High side current monitors for modulable laser diode drivers?

[andythemechanic]

Hi everybody,

after some time just reading I have the feeling this is the right place to ask my question:
In most closed loop modulable laser diode drivers I have seen the current sensing shunt is between diode cathode and ground. IMO this just sucks for diode which are cathode grounded, because one has to isolate the diode assembly from ground. Even for diodes which are not grounded like blurays it's easy to accidentially kill the diode by shorting the sense resitor.
My question is if anybody has experience in using high side current monitors like the ZXCT1009 from zetex for modulation speed up to 100kHz ?
For me this seems like a way to go but maybe I am overlooking something?

Andreas

[FourDee]

I posted something like this a while ago in the stanham driver topic

[shrad]

hi,

I rarely take position on this forum, but here I have to drop my tip

try not to use mosfets in modulated current sources, as they operate in their non linear range

this could result in overheating (I have seen mosfets rated for at least 10 amps heating as hell with a 500mA current) and also, non linear range means non linear modulation, which is not wanted in the case of a modulated laser diode

try to use darlington transistors like TIP142 and the like which will run nicely up to several amps if you add a negative gain in the current loop or reduce the sense resistor value (0.10 ohm for currents up to 5 amps)

also, the opamp used for current sense must be a low noise one, with nicely set bypass capacitors on the power line, resistors should be metal film 1% and all the same value to reduce common mode noise... the PCB tracks should also be kept to the minimal length from the sense resistor to sense opamp and from the sense opamp to the regulation opamp as they easily pick up nasty noise from the power lines, especially with power switching psu's

hope this will be of any help for those wo are not familiar with analog design

[mixedgas]

Zetex has a nice app note on high side sensing using current mirrors.

http://www.diodes.com/_files/product...zetex/an32.pdf


Google high side current mirror



ZXCT1009

ZXTC1021]

MAX4172


Steve

[andythemechanic]

Ok, here is an update: I tried to set up an initial design and simulated it in Orcad Pspice. Result is that at least in the simulation the ZXCT1009 didn't work well, had slow response and tends to oscillate at the wanted modulation rate. So I changed the design to a "classical" high side sensing scheme found in an analog devices application note. This design worked well in simulation. I have attached the circuit and a simulation of step response (1A) at 500kHz. D1 and R8 are there to simulate the laserdiode. I didn't find any suitable rail-to-rail opamps but I thought in a scanner you have the +-12V or +-15V anyway so this is not a big drawback. Of cause if somebody knows a good one I like to know.
What I don't like at the circuit is, that the mosfet is somehow floating but in simulation it worked.
I am no electrical engineer so I really like to hear comments, suggestions etc. from experienced people as it might be that I am overlooking something.
Also I have no experience in making layouts so if somebody likes to jump in and help me to do layouts and proto boards your're welcome to contact me.

Andreas

[mixedgas]

OK, here is a classical late 90s design from DR PHillip C. D. Hobbs From his book

Building Electro-Optical Systems: Making It All Work.

ISBN 978-0-471-24681-7. 760 pages. June 2000.

Posted under the educational use clause.

Probably the best book out there on making optical systems that are high speed or low noise, but at 149$ a copy, I cant afford it. I've sent some of his photodiode speedup stuff to some PLers in the hope that they can get some light feedback working.

This has a current limiter to protect the diode at all costs, high side sense, and a startup relay. Its designed for those diodes you can not afford to lose in the lab, ie the 2000$ custom open can diode at 5.5 microns or something. We will have to work out Cf and Rf and change some values and work on some "Miller Killer" addons to speed it up. It should also have a buffer amp on Vref.

This page is from his catch all chapter on things he thought he should include, but didn't get a chapter on their own, so thats why if you try to read it, its incoherent.

Andy, try LMH6642MA for the opamp. 130 mhz may be too fast.

SEE ATTACHED. Andy, please SPICE this puppy. I'd say your were very close with your design, except for startup conditions. Phil's circuit takes care of that nagging startup state with the 3906 and the relay.

A Lavadrive will beat this one in the speed, waveform fidelity, and flexibility departments. And this circuit will probably need split supplies, leading to endless PL bandwidth waste as we have to explain to many newbies what a split power supply is. 2 x 24 = 48. So accidental connection across both galvo amps would be POOF! waiting to happen. 5 to -24 is a issue as well.

Change Rlim to 2.0 ohms for a 250 mA diode. Its gonna clamp when the voltage across Rlim gets to .6 to say .7 V, and that is temperature dependent.

If Bill ever gets around to it, the relay would be replaced with a Lasorb.

I'm struggling to get my own stuff done and the house fixed, so this is as much help as I can do right now. I do PL at night or when I'm waiting by the computer and phone for a client to call.

[mixedgas]

Quote:

What I don't like at the circuit is, that the mosfet is somehow floating but in simulation it worked.


End Quote:

The mosfet is in the feedback loop, and its gate returns via the opamp sourcing and sinking current, your fine on that if the traces are short and the circuit is well biased. It just means your opamp has to work a bit harder to slew that gate. You may need to add a ferrite bead or two on a real board of that. Spice is great, but it doesnt handle the nagging environment stuff.

Steve

[dave]

If Bill ever gets around to it, the relay would be replaced with a Lasorb.

I believe they are shipping.. could be wrong tho..

[shrad]

be careful, as in my attempts to design a high side monitored current source, I have seen great instability with the sensing resistor directly connected to positive power supply line (oscillating, then a straight 10A across the transistor, melting istelf in 30sec)

with the sense resistor between the diode and the transistor though, no problem and the driver is very stable, as positive rail or ground do not interfere

[andythemechanic]

Thanks Steve for the great circuit. I had some time today ( we have labor day in Germany so no work ) and simulated it. Works great but is quite slow, mainly because of the darlington I think. But it seems to be straight forward to replace this with a low gate charge MOSFET and adjust the resistors and capacitors. Here are my latest simulation results. The Mosfet is not the ideal one yet.
You also mentioned some other tricks to speed the original circuit up.
The possibility I see is to decrease the two 10k resistors at the darlington base or even use some driver circuit to get the charge out faster. Do you have any advice?
Another thing to do is to replace the opamp with a rail to rail one
I think anyway now its time to prototype this one and test it in reality. I will get the parts and start with this. I keep posting here if I have results.