Who makes high current modulatable drivers?

Who makes high current power supplies for laser diodes? I need something that can do up to 30 amps with at least TTL and preferably Analogue modulation. I am trying to resurect this Lightwave Half-Note system that I got from Junktronix. It uses a 20W Coherent CCP diode bar that requires 23.6A for 20W output. I don't think that modulating the TEC will be of any use as the laser head is constructed on an inch thck 5" x 8" block of stainless.

I don't want to spend too much on something like this and I would really prefer something that is plug and play, but I am willing to do some work.

Ben

Analog Modules Inc.

Please send us a pic of the look on your face when you receive the price list.

Steve

HAHA! Is it that expensive?

that can be made with an industrial darlington transistor like those used on big machines, commonly found on ebay for cheap

the current sensing resistor should be a copper trace on the PCB large enough to handle the 30 amps, and the design is similar to any laser diode driver with darlington

a friend at laserfreak made a driver at 17Amps with such a transistor with no heatsink, it didn't even reach 30°C in temperature

betcha lava could make one

Quote:

---

Originally Posted by keeperx

betcha lava could make one

---

Lava needs to work on finishing his Doctorate.

Betcha I could too.

Problem is, can Ben clone it?

Ben, feel like learning something? Or are you looking for a easy fix to see if the rest of the laser works?

If your looking for easy fix, I'm gonna say surplus lab supply with constant current option and a start/stop relay to protect the diode in the undefined states of the lab supply, ie when it starts up and when it shuts down.
Plus a guard transistor to aid during the shutdown state.

Ie a Kepco, a Sorenson, or a HP. Then you shunt the diode with a power FET for the fast modulation and guard function.

Or we can help you make a "real" supply. Your call.

This place puts the PSUs where I can test them and I'll just bill you for the gas each way :-)

http://www.hgrindustrialsurplus.com

Steve

Quote:

---

Originally Posted by mixedgas

make a "real" supply

---

:D

--DDL

Steve,

I know the laser works and I want to have something that will control it. I am fine with making it myself, in fact, I would prefer that just for the learning experience. I have access to one of the best electronic component stores around, so getting components is not a problem.

I am looking at where to start. I know how to do voltage regulation and run a power supply in constant voltage, but I do not know how the constant current setup works.

Ben

Quote:

---

Originally Posted by mixedgas

Lava needs to work on finishing his Doctorate.

Betcha I could too.

Problem is, can Ben clone it?

Ben, feel like learning something? Or are you looking for a easy fix to see if the rest of the laser works?

If your looking for easy fix, I'm gonna say surplus lab supply with constant current option and a start/stop relay to protect the diode in the undefined states of the lab supply, ie when it starts up and when it shuts down.
Plus a guard transistor to aid during the shutdown state.

Ie a Kepco, a Sorenson, or a HP. Then you shunt the diode with a power FET for the fast modulation and guard function.

Or we can help you make a "real" supply. Your call.

This place puts the PSUs where I can test them and I'll just bill you for the gas each way :-)

http://www.hgrindustrialsurplus.com

Steve

---

copper has the downside of a bad temperature characteristic, in the US, you can get a surplus calibrated industrial shunt for 10-20$ with a near zero tempco.

This is not the brand I like, but here is a example.

http://cgi.ebay.com/Empro-HA-100-100...QQcmdZViewItem

Steve

Ok, do you know Vf , the forward voltage for the diode stack?

Your going to do a couple of darlington transistors or industrial fets in parallel.

So, your basic goals are:

I. Protect the diode during the undefined states.
A. startup (diode is shorted out by protection to prevent overvoltage)
b. Shutdown (sense DC fail using a power fail sense chip, common part)
C. prevent Overcurrent failure


II. maintain a constant current

III. Modulate fast, and the easiest way to do this is to short the energy across the diode, while the psu maintains the constant current.

Any other goals???

Steve

The rated forward current of the diode is 1.8V or <2V.

The goals you stated pretty much cover all of what I am looking for in the system. I am curious however as to how current is varied in the system? Where should I start with the design? I have a high current 3.3V power supply that is designed for the laser.

Ben

[QUOTE=Laser Ben;97449]The rated forward current of the diode is 1.8V or <2V.

The goals you stated pretty much cover all of what I am looking for in the system. I am curious however as to how current is varied in the system? Where should I start with the design? I have a high current 3.3V power supply that is designed for the laser.




DESCRIBE THE 3.3v SUPPLY... pICS? Was it the current regulator as well?

2.2 worse case Vdiode + .79 worse case Vce for a NPN = 2.99, ok, that means you can go either FET or NPN, but perhaps not darlington.

Steve

The current is controlled via a MOSFET.

--DDL

[QUOTE=mixedgas;97453]

Quote:

---

Originally Posted by Laser Ben

DESCRIBE THE 3.3v SUPPLY... pICS? Was it the current regulator as well?

