MrHeckles

Out of curiosity has anyone found any advantage in using a switching voltage regulator, rather than the LM317, for a variable power supply?

I set the voltage difference pins on the gate and the source for the FET. Based on that I have moved the diode D6 to connect to the 12v supply, rather than the motor supply. That way it gives about +11 volts over the drain voltage to the highside gate (regardless of what the motor voltage happens to be). On the gate of the highside, I can see that the spike on the gate is about 6v above the source. I am guessing that it is enough to start turning on the FET and allow the shoot through. Slip of the tongue (so to speak). I did indeed mean P channel. No, but you do drive a tank with two hands on the brakes. The whole unit has two motors (one for either side), and castor wheels at the front. The steering is provided by a joystick that sets differing motor speeds. The reason I need quick speed reduction for the motors via the braking action is so that the thing will respond quickly to steering inputs, rather than waiting for the inside motor to coast to a slower speed.

Thats exactly what should be happening. It's a fantastic braking method for the motor. It allows for very tight speed control. With the highside FET then turning back on, it gives very nice regenerative braking. The highside gate goes up to 22v or so, the spike is only around 14v-16v, just enough to start it shooting through, though not enough for a full turn on. How would I go about clamping the spike when the required gate voltage is higher than it anyway? I have seem the 4qdtec site. Thats where I got alot of information about the first motor controller. They dont think switching the highside FET is a good idea, but I cant get decent PNP Mosfets in New Zealand for a reasonable price.

Is there any good way to do that apart from sticking a larger resistor on the gate of the highside FET with an anti-parallel diode?

I have used R13 and L2 as a basic simulation of a motor. M7 is supposed to remain on all the time. The PWM chopping for the speed control is done via M6. The motor is only supposed to drive one direction at this stage. M5 (I think this is other other lowside fet you are talking about) operates inversely to M6 which provides a very tight control on the motor speed. If the motor is driven faster (by the momentum of the unit it moves) than the current setting, a regen braking effect will cause it to slow down and maintain the selected speed. Your right, I did forget the miller effect. Is there anyway I can stop the drain gate capacitance from causing the spike?

Thanks Jim, I tried that, but it didnt seem to make any difference to the spike. I have sorted out some pictures. 1st, the circuit setup 2nd, the voltages from the probes in the first picture [img width=680 height=303] and a zoomed in view of the spike [img width=680 height=298] finally, this is a trace of the amperage through the highside mosfet. You can see how the spike is causing the shoot-through. [img width=680 height=308]

Hi guys. A while back I ask for, and received, a whole lot of help in regards to creating a motor controller. Happily it all worked out well in the end. I have recently decided to redesign the controller. Now I'm using pSpice to cut down on blowing up components. The circuit I have uses an SG3524 PWM chip to provide a pulse train to a discrete highside gate driver, and the inverted pulse to a lowside gate driver. The idea being that I can get very quick speed responses to throttle changes on the motor, and can use it for regenerative braking. The fets are all N-channel. In my pSpice simulations it works reasonably well, however there looks to be a small shoot-through occuring due to a voltage spike on the high side mosfet's gate when the lowside mosfet turns off. I am working on getting pictures - but in the mean time, is there anything anyone can suggest that might cause such a thing?

One simple question about the SG3524 regulating PWM chip. The error amp inputs are pins 1 and 2, the output of the error amp (and the compensation) pin is pin 3. The current limit amp inputs are 4 and 5. The current limit amp will over ride the output of the error amp. My question is, if I was to drive the voltage on pin 3 directly, would the output of the current limit amp over ride it to drop the duty cycle in the event of an over current detection?

Hi guys, I have a motor controller that uses the SG3524 chip to generate a PWM pulse. This is sent to mosfet drivers which run an H bridge setup. The onbard current limiting is resulting in a "jerkiness" of the motor when it starts to hit the 0.2v limit on the CL + pin of the SG3524. The current limiting in the SG3524 is supposed to reduce the duty cycle to 25% as the CL+ pin reaches 0.2v and then further to 0% when it goes above 0.2v. I figure that because a motor controller is going to drop the current as soon as the limiter cuts the PWM duty cycle back, the "jerkiness" is from the duty cycle being dropped to 0%, and the current limiting turning off, then turning the pulse train back on, resulting in a high current, resulting in the duty cycle dropping to 0% again, and so on. I was wondering if anyone had any suggestions for a way to provide some way of smoothing this out? Any ideas much appreciated.

I have found the ICL7660 voltage converter chip to be very helpful. Here is a link to a datasheet for it. http://www.alldatasheet.com/datasheet-pdf/pdf/67436/INTERSIL/ICL7660.html

Got it sorted now. I had changed the comparator to the -5v rail, and I figure that the -5v through the max333cpp was too close to V-. I changed it back to -9v for the comparator and it works fine. I also put in the reverse diodes to protect the transistors bases. Staigen - the -5v comes from a regulator elsewhere in the circuit (part that wasnt important to the original question, so I didnt bother showing the whole thing). Works a treat now - thanks guys.

I'll keep it in mind :) But I had another idea. I changed PIN's 2 and 17 to connect to -5v, rather than ground. That way I figured it would drag the comparators + inputs to a negative voltage (and turn them off). However, it didnt seem to work. When I changed the switch, the LED's would flick off, then come straight back on. On checking the outputs of the comparators, I found that when the switch was changed, they would drop voltage for a moment, then go back up again, in effect turning off for a split second, then back on again. Why would this be happening if there was a negative voltage on the + input to a comparator and 0v or very slightly positive on the - input?

I have hooked the negative 9 volt rail to the MAX333CPP. The voltage on PIN 3 when it is supposidly grounding is now 0.08v. Lower, but not low enough to turn off the comparator. What is this other solution of yours?

Its a bit hard to describe, so heres the setup As you can see, when the switch is open, and as long as there isnt to much voltage on the load signal feed, the forward comparator will latch on, turning on LED2. When the switch is closed, it SHOULD ground the forward comparator (thus turning it off) and turn on the reverse comparator (assuming that again, there is not too much load signal) and latch on, thus turning on LED1 (and turning off LED2). However, the MAX333CPP doesnt seem to ground PIN 3 or 18 very well, and the comparators do not turn off, and so the LED's stay on. Is there some way to get a good grounding through the chip for those forward and reverse limit feeds to the comparators?

Yes, it does work now (with the 10k resistor), with only one small problem. The number 1 SPDT and the number 4 SPDT in the chip are connected to latching comparators that stay on once a mode is selected (by the switch), and indicate through an LED. To turn the comparator off when the mode is switched (and turn on the other one), they are grounded through the MAX333CPP. Problem is that (I assume due to the internal resistance of the chip) even when PIN 3 is connected to PIN 2 (ie grounding it), there is about 0.12v on PIN 3 which keeps the indicator LED on, regardless of the actual mode selected. Is there anyway to overcome this so that when in the appropriate mode (via the switch) that PIN 3 and PIN 18 ground properly through their respective connection?