Startup Circuit for CMOS current source

Without a depletion mode device I do not believe it is possible to have start-up current go _completely_ to zero.

But you _can_divert_ the start-up current away from your reference to reduce power supply sensitivity.

Do you have specifications for desired reference current and total current consumption?

Keep in mind that a reference that uses only CMOS devices sucks for accuracy. It'll be "flat" versus supply, but not constant versus temperature or device process corner variations.

I usually charge a fee for this kind of question ;-)

...Jim Thompson

Hi Joerg,

I have poly resistors. What do you suggest?

Thanks QQ

Joerg wrote:

Hello QQ,

No poly resistors either?

Regards, Joerg

Hi Jim,

Thanks for your advice, I was beginning to think this problem was harder than I initially thought too.

The current source does not need to be very accurate. It is used for bias.

There are no required specifications, this is just a thought exercise. For that reason (and that I cant afford it :) I cannot pay you for your expertise - sorry.

QQ

I was just joking... provide some specs and I'll show you how tomorrow afternoon.

...Jim Thompson

Um... use a capacitor?

John

Sheeeesh, John, That's as bad as a digital guy using one-shots ;-)

...Jim Thompson

Well, how slow can a power supply come up?

John

Hello QQ,

Maybe you could replace the depletion mode device with it to supply the start-up current.

Regards, Joerg

Hi Jim,

As for the specification - current in the startup circuit less than 1nA

- after the current source has stabilized would be good. Supply varies only from 4.5V to 5V - I am hoping for the switching off of the startup circuit to not be too sensitive to power supply voltage

- incase the supply range were wider.

Thanks QQ

Takes some VERY BIG resistors.

What is the bias current value supposed to be?

...Jim Thompson

Nah. One-shots are OK as long as the timing requirements are sloppy enough that a factor of, say, 3 either way makes no difference (even when single-stepping with an ICE).

_Tuning_ one-shots, however, is always an exercise in turd-polishing.

Cheers,

Phil Hobbs

I like HC123's. Nice schmitt inputs, retriggerable, tolerate big timing resistors, complementary outs. Logic doesn't have to be synchronous, it only has to be right.

John

I despise transient-type start-ups... very risky, very bad performance with jerky ramp-up PSU's.

...Jim Thompson

Hello Jim,

Can analog guys use one-shots then?

I did on my last design. And on most of them before that. But no 555,

Regards, Joerg

Show me ;-)

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |

I'm curious--why would you bother auto-tuning a one-shot? Sounds a little bit like a turd car wash.....

Cheers,

Phil Hobbs

"The modified circuit also uses a depletion mosfet..." In my entire lifetime I've encountered processes that had a depletion-mode mosfet only twice.

Many CMOS processes do not include a PNP device.

I regularly do start-up circuits, but the start-up current doesn't go completely to zero. But it CAN easily be diverted so that it has no effect on circuit operation.

...Jim Thompson

QQ's responses seem to imply pondering rather than real application... I've yet to see even a vague specification.

...Jim Thompson