Voltage regulator/ reference

TL431?

Dirt cheap, available thru-hole, relatively robust, available worldwide. Either supply via a resistor or 21mA current source. If the load current is likely to fluctuate a lot during the experiment then I guess self-heating induced TC drift could be a problem. A PNP shunt booster might solve that.

piglet

Most opamps are current and thermal limited.

The classic fix for c-load stability adds a resistor in series with the opamp output, and a cap to ground, both of which further protect the opamp.

They could always blow it up by connecting it to some weird voltage. That would need more serious protections.

Data sheets often specify DO NOT EXPOSE TO STUDENTS.

Hi piglet. OK that was the other reference I found on Digikey. I've hardly ever used a shunt reference (LM4040 a few times) (oh and the LM399.. but that's a little over the top.) It's certainly cheaper than the LT1460! The load should not be changing by much. I'll add some to DK order, Thanks

George H.

Right. The c-loaded opamp worked fine. But there are all sorts of potential (NPI) voltages available... what happens if the output of an opamp (+/-15V rails say) gets connected to +30 or -30 V?

I was playing around with depletion fets and the DN2540 with ~100 ohm of Gate- Source resistance makes an OK ~20 mA current limit*. I could put that on the output of the opamp and feedback around it. (But I'm not sure how well it would work.. and lotsa testing takes time.) I thought about protection diodes on the output to the power rails... but then maybe the students blow up the power supply.. which would be worse that frying an opamp.

Grin, Well some of the professors are worse than the students.

George H.

That depletion fet thing should work fine. I'd trust a Spice sim and just go for it.

Do clamp the opamp output to the rails (or ground) so that 20 mA has somewhere to go, especially with power off.

We use DN2530s, SOT-89s, good for 3 watts on a copper pour.

There are also some self-protecting bidirectional solid-state relays that can be fun as current limiters, but at higher currents.

diodes, resistors. Zener on psu too.

NT

Oh, put the diode clamps between opamp output and current limit... Then the most the power supply has to suck up is ~20 mA... That's smart. I could have wimpier diodes then.

The DN2540 comes in TO-220 packs! I could probably handle a couple of hundred volts, Oh boy, now I want to try bullet proofing the opamp. I'll think about it.

The weekend is going to be filled with trying to fix the brakes on my old tractor. It's my son's idea. We've got a month till "Drive your tractor to school day" and I told him he could drive it if we can get the brakes working... big disk brakes, I've never had them apart.

How we have fun in the country. :^)

George h.

Re power off state: I've tested that some.. and near as I could figure over-voltage (at the input) started to power on the circuit, but with a current limit it's not a problem.

What's the Vgs voltage spread on the DN2530 like? I've got ten DN2540's, I've tested 4 and about the same.. ~+/- 100mV.. but factor of 2 higher than the spec sheet. Are parts in batches/ on tape fairly similar? It would be fine if I could measure some, and pick a resistor for the reel.

George H.

If you really need a stable 5 V, use an LM4140-2.5 and a 2:1 amplifying circuit with an AD706 op-amp. For 20 mA load, you may need a current boosting transistor after the op-amp. For best results, the LM4140 should be run off about 4 V, dropping the input with a resistor is fine. It also needs a cap at input and output.

This is probably way overkill, this is what I did when we needed microvolt stability, with absolute minimum 1/f noise.

Jon

The old standby for a 'reference-class regulator' is the LP2951.

Cheers

Phil Hobbs

Beer check! So here's my 'bullet proof opamp.

The current limit drops a little less than 2 Volts... Hence the ~8V zener.

Do you see any blunders? I tested this at ground and +/-40 V volts on the output. Everyone seemed happy.

George H.

Thanks Jon, I'll try and remember that if I need it.

George h.

Got it, Thanks Phil. GH

no diodes for inductorproofing. Little in the way of input protection. Any power supply protection? i_limit is an unknown. It might do what you need but it's not bulletproof.

NT

Hmm inductor proofing... I hadn't thought about that. (does that mean I owe you a beer?) At 5 V and ~20 mA max out I think it will be OK.

Yeah the input is set by me.. so needs no protection. There is the issue of picking G-S resistors for the dn2540's. The ten I have here all have about the same curve... but?

George H.

if kids are handling the output they'll do their best to destroy it. They'll connect everything including as many batteries as they can find. 2 diodes is all it costs for inductorproofing, plus series R & Z for batteryproofing.

depends how many you're making, and thus how you're assembling them. Cuttable wire links are popular in consumer equipment. If that's too imprecise or large, parallel pads could enable slicing the original R's track and adding a new one after construction.

Enjoy the beer.

NT

Given that creative fellow who blew out your 78xx current source by connecting a Helmholtz coil to it, I suspect you need better notes. ;)

Cheers

Phil Hobbs (Who has similar war stories to tell--we should meet at the Dayton Hamfest)

Better notes...? OK my notes stink mostly. I keep notes, but they are useless to anyone but me. Right. inductors, I'm a slow learner and sometimes it takes me more than once to learn something. :^)

This supply will have something like a (RC) 100 ohm x 0.1uF tc. Inductor-wise I think the 100 ohm in the current limit will help save me. (Kids/students blew up this 'bullet proof' transistor which JT (RIP) said was the same silicon as the LM317, but different metalization. I can't recall the part number... We still use it a darlington with current and thermal overload protection. With a 1N400x across the output, that's a bullet-proof output too. (so far :^)

Dayton Hamfest? Ohio? Soon? Maybe Sphero, or others could have a meet-up?

George H.

Put a diode at the input of the supply, so it?s inpossible to hook it up wrong

Active diode if you cannot live with a diode drop

Cheers

Klaus