----UPDATE-----
While the original question verted on technique to analyze ripple induced by 12VDC to AC 230V inverter on the input 12VDC, THE TOPIC moved on solve the problem.
I've solved it.
Brief recap
- From AmyAlice post I got suggestion to try to eliminate negative ripple with diode SB140 (from aliexpress, maybe fake?). FAIL.
The negative ripple is still present. The problem is the ripple pass over the regulator of Arduino.
- Together with SB140 Ive tried a NANO V3 BOARD modified with one of each .1, 1, 10, 100uF capacitor all together, in parallel, packed in one big cubic cap soldered between pin Vin and GND of the PCB (shortest path possible to reduce inductive downside). FAIL.
The ripple is still the problem and the caps seems not sufficient to shunt HF to GND. The diode doesn't make any difference.
- The ISOLATED DCDC, in this case this general AC/DC input to 5VDC USB-A output (optocoupled) worked. SUCCESS.
- Ive prototyped the ultimate solution BUT probably I will never implement. It was the last thing to do if nothing would worked. Here the layout. It is composed by 2 relays and an 230VAC to 12VDC transformer. When INVERTER turns on, AC trigger the first relay to disconnect NANO from battery source and then connect to the AC/DC 12V source. So when Inverter is off, the ripple is absent and the lifepo4 system power it up. When ripple and so inverter starts, it switches to AC rippleless source. With general ON/OFF switch.
------------------------------------------
Hallo,
I've bought this oscilloscope (ZEEWEII DSO2512G) and set to AC coupling, 10x probe.
But I dont know how proceed in a reliable way.
The problem is not strictly inherent to audio world but other comunities can't help.
I've got a ARDUINO NANO board connected to 8 DHT22 humidity temperature sensors.
Changing PSU influences the output of sensors. My choice PSU (13V>5V dcdc) fails many times each read cycle. The best is achieved with a 5V powerbank.
The situation is:
I would like to investigate PSUs and THAT'S THE TOPIC
So... can you help me to put down a method to investigate in that?
Since supplied current is not the problem, what remains is voltage ripple.
Im not sure if my DSO (120MHz 500msas) can do the job....
I think that FREQUENCY of ripple is the problem that interferes with proper sensor reading.
But Im really newbie in Oscilloscope world and PSU output analysis.
Thanks is advance
Gabriel
While the original question verted on technique to analyze ripple induced by 12VDC to AC 230V inverter on the input 12VDC, THE TOPIC moved on solve the problem.
I've solved it.
Brief recap
- From AmyAlice post I got suggestion to try to eliminate negative ripple with diode SB140 (from aliexpress, maybe fake?). FAIL.
The negative ripple is still present. The problem is the ripple pass over the regulator of Arduino.
- Together with SB140 Ive tried a NANO V3 BOARD modified with one of each .1, 1, 10, 100uF capacitor all together, in parallel, packed in one big cubic cap soldered between pin Vin and GND of the PCB (shortest path possible to reduce inductive downside). FAIL.
The ripple is still the problem and the caps seems not sufficient to shunt HF to GND. The diode doesn't make any difference.
- The ISOLATED DCDC, in this case this general AC/DC input to 5VDC USB-A output (optocoupled) worked. SUCCESS.
- Ive prototyped the ultimate solution BUT probably I will never implement. It was the last thing to do if nothing would worked. Here the layout. It is composed by 2 relays and an 230VAC to 12VDC transformer. When INVERTER turns on, AC trigger the first relay to disconnect NANO from battery source and then connect to the AC/DC 12V source. So when Inverter is off, the ripple is absent and the lifepo4 system power it up. When ripple and so inverter starts, it switches to AC rippleless source. With general ON/OFF switch.
------------------------------------------
Hallo,
I've bought this oscilloscope (ZEEWEII DSO2512G) and set to AC coupling, 10x probe.
But I dont know how proceed in a reliable way.
The problem is not strictly inherent to audio world but other comunities can't help.
I've got a ARDUINO NANO board connected to 8 DHT22 humidity temperature sensors.
Changing PSU influences the output of sensors. My choice PSU (13V>5V dcdc) fails many times each read cycle. The best is achieved with a 5V powerbank.
