Forums

PWM Circuit for multiple heaters

Started by Fibo August 5, 2014
Hello,

I'm trying to make 3 PWM circuits to control 3 different thermoelectric coo=
lers (peltier cooler)

I'm aware that ideally you shouldn't run these with a PWM, that's fine, and=
 I'm only using these in one direction, only as a heater.

So I made a simple PWM, with a Linear Tech. PWM part, a high side FET drive=
r and a FET.

and the thing works ok for 1 heater, but once I plug three in... it goes ba=
nanas. Basically it looks like the PWM signal from one heater couples on to=
 my rails somehow and then starts to impact the other heaters... I'm not ex=
actly sure if that's what's happening, but they defeintely don't play nice =
together, once I plug two in the PWM goes all out of what, bascially become=
s super sensitive and almost floats up to 100% duty cycle when I touch the =
potentiometer

My circuit is on the link below

http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg

My circuit is that image 3 times, all common grounds throughout

I've thought about
1. Increasing my gate resistor to the FET
2. Instead of taking my control signal from the rail take it from a voltage=
 reference (TLV431)
3. Increase my input voltage control from 1V to 5V (it's an option on the l=
inear chip)
4. I don't want to, but maybe add isolated supplies?
5. OR maybe I should buffer my control signal?

much thanks!



You have bypass caps...right?

And, do you really have three individual RECOM thingers?  Those aren't 
cheap.

You should use all the pins provided on the gate driver, even if they're 
the same function.  The chip expects low resistance between equal pins, 
and if it has to draw current through the die instead, bad things can 
happen.

High side?  Where?

Tim

-- 
Seven Transistor Labs
Electrical Engineering Consultation
Website: http://seventransistorlabs.com

"Fibo" <panfilero@gmail.com> wrote in message 
news:460e1b6c-7461-4786-aecf-be1101177891@googlegroups.com...
Hello,

I'm trying to make 3 PWM circuits to control 3 different thermoelectric 
coolers (peltier cooler)

I'm aware that ideally you shouldn't run these with a PWM, that's fine, 
and I'm only using these in one direction, only as a heater.

So I made a simple PWM, with a Linear Tech. PWM part, a high side FET 
driver and a FET.

and the thing works ok for 1 heater, but once I plug three in... it goes 
bananas. Basically it looks like the PWM signal from one heater couples on 
to my rails somehow and then starts to impact the other heaters... I'm not 
exactly sure if that's what's happening, but they defeintely don't play 
nice together, once I plug two in the PWM goes all out of what, bascially 
becomes super sensitive and almost floats up to 100% duty cycle when I 
touch the potentiometer

My circuit is on the link below

http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg

My circuit is that image 3 times, all common grounds throughout

I've thought about
1. Increasing my gate resistor to the FET
2. Instead of taking my control signal from the rail take it from a 
voltage reference (TLV431)
3. Increase my input voltage control from 1V to 5V (it's an option on the 
linear chip)
4. I don't want to, but maybe add isolated supplies?
5. OR maybe I should buffer my control signal?

much thanks!




"Fibo" <panfilero@gmail.com> wrote in message 
news:460e1b6c-7461-4786-aecf-be1101177891@googlegroups.com...
Hello,

I'm trying to make 3 PWM circuits to control 3 different thermoelectric 
coolers (peltier cooler)

I'm aware that ideally you shouldn't run these with a PWM, that's fine, and 
I'm only using these in one direction, only as a heater.

