Is there a way to test the PWM dew controller (using a multimeter) to see if this is a problem? Would you see a voltage between the earth on the input to the PWM and the outer shell on the outlet when the load is turned down to minimum?
Just measure the RCA outer shell to earth with the PWM on or off. If you see a pulsing 12V or 12V then just take care with your RCA connectors heat shrink cover them.
Sounds good - can you give us some more info on the actual specs for your controller?
Cheers,
Peter
Hi peter...a very simple one at that.
I designed it for 5 Amps max. The P channel FETs can take much more but I limited it to this with the design of the PCB tracks in consideration & my needs in consideration This was originally just done for myself, cheap & simple & safe but it has taken on a life of itself.
PCB tracks can be changed but atm around 48 watts max is a fair power draw.
Just uses LEDs & pots for varying PWM output. Output is very linear from in I draw aqcross it's range. I mucked around with lots of PWM frequencies but settled on a low freq, below 10Hz as it provides the best visual feeback on pulsing LEDs. Low freq all so helps reduce noise onto the power lines.
No current foldback only fuse protection. I did consider it but decided not. There are current limit protected FTEs available but the I am really after a cheap simple solution.
No undervolts as well....although this is something I am going look into.
All is up in the air until I receive my cct boards.
I am looking at making this a 6 channel PWM, easy enough & allows more complexity with a bigger ccct board.
High-side switching - meaning you interrupt the 12V rather than the ground return - is the way to go and can avoid some EMC issues as well.
I have made a fair range of dew heaters now, both analog and digital control, with and without thermostatic control ( incl humidity / dew point control )
In nearly all cases, if you use PWM you will get RFI.
Even new commercial boxes marketed as having low RFI are affected
The main problem is not the box, its the wire going to the strap
as that radiates away like blazes, and PNP vs NPN wont fix it.
Secondly, with PNP switching, you need more circuitry if you want to use a microprocessor to drive it. Again a tradeoff
For anyone wanting a very simple 0-100% manual controller
an LM324, a pot, a few resistors and a mosfet is all thats required.
I have attached a piccy of the simplest one i make
Once you have the LM324 configured as a sawtooth generator
its a simple matter to add multiple outputs by just adding more
LM324s set up as comparators to drive the Mosfets.
Very cheap.
Just tested my controller and sure enough as you decrease the output of the PWM unit you get increasing voltage (up to 11V) on the outer shell of the RCA outlet. Looks like I'll be changing out plugs this week! Thanks for the heads up.
Just tested my controller and sure enough as you decrease the output of the PWM unit you get increasing voltage (up to 11V) on the outer shell of the RCA outlet. Looks like I'll be changing out plugs this week! Thanks for the heads up.
It's a simple mistake by some heater designs that has propagated itself somewhat.
Whats a cheap electric blanket worth? On ebay they go for $30. There is about 6 metres of insulated coiled nichrome in them.
That is very true. Although I'm more inclined to work with the thin film stuff for ease of solder & they offer a bigger surface area of heat dissipation.
Peter, I'm doing this in part for the interest in doing so. I have the hope that I can get something out of it which is more effective than I can buy but mostly I'm enjoying the process of learning.
My preferences
- my own involvement for the fun of learning
- minimise power usage and don't do unnecessary heating (or miss heating when I should)
- extendable - I'd prefer not to be limited to a set number of ports although I can't see why I'd need many
- upgradable, if I find a better way of doing things I'd like to be able to go with it.
- I'd like to be able to just add in heater modules and have the controller manage them.
I've not yet decided if I'll add in digital readouts on the controller, I could add in an LCD screen (or 7-segment led panels) and show a few temperatures (ambient, dew point and individual element temps) but I've not put much thought into that yet. I have pondered small PCB's for the heater electronics but I'm still looking at options for that. If I do go PWM then I will probably need to do something for \\
that.
In nearly all cases, if you use PWM you will get RFI.
Even new commercial boxes marketed as having low RFI are affected
The main problem is not the box, its the wire going to the strap
as that radiates away like blazes, and PNP vs NPN wont fix it.
Hi Andrew,
You are right, switching power causes interference. You are also right when you blame the cables for radiating most of it. Another possible path is throught the changing load on the battery depending where you connect your supply.
As far as reducing the radiation from the cable you could use shielded cable and connect the shield to ground - hence my preference for switching the high side and leave the gound connected through.
You could also add some softer switching at a lower frequency and reduce the spectrum you radiate - I guess that alone will take away a good deal of your problems.
