My try at Daves DC load? opinions

[c4757p]

Both are wrong for this application. Try this.

Your first one is no good at all - no DC path for the inverting input! The second one is to compensate a capacitive load on the output by adding a resistor - not what we're trying to do (though if you're driving the gate on a huge MOSFET with a sensitive op amp, you might want to do that as well!)

[ahnuts72]

Thanks for the help c4757p that's how I had originally had it laid out
minus the resistor but it worked both ways on the breadboard so i
left it out.






Sorry my fault should have seen that mistake.
The first one should have been like this.

[c4757p]

Thanks for the help c4757p that's how I had originally had it laid out
minus the resistor but it worked both ways on the breadboard so i
left it out.

It will sometimes work without the resistor, as long as there's some impedance. On a breadboard, you've got impedances up to your ears.

The first one should have been like this.

A shorted capacitor doesn't do anything, so that's no good either.

[ahnuts72]

Thanks very much for the help I really appreciate it.



Would this be correct according to your drawing?

[c4757p]

Capacitor is in the right place, resistor is not. It should replace the trace going from pin 6 down to the sense resistors. Also, remove C1. I don't know how it got there but it's just going to make this even more unstable.

[ahnuts72]

Thanks I almost started to put the resistor where you said but then i started second guessing my self
and put it where it is but as i look it is out of the right feed back loop where its at.

I'll have to check my schematic for c1 it may have supposed to go between pin 6 and 7
like you have it this is only my second board not using the auto-router so I  may have routed
wrong.

[c4757p]

this is only my second board not using the auto-router so I  may have routed
wrong.

What are you using to route it? Most PCB software won't let you route a trace that doesn't match your schematic.

Edit: It's there in your schematic too. Just take it out of both - it's no good.

[ahnuts72]

I'm using DipTrace.

I just checked and it will let you do it if your not careful.
But I may have had it wrong in the schematic I'm sure it
was operator error though.

[Harvs]

Attached is the basic compensation scheme that we've all used here.

R2 serves to isolate the op-amp output from the mosfet gate capacitance so that the output drive doesn't get forced into saturation.

Then the network of C1 & R3 roll off the op-amp response well below the point where the control will hit a 180 degree phase shift.  Or in other words the system becomes unstable as the feedback becomes positive instead of negative as we need.

I don't know if you're an LTspice user, but I've attached the spice file.  It can be quite informative.  If you effectively remove the compensation network, set C1 to 0.001pf and R2 to 0.1ohm, then run the simulation you're see the current levels through R1 ringing at the switch point.  This is a sure sign of a network with marginal stability.  Add the values back in and it'll become a nice clean transition with zero overshoot.

The values can vary a little depending on what op-amp and mosfet are used.  But what I've used in the simulation are a pretty safe bet for the LM324 and a generic power FET.

As a good primer, this series of 5 videos from Analog Devices is quite good.

[ahnuts72]

Thank Harvs I haven't made it to op-amps in Practical Electronics For Inventors
yet don't know if compensation is covered  ill definitely checkout the videos.

Ive messed around with ltspice a little ill try to run the sim files and play around with it.

[ahnuts72]

Ok tried to take everyone's suggestions and include them.

Hows version 2?

[Harvs]

Do you have a schematic you could post as well please?  Makes life a lot easier for everyone.

[ahnuts72]

Here you go.

R5,R6,R7 and c2 aren't on there they are for my panel meter
haven't added them.

[ahnuts72]

Almost finished.
Mocked up how its gonna be when finished.
I still have to hook up the power switch,cut the
side vents,fill some holes and sand and paint
the enclosure and make a label
for the front panel.

[ahnuts72]

Ok here's a shot of the waveform on the output of the op-amp.

I figured i should be getting a pretty flat line.
This is at .500 amps at 19 volts.

Is this the output oscillating?

[c4757p]

Yes, it is. What are the plot scales on that? I can't tell from the photo.

[Harvs]

What are the values of the parts surrounding the op-amp?  (i.e. the resistors that don't have values on the schematic.)

[ahnuts72]

R4=330
R20=10k
R16 and R18 are both 10k

[ahnuts72]

Sorry c4757p missed your post somehow.

t/d 2mS
v/d 50mV

[Harvs]

Attached is a quick plot of the closed loop freq response of the circuit.  You don't have any attenuation down at the frequ you're observing. 

But it's surprising that this circuit would be oscillating at that sort of freq... 

Typically when people have these oscillate it's up in the MHz range, and pretty full on, hence rolling of the response in the 40-50kHz is pretty safe.

I'd have a poke around you're circuit with the scope and see if it's being coupled in from somewhere else...  What's the supply you're loading look like on the scope?

[Rerouter]

you seem to have installed your DC input where your bulk capacitance should live, and now that i look at it, your 0.1uF ceramic should have probably been put a whole lot closer to the op amp, so its very possible a good bit of noise off your supply is being coupled in, the oscillation spikes are about 240Hz, which would line up roughly with a double of the 120Hz ripple being spat out by a cheap wall wart,

[Kevin.D]

Look at the  supply line on your scope or just unplug your fan to see if it improves .That frequency your waveform is at   '200-300Hz' is about the same frequency as what the coils would switch in a brushless Fanmotor . If you haven't done so already you will need to filter the fan noise from getting into your supply rail .

[c4757p]

Sorry c4757p missed your post somehow.

t/d 2mS
v/d 50mV

Look at the  supply line on your scope or just unplug your fan to see if it improves .That frequency your waveform is at   '200-300Hz' is about the same frequency as what the coils would switch in a brushless Fanmotor . If you haven't done so already you will need to filter the fan noise from getting into your supply rail .

Yes, I second this. That looks more like fan noise than oscillation. I didn't realize your horizontal sweep was so slow on that picture.

[c4757p]

I wish I caught that before when I was looking at your design, but I wasn't really paying attention to the fan section. Fans are really, really noisy and need a ton of filtering if you're putting them that close to analog stuff. You can still add filtering inline, since it's not going to fit on your PCB now. I'd say multiple capacitors of different sizes (ranging from about 10nF to 47uF-100uF or so), plus a chunky ferrite bead or choke.

[ahnuts72]

Ill try to unplug the fan when i get off of work and see what happens to the waveform.
I was looking at the schematic and noticed I don't have a bypassing  cap on the supply to the opamp either.

Im using a 5volt 1amp linear wall wart to power it.