Go Down

Topic: PWM with low pass filter do I need a pull down resistor? (Read 1 time) previous topic - next topic

KOR

Hi I want to use one of my PWM pins with a low pass filter as an dac to drive a mosfet. my current design is for a fast pwm with a 0,82uF cap and a 4.7kOhm resistor. My question is do i need a pull down resistor ? and what value should it have? I would put it in front of the low pass filter.
thanks for the Help

Wawa

I don't think a pull down resistor is needed here.
They are normally used to prevent problems during bootup, when the pin is still an input.
The capacitor will keep the gate at ground potential during bootup.

But what is the purpose of this circuit.
A mosfet has a threshold. That could be ~2volt for a logic level mosfet.
The mosfet might do nothing with a PWM value of 0 to ~100, be in a linear region (possibly getting hot) from ~100 to ~200, and saturate above ~200.
Is this for your Peltier cooling? Please explain.
Leo..

KOR

Yes it is for a peltier element. I will use this one: IRFZ44N and you are right it has a 2V threshhold(datasheet: Gate Threshold Voltage Min 2.0 Max 4.0V). what can I do to fix this. in my previous board version I had no filter but I was told peltiers don't like pulsed voltages

KOR

Hi,
I wanted to make a dc with a low pass filter to drive an IRFZ44N which controls the temperature of a peltier cooler. someone way smarter than me pointed out that this n-channel mosfet has a threshold of about 2V.
So my PWM would do nothing up to 2 V than some linear thing.
I have a version of the board which doesn't use the filter but I was told peltiers really hate beeing turned on and off all the time.
Since they will run long times I wanted to improve my previous design by a filter.

(my calculator says with 7812Hz and a 4.7kOhm resistor and a 0.82uf cap I will get a ripple voltage of ~0.04V and a settling time less than 0.009s which would be perfect)
Thanks for your help

Grumpy_Mike

#4
Nov 14, 2016, 11:47 am Last Edit: Nov 14, 2016, 11:48 am by Grumpy_Mike
Quote
Since they will run long times I wanted to improve my previous design by a filter.
Not sure what you mean by improve? What do you want to improve?

Quote
a settling time less than 0.009s
Why do you think that is good? It is way faster than you need, thermal time constants mean that anything less that about a second are going to be no improvement.

Quote
with 7812Hz
Where is this figure coming from? It is not the standard PWM frequency.

How are you going to wire up your FET? What is the maximum voltage you want to drive your peltier device at? What current does it take?

We need to know all this information to give sensible advice.

KOR

the frequency comes from timer 0 divisor 8 http://playground.arduino.cc/Main/TimerPWMCheatsheet. OK I forgot that it is 7812.5.
Quote
Why do you think that is good? It is way faster than you need, thermal time constants mean that anything less that about a second are going to be no improvement.
I know that the settling time is not very critical.
previously I had no filter.  feeding the raw pwm to the mosfet (IRFZ44N) so I got 12V 6A switching on and of hitting the peltier. I used a high devisor and that was it. But I want to improve it by using an analog smooth regulating signal.
What I want to improve is that the peltier won t be hit by 12V 6A 60 times per second. I want an analog smooth suply voltage 50 % pwm equals to 12V 3A...

JohnLincoln

If you just use the standard PWM output to turn your MOSFET, then you are using it in a digital mode, which dissipates little power, as it is either fully conducting  or fully off.

If you were to feed a variable DC voltage to the gate, then you are going to be operating the MOSFET in it's linear region of operation, and it is likely to dissipate a lot of power, which will heat up the MOSFET.  You are likely to need a heatsink to cool the MOSFET. 

You might be better off not going down this route.


Grumpy_Mike

If you look through the history of this sort of problem here, it is by no means that there is universal agreement on if PWM actually harms peltier devices.

Is it important to control the peltier device so finely? A simple on / off control where the on and off times are in the order of several seconds are normally enough for fine temperature control.

KeithRB

Not to mention that you will need at least one OP-Amp to convert the non-linear gate transfer characteristic to a linear control voltage.

DVDdoug

Quote
I have a version of the board which doesn't use the filter but I was told peltiers really hate beeing turned on and off all the time.
I'm not a Peltier expert, but from what I've read they are less efficient with PWM.    However as john said, the MOSFET (or anything similar) in linear ("analog") mode will be super inefficient.    ...Unless you are heating, and in that case any inefficiency is converted to heat anyway, and some of the heat would be coming from the MOSFET instead of from the Peltier.  

But as Mike said, temperature doesn't change instantly, so you don't need PWM anyway.   Virtually every heating & cooling system switches on & off when you are above or below the target temperature.   For example, if you are cooling to 10 degrees you turn the system on 'till you hit 10 degrees, then you turn it off until the temperature rises above 10 degrees again.*    And, I'm sure you've noticed that your home furnace turns completely-on or completely-off...  If you need a little heat, the furnace doesn't put-out a little warmth...  It turns on for a short period of time until the target temperature is reached.





* In real-world systems there is usually some hysteresis or "swing"...   i.e. If you are cooling to 10 degrees, you turn on the cooling when you hit 11 degrees (or 10.1 degrees, etc.) and you keep it on until you hit 9 degrees (or 9.9 degrees, etc.).

Jiggy-Ninja

Hi I want to use one of my PWM pins with a low pass filter as an dac to drive a mosfet. my current design is for a fast pwm with a 0,82uF cap and a 4.7kOhm resistor. My question is do i need a pull down resistor ? and what value should it have? I would put it in front of the low pass filter.
thanks for the Help
Don't do that.

The best use for the MOSFET in this case is as a power switch, driven fully in saturation and fully in cutoff. This puts all the power on the Peltier, and you won't need as much of a heatsink on the MOSFET.

Quote
I was told peltiers don't like pulsed voltages
I've heard that the problem with this is thermal shock. If you cycle them on and off too slowly (like a thermostat does with your oven, with a cycle period of many seconds or minutes), the heating/cooling cycles cause repeated thermal expansion and contraction, which can physically damage the element.

A few hundred Hz PWM will be far too fast for any appreciable thermal shock to occur though, and shouldn't be a problem.
Hackaday: https://hackaday.io/MarkRD
Advanced C++ Techniques: https://forum.arduino.cc/index.php?topic=493075.0

MarkT

The losses in a peltier device go with current squared, the cooling power with the current, I believe,
suggesting analog control of the current is best.  Of course using a SM current source to drive it
would be ideal as switch-mode can be 90% efficient in the electronics fairly readily.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

ad2049q

Possible solution:


PWM provides almost 5V and 0V
as that crosses the 2V of the FET, it could cleanly PWM the FET current.
You want a smooth current through the Peltier.  Can it be done with a substantial inductor coil in between the FET and the Peltier ?  You'd need to check whether your FET has a suitable built-in diode or needs an external "freewheeling diode" around it.

Also check for ways of changing (in setup) the PWM frequency.  If it is many kHz then you have better chance of needing a feasible sized inductor coil. 


Coding Badly


Go Up