More transistors to constant current

Thread Starter

pager48

Joined Nov 25, 2018
160
Does this type of constant current source make each identical transistor in it dissipate identical power or can one of the 4 transistors dissipate much more and be damaged?

In actual PCB each transistor has different characteristics even if they are the same part number but In LTSPICE every transistor is identical since it uses same spice model.

 
Last edited:

crutschow

Joined Mar 14, 2008
23,291
Yes, the tolerance difference in Vgs(th) between transistors of the same part number can cause a large difference in the current carried by each MOSFET.

You can help equalize the currents by putting a small value resistor in series with each MOSFET source terminal.

Better would be to use a separate op amp and shunt resistor circuit for each MOSFET.
Then the currents would match within the shunt resistor tolerance plus the op amp input offset voltage tolerance.
You drive the four op amps from one pot.
A quad op amp, such as the LM324, would require only one op amp package for the four MOSFETs.
 

Thread Starter

pager48

Joined Nov 25, 2018
160
Yes, the tolerance difference in Vgs(th) between transistors of the same part number can cause a large difference in the current carried by each MOSFET.

You can help equalize the currents by putting a small value resistor in series with each MOSFET source terminal.

Better would be to use a separate op amp and shunt resistor circuit for each MOSFET.
Then the currents would match within the shunt resistor tolerance plus the op amp input offset voltage tolerance.
You drive the four op amps from one pot.
A quad op amp, such as the LM324, would require only one op amp package for the four MOSFETs.
Aren't the lm358/lm324 plagued with quirks especially if one channel is unloaded then the others oscillate or do something they aren't intended to?
 

crutschow

Joined Mar 14, 2008
23,291
Aren't the lm358/lm324 plagued with quirks especially if one channel is unloaded then the others oscillate or do something they aren't intended to?
Not that I'm aware of.
But in this case, all channels will be loaded equally.
 

TeeKay6

Joined Apr 20, 2019
377
Aren't the lm358/lm324 plagued with quirks especially if one channel is unloaded then the others oscillate or do something they aren't intended to?
@pager48
The LM358/LM324 opamps are very well behaved...when used properly. If you are not using one of the opamps, it is NOT proper to leave its input pins unconnected. Connect any unused non-inverting input to ground. Connect any unused inverting input to its corresponding output pin. This configures the unused opamp as a gain-of-1 amplifier for the 0V (ground) at its non-inverting input. This configuration prevents pickup of stray signals by the opamp inputs.

An alternative solution to your problem would be to use a single MOSFET of greater current & power rating; then only 1 opamp is needed. Congratulations on realizing that paralleling MOSFETs is not as simple as many people think.
 
Last edited:

crutschow

Joined Mar 14, 2008
23,291
Here's the LTspice simulation of an example circuit using the LM324 (you can use some other quad single-supply or rail-rail op amp if you prefer).
As you can see, all the MOSFETs conduct equal currents even through all four MOSFETs are different types (for simulation purposes) with different Vgs(th).
The different gate voltages provided by the op amp output to generate the desired (identical) current in each MOSFET can be seen in the second plot pane from the top.

upload_2019-8-22_23-4-9.png
 
Last edited:

Thread Starter

pager48

Joined Nov 25, 2018
160
Id rather use the 4 channel/transistor lm324 constant current arrangement.

But can the small series source resistors in the single opamp circuit be under 100mΩ to balance out the current?
 

crutschow

Joined Mar 14, 2008
23,291
But can the small series source resistors in the single opamp circuit be under 100mΩ to balance out the current?
The value of the balancing resistor depends upon the variation in Vgs(th) that the MOSFETs have, and how much unbalance you can tolerate.
A 100mΩ resistor will give 100mV of correction for each 1A of current unbalance.
 

Alec_t

Joined Sep 17, 2013
10,350
but In LTSPICE every transistor is identical since it uses same spice model.
They don't have to be identical. If you want to test the effect of parallelling non-identical FETs, you can use a variety of spice models.
 
Top