Can you help me...it's regulator for a train. I cant make

Your gate signals must be with referenc to aource.
Here your gate signals for high side fets are with reapect to gnd, this means that your gate will most likely see a voltage higher or lower than 20v, this destroys the device.

Look for a half bridge mosfet driver, use one for each side of the H bridge.

When you build the new circuit, use 10-30ohm reaistors in series with the gate. Layout should keep area between wires small.
Use decoupling caps, film and electrolytic for the bridges, and ceramic for gate drive.

Any parts that does not makes, try google, or begin with a less complicated version of the circuit.

>>1194115
Thanks for responding.
I will try with resistors for gates, but where exactly should i put these caps? I got one 100uF for Vcc.
With voltages from avr uc it didnt opened or very small open, not enough to move 12V motor. I power this with liion pack, whatever 9-15V, now its about 11-12V. Signals for ure arent higher, more like at 11-10V just like IRF datasheet says.

File: faulty_bridge.png (4KB, 262x193px)

4KB, 262x193px

>>1194075
This configuration may work with bipolar transistors but not with MOSFETs because the upper part has not enough gate voltage. Please read (to the end):
http://www.talkingelectronics.com/projects/H-Bridge/H-Bridge-1.html
There's a link about how to avoid such errors.

>>1194135
>http://www.talkingelectronics.com/projects/H-Bridge/H-Bridge-1.html
It got 11V on gates, all of them.

Also i got high states on every gate. I dont know how when i dont give signals to the inused pair! I dont get it. Seems to work but all of them! Also thoe not pwm'd!

>>1194166
>It got 11V on gates, all of them.

Yes, but no.

Gate voltage is measured relative to source, which is why it's often called gate-source voltage (Vgs). High-side FETs may get plenty of voltage to their gate, but the source is also going to be above ground because it's on the high side of the load. Thus, the differential between gate and source is insufficient to turn the transistor fully on. You need to either use p-channel FETs or a bootstrap circuit to produce sufficient voltage between gate and source.

Pic related.

File: h-bridge-driver1[1].jpg (267KB, 2312x1616px) Image search: [Google]

267KB, 2312x1616px

>>1194176

>pic related

I'm not retarded, I swear.

Still there is some problem,
looks like not used fets are opened
even when they got no signal from opamp! I dont get it. I added resistors 47R between gates and signals, also i got pull downs 1k on all gates.

I guess one pair up/down is faulty maybe caused by overheating which ones happened. Also they were making the signal from opamp go down to about 5V from given 11. I will exchange them tommorow and will post results.
On outside of one of the upper fets i got amplitude about 7V(out of about 11,8V).
This isnt very good. Maybe i will try some BD140 to drive them. But i thought there is huge resistance from gate, so im confused how to drive n-fets.

File: mostek.jpg (42KB, 603x481px) Image search: [Google]

42KB, 603x481px

Could this work better? I need more voltage cause now im getting just half of vcc at the motor "+" connector that mean source of high side n-fet.

File: MOSFETH.png (25KB, 717x407px) Image search: [Google]

25KB, 717x407px

>>1194608
>im getting just half of vcc at the motor
Of course, that's exactly what your 'faulty bridge' circuit is supposed to do. Read what >>1194176 posted to understand why that is. It's called gate-source voltage for a reason. You need p-channel MOSFETs at the high side. Image source:
http://www.bristolwatch.com/ele/h_bridge.htm
Read it.

File: TO16k.png (5KB, 456x329px)

5KB, 456x329px

>>1194075
Just dont use n channel paired with p channel.Its Like pnp/npn combo and itll work.

>>1194668
Using 10kohm as gate risistor is bound to end in failure..

>>1194751
>risistor
No, it works. Why would it not?

>>1194786
It will work for very light loads and slow switching / dc signals.

The gate of a mosfet is comparable to capacitor. If you want efficiency and thigh switching frequency you need to the gate (capacitor) fast.
This r charge equires current, which is limited by the resistor.

