Hi, I have built the attached circuit. It is meant to control the power consumed by the bulb. The power feedback in the illustration is just a difference multiplied by a gain and the voltage across the bulb. However in the real circuit the power is measured by the MAX4211DEUE power monitor. For low base currents, the transistor will open up and allow current to flow through the light bulb as predicted. However if I saturate the transistor fully and wait a little while, I'll find that I can't pull it out of saturation even if I bring the opamp setpoint low. The bulb resistance in the figure is slightly misleading because it changes with the temperature of the bulb. However, the opamp model is correct, and the transistor part number is correct.

Can I get some advice to figure out what is going on? The current through the bulb when this happens is about 0.5 amps. Is the Vce getting too low and creating some kind of equilibrium that makes the control loop not work? My initial thought to solve this problem is to not let the current through the bulb get so high, but does anyone have any other suggestions?

 

Please show a schematic of the actual circuit.

 

Please show a schematic of the actual circuit.

Can't do any better than a guess from what the TS posted - but I'd say net positive feedback makes it a bistable circuit.

 

Sure thing! I apologize in advance as the circuit I'm describing above is a small portion of a larger circuit board, but here is the full schematic.

Here we go. The analog signal comes through and isolation amplifier, 0-5V. This has been verified to work fine.


Then it goes into the first of two CMOS switches which route the signal. The first switch triggers when the REM_CTRL signal goes high and passes the AN_CTRL_ISO through. Note that this voltage is scaled to about 2V. This has also been verified to work.



Then the signal is passed to a second CMOS switch which enables each source individually. This works fine.


Then the signal is passed to this MAX4164 quad opamp. Here only two of the 4 channels are used. This opamp takes the output of power monitor and feeds it back to the incoming analog control signal.


The power monitor is the MAX4211DEUE which measures the voltage across the load (a lamp) and the current through the lap. The power monitor also has an internal gain of 16.67. So the formula for POWERFB is POWERFB = 16.67*Vr11*Vsense.




Here WL+ is the positive side of the lamp and the negative side goes to ground. I believe R24 is 2k and R29 is 1k.


I know you guys are probably gonna chop off my head for making such a compact and crazy schematic. If you need to refer further, the full on schematic (if you dare) should also be attached.

thanks so much in advance!
windell

 

So I fired the circuit back up and this time the behavior was different, I couldn't get it to dim at all. I figured something has to be up so I turned the input all the way down and traced the analog signal back to the source and found that two of the switches on the board weren't being enabled. Then I realize that I forgot to rename the EN_REM to EN_REM_ISO on the board. I soldered some jumpers in and voila! Success!

I then tuned the voltage dividers a bit to get a full swing on the voltage sense. Then I realized that my opamp doesn't have specs out beyond 2 mA and my base current for the transistor was something over 3 mA. I figured that this wasn't a good thing so I paired the power transistor with a small signal transition to make a darlington pair. I don't need to worry about that current limit now! I should've chosen a opamp with a higher drive current. There is no need to use a fancy low power opamp on this board.

Then at the end of the day while characterizing the circuit, I blew the isolation amplifier because the power supply I was using had a range >>5V. I removed the IC and just jumped around it. Gonna revise the board anyways so it can limp along for now.

Thanks guys!