2.2 worse case Vdiode + .79 worse case Vce for a NPN = 2.99, ok, that means you can go either FET or NPN, but perhaps not darlington.

Steve

---

It's a 3.3V 50A Condor power supply if I remember correctly. The system used it as a source for the on board current control.

We're gonna start with the classic constant current source, however the NPN gives way to a MOSFET, actually a bunch of them in parallel.

See;

http://www.repairfaq.org/sam/glpdrv3.gif

this is a very simple thing, and needs some scaling and addons to do what you want it to do.

Steve

So I want to use an array of Q1 that are all in parallel with each other correct, AKA, wire several of them with the same pins connected together.

What is U1? Do I need to scale up the wattage of the resistors as well? It doesn't look like anything is dropping a major amount of current.

I am going to draw up a schematic and you can take a look at what I came up with.

Alright, so Q1 is a high power NPN transistor. You mentioned I would want to use a darlington transistor. Woud something like http://www.st.com/stonline/products/...261/tip102.pdf work?

certainly not for 30 amperes ;)

mosfets in parallel are a good alternative, something high current like IRFD450 perhaps? I don't have the necessary knowledge to choose the correct mosfet

it could also be doable with a darlington like these:

http://tinyurl.com/q4byge

or

http://tinyurl.com/p78hhu

you can contact djAlex1664 on laserfreak french section who managed to make such a driver with these, but he's really busy these days so it might be good to buy one (very cheap) and test some designs (they won't burn if you stress them, they're really strong parts)

Quote:

---

Originally Posted by Laser Ben

So I want to use an array of Q1 that are all in parallel with each other correct, AKA, wire several of them with the same pins connected together.

What is U1? Do I need to scale up the wattage of the resistors as well? It doesn't look like anything is dropping a major amount of current.

I am going to draw up a schematic and you can take a look at what I came up with.

---

You need to do a lot of scaling and you need emitter ballast resistors in each parallel transistor.


Steve

Any powerful mosfet , right now I'd go with 5 or 10 in parallel, the trick is the degeneration resistors in the source or emitter leads, to force current sharing, you need a little negative feedback between each device and ground. The neat thing about that is it gives you a means to sense current.
If you do not do the degen resistors, one device will hog current and start the self destruction process, which will short out the drive to all the other devices and it shoots up to full current in a microsecond.

Lets do 10 devices. Darlington may be out because you only have 3.3 V of supply and each silicon junction adds between .6 and .8V of drop, so two junctions = 1.2V of drop, this limits you to NPNs and Mosfets.

Electronic goldmine has A11060 The FQP55N06 60V 55A N-Channel MOSFET (Fairchild) on sale, for 79 cents each. At 2.2V diode and 30 amps, you'd need P = IxE = 60 watts. Your regulating device has to dissapate 30 or more watts as you have a 3.3V supply. The spec for the FQP55N06 is
for when it is driven full on in digital mode, not analog, and with a nearly infinate heatsink. which you cannot, by definition, use.

So lets spread that out a bit. 10 devices, means 3.3 watts each device, which is a little easier to get out of a TO220 package, We will use a .1 ohm degeneration resistor in the source of each device, so we know we have 3.3 watts going down each resistor, so P = I^2 x r = 3 * 3 amps * .1 ohm = .9 watts per resistor, so you need a 2 watt .1 ohm resistor to be safe.

So right now we have 10 79 cent fets, and ten .1 ohm 2 watt resistors.

If we take a bunch of 1 K ohm resistors and tie them to the high side of the degen resistors, and then tie the other sides all together, we get a voltage proportionate to the average current of all the resistors.

Now we need a test load to run before we hook a diode to this beast, and we have not yet addressed modulation or the heatsink size.

All the mosfets need their junctions to temperature track to prevent runaway, this means they are mounted right next to each other if not touching each other in a line on the heatsink. This is also a negitive feedback technique, if one junction starts to get too hot it backs off on its conductivity, and by keeping the junctions tracking, if one backs off, the cooler one takes up the slack, and by keeping them close together they average their temperatures and thus tend to share the current.

Now this feeback loop is not the one you need for mosfets, but if you look at the lexel 88 power supply in Sam's Laser Faq, all three drawings, you can see where this is going.

DO NOTE< that the mosfets have one lead connected to the case, and thus you may need a nonconductive sil pad between the mosfets and the heatsink, to prevent fireworks.

More to come.

Steve

really interesting! that is what I like in this forum! constructive technical discussions! :D

I don't completely understand what you mean when you say to incorporate the resistors in with the transistors in parallel. Do you have an example of what you mean?

Thanks!

Quote:

---

Originally Posted by Laser Ben

I don't completely understand what you mean when you say to incorporate the resistors in with the transistors in parallel. Do you have an example of what you mean?

