First Bench Supply

[MikeNye]

Hi Everyone,

I've been working on a design for my first home-built bench supply. The specs I'm aiming for are:

• 0 - 30V DC output
• 0 - 3A current limiting / constant current
• Linear (low noise)
• Controllable via µC / DAC

After quite a bit of reading and experimentation in a simulator, I have come up with the following design:



If you'd like to see this design in the simulator, I'm using "circuit simulator" from falstad.com (http://www.falstad.com/circuit/). You can load this design into the simulator with the following URL: http://www.falstad.com/circuit/#%24+1+5.0E-6+4.818269829109882+44+5.0+50%0Ar+432+80+528+80+0+0.1%0Ag+784+80+784+96+0%0Ax+719+104+757+108+0+16+Load%0Ar+432+80+432+176+0+5000000.0%0Ar+528+80+528+176+0+5000000.0%0Aa+480+176+480+304+0+30.0+-5.0+1000000.0%0Aw+432+176+464+176+0%0Aw+528+176+496+176+0%0Ar+528+176+528+304+0+5000000.0%0Aw+528+304+480+304+0%0AO+528+304+576+304+1%0Ax+454+57+508+61+0+16+Rsense%0Ax+618+310+713+314+0+16+%3D+I+*+Rsense%0Ax+586+61+617+65+0+16+Iout%0Ar+432+176+432+288+0+5000000.0%0At+352+112+352+80+1+1+-17.384541032767196+0.7228104943582956+100.0%0Aw+336+80+272+80+0%0AR+272+80+224+80+0+0+40.0+30.0+0.0+0.0+0.5%0Aw+368+80+400+80+0%0Aw+400+80+432+80+0%0Aa+352+208+352+112+0+30.0+-5.0+1000000.0%0Ax+215+213+248+217+0+16+Vset%0Ax+209+406+248+410+0+16+Imax%0Aa+528+352+432+352+1+30.0+-5.0+1000000.0%0Aw+528+368+528+400+0%0Ar+528+400+432+400+0+9000.0%0Ar+528+400+624+400+0+1000.0%0Ag+624+400+624+416+0%0Aw+432+352+432+400+0%0Aa+336+400+336+320+1+30.0+-5.0+1000000.0%0Ax+177+85+202+89+0+16+Vin%0Aw+432+400+432+432+0%0Aw+432+448+656+448+0%0AO+656+448+688+448+1%0Ax+630+61+666+65+0+16+Vout%0Ax+672+435+813+439+0+16+Measured+Current%0Aw+352+400+432+400+0%0Aw+528+304+528+336+0%0Aw+432+432+432+448+0%0Aw+400+32+400+208+0%0Af+384+288+384+240+1+1.5%0Aw+336+304+368+304+0%0Aw+400+208+400+240+0%0Aw+336+240+352+240+0%0Aw+368+208+368+224+0%0Aw+368+224+368+240+0%0Aw+352+240+368+240+0%0Aw+368+304+384+304+0%0Aw+384+304+384+288+0%0Ad+336+304+336+240+1+0.805904783%0A172+336+208+272+208+0+6+12.0+24.0+0.0+0.0+0.5+Voltage%0A172+320+400+272+400+0+6+0.36+2.0+0.0+0.0+0.5+Current%0Af+240+304+176+304+0+1.5%0Af+240+352+176+352+1+1.5%0Aw+176+320+144+320+0%0Ar+144+320+80+320+0+220.0%0Ag+80+320+80+336+0%0A162+80+288+176+288+1+2.1024259+1.0+0.0+0.0%0A162+80+368+176+368+1+2.1024259+1.0+0.0+0.0%0Aw+80+368+48+368+0%0Aw+48+288+80+288+0%0Ax+54+394+168+398+0+16+Const.+Voltage%0Ax+57+274+172+278+0+16+Const.+Current%0Aw+48+288+48+320+0%0Aw+48+368+48+320+0%0AR+48+320+32+320+0+0+40.0+5.0+0.0+0.0+0.5%0Ar+384+304+384+352+0+10000.0%0Ag+384+352+384+368+0%0Aw+176+320+176+336+0%0Aw+240+304+240+320+0%0Aw+240+320+240+352+0%0Aa+288+304+288+240+1+30.0+-5.0+1000000.0%0Ag+272+304+272+320+0%0Aw+304+304+336+304+0%0Aw+432+288+432+304+0%0Ag+432+304+432+320+0%0Aw+240+240+288+240+0%0Aw+240+240+240+304+0%0A174+688+80+784+80+0+50.0+0.6584000000000001+Load%0Aw+736+64+736+48+0%0Aw+736+48+784+48+0%0Aw+784+48+784+80+0%0Ax+346+71+360+74+0+10+Q1%0Ax+329+145+342+148+0+10+U1%0Ax+461+270+474+273+0+10+U2%0Ax+447+370+460+373+0+10+U3%0Ax+317+332+330+335+0+10+U4%0Ax+269+255+282+258+0+10+U5%0Ax+203+290+217+293+0+10+Q2%0Ax+203+374+217+377+0+10+Q3%0Aw+576+80+624+80+1%0Aw+624+80+672+80+2%0Aw+672+80+688+80+0%0Aw+528+80+528+32+0%0Aw+528+32+400+32+0%0Aw+528+80+576+80+0%0Ax+394+282+408+285+0+10+Q4%0Ax+317+261+330+264+0+10+D1%0Aw+336+304+336+320+0%0A

