LinkBack URL
About LinkBacksMagic smoke released is always a bad sign.
Magic smoke released is always a bad sign.
Wayne Hill
Looking for a cheap alternative for moderate power Rsenses. Just for giggles if i wanted to make my own Rsense using copper wire, assuming a max drive current of 5a, Id need 14-15 awg and at a foot the resistance would be ~2.5mOhm. To accomodate for this in my circuit I can only think of 2 methods for dealing w/this, both of which prolly wont work very well.
If I put a gain stage in to boost Vsense to the gate drive op amp the gain would have to very large. Im guessing this would result in current overshoot as the increased gain would kill the bandwidth and wouldnt be able to respond fast enuff.
If I divide down the Vin after the first buffer the signal becomes smaller than the gate drivers in offset.
Maybe if I combined both of the above and found another material that was more resistive?
I think you are overcomplicating things a bit. I would choose your current sense resistors to give about 0.5v at your 100% current of 2.3A and a safe dissipation of 1.15W max so you can use standard 5W resistors. The average dissipation when running is 0.707 x 1.15W so when motor is running your resistors only dissipate 0.81W which is all easy enough.
2. You don't need super opamps, just some that will work ok rail to rail or at least at the 0v rail as the current sense input will be 0v to 0.5v.
3. What made you put inductors in parallel with the FETs???
4. What voltage specs etc are your FETs? You shoudln't be losing FETs in a lini design unless they are very low voltage or are being physically cooked by underheatsinking.
1. Yeah probably over reaching at this point, Instead of focusing on a 5A driver, just worry about my immediate needs. At .5v for Imax my 8 bit ladder wont reach 1/32 w/5v vdd, 1v tho would make power more manageable and still be able to get 1/32 in hardware.
2. Ive used this circuit before, but as variable resistance to heat aluminum blocks, in my exp some r2r opamps will try to invert if you try to switch them too quickly, hence the high bandwidth req. Tho your suggestion of integrating the input would likely help this. Ive had good luck w/opa2340
3. Oops, screwed up cut/paste from my orig schem. Should still be a low side sink for the FET, good catch. I also found in my orig schem i have a few nodes not tied to ground. Please understand I do most my schems free hand and just use the editor to get stuff on the web.
4. Dunno really what happened, the mosfet is a ndp6060 60v vds 100w. I was using a real power supply 30v/5a and I was scaling the Vmotor voltage not looking at the ps but the motor when it locked up. I looked at the ps and It was at it was clipping the current limit, so I hit the power button before I saw the actual voltage. Mind you it was below Vds. Snooping around the circuit one of the 4 Rsenses was beginning to melt the bread board, sure enuff that fet was shorted. Ive used these mosfets in the circuit mentioned above and have never seen one fail in this manner. Honestly I dont know what the hell happened.
Thanks for the critiqueing, keep it coming!
ps you have paypal account for a donation?
Hmm, re;
1. At 1v and 2.3A you are pushing the sane limit for a cheap 5W resistor as your current sense resistor. rather than go to a 10W resistor or some expensive option I would consider reducing to 0.5v (giving 1/4 of the heat dissipation) even if it needs a simple voltage divider after your R2R ladder or an adjustment to the R2R ladder etc.
4. Those mosfets 60v 60A really don't sound suitable to me. They are optimised for a very low Rds on for switching supplies. That is totally wasted when you are running linear as they are never turned fully on. When the unipolar motor changes phase you will get about 2x the PSU voltage on the open side FET, so you are continually putting 60v on a 60v max rated FET.I would look for a 120v or 150v 15A or 20A FET or even a good old fashioned bipolar transistor, but definitely aim for a Vds or Vce about 4x the PSU voltage. Even if it is a lower current device that won't matter as you are regulating current at all times in the linear system.
5. Nah on the paypal I canned them after experiencing their terrible customer service. I'm here for fun and experience too so don't sweat it.
1, Or just run a lower vdd to begin with, but since Im already buffering the ladder from the IO switch dividing down would only add one resister per.
2, Wow, this hadnt occured to me, Ill will need to research this a bit. Most of my limited exp is w/fets, not too familiar w/bipolars. Dont bjts suffer from drift w/heat? Again I need to look into to this a bit more.
Re the Rsenses, is using wirewounds a concern or should I be looking at non-inductive? Intuitively Im thinking nonL but now my confidence in my understanding of the circuit is a little shaken.
Any transistor with a fairly high Vds/Vce and a decent power dissipation should be ok. They won't be switching fast (not by SMPS switching standards!), and they won't see saturation so a slowly changing linear system is one of the cases where FETs are not always the best. FETs and BJTs will both have drift issues with heat int he linear region, but neither matters as you are controlling current via closed loop in the opamp.
I don't think the extra cost of non-inductive resistors is worth anything here. Once you have the smoothing (integrator) caps in place the rate of current change will be slowish, in the order of milliseconds etc from one microstep to the next so that very low inductance in the R is insignificant.
Alright, well I got some homework to do on this. Your insights have been very helpful. At least I know the general circuit is sound enuff, I feel good enuff to do a proper schem and begin pcb layout.
Unfortunately, anything other than theorizing will have to wait a bit as I just closed escrow on a new place. Gonna have my hands full remodeling for a bit.
Since you dont accept donations, I guess all I can offer is my gratitude. Id really be stumbling thru this w/o your help.
Big cheers to you Sir
Wayne
Posting Permissions