So I made a simple PWM, with a Linear Tech. PWM part, a high side FET driver 
and a FET.

and the thing works ok for 1 heater, but once I plug three in... it goes 
bananas. Basically it looks like the PWM signal from one heater couples on 
to my rails somehow and then starts to impact the other heaters... I'm not 
exactly sure if that's what's happening, but they defeintely don't play nice 
together, once I plug two in the PWM goes all out of what, bascially becomes 
super sensitive and almost floats up to 100% duty cycle when I touch the 
potentiometer

My circuit is on the link below

http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg

My circuit is that image 3 times, all common grounds throughout

I've thought about
1. Increasing my gate resistor to the FET
2. Instead of taking my control signal from the rail take it from a voltage 
reference (TLV431)
3. Increase my input voltage control from 1V to 5V (it's an option on the 
linear chip)
4. I don't want to, but maybe add isolated supplies?
5. OR maybe I should buffer my control signal?

much thanks!

++++++++++++++++++++++++++++

Do you have a scope? have you looked at the rail(s)?

Is there only one bypass capacitor in the circuit? What is it?

Is the driver pulling down the low power regulator?

What are you using for a bread board? Take a picture of your setup and post 
it.


tm



 
> My circuit is on the link below > http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg
Are you sure that you can leave the output power unconnected? What's powering the output of the gate driver?
On Tuesday, August 5, 2014 8:02:48 PM UTC-5, edward....@gmail.com wrote:
> > My circuit is on the link below > > > http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg > > > > Are you sure that you can leave the output power unconnected? What's powering the output of the gate driver?
I put 470uF across the rails, it didn't help. It was on a solderless breadboard, but I got frustrated and took it all apart, I'm going to redo it to try and locate the issue. Yeah the gate driver is hooked up correctly, those pins are connected inside of it
The solution is simple. Take those cheapskate figners out of the pocket, pu=
lling out your wallet with the credit card. Call Digikey and buy some coils=
, caps and diodes. Problem solved. Hundred kHz ? Little low but you can pro=
bably get good coils for like two buck apiece. Caps, half a buck, diodes ev=
en less. We are talking a BOM here of about twenty bucks, and that's if you=
 splurge.

Well actually at 1 kHz it might be mnore for the coils. Why not just switch=
 them faster ? Any transistor on the market today that cannot effectively P=
WEWM at 10 kHz shouldn't be on the market today.=20

Maybe $25.
>"and the thing works ok for 1 heater, but once I plug three in... it goes bananas."
Three separate channels ? I had assumed that - am I wrong ? Either way, it is most likely a ground loop problem. And those chips, I guess they're OK but us oldtimers used to do shit ike that with three fifty cent trsansistors and a power transformer out of an old TV set. Just a matter of selecting thos varibales.
>"Tim Williams 7:32 PM (2 hours ago)
... You have bypass caps...right?" You cannot just do that without sticking some inductance in there. It has not to do with the Peltier, it has to do with the silivcon that must PWM that power.
On Tuesday, August 5, 2014 9:25:00 PM UTC-5, jurb...@gmail.com wrote:
> >"Tim Williams 7:32 PM (2 hours ago) > > ... > > You have bypass caps...right?" > > > > You cannot just do that without sticking some inductance in there. > > > > It has not to do with the Peltier, it has to do with the silivcon that must PWM that power.
inductors, caps, diodes? for what? filter out the PWM? do some kind of buck converter thing? Tim, just 1 recom, I mis-spoke, no high side FET here... i'll externally connect the internally connected stuff
On Tue, 5 Aug 2014 19:02:25 -0700 (PDT), Fibo <panfilero@gmail.com>
wrote:

>On Tuesday, August 5, 2014 8:02:48 PM UTC-5, edward....@gmail.com wrote: >> > My circuit is on the link below >> >> > http://i555.photobucket.com/albums/jj477/panfilero/PWM.jpg >> >> >> >> Are you sure that you can leave the output power unconnected? What's powering the output of the gate driver? > >I put 470uF across the rails, it didn't help. It was on a solderless breadboard, but I got frustrated and took it all apart, I'm going to redo it to try and locate the issue. > >Yeah the gate driver is hooked up correctly, those pins are connected inside of it
You still want to connect all of the pins, *particularly* on a solderless breadboard (such a kludge may never work properly). They're there for a reason.