High Current Laser Diode PSU / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

High Current Laser Diode PSU

first previous 2 3 4 5 next last

Move Thread

LAN_403

EEYORE

Tue Aug 12 2008, 06:51PM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Hello,
I cant get the circuit to work without all the components (vref, op-amp, and the mosfets) being at the same voltage. (12volts). I even tried to use a 12 volt battery. (That ought to be pretty steady). Some thing happend. Current climbing. I simply dont understand how to get a bandgap ref to work like this. If I dont keep all the components at the same voltage, everything goes crazy. Is this due to high side current monitering? I guess what I am getting at is that I need a bandgap reference that can keep things at 12 volts FROM a 12 volt supply.

If I switched to lowside with NPNs, would that allow me to work the Vref from a lower voltage? I honestly dont see how the original circuit works so well. He is doing highside with a voltage regulator. He no doubt cannot be getting it to go to 12volts AND keep everything else at 12 volts!

Why is my op-amp not working the same way everyone else's works? Why does mine want to adjust current based on the difference between 12volts and Vref? If it were simply as everyone else says, then it would be a piece of cake! Could it be my op-amp?

Looking at the dummy load circuit, your powering your op-amp from a 9volt source, but your Vref will go from 0-1.25volts. Right? How does yours not go off the deepend? Is it this lowside sense business?
UPDATE: I got real frustrated and gave the circuit a "smack with a hammer". Really just flicked the wires around abit. It now behaves a little better

It still walks up and down, but stays around +/- 5 mA or so. Seems that some wires were not making as good a connection as they could have been. (Still on a breadboard here). There is an awful lot of heating going on. The breadboard is getting very warm. The fets are hot despite being on a heatsink with fan attached. Im guessing that my problems MAY be due to poor wire connections (ones that carry alot of current) and fet heating. (rds is .5ohms I think). I have new fets coming with much lower rds. That SHOULD mean less heating right? (Less power dissipation). If they improve the operation even more, then I tihnk I will call it problem solved.
Thanks
Matt

Electroholic

Wed Aug 13 2008, 01:25AM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

ditch the breadboard, they are no good for power electronics.

Rdson doesn't not matter when the fets are running in saturation mode.
It gets hot because you are using 12V in instead of 5V and the fet has to drop all that voltage.

and because you are using highside current sense, your feedback is referenced to the positive rail on the high current supply.

why dont' you just solder down a single channel, and maybe test out different opamps and stuff.

EEYORE

Wed Aug 13 2008, 06:38AM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Dang it, oh well I needed some more FET's anyways...
I cleaned up the setup some with thicker nickel plated wires and shorter paths. Now everything seems to work right! Current stayed where I set it. Every 10mins or so, current rises by 1mA and then settles back again. It has been running the laser diode at 4.013Amps now for an hour. Sooooo, I hope I can say problem solved. Crappy wiring was to blame. I will keep testing it for a couple more days and then finalize an expresspcb board.

Here is a question for the pcb: What would be better? Two more FET's (and another dual op-amp) or the temp control circuit? As of now, the temp control is pointless, as the fan needs to run non stop. It is a little neat though to have, but I doubt would be all that useful for anything but turning on a fan/TEC. (Doubt its use for temp tuning).
Thanks for the help!
EDIT: So I decided to try out the LMC660 quad op-amp. It should be good on the 5 volt supply. I cant get more than 1.5amps or so out of the mosfet. The Vcm of it is 1.5volts. Doesnt that mean that it can accept up to 1.5volts of feedback from the sense resistor? With a 0.1ohm resistor, that should NOT be a problem, yet it is?1.5 volts ought to mean 15 amps max...Whats going on here now? I checked that my vref can go from 0-1.5volts and it does. (Still using high side current)
Matt

Electroholic

Thu Aug 14 2008, 03:53PM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

congrads on figure out the stability issues.
However I still think you should be using low side current sense.
Do you understand that part where your sense resistor is giving you 5V to 4.5V depending on current? and that Vcm of opamps will always be lower than V+?

when current is 0, your high side current sense resistor will give you V+, because no current = no voltage drop. when current is I, then its V+ - (I*R).
so 4.5V at 5amps. for V+ = 5V and R = 0.1ohm.

So basically, you need to look for something with a rail to rail input.

For low side however, the range would be 0V to 0.5V.

EEYORE

Thu Aug 14 2008, 04:30PM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Ahh... That makes much more sense to me!
Things just got much more clear for me. With low side current, do i need an nfet instead of pfet?
thanks so much!

