Unexpected digital noise on a linking cable

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
Right now I have an LMP7721 with the ion collector electrode connected to the (-) input through a 10M resistor and a 10M feedback resistor. The (+) input goes to earth/ground. This is followed by an inverting buffer opamp with an offset voltage (small) applied to the (+) input and unity gain (10K feedback and input resistors). The output goes to the AD through a 10K resistor and there's a 1MFD ceramic capacitor to ground to give 10mS smoothing. The setup is a little haunted by the fact that the lamp is powered by 100MHz RF and I've done a lot to try to exclude this - it isn't appearing in the noise level at present however.

I'm planning to systematically go through the analog to tie down the sources of noise - not done it yet.

Currently, of course, 1 picoamp ion current will appear as 10uV at the AD1230 input - 640 uV at the actual AD and utterly invisible at present but an eventual goal.

I am contemplating inputting the signal on the (+) input of the LMP7721 and simply connecting the (-) to the output; thereby reducing resistor noise but maybe seeing charge buildup on the input as the device has such a huge impedance.

The ultimate goal will be a sensitivity (benzene) of better that 0.1 PPB in air for an injected sample of 1 mL - I have done this routinely in the past but with older components and non digital.

The above stuff isn't relevant to the theme of this group, though - I apologize for that.
 

Ian0

Joined Aug 7, 2020
715
Sounds like a completely fascinating project.
Resistor noise currently dominates - 2.2uV for a 15Hz bandwidth, even at 10fA/rootHz amplifier current noise comes second with a 10M resistor on the input.
The non-inverting amplifier should be quieter, as the noise source of the resistor is no longer in series with the input.
I'd be very suspicious of that ceramic capacitor. That will be one of those dreadful X7R dielectric types which are horribly piezoelectric. I think a plastic film capacitor would be a lot better. Probably polypropylene rather than polyester, but that might make it huge.
 

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
Thanks. But do you agree that charge buildup would be a problem when using the non inverting input? There is simply nowhere for it to go, is there.

As for the 1uF smoothing capacitor, thanks again, I'd not factored that in at all and, as well, I am using an RF feedthrough between the analog and the digital compartments which has who knows what God-awful stuff in it. Polypropylene would be way too large - suppose I up the resistor for that RC?
 

Ian0

Joined Aug 7, 2020
715
I would assume that there would still be a resistor from non-inverting input to ground. The resistor is in parallel with the source impedance of the signal (whatever that may be) rather than in series, so the Johnson noise is much lower.
Are you familiar with the design of moving-magnet phono preamps? The inverting version is 14dB noisier than the non-inverting version for that reason.
There's a neat paper by Marcel van der Gevel* in which the input resistor is not returned to ground, but a 10x larger resistor is connected to a signal -10x the output of the preamp. The input sees the original input impedance but the noise is 1/10 the noise of a resistor 10x as big. As noise varies with the square root of resistance, noise is reduced by root 10.
A few other points:
the LMP7721 probably has the lowest bias current, but doesn't have the lowest current noise. LTC6240 can manage 0.5fA/rootHz
the noise of the AD1230's programmable gain amplifier isn't stellar - 12nV/rootHz.
One normally put the gain at the first stage, in order to avoid amplifying the noise of the first amplifier.
Designing in audio means I know about capacitors! You can get NPO ceramics up to 100nF, and they have none of the problems, in fact, they are probably better than polyester but maybe not polyproplylene. You could put a few in parallel.

*Electronics and Wireless World, October 2003, pp.38-43
 

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
Thanks Ian

I would assume that there would still be a resistor from non-inverting input to ground. The resistor is in parallel with the source impedance of the signal (whatever that may be) rather than in series, so the Johnson noise is much lower.
I'd guess the capacitance of the whole input assembly to earth is not more than a few tens of pF so I might try with a 10M or 100M and see.

