other contribution they have is leakage.

On resistance, I don't think is a problem in this area. Most integrators have a resistor either before or after the switch. The sd2000 has 10k resistance between the switch and opamp which sets the maximum gain of the integrator at 10. An extra couple hundred ohms is neither here nor there, as long as it remains constant.

Better bypass caps also help reduce the ringing when an analog switch is turned on. I noticed the best improvement when a 47uf? or .47uf(can't exactly remember!) tantallum was added in parallel with the existing 100nf caps. As for an actual improvement in noise performance.... I do not know if it made the slightest difference as I was not as clued on to measuring noise as I am now. Every time I make an "improvement" I always check it by connecting the preamp to ground and measuring the noise at the end of the signal chain, before it goes to the audio both before and after.

The lowest noise is always obtained with the preamp connected to ground and no coil connected, which tells me that most noise is from high currents flowing into the coil causing voltage variations in the analog supply and ground plane.

Cheers Mick

[mario]

Hi Mick,

Good suggestion!

It never occurred to me that those trimmers should be readjusted when sampling faster (at the higher frequencies).

I had to readjust them after replacing the 4066s, then later after changing the LTEs. Those trimmers do seem to cause the effecting components to run very noisy if misadjusted.

[moodz]

and here mono coil solution:

You forgot the shielding. :-)

[mikebg]

Unfortunately, the sensitivity of a metal detector is not limited by noise because a large AIR&GND signal exists in input of RX. To make sensitivity limited by noise, the designer should suppress AIR&GND signal. A competent designed synchronous demodulator can suppress the noise in its input, but this can not solve the problem.
Note that WM6 uses the worst sensing head - MONOCOIL. It generates the most large AIR&GND signal, thats why the modulation index of TGT signal appears minimal and the gain of preamp is most limited.
Here is a visual explanation:
http://www.geotech1.com/forums/showp...10&postcount=6

[Funfinder]

Low noise is not the problem, but if the detector creates very weak EM-fields in combination with low noise stages, amps, filters etc. which are disturbed by EM-interference of all kind.

There has to be found a good compromise between consumed battery power and strong EM-field which is able to create significant changes concerning the metals-surface eddy-currents.

Low-noise amplification is nice in theory but useless if the stages receive EM-interference of all kind.

The S/N ratio hat to be good to get a high fidelity detection signal. Compared with a good mic - you can't use a very sensitive micro at a rock concert because of the feedback whistling and the same fatal combination has a metal detector with very weak detection field plus ultra-sensitive stages.

Another problem is ground balance:
If the detector is very sensitive so in theory it can work even with his own very small created EM-fields not only electro smog will heavy interfer, but especially even the slightest changes in ground mineralisation - which destroys the possibility of good depth detection.


That's why I prefer the sledgehammer method:

Use a huge amount of power and medium sensitive detection- and filterstages and you can ignore most of this weak stuff that otherwise will disturb your detection like nearby cars can destroy the reception of very weak radio stations with a very sensitive radio-receiver.

[Sean_Goddard]

Hi Skippy,

Is there an improved drop-in replacement for the CD4066? My searches have not been very successful.

After reading its datasheet, it would appear that this part is a general purpose switch that serves a dual purpose as multiplexer.

There's definitely room for improvement by using true multiplexers. I seriously doubt the newer Minelab models have retained the design since these 4066s are noisy when compared to a dedicated mixer solution.

I like your idea of using the INA132 -it does have some nice specs.

To Simonbaker,

It's my old Minelab PI, you can find some of our findings here at geotech. Rather than overhaul the front end, I would like to eliminate the remaining internal noise.

At some point in the future, I would like to digitize the signal. To get a good resolution, I need the detector to be relatively quiet. Is there a proven circuit you would recommend?

Check out the schematic I posted with the "Flindulators" in it. As to "Where does noise come from" the simple action of "mixing" produces vast amounts of noise inherently in the process.

[simonbaker]

Low noise is not the problem, but if the detector creates very weak EM-fields in combination with low noise stages, amps, filters etc. which are disturbed by EM-interference of all kind.

There has to be found a good compromise between consumed battery power and strong EM-field which is able to create significant changes concerning the metals-surface eddy-currents.

Low-noise amplification is nice in theory but useless if the stages receive EM-interference of all kind.

