motor driver circuit
Started by ●March 29, 2004
Reply by ●March 30, 20042004-03-30
Reply by ●March 30, 20042004-03-30
If you want a nice clean square wave, drive a resistor. Theoretically the inductive kick back is infinite. I don't think that 20 khz is fast enough to need external diodes. I have found transzorbs at 1.5X the operating voltage and 2X the motor power to be fairly bullet proof protection for everything. Chad --- Vecheslav Silagadze <> wrote: > Thank for the replys, > > I'm not using discrete FET's for the circuit - I bought a few L298's, > which > are an integrated package with 2 H-bridges - 2A each. Takes up > *a-lot* less > space on the board than 8 MOSFETs. > Regarding using caps on the motors - interestingly enough I tried > this and > found that the motor with the caps runs much slower at the same > voltage/current. I guess this is to be expected since you're bound to > have a > lot of losses through the cap at 20kHz PWM. So I just decided to > avoid > putting them in at all. > Regarding the diodes, I am using 1N5819 diodes - they are 1A and > looks like > they are just the lower current counterpart for the 1N5822's (which > the H > bridge.) > > I had a MOSFET bridge a while ago as well, before I decided to switch > to the > L298, and I found that it was giving the same kind of kickback as > it's > giving me now - even though the FETs had built in zeners. I don't > know if > this is just something standard, or what, but have anyone looked at > the > output of their bridge with a scope and confirmed a nice clean square > wave? > > Mine seems to have a lot of kickback on the on-off switch, and > virtually > none on the off-on. I checked the diodes a 1000 times and I'm sure > the > circuit is fine. > > Vecheslav Silagadze > <<< snip >>> ===== My software has no bugs. Only undocumented features. __________________________________ |
Reply by ●March 31, 20042004-03-31
Reply by ●March 31, 20042004-03-31
The spikes are really only a problem if the driver is failing. Even the
simple 1N4000 should easily
clamp the spikes. But if the main problem is lack of torque I suggest you look elsewhere...the spikes most likely has nothing to do with it. Do expect 100% torque at 100% PW. Do NOT expect 50% torque at 50% PW. Depending on frequency and motor inductance the PW and current is not linear.... To get 50% torque it is not uncommon to need 80-90% PW. Steppers are basically 'current driven', so as RPM goes up so does the voltage.... to get good speed and torque from a common '5V' stepper you may need 20-40V ! Vecheslav Silagadze wrote: "A much better choice is the National LMD18201 MOSFET H-bridge. It can-- ******************************************* VISIT MY HOME PAGE: <http://home.online.no/~eikarlse/index.htm> LAST UPDATED: 23/08/2003 ******************************************* Best Regards Eirik Karlsen |
|
Reply by ●March 31, 20042004-03-31
On Tue, 30 Mar 2004, Dave Mucha wrote: > --- In , Eirik Karlsen <eikarlse@o...> wrote: > > Well....I'm not sure what you mean with 'kick-back', but if it is > only a small spike observed on the scope > > then I'd say 'don't worry'. If you've made a sensible choise of > output diodes then the spikes will be > > limited to VCC and GND. > A motor coil is an inductor and will act just like one. > > Put power on it and it will resist the current by storing it in a > magnetic field. Remove power and as the magnetic field breaks down, > it will release the stored power. > > Depending on the size of the coils in the motors, you would probably > get about 30% more voltage than you were supplying and could easily > get a few amps. > > I'd expect way more than a blip on a scope. > > In fact, when using stepper motors, the back EMF offsets a drop in > the power supply when switching from one coil to another. > > I like to select the diodes by a factor of 2 regarding both voltage > and current as to what is supplied to the coil. > > This is similar to relays, the colapse of the coil will deliver a > spike back into the device driving it. > > Dave > More than that, a real problem with steppers is not the voltage amplitude on the coils, or the "self-induction" which can be suppressed, even for high frequency driving (supplying the coils under constant current from a potential much higher then the nominal voltage of the motor, and switching very fast that current into the coils) but the natural resonance frequency of the motor, which could kill your driver stage, (or just shock your ears...) only when the command pulse match with the motor resonance. best regards, Vasile |
Reply by ●March 31, 20042004-03-31
On Tue, 30 Mar 2004, Eirik Karlsen wrote: > Dave, > yes I'm aware of all this, infact I once exploited the back EMF to make a simple HV generator. > It ran on 12VDC and had a transistor switch a simple inductor (not a transformer). > Then rectified the spike on the CMOS drain and got about 2KV. > It was a 1200V transistor so I think the voltage were limited by the transistor's breakdown voltage! > > But if you look at the L298 application note you'll se that each of the 4 outputs is clamped > by two diodes, one to VCC and the other to GND. > So if suitable diodes are selected, they are mounted in the correct direction, and the PSU > has some minimal capacitance across it, then the EMF-spikes will be limited to the > PSU rails (+/-0.7V or so). There is no necessary connection between PSU output capacitance and EMF here. Any PSU have a small output impedance which will "eat" the switching pulses through that diodes you are talking about. The output capacitor of a regulated PSU is just for stability and does not affect significatively the output impedance. Usualy it has a much larger value that it's required. best regards, Vasile |
Reply by ●March 31, 20042004-03-31
Not necessary, but possible......