Are you familiar with the design of moving-magnet phono preamps?
No, I'm not.

the LMP7721 probably has the lowest bias current, but doesn't have the lowest current noise. LTC6240 can manage 0.5fA/rootHz the noise of the AD1230's programmable gain amplifier isn't stellar - 12nV/rootHz.
I'll take what I have for now and if I make progress I'll upgrade.
One normally put the gain at the first stage, in order to avoid amplifying the noise of the first amplifier
In the old days (1980s), I got acceptable S/N using a standard 10M input resistor (through hole of course) and a 10G fancy glass feedback resistor - forget the chip but it had a JFET input - those were the only components and it was rock stable. There is a problem with offset however as there is a background ion current depending on the purity of the carrier gas.

You can get NPO ceramics up to 100nF, and they have none of the problems, in fact, they are probably better than polyester but maybe not polyproplylene. You could put a few in parallel.
Do you see any problem using 100K and 100nF on the input to the AD chip?
Richard
 

Lo_volt

Joined Apr 3, 2014
123
Was this USB3? Interesting about the drain wires soldered at either end - but I assume the longitudinal current between them would be pretty small - I have a pretty long USB cable to my printer and, of course the limit is, what, 5-8m?
No, it was USB 2.0. Ordinarily in communications the shield should be connected at only one end. This is especially important in long cable runs and in industrial environments as there could be voltage potential differences between the two devices that might cause current flow through the shield. This is a sure way to introduce noise into your signals.

But USB is a different animal. The USB spec's limit the cable length and very often the device is connected to the same ground as the host.

In our case, the ground connection from the host to the device was necessary and the device was not specifically grounded itself. We found that a third party module that we used, a spectrometer, was very susceptible to static discharge. That's why it would reset. The solid ground through the USB cable eliminated the issue by providing a safe path for static.
 

BobaMosfet

Joined Jul 1, 2009
1,222
I took a length of braided copper sheath off a piece of co-ax, cut the SCK and DO lines at one end and twisted them (just as Alexander Graham Bell did!), threaded them through the sheath and grounded this at both ends; rejoined the two lines and put a shrink over the joins. I had a thicker shrink over the braided section too. I find a big improvement; almost perfect.
But I want to leave the question on the table - why did jiggling the original cable cause "noise"? Flexible USB links operate at far greater speeds yet we seldom get data corruption - note: "jiggling", rather than RFI which would be more easily explained.
As for your original 'jiggling' issue- what did you jiggle it with... your finger? You are a big capacitor, and you may have coupled additional energy into the cable. Just a thought.
 

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
As for your original 'jiggling' issue- what did you jiggle it with... your finger? You are a big capacitor, and you may have coupled additional energy into the cable. Just a thought.
It's fixed now but the data got messed up only when I actually moved the cable; not, to my recall, when my finger came near: I think I would have picked that up. But it seemingly was a grounding issue so changing capacitance had to be in there somewhere.

Lo_volt said:
In our case, the ground connection from the host to the device was necessary and the device was not specifically grounded itself. We found that a third party module that we used, a spectrometer, was very susceptible to static discharge. That's why it would reset. The solid ground through the USB cable eliminated the issue by providing a safe path for static.
I used to have one of the old CRT projection units which had the receiver inside. It was at the end of a long coax and audio line and I got hum on my audio system. I only got rid of that by taking the mains ground connection off the projector so that its ground got defined by the long cable(s).
 

Ian0

Joined Aug 7, 2020
715
Thanks Ian - I'd value your critique of the attached.
Looks viable to me.
Having given it some thought, I suspect that the input signal is very much a pure current source, so a transimpedence amplifier might be just what you need. The circuit ends up the same as your original without the 10M input resistor, which will remove some of the noise.
I only got rid of that by taking the mains ground connection off the projector so that its ground got defined by the long cable(s).
Don't do that! I very much doubt that your coax braid can withstand the full prospective fault current of a 30A ring main, and can get the fault current safely to earth without damaging other equipment connected to it!
 

metroboyd

Joined Sep 28, 2020
4
HI Spottymaldoon, here is my two cents. the initial read says to me you have a bad cable. If you wiggle it and the signal changes or distorts, either your connectors or your cable is bad.