The situation is:
- powerbank, no reading fail at all
- 230v to 5v 1A apple transformer, fail just one sensor, like 1 time in 30 cycles or more
- dc dc fails a lot, like multiple sensor each cycle every cycle
- DIY L7805 regulator board fails like the dc dc
I would like to investigate PSUs and THAT'S THE TOPIC
So... can you help me to put down a method to investigate in that?
Since supplied current is not the problem, what remains is voltage ripple.
Im not sure if my DSO (120MHz 500msas) can do the job....
I think that FREQUENCY of ripple is the problem that interferes with proper sensor reading.
But Im really newbie in Oscilloscope world and PSU output analysis.
Thanks is advance
Gabriel
Last edited:
What do you mean?
The ground is the same for every configuration, except for battery and 230v>5v phone charger.
The circuit is:
- lifepo4 12.8V (+) > switches and fuses > dc dc > 5v > arduino > sensors
- negatives are in common, continuity is checked
In my opinion, frequency above 10Mhz dont affect it so much like Apple 5V1A PSU.
Frequency below interferes with sensor read timings... but again, Im a newbie and that is just a supposition.
if you want a low noise power supply search LM317 Denoiser. It works very well. Any Dc to DC switch mode power supply will be noisey to an extent. The fast switching of the chips on the ardunio dont help. Try isolating the sensor with a Pie filter ie inductor capacitor inductor or CLC or giving the sensor its own low noise power supply
Have a look at Mark Johnson's threads on the PO89ZB or AmyAlice filters - both are designed to take the noise out of switching power supplies and are very effective. PO89ZB is available in the DIY Audio store as a completed module; AmyAlice gives even better filtering but you'd need to order some PCBs plus parts and solder them yourself.
Are you referring to what? Arduino NANO... no. And neither shielded cable if we talk strictly to real shielded cable.
The situation is:
From Electric Box of my van, behind a 1cm poplar wood runs these 10 1mmq wires (I get 1mmq to ensure that they would be resistant to tear and wear).
10 = 8 signal and + and - shared.
But all together are sandwitched (like a strip) in aluminum tape, taped to aluminum face polyurethane solid insularor.
Since with battery they works fine I would exclude ext. interference troubles.
Grounds are all connected properly. Continuity test always bips showing 0.
https://www.diyaudio.com/community/...o-retrofit-upgrade-any-317-based-vreg.331491/ is it this?
PO89ZB
AmyAlice
I will try LM317 first then these fikters. Thank you all
yes you found it. To power your sensors you may wish to try galvanic isolated dc to dc power supplies. I used one to power a bluetooth module for an old Hi fi receiver. Here is one. The prices must have gone up a lot. I paid about $1 for mine.
https://www.ebay.ca/itm/15423893982...vcfrjRNICq6cH0qh0LMshdYQ==|tkp:Bk9SR4KhjvHAZQ
https://www.ebay.ca/itm/15423893982...vcfrjRNICq6cH0qh0LMshdYQ==|tkp:Bk9SR4KhjvHAZQ
I just made the De-Noiser (with 2N3904/2N3906 instead of BC337/BC327). I will update when I will try it.
(R2 680R / R3 1K2)
Meanwhile here the newbie made analysis (correct my method please!).
(The LM317 is soldered on back of PCB to use the pcb as heatsink. As always seen in SMD stuff regulator and similar. Obv, I will keep current abs low.
(R2 680R / R3 1K2)
Meanwhile here the newbie made analysis (correct my method please!).
(The LM317 is soldered on back of PCB to use the pcb as heatsink. As always seen in SMD stuff regulator and similar. Obv, I will keep current abs low.
Yes I will. In fact Apple and other phone charger should be galvanic isolated (flyback?) circuits. Correct me if I'm wrong.
The DieNoiser (2 BJTs) works but inconsistent.
It loses reads randomly. Like each cycle 2 random sensor readout is invalid. But I can work around implementing the software to exclude these reads (now they print "Nan", but if simply don't print when they fail I just lose speed of refresh. A parameter idc).
