I'm looking to make a "Current Source" I'd like to supply a maximum of 10A, I'm thinking about putting a transistor at the output of an Op-Amp and using the transistor to pull the 10A through my load.
My question is, would I be better off using a MOSFET or a BJT for this (darlington or maybe even IGBT) it really just has to sit there and handle 10A... does this point towards which transistor has a better SOA profile?
much thanks!
Actually, to make a BROADBAND current source into the 10's of MHz, it's better to use a good resistor. Something like 100 ohms would even be better than an active component. Go look at the schematic for HP's 8552 plug-in, part of the Spectrum Analyzer line and you'll find such a resistor. albeit a bit high wattage. [End of tongue-in-cheek.]
I'd preference towards the IGBT, as yielding a bit better control for those 'unexpected' scenarios, ...that ALWAYS happen. Be sure to match a passive network to make the output Z stay relatively constant over your spectrum, else the capacitance starts lowering it dramatically as you go up in frequency.
You're starting to get into 'component' characteristics, maybe Jim will jump in with a definitive answer instead of an opinion.
Wow.. 10A current source.. no matter what you choose you'll need a monster heat sink. How about using a current regulated power supply. You can get a 10 Amp 30V Mastech for less than $200. (I was looking the other day.) Do you need to do something fancy with the 10A? like modulate it or something?
George H.
Mosfet. Better SOAs, no base current error, easy to drive.
Post your schematic and we'll check it for you.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation
0A, I'm thinking about putting a transistor at the output of an Op-Amp and using the transistor to pull the 10A through my load.
darlington or maybe even IGBT) it really just has to sit there and handle 1
0A... does this point towards which transistor has a better SOA profile?Thanks, it's just to control a heater, just a resistive load. Here's a sket ch of what I'm thinking.
I was looking at the STY139N65M5 Mosfet, thinking it would need minimal hea tsinking
It has thermal resistance of 30C/W (j-a), if I double RDSon from 17mOhm to
34mOhm, at 10A that's 3.4W, which at 30C/W is 102C above ambient... which i s very hot but I think the part should survive right? I'll be heatsinking i t anyway. 10A is my max which I probally wont use muc.much thanks!!
Your calculations are wrong, the STY will dissipate Vcc*10-(RL+Rshunt)*10*10 watts.
-- Saludos
For just a heating element I would switchmode it. It's such a piece of cake and won't use a ton of aluminum. You don't even need any inductance. Any f requency that doesn't bother your ears should be fine.
Then your source resistor can become a small inductor for current feedback making the whole thing a tad more efficient.
I also question why to use a regulated current for a heating element. If it is well insulated I guess that is a good method of temperature control but if there are variations in ambieent that will affect it, then controlling voltage might be better because it will tend to self regulate as ambient ch anges. Of ocurse there are some applications that would require it, like fo r measurments or something. If that's the case and you're stuck with curren t mode control I think switching would still be much better.
Otherwise that FET may produce more heat than the heater. If you don't feel ike designoing the PWM you could probably adapt someting that already used a standard old (read cheap) IC like a upc494, something like that. Of cour se there are better ones now. With any luck someting will drive the FET jus t right and keep your component count down.
eatsinking "
If by that you mean that it can dissipate 625 watts, don'e think you autoat ically get all kinds of headroom. the maximum operating temperature is no h igher so a 625 watt device needs as much heat sinking to dissipate 100 watt s as does a 250 watt device. You simply don't gain that way.
With it not being switched, the only time it will run col is all the way on or off. Sure it will run cool switched but that is not what the drawning i ndicates. If you use 24 volts DC in and want the current say, developed by
12 volts at the element you let's say 7 amps at 12 volts. You need the heat sink for 84 watts even if the FET can dissipate 10,000 watts.A higher dissipation device may derate more slowly than a lower one, but th e curve still ends up at the same place - zero watts at 200C or something l ike that.
Whatever you use, read the data sheets very carefully. SOA profile has as much to do with packaging as with the basic technology, and most transistors these days, of whatever stripe, are made to be power switches, not current-passing devices.
I'm going to go against J.L.s suggestion and say that maybe you should use 2N3055's, or other devices using that die. If you want the "no base current error" then use a MOSFET from collector to base in a Darlington configuration.
Even if you use someone's xxx3055, though, you should still make sure it's not going to burn up.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
I second that. Please explain why you feel you need a controlled current!
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Heaters act as constant current sources to some degree because the metal conducts less as it heats. You can actually buy old constant current light bulbs in surplus stores. That means that controlling it with another constant current source will produce some instability the heating element's voltage. I wouldn't do it that way unless the regulator was part of the heat source too.
Pulse width modulation is more commonly used for heating element control.
!
I don't need a controlled current source, this was just the first thing tha t popped into my head, but if PWM is better, I'm all for it. Would the PWM people are suggesting be like setting up a buck converter where I can cont rol the PWM to adjust the voltage across my heater?
Why do you want to control voltage?
Using PWM with a constant supply voltage gives you linear power with duty cycle.
Best regards, Spehro Pefhany
-- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
With the feedback from the source resistor gou g to the inverting input of the OPAMP, it kinda makes people think you want to have a current set point for some reason. Someting like that may be used in a laboratory where the are measuring, for example, the heat output o an exothermic chemical reacti on but the measurement must be made at a temperautr much higher than the no rmal ambient room temperature. Hey, in a group like this anything is possib le.
In your case, if things are really not critical then that makes it really s imple. No need for anything at the source, and whatever percentage of input
enerator and sonmething to slice it on and off with the voltage from a pot of whatever. If you want some sort of digital control all you have to do is make a variable voltage out of it.