...or if that doesn't work, you can import the SPICE that I have pasted at the end of this post.

I would like to explain how I envisage the supply working:

• Power is delivered into the supply by way of "Vin". I envisage that this would be the output of a switchmode supply, or a transformer/rectifier/smoothing_cap arrangement.
• Series pass transistor Q1 is responsible for voltage & current limiting.
• At the emitter of Q1, there is a 0.1R current shunt resistor.
• Op amp U1 is responsible for driving the base of the series pass transistor Q1.
• During constant voltage operation, the non-inverting input of U1 is connected to the output of a DAC, and the inverting input (sense line) is connected to the output where the load is connected via p-channel MOSFET Q4 (more on this later...).
• For current limiting / constant current operation, we have a difference amplifer (U2 & surrounding 5MR resistors) with unity gain, which outputs the measured voltage drop across the current shunt resistor. This is fed into U3, which is configured to apply a gain of 10, thus the output of U3 is the current measured, and is fed into the non-inverting input of U4.
• U4 determines the mode of operation of the supply, between CV & CC. The inverting input is connected to the output of a DAC. If the current measured across the shunt resistor (non-inverting input of U4) is greater than the set current (inverting input of U4), then the output of U4 goes maximum positive. The positive output from U4 causes Q4 to turn off (thus preventing the sense line from being connected), and the voltage is fed through diode D1 into the inverting input of U1, which in turn causes the supply's output voltage to drop to prevent the current from exceeding the set value.
• D1 is necessary to prevent current from the sense line flowing into the output of U4 in CV mode.
• U5 is simply used as a comparator to drive a couple of MOSFETS (Q2 & Q3) which drive LEDs which will tell the user which mode the supply is operating in.

My concerns with this design are as follows:

• BJT Q1 will need to handle a decent amount of current, and will have to have a relatively high gain to prevent overloading the output of U1. Accordingly, when I'm building up the first iteration, I'll be using a darlington pair or a device with a built-in darlington configuration.
• MOSFET Q4 will need to have a really, really low Rds(on) otherwise will skew the output voltage accuracy...? I guess I could adjust for this by adding a trimpot on the non-inverting input of U1?
• I'm concerned about oscillations between CC & CV modes when Vset & Iset == Vout & Iout respectively...

Any comments/thoughts on the design and the concerns above would be greatly appreciated.

My next step will be to breadboard this and see how different reality is from simulation... 

Thanks for taking the time to read my post, and I look forward to hearing from you all.