If he'd used 6ft of thin wires from the PSU and and no decoupling cap. across the bridge then spikes is only to be expecte...diodes or no diodes. Vasile Surducan wrote:
|
Reply by ●March 31, 20042004-03-31
--- In , Eirik Karlsen <eikarlse@o...> wrote: > The spikes are really only a problem if the driver is failing. Even the simple 1N4000 should easily > clamp the spikes. > But if the main problem is lack of torque I suggest you look elsewhere...the spikes most likely > has nothing to do with it. > Do expect 100% torque at 100% PW. > Do NOT expect 50% torque at 50% PW. Depending on frequency and motor inductance > the PW and current is not linear.... > To get 50% torque it is not uncommon to need 80-90% PW. > Steppers are basically 'current driven', so as RPM goes up so does the voltage.... > to get good speed and torque from a common '5V' stepper you may need 20-40V ! > If you up the voltage, you need to limit the current. Also, I'm not sure if this is a stepper motor as I missed the beginning of the thread, but a stepper motor is a constant power device and will deliver all it's power in one step, but the torque will be spred over the number of steps. That means a motor at 1 RPM is delivering the same power as a motor running 1,000 rpm. The difference is that the torge per step is huge at 1 rpm, and you only get 1,000th of that when running at 1,000 rpm and at some point, the motor will not have enough torque to spin itself and it will stall. Dave |
Reply by ●March 31, 20042004-03-31
OK. I can see from this and some other post that there is a misunderstanding about what a fast recovery diode is. The "fast recovery" refers to the diode's ability to switch from reverse bias (which is what it is in when the PWM output to the motor is high) and forward bias (which is what it is in when the inductive kick-back is being absorbed by the power supply). This must happen very fast in a PWM motor driver, regardless of the PWM frequency. The time between when you are driving current through your motor coil and when you shut off that current is very short indeed. The diode must switch between reverse bias and forward bias very fast so that it can pass the current that suppresses the inductive kick-back voltage. If the diode is too slow to recover, you will get a voltage spike before the diode switches to forward bias. Sorry for the long narrative. I hope it helps the group understand PWM circuitry a little better. On your other zero torque issue - it is common to see almost no torque down around 30% duty cycle. The last gearmotor that I used did not even start to rotate until about 50% duty cycle. When it did start rotating, it had a high torque. > I am getting a voltage spike way over my Vcc actually - nearly 2xVcc If your circuit is correct, you can see that this voltage will never exceed a diode forward drop voltage above Vcc IF your diode recovers fast enough. > in fact. The motor still runs, but it gives pretty much zero torque > at around 30% duty cycle. > I will assume it is the clamping diodes that are causing the problem > and look for faster recovery ones (even though at 20kHz I can't > imagine any kind of diode won't switch fast enough... the 1N5819 I'm > using should be more than adequate.) Yes it should. If you try new diodes, try the 1N5822. I've used these with great success (basically the same as the 1N5819 but with more forward current and fwd surge current - this might make the difference). I've checked the circuit many > times over and I'm certain that the topology is fine. > > Vecheslav Silagadze > |
|
Reply by ●March 31, 20042004-03-31
> On your other zero torque issue - it is common to see almost no > torque down around 30% duty cycle. The last gearmotor that I used did > not even start to rotate until about 50% duty cycle. When it did > start rotating, it had a high torque. That's interesting. Why is the power being eaten up like that? Conceptually it would seem that at 30% duty cycle I should be getting 100% torque with 30% speed because I'm delivering 30% of the power I would be delivering at 100% duty cylce. But I guess this is one more of these places where theory meets reality :) I will try a fast recovery diode, but I'm basically limited to what the local Sayal Electronics carries - from what I remember the "best" diodes they had were the 1N5819s. It's unfortunate, but it seems that the datasheet on the 1N5819 doesn't specify the recovery time (the L298 driver I'm using asks for <200ns...) Vecheslav Silagadze |
|