Low current source

"Fred Bartoli" wrote in message news:432b4088$0$14830$ snipped-for-privacy@news.free.fr...

It's very simple! As you say, it's not very accurate, but that's not really a big problem for me. It's a little awkward to adjust over a large range, too; for any particular value of R, the range of adjustment (by changing the

How do I clamp the voltage across the capacitor?

It won't do to just clamp the output of the circuit, because then it wouldn't immediately start ramping down when the logic changes.

One end of the capacitor is at ground (the inverting input of the opamp), nothing to clamp there. The other end is at the opamp output, a low-impedance point so no point clamping there either.

I could put a resistor in series with the opamp output, thus making it a relatively higher-impedance point that I could then clamp. Since it would be inside the feedback loop, it shouldn't affect the integrator. Like this:

--/\\/\\/------o-----||---------. | | '-|\\ | | >--/\\/\\/----o--- OUTPUT (clamped to 0V < Vo < Vmax) ,-|/ | ===

As long as the clamps are low leakage, and the output is driving a very high impedance (another opamp input), this should work. Right?

By the way, thanks very much, everyone!

-walter

Walter,

Could you repeat your desired voltage limits/bounds? I don't think I got all the thread. Also, what power supplies do you have available?

Thanks!

...Jim Thompson

If you clamp the output as shown above, the OpAmp output will rail, then the summing node will wander off somewhere, giving the same kind of unknown-starting-point problem.

...Jim Thompson

The thread is getting pretty spread out... there are a lot of ways to skin a cat.

Goal is to make a linear voltage ramp from 0v to 1v, and back down, with separate time constants for up and down, each adjustable by trimpot from

I can make whatever power supplies I need, but I will be starting with

This is a one-off, so parts cost is not a big issue; important aspects are simplicity, reliability, field serviceability, and low parts count (the latter just so as to minimize assembly time and physical size).

Linearity should probably be no worse than, say, 20%, and the ramp time (once set) should not vary more than 20% over a time period of several years and over a temperature range of 20C to 35C.

Jim, you also said:

Oh, duh. Dunno what I was thinking. The current through the resistor has to come from somewhere, and that somewhere has to be through the capacitor, clamps or no. The only way to clamp the capacitor voltage is to prevent current from flowing through it. Sorry, thinking out loud.

Walter Harley schrieb:

I think the easiest solution to this problem is using a simple integrator (opamp with capacitor in the feedback).

For separate adjustability of the slopes, use analog switches that either connect the input to different positive/negative voltages, or use one analog switch to reduce the input resistance for the shorter ramp, or use different positive/negative voltage amplitudes of the driving square wave.

You'd better written this in the OP already... You really don't need an explicit current source. :-)

Here's one way, with empirically-determined setpoint voltages:

. ,----- setpoint - 0.4V . b . ,------c e -, . | | . --/\\/\\/------+--+----||-------+ . | | . '-|\\ | . | >--/\\/\\/----o--- OUTPUT . ,-|/ . | . ===

If you want more control over the clamping voltage, it's necessary to define the clamping-transistor's current, which is easily done with a diode plus a sinking current (e.g., a resistor to V-):

. ,----- setpoint - ~0.4V . b . ,--|

junkbox,

currents

every

drop

MAT03),

This gets to be somewhat complicated for the purpose, I believe. And if accuracy is not of great importance he could use the much simpler

calculate R for IC about 0.5/1uA ___ +Vs-----|___|---+-----. R | | | .-. | | |10M | | | | '-' >| | 2N5087 |---+------> 50nA /| | | === GND

That one has about -0.4%/C tempco, needs not matching, and can also be temp compensated with an additional R and a PTC.

This is a simple 555 circuit with simple frequency control- I got 300 seconds with the LMC555 and Radio Shack 0.22u 50V films with less than

. ,------ R2 --------------------------------, . | C | . | __ A1 ,---||----+ . -- R1 --+--|+ \\ | __ A2 | . | >--+--|>|-- pot --+-- R4--+--|- \\ | . ,--|-_/ | | | >--+--- out . | '--|--+--- out . | | ,--|+_/ G = -R2/R1 . gnd | | . program max | gnd . slew rate | . -- R3 --+------ e c -' Vprog . | __ b -> I_slew slew = ----- . '--|- \\ | Q1 R3 C . | >--' PNP . ,--|+_/ . | . gnd

During slewing the LM13700 steers a positive or negative version of the programmed slewing current into the integrator. This works well at low slewing currents (even well under 1uA), which you can get with high values of R3 and/or low programming voltages to R3. R3 can be made from three resistors in a divider setup.

Thanks Fred, you're too kind! You're correct, the slew rate goes as 1/R (nonlinear), but the time to voltage goes as R (assuming Rpot >> R3), so maybe Walter will find the concept useful.

Some might wonder how two opamps inside one feedback loop can be stable, without extra compensation components. My conjecture is the loop will be stable, because of the low-frequency integrator pole, and because the A1 stage gain is attenuated by ~ R3/pot and the diodes, whose dynamic resistance rises dramatically as their current approaches zero at loop balance, so the extra gain should be low enough to be safe at equilibrium. If I'm wrong, it's easy to fix: a small resistor in series with C, etc.