Kind regards,
Mike

SPICE:
$ 1 5.0E-6 4.818269829109882 44 5.0 50
r 432 80 528 80 0 0.1
g 784 80 784 96 0
x 719 104 757 108 0 16 Load
r 432 80 432 176 0 5000000.0
r 528 80 528 176 0 5000000.0
a 480 176 480 304 0 30.0 -5.0 1000000.0
w 432 176 464 176 0
w 528 176 496 176 0
r 528 176 528 304 0 5000000.0
w 528 304 480 304 0
O 528 304 576 304 1
x 454 57 508 61 0 16 Rsense
x 618 310 713 314 0 16 = I * Rsense
x 586 61 617 65 0 16 Iout
r 432 176 432 288 0 5000000.0
t 352 112 352 80 1 1 -17.384541032767196 0.7228104943582956 100.0
w 336 80 272 80 0
R 272 80 224 80 0 0 40.0 30.0 0.0 0.0 0.5
w 368 80 400 80 0
w 400 80 432 80 0
a 352 208 352 112 0 30.0 -5.0 1000000.0
x 215 213 248 217 0 16 Vset
x 209 406 248 410 0 16 Imax
a 528 352 432 352 1 30.0 -5.0 1000000.0
w 528 368 528 400 0
r 528 400 432 400 0 9000.0
r 528 400 624 400 0 1000.0
g 624 400 624 416 0
w 432 352 432 400 0
a 336 400 336 320 1 30.0 -5.0 1000000.0
x 177 85 202 89 0 16 Vin
w 432 400 432 432 0
w 432 448 656 448 0
O 656 448 688 448 1
x 630 61 666 65 0 16 Vout
x 672 435 813 439 0 16 Measured Current
w 352 400 432 400 0
w 528 304 528 336 0
w 432 432 432 448 0
w 400 32 400 208 0
f 384 288 384 240 1 1.5
w 336 304 368 304 0
w 400 208 400 240 0
w 336 240 352 240 0
w 368 208 368 224 0
w 368 224 368 240 0
w 352 240 368 240 0
w 368 304 384 304 0
w 384 304 384 288 0
d 336 304 336 240 1 0.805904783
172 336 208 272 208 0 6 12.0 24.0 0.0 0.0 0.5 Voltage
172 320 400 272 400 0 6 0.36 2.0 0.0 0.0 0.5 Current
f 240 304 176 304 0 1.5
f 240 352 176 352 1 1.5
w 176 320 144 320 0
r 144 320 80 320 0 220.0
g 80 320 80 336 0
162 80 288 176 288 1 2.1024259 1.0 0.0 0.0
162 80 368 176 368 1 2.1024259 1.0 0.0 0.0
w 80 368 48 368 0
w 48 288 80 288 0
x 54 394 168 398 0 16 Const. Voltage
x 57 274 172 278 0 16 Const. Current
w 48 288 48 320 0
w 48 368 48 320 0
R 48 320 32 320 0 0 40.0 5.0 0.0 0.0 0.5
r 384 304 384 352 0 10000.0
g 384 352 384 368 0
w 176 320 176 336 0
w 240 304 240 320 0
w 240 320 240 352 0
a 288 304 288 240 1 30.0 -5.0 1000000.0
g 272 304 272 320 0
w 304 304 336 304 0
w 432 288 432 304 0
g 432 304 432 320 0
w 240 240 288 240 0
w 240 240 240 304 0
174 688 80 784 80 0 50.0 0.6584000000000001 Load
w 736 64 736 48 0
w 736 48 784 48 0
w 784 48 784 80 0
x 346 71 360 74 0 10 Q1
x 329 145 342 148 0 10 U1
x 461 270 474 273 0 10 U2
x 447 370 460 373 0 10 U3
x 317 332 330 335 0 10 U4
x 269 255 282 258 0 10 U5
x 203 290 217 293 0 10 Q2
x 203 374 217 377 0 10 Q3
w 576 80 624 80 1
w 624 80 672 80 2
w 672 80 688 80 0
w 528 80 528 32 0
w 528 32 400 32 0
w 528 80 576 80 0
x 394 282 408 285 0 10 Q4
x 317 261 330 264 0 10 D1
w 336 304 336 320 0

[Mint.]

Bumping this up to help the guy get more replies.
I don't know much about op-amps, but I gotta say that this looks like a very solid design. Good job!
I really like the 2 LEDs that you have for constant current/voltage, awesome add on there.

[MikeNye]

Hi Mint,

Thanks for the bump. 

I think I'm going to use LTSpice to model the circuit before breadboarding, in the hopes that someone more experienced than me can comment before I get to breadboard stage.

Cheers!

-M

[MikeNye]

OK, so last night the wife had a group of girl friends over so it was a good chance to lock myself in the lab and set up an LTSpice simulation of my circuit above.

It turns out that there are some major stability issues when the circuit enters constant current mode. The whole circuit oscillates quite nicely....

I will perform some debugging and see what I can come up with, however if any more experienced people could comment, it would be greatly appreciated.

Thanks.

-M

[Rerouter]

interesting design, i for one was caught off guard that falstad could import spice :/ (missing only 1 wire link above u4)

though your missing a whole mash of parasitics, for a start all of your op amps have some capacitance between pins, throw a few picofarad between your input pins and it sings in the low MHz on the output,

u4 output to u5 input with q4 in the system sets up some rather horrid oscialltions on non ideal op amps,

u2 and u3 are ok, and look fairly stable, its primarily u4's output to worry about,