A common dc dc 2x USB-A 1.5A charger module DOESNT TOTALLY WORK.
The DcDc instead works but bad.
The DieNoiser works passable.
The APPLE charger is still the best.
Two more I wanna try.
- galvanic isolated dc dc (as suggested: may I use bigger generic module or this one is State Of Art one for my power needs?)
-AmyAlice that is what I thought when thinking how to dampen ripple, in a physic way. But Im happy to have all ready 🙏
While Im glad to find a solution to the problem, I would also know how to analyze - if possible with my DSO - the situation and a methodology to.
It loses reads randomly. Like each cycle 2 random sensor readout is invalid. But I can work around implementing the software to exclude these reads (now they print "Nan", but if simply don't print when they fail I just lose speed of refresh. A parameter idc).
A common dc dc 2x USB-A 1.5A charger module DOESNT TOTALLY WORK.
The DcDc instead works but bad.
The DieNoiser works passable.
The APPLE charger is still the best.
Two more I wanna try.
- galvanic isolated dc dc (as suggested: may I use bigger generic module or this one is State Of Art one for my power needs?)
-AmyAlice that is what I thought when thinking how to dampen ripple, in a physic way. But Im happy to have all ready 🙏
While Im glad to find a solution to the problem, I would also know how to analyze - if possible with my DSO - the situation and a methodology to.
Hi. I have some questions. 1. How do you power Arduino board? Because it has internal +5V regulator, voltage 8...9 V is required if you supply external power to Raw input, dc jack. If you supplying external +5V to raw inpit pin , arduino will no longer get +5v, intetnal regulator has about 1,5v loss and also diode in series ,for reverse polarity protection. So if you use +5V , connect it to pin marked +5V and GND pins of Arduino. So how you have connected power?
2. What you mean with sensor fails to read? Is it reading analog value , very much out of nominal voltage sometimes? In some cases, with analog sensor, simple rc filter helps, like 10k in series with signal and 100nf to gnd. Values depend on sensor speed required.
If sensor is i2c or another digital, and you get connection errors, then it's different case. Check supply voltage on arduino pin marked +5v first. I've replaced internal poor regulator AMS1117 to LDL1117, same pinout, low drop, all specs better.
2. What you mean with sensor fails to read? Is it reading analog value , very much out of nominal voltage sometimes? In some cases, with analog sensor, simple rc filter helps, like 10k in series with signal and 100nf to gnd. Values depend on sensor speed required.
If sensor is i2c or another digital, and you get connection errors, then it's different case. Check supply voltage on arduino pin marked +5v first. I've replaced internal poor regulator AMS1117 to LDL1117, same pinout, low drop, all specs better.
Couple things to verify. Did you connect a 10K resistor between dataout of the sensor and VCC? Adafruit has a pretty good writeup, https://learn.adafruit.com/dht/connecting-to-a-dhtxx-sensor Another thing they said is sampling is limited to once per 2 seconds. They also suggest powering the sensor from 5V not 3.3
You need to think about grounding problem in your household mains wiring, not just your devices. It is common for the building safety ground to carry noise currents from motors etc. So, the safety ground voltage is not exactly the same from one outlet to another, causing noise currents through the ground connections between your devices, especially your scope.
Another issue is switch-mode power supplies that only use two prong power plugs. In that case, the power input filter y-capacitor is grounded at the output, so noise current travels through your device ground. Left open, the Y-cap is a voltage divider that puts half the line voltage on the floating ground. Generally speaking, 2-prong switch mode wall-warts are not useful for any device that connects to another device.
These problems are not related to how well you filter the DC. The noise voltage is on both (all) DC power connections, positive and negative. You need to block/avoid any ground noise currents. The standard "ground lift" (diodes and small resistor) circuit may be useful.
Another issue is switch-mode power supplies that only use two prong power plugs. In that case, the power input filter y-capacitor is grounded at the output, so noise current travels through your device ground. Left open, the Y-cap is a voltage divider that puts half the line voltage on the floating ground. Generally speaking, 2-prong switch mode wall-warts are not useful for any device that connects to another device.