Thanks!

---

See attached BMP

Note Ben Crazy is a late night Cleveland TV thing:

http://www.youtube.com/watch?v=XEUSJ...eature=related

Steve

Hi Guys,

Just a couple of notes on this one....
I am just finishing a design for a bar PSU for a little project I am planning.
I have gone down the line of a multiphase buck converter - 12V in, 0 to 60A out & will drive 1, 2 or 3 bars in series at up to 60A.
Whilst massively overkill for what Laser Ben requires I did most of the design work for a linear solution before deciding to go switched mode.

If you only want 25A or so there is no need for an array of MOSFETs - a single device will do fine.
Using a 3.3V supply for a diode with a Vf of 2.2V at 25A you have 27.5W dissipated in the driver. Go for a 10mR sense resistor and creates a split of 6.25W in the sense resistor and 21.25W in the MOSFET.

20W can easily be removed from a TO-220 – you are only forced to arrays or larger devices once you are getting towards the 40W+ mark. Typical thermal resistance Jn to case on a TO-220 is just under 2C/watt. Add another 0.4 C/watt for the thermal transfer sheet and your die will be running max 50C hotter than your heatsink.
Chose a heatsink / heatsink and fan combo with a thermal resistance of 1C per watt and your heatsink will be 30C above ambient. If ambient is 20C the die will be running at a casual 100C when you run CW at the full 25A.
Depending on manufacturer the Tj max operational is usually either 150 or 175 degrees C. With an ambient of 50C the junction will still be <125C so a reasonable headroom to guarantee long term reliability.

Choice of MOSFET and trying to run the whole driver circuit from 3.3V…
I thought you might struggle getting a MOSFET to sink 25A whilst only having a Vgs of 3V (250mV lost on the sense resistor) however I think I have found one –
IRL3402 - http://www.irf.com/product-info/data...ta/irl3402.pdf
See the Id Vgs curves on page 3.
Tj max is 150C but the thermal resistance Jn to case is only 1.1 C/watt making Tj <110C with 25A out and a 1degC/W heatsink.
If you want a little more margin chose a heatsink of 0.5C / watt.


Sense resistor
I would go for a 10mR 1% wire-wound such as the Vishay Dale RH010R0100FE02.
This is an aluminium clad resistor and can share the heatsink with the MOSFET.
Due to the low resistance of the sense resistor it will be essential that you make a four terminal connection to it.
I assume the 3.3V supply is floating (0V not connected internally to mains ground)
In which case the PSU 0V should go direct to one eyelet on the sense resistor.
Just fractionally in-front of the eyelet (closer to the wound section) take a 0V line to the driver. This line is the driver 0V supply and 0V reference that any input signal to the driver will be with respect to. The +3.3V needs to go directly from the PSU (via normal hookup wire) to the driver +ve supply. The +3.3V feed to the LD bar needs to go directly from the PSU to the LD bar via heavy cooker wire. The Cathode side of the LD bar needs to go via cooker wire to the
MOSFET Drain then the MOSFET source goes via heavy Cooker wire to the other eyelet on the sense resistor. Fractionally in-front of this eyelet take a hookup wire to the driver – this is the sensed current voltage of 10mV/A.

In that way voltage drops down the cabling to the sense resistor will be excluded from the "sensed" voltage. Also the driver’s supply voltage is not lessened by voltage drops down the +3.3V feed cable (important since if the supply is much lower than 3.3V you will not be able to sufficiently enhance the MOSFET. Unfortunately the driver’s supply voltage is still lessened by voltage drops down the PSU 0V connection to the sense resistor so try to keep this cabling length to a minimum – or if the 3.3V PSU supports remote voltage sensing tie the PSU 0V o/p sense line to this point.

Since the feedback voltage on the sense resistor will be 10mV per A you want to use an op-amp with a low input offset voltage. Also since you need the better part of 3V to drive the MOSFET to 25A you need an op-amp with R-R o/p and an input common mode range which includes its –ve rail.

I have sketched out a quick schematic here:
http://www.die4laser.com/misc/25A_La..._Schematic.pdf
If wanted I could easily hack this together using a standard die4drive PCB.

What did you do to protect the diode during PSU startup and shutdown?

For the record, I'm a very careful , conservative designer and was going with a design I know to work in ion laser service at much higher currents, with parts I could get cheaply.
Also when using a one of a kind diode I like a colder mosfet, and six sigma reliabilty. I know one device can handle it, but a 125'c junction temp is a little hot for my liking.
Steve

Quote:

---

Originally Posted by mixedgas

What did you do to protect the diode during PSU startup and shutdown?