Since you have almost no issues, just make a square wave and use a resistor and cap to turn it into a triangle wave. A sawtooth would work as well but I thibnk you can make the triangle wave with less components. Then a simpl e comparator will give you the PWM based on whatever DC voltage you put int o the other pin. Inverting and noninverting don't really matter, depending on which way you want the control voltage to go. It probably makes more sen se fo the increased voltage to result in increased output, for some reason. So you put the triangle into the inverting input. The DC on the other inpu t can be modified by the output of anything that'll sense temperature if de sired. That can come later.
you can probably switch this thing at like 1 Khz, but some things do become speakers. If that happens just take it up to 10 or 20 Khz. Much higher and you might have to screw around with drive optimization. At 20 Khzz and a t otally (hopefully) resistive load, anything that'll put about 12 volts of s quare wave on the gate should work.
Always use the KISS theory, keep it sinple.
Got it, thanks! PWMing this makes sense, KISS!
NO!
IGBTs only have positive temperature coefficient of saturation voltage when driven HARD into saturation. When run in the linear region, they have NEGATIVE temp co., therefore the center (usually) of the die will go into thermal runaway. I was told by an IR engineer that even allowing an IGBT to operate in the linear region for sub-us periods would lead to failure.
Power FETS are vastly more robust, as they generally are always running in the pos tempco regime, even in the linear region.
The only advantage of an IGBT is low conduction losses in saturation, and a current source is pretty much by definition never running in saturation. Also, unless over 200 V, IGBTs have no advantage at all over FETs.
Jon
Umm, don't you have that backwards?
IGBT saturation is like BJT saturation, or diodes: negative tempco can cause current hogging. I have seen newer devices that claim 'ease of paralleling', i.e., zero or positive tempco at useful currents.
IGBTs and MOSFETs have the same linear characteristic, because the same phenomenon is at work. Vgs(th) drops with rising temperature, whether in a spot or the whole device. There is no positive tempco region (outside of saturation). As far as I know, MOSFETs have always exhibited "2nd breakdown" somewhere, it's just that the point at which it occurs varies widely. Oooooold lateral FETs are supposedly the best; presumably, breakdown occurs at temperatures and power densities impossible to achieve in a still-functional device.
Modern VDMOS stinks (if you want to cringe, check the DC SOA of an IXYS PolarHV type, I think?), but can be improved (e.g., IXYS' L2). Apparently, superjunction types are good (Infineon CoolMOS, etc.).
I don't know that I've ever seen an IGBT rated for linear operation though. I would expect they exhibit 2nd breakdown as BJTs do. One notable difference, the RBSOA is usually square (whereas a BJT typically only goes up to Vceo when turned off sharply, eventually making it up to full rated Vcbo as the junction clears out -- the RBSOA of MJE1300x series transistors is interesting), meaning you can switch it off 'instantly' without concern, which suggests different things at work.
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com "Jon Elson" wrote in message news:QuidnfqrmrjoOm_PnZ2dnUVZ_sudnZ2d@giganews.com... > RobertMacy wrote: > > >> I'd preference towards the IGBT, as yielding a bit better control for >> those 'unexpected' scenarios, > > NO! > > IGBTs only have positive temperature coefficient of saturation voltage > when driven HARD into saturation. When run in the linear region, they > have NEGATIVE temp co., therefore the center (usually) of the die will > go into thermal runaway. I was told by an IR engineer that even > allowing > an IGBT to operate in the linear region for sub-us periods would lead > to failure. > > Power FETS are vastly more robust, as they generally are always running > in > the pos tempco regime, even in the linear region. > > The only advantage of an IGBT is low conduction losses in saturation, > and a current source is pretty much by definition never running in > saturation. Also, unless over 200 V, IGBTs have no advantage at all > over FETs. > > Jon
Right. Current control will result in heater power that is proportional to the square of the DAC output, which can be nasty to close a loop around.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
I **THINK** I said it right. They have negative tempco when not saturated, which is what would happen in a linear use like a current regulator. The IGBT is hundreds of thousands of parallel structures, and they won't current-share unless you get some amount of positive tempco of the Vce. Supposedly, they are designed to develop a small positive tempco when saturated hard, this is at least good enough to prevent current hogging WITHIN each transistor. Since the OP suggested 10 A currents were involved, this would be a serious problem.
OK, I see your point, up to a point (pun intended). In any linear application, it would be VERY important to check the data sheets for hints of SOA, although they often don't give you all the data you might want. But, I still think the SOA concerns are MUCH worse in an IGBT that a FET. With the IGBT, it is VERY easy to get secondary breakdown in linear operation, the IR guy I spoke to said you could demonstrate failure under 100 mA with a device that was running stone cold to the touch, if you knew how to tease it just right. Power FETs seem to be a lot more robust, and you really have to push them to the limits in several ways simultaneously (current, voltage, case temp) etc. to cause outright failure. (Now, I tend to stick with pretty robust devices from IR, so there may be others that are not so tolerant.)
Jon
an IGBT to operate in the linear region for sub-us periods would lead to failure. "
I never heard that. I understand about linear mode paralelling. Are you tal king ONE device within its ratings here or something else ? Perhaps I'm rea ding it wrong.
One reason I ask is that I was considering using them in a linear applicati on. If I should just forget it then I will use something else. It would hav e been audio. I wasn't planning on paralelling them at all. If I can't use them there are always regular MOSFETs. (I do not even want complementaries, this is a center tapped choke output which of course has absolutely no nee d for them)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.