EDIT: So I have been trying out different op-amps as well as looking up others. Most have been dissapointing. Some work alright I guess, others not. The last one I tried was the LMC660. It worked somewhat ( did not respond to feedback linearly...100mV across a 0.1ohm resistor did not result in 1amp through fet) on the 12 volt supply, but would not work past 1.3amps through the fet on the 5 volt supply. (Both low side current). Seems the TL082 is still the best performer!
I have decided to stick with high side current since I have had most success with it. Now I need to figure out a way to regulate the Vref to make it more solid. Even though a little up and down current wont kill a laser diode if not run near its max, it just bugs the piss out of me to watch!

I need to help figuring out how to regulate the Vref, when my power supply is a regulated 11.99volts DC. (vicor flatpac).
Since the feedback with go from 12-11.5 volts, I have no idea how to do this. I dont even know why the vref is so sloppy if the voltage across the divider COMES from a regulated supply? Might adding smoothing capacitors to the supply help?
Are you guys about tired of me yet ? :)
matt

EEYORE

Sun Aug 17 2008, 12:16AM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Some updates:
Design topology finalized; low side current sense.
I got the driver to work with an op-amp from maxim; MAX4167
My FETs are NTP15N40, but I am only able to get 4.7amps combined from the two of them. I am going to try out some IRF540's soon and see how they do.
There is still significant heating of the FET's, but not nearly as much now that I am running from 5 volts. Stability is still an issue, but not as bad as before. Sometimes current will drop some 50mA or so. Sometimes it wont. I am now using a 25amp DC ammeter and the needle stays put. Could the fluctuations on the DMM be "ghostly"? The voltage on Vref also stays put, yet the current creeps up and down (mostly down) according to the DMM. Or is it just that the ammeter is much too inaccurate to show the "small" current flucuations? If my Vref remains solid, should the current too? Could the inputs need more current or something? What could cause this instability? It shouldnt matter how hot the sense resistors or the FETs get, because Vref remains solid. Shouldnt the op-amp adjust accordingly to changes on the "other side" of things?

I am currently using an LM350 along with a voltage divider for Vref. A precision 2.5volt reference is coming to replace the LM350.

I will be redoing the PCB to suit the new low side current sense topology. I hope to be able to reliably control 10amps with this setup so that I can scale it up for a 20watt laser diode. My future plans:
1-improve Vref
2-Much better heatsinking of the FET's
3-Better FET's = more current? The rds of the IRF540 is some 26mOhms, whereas the current ones are some 260mOhms....I will be editing this post as I make improvements...
Matt

Electroholic

Sun Aug 17 2008, 05:31AM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

The diode is your bandgap reference, top goes to power and bottom goes to ground.

In this configuration, the pot will output any voltage between Vref and ground.

power dissipation for the fet is the product of current and voltage drop across the fet. Pick one that has a low Rjc(junction to case thermal resistance), and in a nice easily heatsinkable package, such as the to3p or to247. Look for something low voltage and high current. If all fails, IRFP250 is good if not
overkill. CPU heatsinks are good too.

i am using some IXFH50N20 in my dummy load, 5 channels, each can sink up to 12A, limited by my 1.25V reference.

As for the drift you are seeing on the dmm, if the internal shunt heats up or cools down it might mess the reading up. I saw the same problem testing my dummy load. Until I found a 100A shunt. The thermal mass helped a lot.
Also, doubt you can see a 50mA change in on a 25A analog meter.

EEYORE

Sun Aug 17 2008, 07:15AM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Hello,
how did you get your current up to 12 amps? Mine now will not go past 4amps? Is it the range of feedback for the op-amp? Or is it limited by the FETs? I can go from 0-2.5 volts on my Vref right now. But current stops around 4amps or so. (Vref is around .367volts or so at around 4amps).
EDIT: It was the FET's. I acutally switched to an IGBT because I had alot of them and was curious. Some beer, some curiostiy, and now I just need to change my Vref divider a bit. The IGBT's work. I will see what I can get out of them...
Thanks!
Matt

Electroholic

Sun Aug 17 2008, 03:54PM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

are you running the opamp off the 5V supply?
if yes measure your gate voltage at 4amps, i bet the opamp output is maxed out at 5V. I used a 12V rail.

EEYORE

Sun Aug 17 2008, 05:44PM

Registered Member #99 Joined: Thu Feb 09 2006, 06:10PM
Location: florida, usa
Posts: 637

Electroholic wrote ...

are you running the opamp off the 5V supply?
if yes measure your gate voltage at 4amps, i bet the opamp output is maxed out at 5V. I used a 12V rail.

Hello,
yes I am. The op-amp is speced to run at 5 volts, with 6 volts being its max. I managed a good 11amps out of the IGBTs. That seems pretty decent for 5 volts. Now I will try out some t0247 devices to minimize heating. I need to better heatsink the current sense resistors too. They get super hot at 11amps! The transistors will have a fan attached to the heatsink as well...Anyone know of a fan that can run from 5 volts?
Matt

first previous 2 3 4 5 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob