Constant Current Dummy Load

[henson]

I got bit by the electronics bug about a year ago, and have since watch every video on the blog. For my first post to the forum I would like to share my build of Dave's constant current design. This marks my second project to date, the first being a simple LM317 power supply. I followed the original circuit faithfully, only changing out the LM324 Op Amp with a MCP6004 Rail to Rail Op Amp and I added a couple of trim pots for calibration. For this project I wanted to try my hand at designing a circuit board and getting it professionally manufactured. To have the board manufactured I designed it using CadSoft's Eagle PCB application and sent the board file to Osh Park for manufacturing. Osh Park made it really easy to just upload the board file online with no need for producing Gerber files or anything else. Turn around time on the board was pretty good to, I think it took about two weeks.

A few additional consideration in the development of this device:

•    The MOSFET is mounted directly on an external heat sink and not on the PCB.
•    The onboard 7805 regulator feeds both the Op Amp and adjustment pot.
•    The device is powered via an external wall adapter. I am using a 7.5 volt 800 mA adapter from my parts bin (I have no idea what it went to originally).
•    The 10 resistors get a lot warmer then I thought they would. I have a heat sink I can put on top of them, but need to get some thermal compound that will hold it in place.
•    The only way to know what the current draw is putting a load on it. In future designs I would like to have this as an independent control.

Download Eagle Files Here

[MrAureliusR]

Sorry to bump an old thread but I didn't want to start a new one.

I'm trying to build my own based exactly on Dave's original schematic. I'm using an LM1458 Dual Op-Amp instead of the LM324 (though I do have an LM324 around if that's what the problem is). I have everything else the same except I'm using a 10 ohm resistor (10 100 ohm in parallel) just to test with until I get over to the electronics shop this week.

The problem is, the op-amp is oscillating like crazy! I've triple checked my circuit to make sure I have everything right and I do. I've tried playing around with the circuit a bit but it doesn't help. This almost makes perfect sense because the amount of time it takes for the MOSFET to change its ON or OFF state is too long, and the op-amp is constantly over-compensating. Does that make sense? (Now that I'm thinking of it, it doesn't really)

Kind of stumped here. Is it something inherently wrong with the LM1458? I don't really see why it should matter too much, one half is just a voltage follower and the other is just essentially a simple negative feedback loop?

Any advice is appreciated! I'm a digital guy and this analog stuff needs to sink in a bit. I am eager to learn more about op-amps (I've got a whole pile of them -- no kidding -- just sitting on my bench) and so I'm constantly trying to find things I can breadboard and then poke my scope probe into and see what's going on. This one is making me scratch my head a bit though.

[oldway]

LM324 is a 4x rail to rail OP amp and LM1458 not.
Is this not the problem?

[Ziltoid]

How about some pictures and a shematic?

[psycho0815]

try putting a cap between the gate of the mosfet and gnd. that worked for me. 10uF or so schould do.
you can probably get away with less.
Also have a look at http://www.ti.com/lit/ml/sloa079/sloa079.pdf.

Generally dave's design should work with any opamp that can operate on a single supply, which should be pretty much everyone.

BTW: LM324 is very much not railt to rail it does swing to gnd though, iirc, not sure about the LM1458, but as long as it can go below VthG of your MOSFET that shouldn't matter much.

[ConKbot]

try putting a cap between the gate of the mosfet and gnd. that worked for me. 10uF or so schould do.
you can probably get away with less.
Also have a look at http://www.ti.com/lit/ml/sloa079/sloa079.pdf.

Generally dave's design should work with any opamp that can operate on a single supply, which should be pretty much everyone.

BTW: LM324 is very much not railt to rail it does swing to gnd though, iirc, not sure about the LM1458, but as long as it can go below VthG of your MOSFET that shouldn't matter much.

Putting a bunch of capacitance on the output of the op-amp, while it would lower the freq of the RC pole formed by the output resistance and gate capacitance, could cause its own kind of instability. Many op-amps have limitations on how much capacitance they can drive. The LM1458 datasheet doesnt specify, so I'd tread carefully.  Series resistance (a mosfet gate resistor) would lower the frequency of the pole as well, and the resistance would help isolate the capacitive load from the op-amp.

Also, is this on a pcb or breadboard or protoboard?  If you have long leads on a breadboard, the inductance of the leads can ring with the gate capacitance of the mosfet and play havok with switching losses. A gate resistor can damp these. Some applications can use ferrite beads, as they are lossy at high frequencies and can damp HF oscillations also.

If you want to just try and see what works, I'd start with a 10R gate resistor and work up until ~300 ohms or so to see if it stabilized.  If it does work, then poke around in the TI app note asne see if you can make sense of why adding the gate resistor helped, that way its a learning experience and not just a parts shotgun fix.

[dannyf]

Putting a bunch of capacitance on the output of the op-amp

Should be avoided at all cost, unless you know exactly what you are doing.

[dannyf]

The problem is, the op-amp is oscillating like crazy!

Without the schematic of your particular build, it is hard to say why.

However, assuming a general topology where the opamp is driving the base / gate of the current regulator, the issue here is likely the gate capacitance (typically 3000pf or more if a logic level device is used). Put a small gate stopper there usually relieves the symptom but doesn't fundamentally solve the issue.

This is a case where a slow device is preferred over a fast device.

LM324 is very much not railt to rail

It can still be made to work, depending on your design.

[megajocke]

The input common mode range of the 1458 doesn't gp down to its negative supply. If you want do do the syraightforward implementation you'll need a bipolar supply for the opamp.

[Harvs]

Do a Google search on this site for the problem and you'll come up with I'd guess at least a dozen threads where people of asked the same question, and all got the same answer to solve the problem (and I can't be bothered re-drawing the schematic.)

The root cause of the problem is almost never the loading of the gate capacitance on the op-amp.  The gate capacitance doesn't make for a nice response, but it doesn't usually take the opamp into instability on it's own.

Rather the cause is the Drain to Gate capacitance of the mosfet interacting with the inductance of the leads connecting between the mosfet drain and the output capacitance of the DUT (normally a few uH.).  This forms a feedback mechanism that quickly gets to zero phase margin as the current increases, and hence the mosfet gain dIds/dVgs increases.  If you want to make a load with reasonably high bandwidth, the answer is to use a snubber on the output to dampen this inductance, have a look for Jay_Diddy_B's dynamic load thread, or go download the service manual for the HP 6030 electronic load and have a look.

The fact the LM324 is not rail to rail is of no significance, as long as it can go well below the gate threshold of the chosen mosfet.  However, what is significant is the chosen opamp's input range includes the negative rail (unless of course you're using a bi-polar supply.)

[oldway]

Rail-to-rail op amps

    When doing single supply design it is common to use what are called rail to rail op amps.  These amplifiers can output voltage very near the power supply voltages (or rails).  You have already seen an example of one such amplifier, the LM324, whose output can go almost to the negative rail (ground), but still can't get close to the positive rail (Vcc).  If you do you design carefully, you could use a 324 in this lab.  It has the advantage of being quite cheap. 

http://www.swarthmore.edu/NatSci/echeeve1/Ref/SingleSupply/SingleSupply.html

With single supply design, if inverting input can't go almost to the ground, gain became very high for low current of the electronic load (no feedback anymore) and it will "oscillate like crazy"...
Use a LM324 and that's all right.

[nctnico]

The easiest way of making a constant current dummy load is to use the collector of a transistor and very slowly adjust the drive current to compensate for thermal runaway. The collector of a transistor can be seen as a current source! Just make sure to stay within the safe operation area.

edit: not resistor but transistor. Need to get some serious sleep...