Not so DC waveform of minus 5 volt rail at the ne5534 - could this be cause of noise problems

[6666]

That's ok for me, fits in easily and weight is not much of a concern since it's for underwater.

Have you actually tried this thing where you are going to use it ?

[Polymer]

Have you actually tried this thing where you are going to use it ?

Yes, I have. Found a few odds and ends in the short time I was testing it. Mask was leaking, water became milky - wind came up. It is waterproof to 1.5m at least .
It turned out to be buoyant, so my calculations to volume and displacement were off. Will put more weight in.
This first build is not as ergonomic as I thought it would be, but it works. Will be trying different approaches to this in the next few months.

[Polymer]

I am still chasing the noise issue and am learning a lot in the process.

I put my noisiest surf pi in a nirosta (magnetic stainless steel) box.
There is NO coil attached, only the damping resistor ist connected.
I am not measuring anything externally, just listening to the threshhold wobble.
It stays the same even with the lid on.




Question: is this a proper way to reduce emi in any way or is it just wishful thinking?
In other words, is this a bollocks way to "measure" EMI before after?

Here some random notes on my noise quest ...

I remember seeing a picture of an original Surf PI board.
It DOES have a star ground.
Many designs here have daisy chained supply rails/gnds
and/or mixed digital/analog return paths.

I've read that using low ohm resistors and C's and L's at the opamps power pins usually works,
but is more of kludged degraded circuit not adressing the real problem which supposedly is not well known in the EE world.

Having the Mosfet banging away loudly so close to everything without shielding is also, well ...

I have found that the cheap battery holders with springs at the back are sucky at best.
The wires are just sorta jammed in with the metal contacts, you can't really solder them well without destroying the plastic.
After a while those springs turn into spaghettis, I mean limp wet noodles.
All that looks like a shifty ground connection to me.

I am now getting proper spring steel contacts and making my own battery holders.

At the moment my EMI issues seem more hardware/layout related than external.

I believe I now understand why Sean_Goddard wanted to create a new Surf PI from scratch.
Maybe its time to resurrect this starting with some really basic stuff instead of
quibbling about which OpAmp to use or other fancy stuff.

[waltr]

That SS contained should be ok for electrostatic shielding and maybe magnetic shielding. Mag shielding is tough to do.

PCB lay, power/ground routing is critical for circuits with High Gain and high current switching.

Star connections, low impedance ground planes, etc are important.

Decoupling op-amp power with a small R does help as long as the cap is very close to the op-amps power pin and returned to a quite, low impedance ground. Otherwise the cap just couples the ground noise into the op-amp.

Study the PCB I did for my HH2 here.
http://www.geotech1.com/forums/showt...ake-on-the-HH2
Note the solid cooper ground on the bottom and the power traces for the circuits are a star at the battery connection.
I do need to do a screen capture of the PCB from the PAD display showing both sides of the PCB in two colors. I'll try to remember to do this (PM me if I forget).

At work I design highly sensitive analog instruments so do know about how critical the PCB layout can be. We do use lots of op-amp decoupling and careful power and ground routing.
You could try adding more caps to op-amp power pins.
Also try cutting power traces to op-amps and wire in star.

However, if the coil is not connected then there shouldn't be high current switching to create noise.
The damping R would pull only 10's of mA. You could try pulling the MOSFET off to see of there is an improvement.
It is possible it is the Digital timing circuits creating the noise. Do you have good bypass on each digital chip?

[dbanner]

I'm surprised no one has ever traced over the original surf PCB.

[kt315]

i can repeat.

i asked Joe to sell me SURF PI board. then i could make it. Joe does not show involving. no chance to make the board
if there is not full info (full info meaning this is the board in my hands).

Hi Kt, hope u can make some sense of this. If you need other pics will have to wait til after new year when I can get back to it. Thanks for looking and taking time. Joe

[Polymer]

Thank you for your help waltr!

... Decoupling op-amp power with a small R does help as long as the cap is very close to the op-amps power pin and returned to a quite, low impedance ground. Otherwise the cap just couples the ground noise into the op-amp.
I put them directly between pins on the underside of the pcb.