These problems are not related to how well you filter the DC. The noise voltage is on both (all) DC power connections, positive and negative. You need to block/avoid any ground noise currents. The standard "ground lift" (diodes and small resistor) circuit may be useful.
Yeah, agree.
But even with a good household ground, using the SMPS will present problems due to their design: there is no good low-impedance ground plane on the DC side.
Due to the ways the SMPS operate, you'd want to isolate the DC side ground plane from the 240V (110V) mains ground. However, to couple the DC side noise back to that (good quality household ground), you need a low impedance path.... hence the issue.
But even with a good household ground, using the SMPS will present problems due to their design: there is no good low-impedance ground plane on the DC side.
Due to the ways the SMPS operate, you'd want to isolate the DC side ground plane from the 240V (110V) mains ground. However, to couple the DC side noise back to that (good quality household ground), you need a low impedance path.... hence the issue.
DHT = digital hum temp
5v pin and GND pin with ext 5v source.
Arduino clone NANO V3.0
Yes, on each sensor. In fact with a powerbank it works flawlessy (and its not a 5v battery but a 3.7v battery dc dc to 5v).
I sample every 5 sec. Its a cycle, sensor per sensor.
That is what I'm doing.
Im in a camper van. Lifepo4 12.8V directly connected to 5v regulator or to inverter then apple charger.
My scope is battery powered. Dso2512g google it please 🙂
I got it. But a galvanic isolated dc dc would be the best, right?
Circuit is that. Power in (usbC or 5v pin) > arduino > display and sensors.
The 5v sources are described in previous posts.
Thank you all guys 😀
Ok... I've checked for some datasheet of this sensor dht22, it has single wire communication to arduino, and may need shielded wire for long distances. So next question would be, how distant are sensors from board ( picture says less than 10 meters)? Also , if using long wire, it may have capacity to gnd wire, also catch variuos noises, so you may need to try stronger pullup, lets say 4,7k to +5V , or just try another 10k in parallel to existing ones. Shielded cable will have it's own capacity too, so that may not improve result from first try. Check with oscilloscope data pin from one of sensor, how good waveform you will have? Try to reduce pullup and see how it changes. With digital signals, power supply ripple of few hundreds millivolts is not a problem, problem is noise getting to data lines. Datasheet don't specify how sharp data pulses must be, so we can assume it should look nice. As try you may connect one sensor close to arduino and check with oscilloscope what rise/ fall times pulse has. Digital scopes can record pulses and that makes analysing them later easy. So i think loss of sensor reading means communication is disturbed at that moment, maybe electric pulse or noise by turning on off some equipment makes small voltage jump. If this would be caused by dc dc converter or mains, you would have errors constantly.
OK, another thought. Are the resistors pulled to 3.3 or 5? The DHT should be powered by 5 whereas the resistors pulled to 3.3 according to ada. I use smps for all kinds of i2c sensors and have never had a problem. I have had weird issues with things like pi wifi triggering motion detectors too close to the pi as an example. I did some googling and some have suggested for long distances to drop the pullup down to as low as 2.2K. Some also suggested using drivers on longer wires with better slew rate control on the outputs. A long unterminated wire is going to be pretty wonky. I know for Carrier's infinity hardware, Carrier uses RS485 hardware.
Weird stuff can happen, it may take time to get to the reason. My latest was the new pico's have a reset line. Nominally not used. I used it. Big mistake. The thing is ultrasensitive responding to pulses of like 40ns. Noise was triggering reset from the wire that connected it to a pi pin to reset it. My final resolution was a 400 ohm pullup on the reset line with a 20uF cap. Both soldered on the pico board. I also noted a 3/4" wire dangling from the pico board pin that was unconnected was enough to get spurious triggers. Good luck.
Weird stuff can happen, it may take time to get to the reason. My latest was the new pico's have a reset line. Nominally not used. I used it. Big mistake. The thing is ultrasensitive responding to pulses of like 40ns. Noise was triggering reset from the wire that connected it to a pi pin to reset it. My final resolution was a 400 ohm pullup on the reset line with a 20uF cap. Both soldered on the pico board. I also noted a 3/4" wire dangling from the pico board pin that was unconnected was enough to get spurious triggers. Good luck.