A mosfet can easily have 200pF or more. Use RC time constant to calculate the rise time from 0 to 4 volts (assumed threshold of 1 A) with a 10k reaistor and 10v gate drive voltage, you will spend 1us in commutation.
Here im not accounting for the miller effect..

It is not without reason mosfet drivers are capable of supplying several amps peak. You want to charge the gate fast.
10k would also cause problems with false triggering if fast voltage changes would occur between drain and source (inductive load/motor), this is again due the miller capacitance (drain to gate)

>>1194803
I'm well aware of all that but OP wants to control a model train motor (relatively low voltage and current) from the output of a slow 358. This is neither a kW nor a MHz affair. The gate drive is certainly sub-optimal (and asymmetric, too) but it needs not "end in failure". At least he will have less EMI trouble..

>>1194840
I run this with 4Khz which is max for LM358 at 10-11V aparently. I still see pretty long edges at this frequency.

>>1194803
Also i can not buy p-fets now, i have BD139 BC547 or BC560 and resistors, all i can have now.

And motors for trains are usually 12V, which i have. It should do at least 1,5A so its able to run let say 3 elektroloks.

File: L298N_motor_bridge.jpg (39KB, 501x546px) Image search: [Google]

39KB, 501x546px

>>1195109
>can not buy p-fets now
Meanwhile consider this. Instructions at http://www.instructables.com/id/Control-DC-and-stepper-motors-with-L298N-Dual-Moto/

File: mostek.jpg (39KB, 603x481px) Image search: [Google]

39KB, 603x481px

>>1195166
So this werks enough, but mosfets doesnt stop when low on gate and base of bd139...and still little to low voltage and ampers propably

File: Bootstrapped Mosfet.png (35KB, 655x322px) Image search: [Google]

35KB, 655x322px

And this one doesnt stop too, but it have best voltage and amps propably

File: motor_control.png (7KB, 360x273px)

7KB, 360x273px

Fast gate drive, relay for reverse.

>>1195431
I was thinking about relay, but i dont have big one for more amps, but maybe the one i have will be enough for one lokomotive at least.

File: HFD27_012_S.jpg (14KB, 450x375px)

14KB, 450x375px

>>1195591
>big one for more amps
Modern relays can be small. If yours is not smaller than this it should work.

>>1195294
>mosfets doesnt stop
because pulldown resistors (emitter/gate to ground) are missing. There's no current path to discharge the gate, so the FETs stay ON for a while and heat up.

How do you protect against short circuits in case something derails or shorts the tracks?

>>1195619
>because pulldown resistors (emitter/gate to ground) are missing.
Oh no, i did gave pull downs, thats first what i did, but somehow it didnt help or it didnt wanted to start...i was screwing with this for 2 hours yesterday.

>>1195431
I made small mods to this scheme and made pins
paralell, so on small eb2-5 relay i got i can reach 2 Amps tops, normaly they say its just 1. Then i will pwm n-fets normally like it is bridge. So the relay is clutching plus to the motor pins / rails.

File: mostek.jpg (34KB, 531x343px) Image search: [Google]

34KB, 531x343px

>>1195619
>>1195431
I did this and works well, im just not dure what kind of fuse to install to protect it. Its not very far from 3 loks amper use to short. And dont know what wattage can 0,47mm wires take.

File: mostek.jpg (34KB, 531x343px) Image search: [Google]

34KB, 531x343px

I just tested it and works strange, its like signals fighting with each other, at one pin of motor i got PWM like signal which is expected, but on the other pin of motor i got full VCC, like 11V. Just like it was doing this on PICREL.

I just dont know anymore. motor is running, but no power, to low amps, maybe still something wrong with mosfets?

File: RXEF250.jpg (4KB, 228x225px)

4KB, 228x225px

>>1196055
>works well
Congrats, you finally made it.
>what kind of fuse
I like resettable PTC fuses (see picture).
>what wattage
Wire diameter is selected for current, not power. Rule of thumb for copper wires: not more than 20A per mm^2 (current density). A wire with d=0.5mm has a cross section of πd^2/4=0.2mm^2 and should be good for 4A.