The S/N ratio hat to be good to get a high fidelity detection signal. Compared with a good mic - you can't use a very sensitive micro at a rock concert because of the feedback whistling and the same fatal combination has a metal detector with very weak detection field plus ultra-sensitive stages.

Another problem is ground balance:
If the detector is very sensitive so in theory it can work even with his own very small created EM-fields not only electro smog will heavy interfer, but especially even the slightest changes in ground mineralisation - which destroys the possibility of good depth detection.


That's why I prefer the sledgehammer method:

Use a huge amount of power and medium sensitive detection- and filterstages and you can ignore most of this weak stuff that otherwise will disturb your detection like nearby cars can destroy the reception of very weak radio stations with a very sensitive radio-receiver.

I agree with your main points. However, when it comes to the ground mineralization fluctuations, increased power (sledgehammer) won't help there either.

Where I live, EMI is so bad that low noise amps probably won't make a difference. But out in the wilderness, maybe could benefit.

The real final limit is the soil fluctuations -- that one is really hard to overcome. It is the old "fog in the headlights" problem. It might require some weird geometry solutions, like sweeping from different angles, or combining sweeps from search heads of different sizes, etc, to try to differentiate a deep target from soil fluctuations.

Regards,

-SB

[Funfinder]

thx for your intererst, simonbaker.


Yeah, this soil fluctuation really is terrible at high detection sensitivity / recognition / distinction levels.

The biggest problem is the large balloon-shape spreading of EM fields the usual coil needs for good depth.

What we would need is a linear detection circle of perhaps 1m in diameter that has the same size near the surface and still at 3m depth. Without any spreading.

Using two very small but extremly powerful special shaped magnetical loop antennas like ferrite-coils that create a very small detection stripe and have both a 45% angle directed to the center or both vertically. There has to be found some focusing shielding comparable with small sat-dishes.

This way MDs could find very small and still very deep stuff.

Very strong focused magnetical (or whatever...) power is the solution - but the question is:

How to use, bundle and concentrate that energy?

[simonbaker]

thx for your intererst, simonbaker.


Yeah, this soil fluctuation really is terrible at high detection sensitivity / recognition / distinction levels.

The biggest problem is the large balloon-shape spreading of EM fields the usual coil needs for good depth.

What we would need is a linear detection circle of perhaps 1m in diameter that has the same size near the surface and still at 3m depth. Without any spreading.

Using two very small but extremly powerful special shaped magnetical loop antennas like ferrite-coils that create a very small detection stripe and have both a 45% angle directed to the center or both vertically. There has to be found some focusing shielding comparable with small sat-dishes.

This way MDs could find very small and still very deep stuff.

Very strong focused magnetical (or whatever...) power is the solution - but the question is:

How to use, bundle and concentrate that energy?

Yes, that would be cool if we could focus the mag field like CT scans.

MRI equipment might do something similar -- pretty complicated. I don't know how they "focus" a mag field -- or if they really do, or just mainly focus the detectors and ignore other areas.

Or just use ground radar, which can be truly focused as it is a radiated field. Ya...

Regards,

-SB

[Davor]

@mikebg
Hi mike and all, my first post here...
Anyway, you often mentioned that Rx BW is limited to ~12Hz, and Tayloe is in essence a tracking filter as well. Each branch of an analog switch is filtered by a capacitor that acts as filter/integrator/SAH, whatever. It is easily dimensioned to whichever LP corner frequency you wish, say 12Hz, and obtain zero IF I&Q outputs that can be anti-Tayloed to a desired frequency, say 1kHz audio. Or perhaps audio with pitch related to sensing atributes.

Say, if I arrange a regen TRX as per your design, and couple it with PLL tracking oscilator, and mix RX in Tayloe with PLL and filter it at 300Hz, anti-Tayloe it at 1kHz, I can get a nice BFO-like output with straightforvard audio indication of 1kHz +and- 300Hz.

I'm just thinking out loud

[golfnut]

Noise etc

With Dan Tayloes switching integrator style mixer - the whole signal fed in is 'sampled' 4 times.

All 4 of the sampled phases contribute to the output signal. This gives more output signal - More output signal = Less Noise. (as signal to noise is better)


Conventional non linear mixer type only output on half cycles of the input signal waveform.

1/2 of the input signal power is lost (into a discarded 'Image' actually).
We throw 1/2 of the signal power away. Less signal = More NOISE (as signal to noise is worse)


S