Your thoughts on design testing?
Posted by Boss at
Hi All,
Just wonder what your thoughts and guidance are on the following:-
I have a battery powered microcontroller with analogue circuitry and an LCD in a plastic case. The product works and meets my design goals, but I decided to do some additional test and measurements with a portable digital scope for documentation purposes.
This causes problems I have not observed before (with mains powered earthed designs and test equipment).
1. The battery powered scope when connected to the circuit injects noise. This is made worse when charging/ operating via a mains powered supply with the power supply adding to the noise.
2. Adding a mains earth connection to 0V also makes the noise worse as does placing the unit above an earthed sheet.
Clearly the scope and probe are acting as an aerial injecting noise into the sensitive parts of the analogue circuit. The scope and its power source are also acting as noise sorces. Adding an earth to 0V makes matters worse, by providing a good ground.
In normal use the sensitive parts of the design are of small area on a ground plane and unaffected by the normal environment.
So what are good testing techniques for battery powered portable circuits?
Replies
-
Hi Boss!
My first feeling is that your circuit is nearly unstable. Connecting grounds or approaching ground planes who generate “noise” is revealing. As connecting “something else” to the circuit, in fact the probe.
What do you try to measure, exactly? What is the input impedance of the scope w.r.t. the impedances inside your circuit?
When everything is battery-operated, and thus floating, nothing should happen...
Kind regards,
B.B.
"Adding an earth to 0V makes matters worse, by providing a good ground." = ???
-
Use probe clip and ground clip to your test point . Power on circuit and don't hold or touch the device and oscilloscope. Your hand capacitance might injecting 60 hz noise to test probe or ground.
-
Hi Rleo6965,
Sorry, it's an unreliable way to work. A well-designed circuit MUST operate properly in ANY circumstances. We have to help Boss to find what's going wrong. I suspect phase rotations induced by the equivalent input impedance of the scope to pull the circuit in such a strange behaviour, what should never happen.
Predictable hum pick-up is not noise.
Let's wait for Boss reply.
Kind regards,
B.B.
-
Hi all, thanks for your comments so far.
Just to clarify.
I have an mbed module which is well known for noise on the DAC and ADC I/O which apparently is generated in the module, but I'm not convinced this is my actual (main) problem.
My noise is 'spikes' which I have not actually identified, they are narrow and range from a few mV to 100's mV. They seem to around everywhere, not just at my circuit, so I'm still trying to locate. Even a shorted scope probe with a piece of wire shows noise, hence I believ the scope itself is generating noise. It's a good but old one I borrowed a Fluke 100Mhz Scopemeter (un calibrated).
The mbed is programmed by the usb from the laptop (again not earthed) - Hmmm, I think I'll try to find an earthed power supply, I know Dell had problems with floating laptops and brought out an earthed version. Even here putting the scope on the mbed usb connector (should be at 0V) shows noise.
I also have an isolated 5VDC to +and -15V DC to DC converter to power the analogue circuitry which generates some noise, but I have minimised this with good decoupling and pcb layout so it only adds a miniscule amount of noise/hf hum. The spike noise is around even when the DC-DC converter is off and I'm only using the mbed to output a single DAC value and then sit in a infinite loop.
So I still think the problem is the scope and I see different results with a x1/x10 probe (worse) compared with a HF x10 (passive) probe.
The 0V supply connected to mains earth as I say added to the problem as did this connected to a sheet of pcb material as a base to work on.
I do not think it's circuit instability although I am minimising the bandwidth of each opamp so as not to agrevate the issue. It is all high impedance jfet wide band amplifiers which obviously doesn't help. Due to customer confidentiality I cannot say much more.
So back to my original question, is there a 'best' method of testing battery powered equipment as I cannot see that connecting floating instruments can do anything else than inject noise and act as an aerial?
Look forward to your comments.
P.S. I'm starting again with my measurements and methods. Only measure with scope power supply unplugged. No USB connection plugged in. Use instrument on batteries (no remote power option connected). ETC....
-
Ultimately you have to test with the system set up as it will be used in practice. This may well be no good for investigation and debugging.
I would start with an earthed metal plate with a low impedance ground connection under the DUT. It's a very good idea to insulate it ! Power the device from it's usual batteries (for noise measuring) .
If the scope earth, ground plate and circuit 0V are connected together you shouldn't have any extra noise unless the scope is rather poor. If you get noise with the scope earth and probe tip touching you have a problem which might be the scope, or much more noise from your cicruit than you are expecting. You can tell if it goes away when you disconnect the batteries.
If it's the scope then you'll need to find annother - or possibly connect it via a low noise differential amplifier.
Michael Kellett
-
Have you tried adding large ferrites to the wires connecting test equipment?Its sounds to me like you have some strong sources of radio interference (radiated and conducted) or hum.I suggest getting some large ferrite beads and running the GND and SIGNAL connectionsfrom your DUT through them to create a wall to the radiated or conducted EMI ( electro-mag interference). Try about 10 ferrite beads, and keep other cabling away from them so EMI cannot bypass them by capacitive coupling.This practical use of this method of course depends on the maximum frequency of what your trying to test, if the frequency of the signal your trying to test is not RF then the use of ferrites will be practical.
-
Hi all
Instead of “filtering” the noise, I would suggest to cure the problem at the source. Getting rid of interferences would solve everything. Shut down all your equipment - even what isn't in immediate proximity, as the levels are rather high, excepted the scope. Restart them one at a time to identify the guilty. As noise appears as spikes, anything “switching” is suspect. Including e.g. TL lightning with standard or electronic ballast, if any.
I really don't know better testing method than using battery-powered equipment... if this equipment isn't itself defective, of course. As I still don't understand why ground planes act as aerials. Never met this problem.
Kind regards,
B.B.
-
Yesterday I ordered some beads and clamp on ferrites, so once these arrive will initially apply to the scope probe to check the effect and apply beads to strategic points as suggested.
Thanks
-
B.B. thanks, yes have started doing this. I have already identified one problem.... the "mbed"! Even just setting the DAC to 600mV and the code then in a "do nothing loop for ever", I have found +/-6mV noise spikes at 1mS intervals and a burst at 10mS (very precise times) so this is being erradicated/investigates as the first step.
Think I might have to 'design out' the mbed for next versions......
Now on with the rest of the culprits......