Motor reversing with relay

So I've followed the following guide:
http://www.instructables.com/id/Sup...ino-/?&sort=ACTIVE&limit=40&offset=40#DISCUSS

But used 24v motors, which means I need stronger parts. For the transistors I have NTD5867NL and for the relay I have 845HN-2C-S. Everything works, except the fact that the transistors don't have built-in diodes and therefore don't stop EMF from blowing up the transistors. So, I have the 80SQ045NG diodes.

The problem is, I don't know where to put the diodes. I've tried to set it up like this:
http://www.instructables.com/files/deriv/F9L/KDFG/GU7FXUMH/F9LKDFGGU7FXUMH.MEDIUM.jpg
which killed my transistor.

I tried putting the anode toward the collector and the cathode toward the voltage source, which seemed to immediately kill my transistor. I tried reversing it and the transistor survived, but the EMF still killed it (I think, it doesn't work anymore anyway). So far i've mostly only been messing with the Enable transistor. I don't bother enabling the Direction transistor since that just increases my chances of ruining something else.

So, can anyone point out exactly what I'm supposed to do to make this work? I'm running out of transistors and my relay isn't liking the amount of soldering and desoldering I'm applying to it. I feel like this should be obvious but I don't want to get too experimental at such high wattages.

 

At this point I don't remember if I had the arduino attached to ground or not, I'm sure I did but if I didn't would that really make all the difference in this situation?

Anyways thanks for the diagram. I do have someone else looking at this for me who suggested to use a digital to analog converter, basically as a way to smooth out the PWM waves so there's no opportunity for the EMF to flow back into the transistors (assuming the motors aren't spinning too fast). I'm sure your diagram would help him out.

 

What I meant by smooth the PWM is convert the square wave into a sine wave. The way I see it is when using a square wave, during the few microseconds the wave is at it's low point, the EMF gets an opportunity to push through. This doesn't seem to be just theoretical because if I use a constant PWM, the transistor eventually burns up. If I run the motor at full speed, the transistor doesn't seem to be affected at all (until I shut off the power and then sparks fly).

So, by reducing the voltage (instead of using a square wave), as long as the motor is not moving faster than what the current voltage would normally offer, it shouldn't produce enough EMF to push back into the circuit. So for example lets say 6v spins the motor at 1000RPM. If the motor is spinning at 1000RPM or less, it shouldn't produce enough EMF to fight through the continuous 6v supply source.

The analog output does not solve the problem of the motors coming to a hard stop, coasting, or moving faster than they should. For this, I have a 2nd relay set up that simply redirects the motor away from the circuit, where I can use the EMF to power something else if I like.


So, does that clear things up? And if so, does it still seem wrong?