Study the PCB I did for my HH2 here.
http://www.geotech1.com/forums/showt...ake-on-the-HH2
Note the solid cooper ground on the bottom and the power traces for the circuits are a star at the battery connection.
Could not find this picture!?

At work I design highly sensitive analog instruments so do know about how critical the PCB layout can be. We do use lots of op-amp decoupling and careful power and ground routing.
You could try adding more caps to op-amp power pins. Also try cutting power traces to op-amps and wire in star.
Have already done this.

However, if the coil is not connected then there shouldn't be high current switching to create noise.
The damping R would pull only 10's of mA. You could try pulling the MOSFET off to see of there is an improvement.
It is possible it is the Digital timing circuits creating the noise. Do you have good bypass on each digital chip?
Yes, I have 100nf ceramics on the bottom side of the pcb of the digital chips with the shortest possible path to power rails.
The MOSFET I did pull, alas the problem is still there. Even without the coil connected, but not with such a high amplitude.

I am slowly suspecting that I am in a big area of "all kinds of emi". The 50 Hz emi is easy to mitigate by varying the frequency of the PI.
The other disturbances apparently come and go. This has been driving me nuts. Changing something in the circuit and seeing good results
and then hours or days later the problem is there again

I did a slow trace for noise on pin7 on my oscilloscope last night and was happy with the results and just let it run.
Then I saw lots of noise on the scope when I looked again later. I then started taking photos.
Here are the photos, maybe you can make sense of them. At first the noise was only around -+50mv when I wasn't taking photos.
There is a small "gap" in the first photo at div 4. That was my "happy" noise level before the noise came.




Thought the following article might be interesting for you. What is your take on this?
AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf

Wishing you a wonderful Year for 2019 ... Polymer

[waltr]

The PCB is in this file in the first post of my thread:
WR-PIdet-PCB.pdf
It shows component placement, Top side copper and bottom side copper which is where the Ground plane exists.

Putting bypass caps directly on the pins is best.

The article on amplifier oscillation is good.

Very hard to tell if it is picking up external noise or if breaking into oscillation.
The noise you show in the scope is pretty bad. Is this is on pin 7, +5V, on the pre-amp? That indicates something is wrong in the DC-DC converter that generates the +5V.
Possibly the 78L05 is oscillating. Or maybe issue is the LT1054.
Try pulling all the opamps and other chips and leave just the power supply regulators. Do you see the noise?

If all is good without the other chips them put back just the pre-amp. Does the noise return?
If so then maybe the preamp is oscillating. Two things to try:
1- Decrease the value of the feedback R. The 1M gives a gain of 1000, try a 500k (470k is fine) to drop the gain.
2- Try adding a small value Cap across the feedback R. This would be in the pF range.

[nonkapo]

These are the best photos I've got of the original Whites PCB. Might be helpful to someone.

With regards to the noise I've read topic on the Russian forums where a person put the decoupling capacitors between positive and negative rails, rather than ground as the ground rail is more noisy. Same person add a mod for deeper threshold which if i found I'll show you here so you evaluate it. Note i've not try any of these.

[Polymer]

The PCB is in this file in the first post of my thread: WR-PIdet-PCB.pdf
It shows component placement, Top side copper and bottom side copper which is where the Ground plane exists.
Thank You, got it. Noticed the 5pF cap parallel to the feedback resistor. You mention it further down.

Very hard to tell if it is picking up external noise or if breaking into oscillation.
The noise you show in the scope is pretty bad. Is this is on pin 7, +5V, on the pre-amp? That indicates something is wrong in the DC-DC converter that generates the +5V.
Possibly the 78L05 is oscillating. Or maybe issue is the LT1054.
Try pulling all the opamps and other chips and leave just the power supply regulators. Do you see the noise?
I am sorry, my "Pin 7" was a mistake and misleading. I measure the signal coming from PIN 1 of the final opamp at the diode.

If so then maybe the preamp is oscillating. Two things to try:

1- Decrease the value of the feedback R. The 1M gives a gain of 1000, try a 500k (470k is fine) to drop the gain.
Decreasing gain (475k instead of 1M) doesn't change much, even the detecting depth is still good. Noise still there.