Are we having fun yet?

The OP seems to be having problems recognizing his requirements, but this is just a bog standard slew-rate controlled amplifier with three input parameters: final value, sr(+), and sr(-). No inaccurate clamping required and very uncomplicated, like so: View in a fixed-width font such as Courier.

. . . +----------------[R]--------------+ . | | . | | . | | . | | | . | - | . | ||| | . | V ---> |V| BCS1 | . | SR1 ||| | . | - | . Vin>--[R]-+--|+\\ | C | . | >--+--||--+----||----+--->Vout . +-|-/ | | | . | | | | . REF | | | . +--|>|-+-|-+ . - +-|+/ . ||| | . V ---> |V| BCS2 REF . SR2 ||| . - . | . . . . . BCS2 BCS1 . Vout= - Vin SR(+)= ---- SR(-)= ---- . C C . BCS can be the Bartoli Current Source under rheostat control or voltage control.

Maybe a little capacitor multiplication for reasonable CS numbers: View in a fixed-width font such as Courier.

. . . +----------------[R]-----------------------------+ . | | . | | . | | . | | | . | - | . | ||| | . | V ---> |V| BCS1 | . | SR1 ||| | . | - | . Vin>--[R]-+--|+\\ | C | . | >--+--||-+-----[R2]----+-----+--||--+->Vout . +-|-/ | | | | | . | | | [R2/n] | | . REF | | | | | . +--|>|-+-|-+ . - +-|-\\ | +-|+/ . ||| | >-+ | . V ---> |V| BCS2 +-|+/ REF . SR2 ||| | . - REF . | . . . . . BCS2 BCS1 . Vout= - Vin SR(+)= ---- SR(-)= ---- . nC nC .

That's not going to work worth a damn: View in a fixed-width font such as Courier.

. . . +----------------[R]-----------------------------+ . | | . | | . | | . | | | . | - | . | ||| | . | V ---> |V| BCS1 | . | SR1 ||| | . | - | . Vin>--[R]-+--|+\\ | C | . | >--+--||-+-----[R2]----------+--||--+->Vout . +-|-/ | | | | . | | +----[R2/n]---+ | | . REF | | | | | . +--|>|-+-|-+ . - +-|-\\ | +-|+/ . ||| | >-+ | . V ---> |V| BCS2 +-|+/ REF . SR2 ||| | . - REF . | . . . . . BCS2 BCS1 . Vout= - Vin SR(+)= ---- SR(-)= ---- . nC nC .

If he knows that it's always between 0 and 1V then he can get away with a digital pot and using fixed current sources and Vin. The digital pot might be loaded every time he goes to make a transition, and that's about it: View in a fixed-width font such as Courier.

. . . +----------------[R]-----------------------------+ . | | . | | . | | . | | | . | - I=W*CS | . | ||| +------------+ | . | CS1|V| | | | . | ||| -- | | . | - | / | digi | | . Vin>--[R]-+--|+\\ | W | \\ | pot | C | . | >--+--||-+--->| / | +--||--+->Vout . +-|-/ | | | \\ | | | . | | | -- | | . REF | | | | | . +--|>|-+-|-+ . - | +-|+/ . ||| | | . CS2|V| REF REF . ||| . - . | . . . . . W*CS2 W*CS1 . Vout= - Vin SR(+)= ---- SR(-)= ---- 0< W < 1 . C C .

Very good- I added REF in case someone was thinking of single supply. I prefer more certain protection for the output- not as nice a CMOS input- but that's just a shift- all Radio Shack stuff, TL082- protection only kicks in when you have hardware failure of some kind- these values should get him 0.1->5sec slews, and assumes VCA input is Hi-Z: View in a fixed-width font such as Courier.

. . . . . 0.33U . +15V +----+----------||-----+ . | | | | . | --10K | | . [1.5M] | / | | | .Vin>-[10K]-+----|+\\ | | \\ | | | . | | >+-||----->| / | | | . | +-|-/ | W- | \\ | | | . | | | --- +--------|-\\ | . | | | | | | >-----+ . | | 0.1u| +--------+----|+/ | . | +-||--+ | | | | . | | | --- | | [1K] . | | | | / | | | | . | [5k] | W+ | \\ | | +-----|>|-----+ . | | +-|>|-+--|| / | | | | . | | | | \\ | | | +-|

This circuit works just right, and is what I will use. I implemented it with an LMC6082 (150uV offset and 10fA bias!), using a +/-7V supply, 1MEG trimmers in series with a 20k resistor for the pots, 3.3k for R3, 1MEG for R4, 0.1uF for C, and 10k resistors for R1 and R2. With those values, a 1V ramp is adjustable from 0.1s to 5s, and looks linear at both extremes. I didn't see any signs of instability.

When I simulate it with LTSpice, the output of A1 rings wildly each time it comes off the rails. I didn't see any sign of that in the real world, though, so I assume it's just a simulation problem. I explored some different settings (e.g., small timestep, different opamp models) without any success, but I didn't try very hard.

Thanks again, to everyone who helped me. A non-profit community radio station (KSER-FM) will sound a little better, thanks to you!

-walter

That's economical but you lose linearity over his 50:1 range with the pots like that. I hope he uses it though:-)