Inverter Amplifier Circuit

Yes, that lowers its self-capacitance and makes it more linear at very high opto-currents. But into what load do you observe a 100mV signal? And is this at up to 15-20MHz?

The Inverter CMOS linear amplifier in Figure 3.64 on page 158 in the book the art of electronics. The schematic seem to be unclear, for eg. the capacitor has only 0.001 value. Is this 1000 uf or 1 nf?. Also what is the circle connected to the ground just before the 51 ohm resistor. Can this circuit amplify a 100mv pk-pk signal from a photodiode at approx. 15Mhz-20Mhz. Also is the clamping diodes necesarry for the correct operation of the circuit.

Best Regards

The detector is biased over a 5V supply.

mF and uF were used interchangeably. mmF was never uF. mmf and pF were interchangeable. Not that anyone but me gives a rip, but I think the use of nF is a pain in the ass.

It seems to be catching on the US as well. Perhaps due to chip designers deciding "enough is enough" (with stuff like mmF, .001uF, etc.) after finding their need to routinely use values between 1 and 999 fF in their work?

I routinely use fF, pF, and nF on-chip. ** That's no typo, I used 1nF gate capacitance recently on a heavily transient-loaded regulator **

...Jim Thompson

M=150 (stripes), each L=2u x W=380u (thin oxide, 1.8V device)

on a HUGE RAM chip, so it's barely noticed ;-)

BTW, "compensation" whining is BS. The regulator only needed 5pF compensation... 1nF was for output energy storage (BIG transients)...

...Jim Thompson

The 5pF WAS an extra processing step (metal-oxide-metal cap)... could not use gate capacitance for that, due to no voltage drop at the compensation point. So maybe it's a cost issue for the LDO people, my chip was MIL, or maybe it IS lack of design skills ;-)

...Jim Thompson

Hello Winfield,

In the US that has always been the convention, along with some rather bizarre ones such as mmF for uF. But since AoE sells everywhere from here to Mars maybe it would be worth to actually place uF or pF behind these numbers. Much of the world uses nF and I have seen a lot of confusion about it in my consulting life. Like caps being three orders of magnitude off of what they should be because the local engineer took a US app note and misinterpreted the (missing) designator. Or had given the whole chebang to a local tech and just said "Wire this up for me, please".

This will also help if the book gets translated. Then you could just leave the schematics as they are, assuming that any EE will master enough of English to comprehend the few words that may be on the drawings.

Regards, Joerg

Hello Jim,

And then the LDO chip designers are whining about not being able to place proper compensation on board. After the last debacle I am not going to use LDOs anymore until they figure that out.

Anyway, how many acres did that cap swallow up?

Regards, Joerg

Aha! Bipolar! Makes it easier! My requirement was rad-hard, thus all-CMOS.

...Jim Thompson

Hello Jim,

Yes, that's like a parking spot on a Texas ranch.

Then why are many of these regulators so lousy? Is that 5pF already too much real estate for the bean counters or is it a lack of design skills?

Regards, Joerg

Hello Jim,

Avoiding a processing step is certainly understandable. But they didn't even provide a couple of pins where one could connect an external cap. Maybe that is expecting too much these days.

The real problems that mfgs don't see (usually) aren't at the circuit designers. They can select output caps where the ESR falls into a certain range. What they often can't control is the client hanging some additional stuff onto the rail that comes with more decoupling caps. Or a "smart" purchaser who hasn't read the module spec and got cajoled into buying this wonderful new ceramic cap for three cents less. What we hear a lot is "Yabbut it had lower ESR". The last LDO circuit even drowned out Rush on the local AM station after that happened. A solid 2Vpp on a heavily loaded 4.7V rail, pretty impressive.

Regards, Joerg

Analog Devices makes some LDOs that are supposed to tolerate AnyCap (tm) on the output. I've used the ADP3300 and ADP3330 and I haven't managed to make one oscillate yet but I have not been deliberately trying. Only one hint: If you do not use the noise reduction pin (NR) then don't hang a long trace off the pin, just put a pad and no track.

Chris

Hello Chris,

Yes, there are some LDOs that are designed right. The fly in the ointment with these is that they are usually at least twice as expensive as the vanilla brand. The ADP versions are north of 70c while the TI TPS77001 is around 35c. In most apps 70c is just too much for a regulator. I can do it with discretes or design a little switcher for less (which I usually do).

It's almost bizarre. If you want a part that works right you have to pay extra.

Regards, Joerg

Google: 114000 square microns in acres ->

Best regards, Spehro Pefhany

I read in sci.electronics.design that Jim Thompson wrote (in ) about 'Inverter Amplifier Circuit', on Wed, 5 Oct 2005:

No, we want the answer in acres. Or nano-acres.(;-)

Hello Spehro,

Regards, Joerg

I read in sci.electronics.design that Frank Bemelman wrote (in ) about 'Inverter Amplifier Circuit', on Thu, 6 Oct 2005:

I doubt that many will appreciate that. Has the spelling reform changed it from 'Fahrrad'?