I believe you used to call this muMeet - apologies for not using the original greek. I know you have just had one, so I was wondering, how did it go, what was achieved and what are people working on?
Come to that, what are the rest of you working on? Does anyone have any pictures of their projects, electronic or otherwise? I’m currently stuck in my “day job” on an offshore construction ship in the gulf of guinea - so can’t access my own workshop and would be interested to see what you’ve all been doing.
It sounds like it will be. I have been doing some more work on taking temperature measurements with an LM35 via an Arduino and Labview/LINX. I’m in an electrically very noisy environment, and I’m getting a lot of jitter on the readings.
I’ve put in an RC low pass filter across the LM35 as per the datasheet. I have also put in a mode averaging routine in Labview that has reduced the noise greatly. I want to use the Arduino’s internal 1.1 v reference to get more accuracy so I tried to set this up using the LINX analogue set ref. v.i., but it didn’t work…need to have a re-think, methinks. I’m also going to try to get a Butterworth filter done in software, which with Labview is fairly straightforward (well, in theory…)
It was great to hear what you have been doing. When I am back in the UK I’ll be meeting in with the Moray Firth makerspace people. My friend just got a new 3D printer so looking forward to seeing it. He has an idea to make a 1/4 scale Merlin engine kit using it.
Anybody made anything out of wood? I’m terrible with it so I like to see nice woodwork!
The ultimate use of the LM35 is to measure various fluid temperatures for my rocket engine project. At the moment I am just concentrating on getting it to read as accurately and noise free as possible.
In my day job I do subsea robotic intervention, which is where I am now, on a ship called the Deep Pioneer off the coast of Ghana. That is the electrically noisy environment. In a way it is a good place to develop something like this as it is probably the worse case scenario in noise terms. The LM35 is on a breadboard at the moment so no twisted pairs or ferrite beads as yet. Interestingly when I switched off the fluorescent lighting the noise goes down - not that unexpected.
Looking on the net, noise on the output from these sensors seems to be a common problem. The datasheet shows a power supply decoupling capacitor and a low pass RC filter, both of which I have on the breadboard. That has helped a fair bit. I have also put in an averaging routine in the software that has stabilised the readings somewhat.
What I want to do next to improve accuracy is to use the internal 1.1 volt reference on the Arduino. I had a go at that today but it didn’t work, I now know why though, so that is something to do tomorrow. Just reading around it seems that the on board reference is not that accurate either, often the 1.1 volts is more like 1.084 or something around there. With the 5 volt reference, the ± 2 LSB accuracy of the ADC gives an error of about 0.25%, whereas with the 1.1v it is more like 5% - see here:-
So I reckon the way to go is to build a precise external reference of about 1 volt which will scale the low voltage output from the LM35 better, and do all the noise filtering I can. As I said I’m going to try the butterworth filter, when I get my head around it.
Any input you can give would be hugely appreciated, as is your offer of help and interest in my projects.
If the light makes a difference you have probably got a ground loop problem and twisting wires together and star earth’s always the best way to deal with that, and carefully thinking about where you ground ref is . I would put a ceramic .1 decupaling cap tight to the package andtwist my ground and signal together and run the power in very close proximity . I would also try to terminate ground ref as close to the adc.'s analog ground as possible all to try to minimise the area of the wiring .
I find it best not to let noise in as aposed to trying to find a way of getting it out .
But you might be better off with a simple ntc sensor as the lm35 is micropower and may not be able to overcome the noise , but saying that i have never had a problem with them
Thanks for your advice and help. I will connect the LM35 with twisted cables and re assess my grounding. I have a 0.1 micro farad decoupling cap on across the LM35 power and ground, right next to it at present, as well as the low pass filter.
The environment I eventually use this sensor in should not be anything like as noisy. As I work on the ship for 6 weeks at a time, I take little projects like this with me to kill some of the time when I am not asleep after my shift. There is all kinds of electrical noise here 380 volt 50Hz, 440 volt 60Hz, good old old 220 volt 50Hz, 600 VDC, 3000 volt 60Hz…you get the idea.
As you say, best not to let it in in the first instance.
If noise is an issue, why not steer away from analog and go digital?
The DS18B20 is more accurate than the lm35 when used on Arduino and if run in 3 wire mode is almost immune to noise. The one drawback could be the slower sample time but even in a rocket fluid temps don’t change That fast…
Those ds ics are good , but you still need to look after your ground loops , in the digital world it’s possibley more important all those fast edges require keeping in check .
Thank you for the interest in my mini project and for taking the time to post. Courty, you are right about going digital, and I will most probably do so eventually. You are also right about the fluid temps not changing that quickly. Several of my measurements on the rocket will be made with K type thermocouples and I have Maxim modules got from Adafruit to digitise the output from these (can’t remember the IC part number just now…). I like the idea of the one wire bus on the DS18B20. Parasitic power though will be a no no as I will have temps +100 celsius.
That said digital signals are not always immune to noise…And the eventual home for these sensors shouldn’t be all that electrically noisy. It will, however, be very acoustically noisy…
The reason I’m using the LM35 at the moment is because it is all I’ve got and I’m just trying to get a handle on using the Arduino with Labview and LINX. I am more likely to go with the TCN75A as it is easy to implement with LINX.
Here is a screen shot of the Labview/LINX software that my Arduino and laptop are executing:-
Just remembered what the Maxim IC is for the thermocouple, it is a MAX6675. It takes care of cold junction compensation and uses the SPI protocol.
Here are a couple of photographs of my test set up for this. The microcontroller is an mbed, this was before I started using the Arduino. The dark times.
