Negative logic

It's also useful as a technique for designing rad-hard circuitry. Radiation kills electronics by causing certain parameters to drift as a function of dose. When a critical parameter drifts far enough, your gadget stops working.

Old IC processes yielded poorly controlled device parameters, and designs were dimensioned to accommodate that. Such designs were often also pretty rad-hard. Modern IC processes control device parameters much better, and circuit designs have evolved to expect narrower tolerances. They still drift under irradiation. As a result, modern ICs often have very poor radiation hardness.

It matters to me.

Jeroen Belleman

I don't think I have ever seen any of that. What I believe in is progress, and getting rid of the horrible mess of wires and providers and protocols that we have now.

Sure. If the signal is weak in your basement, buy a repeater at Safeway for $10.

I bet you have a fairly fast edge driving them though, you're not relying on the Schmitts to speed up a slow edge.

Clifford Heath.

For rad. hard, big old fat leaky* transistors are better. (?)

George H.

Am 21.09.2018 um 22:06 schrieb George Herold:

A temp-compensated VBE multiplier.

Cheers, Gerhard

It's really not a good idea to put current into ESD diodes - it ends up flowing into the substrate and can come out anywhere, and while that's happening you can't have any confidence about logic states anywhere else on the same substrate.

Been there, been caught by that.

I'm sure the hackers can't wait.

upsidedown wrote

Yes, but I myself will only need 50 Mbps... nice if I can take it with me as a small wireless router access point, say on boat or camping. They have those 'mifi' 4G things these days with 10 hours or so battery life. All of Europe is covered by the same SIM card subscription / cost these days.

George Herold wrote

I think it is the 1.2 V bandgap reference, the good old TL431 uses that, and 2 resistors, to make a parallel stabilizer (= zener). It is a bit more stable than a Vbe multiplier with one transistor and 2 resistors.

to 1k

t.

d base PNP.

th the NPN:

going 'on',

lowing into the substrate and can come out anywhere, and while that's happe ning you can't have any confidence about logic states anywhere else on the same substrate.

plenty of logic, mcus, fpga that specify safe input as either max voltage or max current

Widlar diode is another name...

piglet

Some older logic, especially 4000 A-series, would latch up and fry if you forward-biased the ESD diodes just a little. HC parts are usually specified for 50 mA into the diodes, and I've never seen one latch up.

It's fairly common these days for data sheets to explain how much current you can dump into the ESD diodes of powered-up chips. Opamps often let you dump a lot. See AoE3 p 361.

Thanks, I've used these as voltage references, but never thought about using them as voltage 'droppers'.

George H.

The bandgap isn't biased right if it only has a CMOS input on the anode end. So, the transistor is complete with two resistors, +5, -5V, and no other power, but the bandgap solution will require a milliamp or so of current, and long LOW duration means it'd benefit from a third (-12V) power source.

e.org:

e:

te:

es to 1k

get.

er,

nded base PNP.

with the NPN:

or going 'on',

ate?

nd

p flowing into the substrate and can come out anywhere, and while that's ha ppening you can't have any confidence about logic states anywhere else on t he same substrate.

Latching up in't the problem I was talking about (though it's one more odd thing the chip can do). The logic states on the chip can stop conforming to the input-to-output relationships specified in the data sheet. That's a mo re subtle problem and is usually non-destructive, but can be equally fatal to the proper operation of a circuit.

They tend to be less forth-coming about the way the chip behaviour changes while you are dumping that current into the ESD diodes.

The business of stopping a op amp blowing up is less subtle than having it behave correctly while the ESD diodes are carrying current. Win Hill's flat tery didn't go into that.