If all that doesn't work and RFI is still critical you may have to suffer the loss of a linear regulator design.
I will be aiming to preserve battery power AND keep the design quiet. I am sure it can be done.
Most antennas use a standing wave design of 1/4wavelength. Thus making the power cable longer will reduce any standing wave forming even cancelling any waves forming.
Using 10Hz PWM and a 1/4 wave antenna (to optimise RFI) you would have a wave length of: C = fλ/4, if f = 10hz thus λ= 4C/10 = 1.2e^8m.. ( Edit: to form a 1/4 standing wave).
Obviously harmonics of 10Hz would form their own standing waves but I suspect it's more a EMI than RFI problem (edit: meaning noise feed back to the power supply as spikes rather than radiated interference).
Maybe using twisted pair power cable?
Any thoughts?
Last edited by wasyoungonce; 09-06-2009 at 08:57 PM.
Yep. Higher frequency.
High frequency driver, say 10 or 20 kHz with decent low-pass L-C filter network on the output, all contained in a shielded case with shielded power cable = efficient, noiseless, variable DC power supply.
And why control a dew heater at higher frequency??. The thermal inertia would render that uneccesary I would have thought. More RFI, less efficient, no gain.
Less RFI actually, almost zero.
Waaaay more efficient than a linear regulator.
Thermal inertia? where did that come from... the heater would see close-to-pure DC.
You are right Simon - higher frequency and filters and shielding would work to give you a near DC output - one small step to using a buck converter - meaning a type of switch mode power supply. Not the first thing I would consider for a heater given the more complex design, but actually worth looking at since you can design them to be reasonably quiet. (not on Veroboard though!)
Cheers,
Peter
P.S. twisted pair actually works quite effectively in reducing radiated emissions. Not much good for conducted emissions though, that needs filter components tailored to requirements.
Yes, exactly that, a buck converter.
Shouldn't be any more complex than what is already being discussed here.
The uC is already being discussed to generate the PWM.
The switching MOSFET is already there too, all that is needed is a freewheel diode, a cap and inductor(or more) for the filter network and it's done.
The purely resistive load simplifies things.
P.S. twisted pair actually works quite effectively in reducing radiated emissions. Not much good for conducted emissions though, that needs filter components tailored to requirements.
Yep...I was meaning "if the problem is unwanted RFI try twisted pair". All good.
higher frequency and filters and shielding would work to give you a near DC output - one small step to using a buck converter - meaning a type of switch mode power supply.
Just for info, i have also played with these as well
I made a unit to drive my ETX to give a constant output irrespective of input ( piccy attached for info )
The problem with these for dewheaters is designing them to be "efficient" over a wide voltage range. ( And you still get RFI if you aren't careful )
I have some std purchased plugpacks that are noisy as hell.
I still reckon a high freq PWM with correct output filtering is a better bet for a dewcontroller. ( And its how nearly all the commercial ones work )
And as to RFI, i don't have any "problems" with it so far.
I know its there, but it hasn't affected my scope ( yet )
Finally got around to comparing the Simple Approximation and the Better Approximation for dew point calculation. Below 50% RH, as stated for the Simple Approximation, the numbers get significantly out of wack. Example (for our somewhat extreme observing conditions on the praries of Canada - during the winter):
Ambient Temperature: -20 C
Relative Humidity: 10 % (not uncommon in the winter)
Simple Approximation: -38 C (frost point - not dew point)
Better Approxiation: -44 C
And of course the number diverge even more as the temperature drops further and the humidity gets lower.
I like observing during the winter months better than the summer as I can have the scope setup and be viewing by 6:00 PM. During our summer months, like now, it does not get dark until 10:00 PM.
I will be continuing on my project to build a dew / frost point dew heater and find a way to do the math with a uC. I have ordered an Arduino. It should be here next week. As I already have the sensors, I may have something breadboarded in a couple of weeks. Then I will make the decisions on the final temperature sensors and board modifications. I will produce PCBs (one off) once the design is settled. I expect I will have to make a further revision farther down the line with features I have overlooked.
I am using CAT-5 cable with RJ-45s to connect the heaters / temperature sensors. Using 3 conductors each for the power and ground to the heater and using 1 conductor for the sensor power and 1 conductor for the sensor data. The sensor ground is connected to the heater ground. Each RJ-45 contact is rated at 1A and the CAT-5 is already twisted pairs. The heaters are not drawing anywhere near what three conductors within the cable can handle.