2- Try adding a small value Cap across the feedback R. This would be in the pF range.[/QUOTE]
Tried this with an 18pF capacitor, my smallest one. Behold a miracle, the violent wah-wah noise was gone.
BUT, the detecting distance went way down. From 8 inches to 2 inches with a nickel.

Put 3x 18pF in series, not much change. It was a small wire mess, perhaps to much L?
I will be getting smaller caps soon.

I am so happy to at least not be hearing the wah-wah anymore ... Thank You! Just got to get the depth sorted.


Originally I thought my speaker was somehow magnetically feeding back through the detectors coil.
That is when I used the LED instead of a speaker. Similar result.
Checked when nothing is connected to the audio output of the detector. There is a visible difference oddly enough.

Here a composite of all four measurements with what is what in the picture:

[waltr]

18pF across 500k Ohm = 17kHz roll-off so noise/oscillation lower but so is response.
1-2pF is most to get needed BW and this may be too much C.
This was to see if the noise is coming from this amp and it seems so. But you need more BW (why response reduced).
Best to find where noise is coming from instead of reducing opamp BW.

Another thing to try.
Put a cap across the op-amp's power pins (4 to 7) directly (solder onto bottom of pins).
Try 100nF but other values may be better and may need more than one cap of different values.
If the 'ground' is noisy decoupling the opamp power pins to the 'noisy ground' doesn't help.

[Polymer]

PROBLEM SOLVED - FINALLY

Thank You Everyone for your help! Especially to waltr who I felt kept egging me on.

The violent wobbles in the signal are finally gone now ... should I face palm now ... or later?

I will write down how I got there in the end and to have it here, lest I become forgetful.

The main problem in finding the fault was that the issue was not consistently there.
So sometimes I tried something on the board and presto the problem was solved, so I thought.
That particular problem came and went no matter what I tried, again and again.
Was it EMI or the board or both? I still didn't know.

I put more capacitors on the board as waltr mentioned. Put a big one on the battery pack, WOW, the problem was gone.
So I thought ... again. Didn't last long. Hmmm, WTF. Adjusting the PPS didn't help either.

Anyway, I then had another close look at the oscilloscope pictures I posted in my last post #70.
I was pondering, why is this wobble gone when nothing was connected to the audio/led output of the detector.
The wobble was only visible through about 1/3rd of the threshold adjustment range when audio/led was connected.

"Follow the money" is well known ... here I was embarking on follow the wobble rhythm ... wah-wah-wah ...
Plus and minus 5v supplies were just fine. So I started probing the pins of the 4093 & 4066.

Did I just see a signal flank jumping left-right-left in rhyme? Was it jitter or scope not quite decisive where to put the pixels?
That was at 5us/div. I went down to 200ns/div. Oh golly gosh. The flanks were indeed dancing to the tune. 220ns worth of jumpiness.

What was I to do with that? Probed further and further upstream. Oh, the 555 output flanks had the same tune too.

Got to measuring the 555 power rails. 12v! Good!? I then measured that in AC at 50mV/div. You know what comes next.
It was dancing too. 80mV worth pp. Can't normally see that when measuring at 5v/div.

Disconnected speaker, wobblies gone ... it could only be the battery holder, batteries or gremlins.

I remember mentioning those cheap battery holders with limp springs and jammed in wire.
My feeling about those was right and I am waiting for proper Keystone contacts.

Looks like just enough power got through to power the detector plus signalling, but at other times apparently not enough
and the 12v supply voltage went funny by a bit. Strangely enough I never had any drop outs of power though.
I have heard of contact rectification and other funnies with different metals/oxidation, maybe this is related in some way.

Now I have "normal" noise and EMI which I can mitigate through rotating the coil when testing.
Changing PPS now also has some effect on residue noise too. GOOD TIMES!

[waltr]

Oh my.
Good troubleshooting a great write up.

So it seems to be the Current draw of the speaker causing a Voltage drop due to high resistance connection to the battery.
That is not an easy issue to find.