---

Hi Steve, Good point. Since the design is a current sink it is relatively immune from PSU voltage fluctuations however there are 4 things to watch for as far as the PSU startup and shutdown and running are concerned:

1) op-amp power supply rejection ratio (how much of any supply ripple appears on the o/p) this is >70db for all frequencies of note so should not be an issue.
2) PSU transient response. You are presenting a dynamic load to the PSU. If the diode is held at threshold you might have a step from 5A to 25A. It the diode is held at system laser threshold it may reduce to a step of 10A to 25A. There will be an associated o/p undershoot and overshoot as the load goes on / off. Again high PSRR should mean this is not an issue.
3) Op-amp start-up behavior. This is an issue. The opamp is rated for operation at supplies over 2.5V. Obviously at voltages lower than this its o/p behavior is undefined. Since the MOSFET Vgs threshold is 0.7V the op-amp must have zero o/p or a controlled o/p for supply voltages of 0.7 to 2.5V. You could just plot Iout of the whole assembly vs Vsupply for various DRV voltages. If this seemed OK you would have to repeat Iout Vs Vsupply with a dynamic signal on DRV to check transient at low voltage was acceptable. A better engineered solution would be a "mute" circuit to clamp op- amp O/P (and +ve input) during Vsupply <2.7V. This could still be done from a single +3.3V supply.
4) Supply fast transients. Here I don't mean the 1ms or more startup ramp of big SMPS - I mean microsecond order or less fast edges from a relay switched o/p or a PSU with a solid state switched output after its output reservoir caps.
This IS an issue for ALL drivers using a MOSFET as a current sink. Fast transients on the drain couple directly through the Cds capacitance and indirectly through Cdg, raising the gate voltage. Whilst a big bar would likely be immune to this I have seen it kill a 405nm diode (twice - I repeated the test because I didn't believe it could have killed the diode. Guess what it did!)
Putting a capacitor across the LD provides a path for this current other than through the diode. It just needs to be a couple of orders in magnitude bigger than Cds - preferably paralleled with a cap 1/10th of this value to increase the self resonant frequency of this capacitive shunt.

Quote:

---

Originally Posted by mixedgas

For the record, I'm a very careful , conservative designer and was going with a design I know to work in ion laser service at much higher currents, with parts I could get cheaply.
Also when using a one of a kind diode I like a colder mosfet, and six sigma reliabilty. I know one device can handle it, but a 125'c junction temp is a little hot for my liking.
Steve

---

Likewise - my day job entails designing high-rel equipment for adverse environments. I have designed analogue kit for operation at 90C ambient (100C PCB temp) where we had to get a concession for any junction temperatures over 105C.

I suggested an IRL3402 running with absolute max Tj of 110C based on an ambient air of 50C and also recommended a 0.5C/W heatsink if there was any concern over Jn temp. This would give Tj max of 95C with 50C ambient.
If running the Jn on a power MOSFET 45C above ambient is not conservative then I'm a piece of Swiss cheese.

You were advocating an array of FQP55N06s. They have a Vgs_th of 2 to 4V so will not work from a single 3.3V supply. By the time you have mounted the transistor array, added the sharing resistors and bought a 12V PSU I can’t see this solution being any cheaper.

Best regards,
Robin

wow, I guess I found how I could use my two 5V 90A psus... they have +/- sense inputs and everything needed to power adjacent circuits :) (+/-12V,3.3V,24V,etc..)

thank you uk-laser, thanks to you I now have some plans in my head for these :D

If you try to run one bar at 25A from a 5V supply your driver will have to dissipate 75W or so.
You will have to go for a MOSFET array as Steve has described.
Still the +12V will be OK for the op-amp part of the circuit.
Also if you chose a MOSFET with a Vgs_th_minimum of over 2.5V (ie NOT a logic level drive type) you are not concerned with op-amp start-up issues.
And on a final note - 3.3V fans are somewhat rare. A big 12V fan will be cheap as chips on the surplus market.

You were advocating an array of FQP55N06s. They have a Vgs_th of 2 to 4V so will not work from a single 3.3V supply. By the time you have mounted the transistor array, added the sharing resistors and bought a 12V PSU I can’t see this solution being any cheaper.

The ones on my bench are turning on just fine....

Oh well...

Steve

Quote:

---

Originally Posted by mixedgas

The ones on my bench are turning on just fine....

Oh well...

Steve

---

Can't comment on the ones on your bench - Just going by the datasheet.
http://www.datasheetcatalog.org/data...d/FQP55N06.pdf
"VGS(th) Gate Threshold Voltage VDS = VGS, ID = 250 μA 2.0 -- 4.0"
Supplied devices will have a threshold of somewhere between 2V and 4V
The datasheet doesn't even have curves for Vgs less than 4.5V
If Vgs_th is higher than your supply there is no chance it will work.
I guess you have some devices which came in much closer to the 2V level.
Well if you want to design outside of the datasheet don't let me stop you.
If it is a one off select on test can work fine.