Back to your question of data and clock. Separate them. Use dedicated data cable rated for data. Twisted pair is fine for clock less than 1 MHZ. IF your clock is in the MHz use coax.
 

Ian0

Joined Aug 7, 2020
715
Back to your question of data and clock. Separate them. Use dedicated data cable rated for data. Twisted pair is fine for clock less than 1 MHZ. IF your clock is in the MHz use coax.
What’s the point of a twisted pair here? It’s not a differential signal so it won’t help in the slightest.
The only way it would help is if the signal were driven by a differential driver, sent down the cable, and turned back to single ended at the other end.
If the clock frequency is 615kHz and the analogue bandwidth is 15Hz, a single pole RC filter should deal with any clock breakthrough. It should be 90dB down at the Clock frequency.
And if the errors show up ONLY in the least significant bits, then it’s not data corruption, now is it?
 

andrewmm

Joined Feb 25, 2011
632
May be a little off topic,
but its been mentioned to use coax for clocks above 1 MHz,

Definitely do not do that,
use differential pair, with a differential transmitter and receiver,
 

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
Looks viable to me.
Having given it some thought, I suspect that the input signal is very much a pure current source, so a transimpedence amplifier might be just what you need. The circuit ends up the same as your original without the 10M input resistor, which will remove some of the noise.
Thanks again Ian.
I had never considered this but it's easy enough to simply remove the 10M resistor and, if that's promising, replace the other 10M with a 1G or even 10G.
I'll probably open a post on this separate issue in a more appropriate place but may I contact you directly?

Don't do that! I very much doubt that your coax braid can withstand the full prospective fault current of a 30A ring main, and can get the fault current safely to earth without damaging other equipment connected to it!
I lived a charmed life and enjoyed hum-free video for many years. The unit has long-since been replaced & I now have an immeasurably better picture costing a fraction of the Sony CRT model.
 

soyez

Joined Aug 17, 2020
51
I have a project that uses a very low offset voltage op-amp as a preamplifier and the buffered output from this is passed to an ADS1230 AD converter chip.
Currently I'm using 16 bits of the 20 bit output and passing the digital stream via an 8 bit SPI to a 16 bit controller chip. The SPI is running at 615KHz.

The digital link between the AD converter and the computer board is done with a flexible unshielded multi-stranded cable 10 cm long.

I don't come from a digital background so I was shocked to find that I get noise when I wiggle the cable - I am stuck with the arrangement but I would like advice on the best configuration for the cable - obviously I can shield it but would there be a benefit in using a USB type cable for the clock and the data line? Or having them in a shielded twisted pair? Obviously I'd like to use the full 20 bits.
Galvanic (direct electrical contact)
Electrostatic coupling
Electromagnetic enlistment
Radio recurrence impedance (RFI)
On the off chance that two sign stations inside a solitary information link share a similar sign reference conductor (basic return way), the voltage drop brought about by one direct's sign in the reference conductor can show up as a clamor in the other divert and will bring about obstruction. This is called galvanic commotion.
 

Thread Starter

Spottymaldoon

Joined Dec 4, 2015
82
Galvanic (direct electrical contact)
Electrostatic coupling
Electromagnetic enlistment
Radio recurrence impedance (RFI)
On the off chance that two sign stations inside a solitary information link share a similar sign reference conductor (basic return way), the voltage drop brought about by one direct's sign in the reference conductor can show up as a clamor in the other divert and will bring about obstruction. This is called galvanic commotion.
An excellent and descriptive term - and a phenomenon which has pursued me, in countless guises, ever since I could tell the difference between amps, volts and ohms - and will doubtless continue its pursuit, laying traps whenever it can, for the remainder of my time here.
 
Top