File: switching.png (14KB, 600x480px)

14KB, 600x480px

>>1196145
You cannot parallel the two relay contacts, both switches are required to reverse the motor. You need only one MOSFET, not two. You do not need a bridge to control the motor speed. PWM goes to gate, H = ON, L = OFF. It's all in the picture.

>>1196216
>You cannot parallel the two relay contacts, both switches are required to reverse the motor.
Yes i can, it works as fets high side. But it can be also made just as you show. I was thinking this way but i was scared my relay would burn if i get more than 2A, so i did paralell them pins.

>>1196148
Yea i maybe buy this. Now its importanter to buy faster opamp cause i see triangles instead square waves at gates, 10V / 15,6Khz. Almost doesnt matter what pwm proportion is.
Someone suggest some opamp? I just saw NE5532p act strange or it was normal for non single supply opamps but it completely inverted signal up side down, when i had 0 PWM it was full DC at the vcc level 11V, 128 PWM gave about VCC/2. Is it cause its simetric supply?

>>1196295
Can you post a diagram of the circuitry between the controller's PWM port(s) and the gate(s) of the FET(s)? Of course there are fast opamps but I still don't understand why you would need an opamp at all.

>>1196316
>why you would need an opamp at all.
To make 5V PWM a 11V PWM?
There is just normal opamp non inverting signal.

>>1196572
For this you would use a logic mos or bjt as switch which would amplify the signal, then drive a totempole (ab output stage)

The easiest way is just to use a mosfet driver. Solves a lot of problems.

The issue you observe with the opamp is most likely slew rate limitations..
Just because the opamp state unity gain at 1Mhz does not make it useful for 10khz sine waves if the amplitude is large.

Reading this threads leads me to belive you know just enough to get in to problems, but not enough to realize the problem.

You should play more with simple circuits, and understand Why they work, not how they are wired.

>>1196664
this is why i ask for help, cause i didnt know exactly. This is more like simple circuit.
Thanks everyone for help.

File: pwm-motor.png (7KB, 417x216px)

7KB, 417x216px

>>1196572
>5V PWM a 11V PWM
That's done with a simple transistor or two, not a slow, linear amplifier. Digital is about switching. Look at the diagram, this is a complete interface between PWM port and the FET. There are two (inverting) transistors such that L at the port means OFF for the FET. This allows you to disconnect the input (or switch off the controller) without the motor going full speed. Can you imagine what the two diodes are there for?

>>1196748
No i dont know what they do. But the bigger problem for me was the high side of the bridge.

>>1196748
The lower one prevents the pullups to swith the N fet right?

File: 1382433324854.png (331KB, 625x370px) Image search: [Google]

331KB, 625x370px

>>1194668
shoot-through imminent
but you really can't expect more from someone who uses eagle

>>1194751
>>1197245
Not my schematic. Anyway with 100R at gates really worked for me for reversing application.

Where does this myth come from that you need a H bridge to set the speed of a DC motor? You don't reverse the direction every millisecond after all. Bridges are required for stepper motors, not for model train DC motors.

File: Tyco-HO-Model-Train-Transformer.jpg (32KB, 400x362px) Image search: [Google]

32KB, 400x362px

>>1197348
my guess he wanted it originally for the reverse function
he's now down to relays which is where he should have started
>variable output transformer - dpdt reverse switch

>>1197245
>who uses eagle
better do pcbs in clay...making life easy is soooooooo stupid and unpro.

File: easy.png (33KB, 350x555px) Image search: [Google]

33KB, 350x555px

>>1197374
>TYCO transformer
Rock solid. Tons at ebay, longlife product?
>dpdt reverse switch
It could all be so easy..

>>1197672
I have 6 pin switch like dat, i didnt want to use cause i got no special disk knob to put it on the potentiometer.

>>1198335
6 pin switch like what?
What's the potentiometer good for?