relays

It has been my experience that the consequences of making mistakes in calculations in education are far smaller than they are in industry. When those instructors start expelling people instead of just giving zero credit, let me know.

You can wait forever for all I care. You attempted to distract everyone from noticing your gross errors concerning latching electromechanical relays by introducing non-latching solid-state relays into the discussion, but you let the cat out of the bag when you said you are using them in a current project. It's pretty easy to make correct calculations from a vague datasheet if you have an actual circuit and some test equipment in front of you that allows you to discard your errors if they don't agree with real-world measurements. Meanwhile, your original howlers are still there and you still refuse to use standard engineering units for power and energy.

So you admit -- at last -- to your many errors?

Liar.

If you think that I don't do sanity checking, stay off of commercial aircraft -- some of them have components that I designed and many of them were tested with systems that I designed.

As for "can't do simple math", I consider using joules as a unit of power to be a fine example of not knowing how to do simple math.

That's just shoveling numbers around. You can divide the abs max allowed power by the abs max allowed current, and get something that's dimensionally volts, but it has no meaning here. The maximum on resistance is spec'd at 1.2 ohms, so at 400 mA the max voltage drop is

But I wouldn't bother computing the switch loss. It's a small error in a calculation that barely justifies two decimals of precision.

John

Incredible! The abs max load current is specified as 400 mA max "continuous." It's right there on the data sheet.

Yeah, tell them that their part doesn't meet their specs. Maybe they'll offer you a consulting contract.

An instant's glance will show that it's small number compared to 40 watts. As you noted, the abs max power dissipation in the fets is 300 mW, so you'd figure it would be below that in use. And it doesn't take long to compute (0.4)^2 * 1.

With 40 watts into the load, at the "typ recommended" drive of 10 mA, power gain is about 3.3K. But 2 mA drive is legal, if slow, and there the gain is about 18K.

Led Vf isn't precise enough to justify a bunch of digits.

It's bog simple. You don't really need a calculator. None of the three guys who bothered to debate this could get it right. No coincidence.

John

Engineering is about being pragmatic, not prissy. 39.8 watts is close enough to 40 as doesn't matter here. As noted, led Vf isn't known precisely, to this is inherently a ballpark calculation.

Using 40 makes it easy to do the math in your head.

"40 over 12 is 20 over 6 is 10 over 3, which is 3.333, times a thousand." Faster to think than to type. It's sort of a game we do here, estimating things to engineering accuracy, instantly, standing up, without benefit of calculator. It's a quick way to decide if some effect is important or not, and a good way to crosscheck calculations for gross errors. Often just the order of magnitude is enough.

Why not compute the maximum legal power gain? The thing is spec'd to operate with 2 mA into the LED. You can't compute the power gain of a nonlinear thing like this without assuming some input level.

FET on resistances are usually measured in pulsed mode, to avoid self-heating drift. The part is clearly specified to work at 400 mA continuous switch current.

John

Look at the led Vf spec.

You can't compute power gain without knowing the input power. Or, at least, I can't.

I don't see any. NEC does very good stuff. For a lot of their RF phemts and stuff, they provide both s-params *and* Spice models.

Hey, did you know that you can often just eyeball the S11 v frequency list, and estimate input capacitance?

John

The order of magnitude is usually the part I get wrong. I have to write the numbers by pencil or pen and paper, in fixed point, and physically count decimal places. ;-)

Cheers! Rich

If that's what you want to calculate, yes. If you want to calculate typical gain, use the 1.2 volt number. Just state what you're calculating, and do it.

Get a life, Jack. You're now nagging me about the third or fourth decimal point, when your initial calcs were, what, 78 and 280?

John

Since when was it not?

If I decide to compute typical gain, that's what I compute. If you want to compute minimum guaranteed gain at 120C on Tuesdays, state so and do the math. There's no dogmatic definitions here.

The computation of the power gain of an SSR is pretty much a whymsical idea anyhow, interesting but mostly useless. It doesn't justify a lot of angst or a lot of decimal points. I *was* impressed by the 18K number.

Just checked Digikey: first old-fashioned DC em realy I tried has a power gain of 7500. Not bad. AC relays are probably even better, since their coil current drops a lot once the armature seats.

I never posted my assumptions, so I had nothing to switch from. I did use 1.2 volts led drop because it made it easy to do the math in my head. I was just shooting for the ballpark. 78 is not in the ballpark, nor is 280. 3300 certainly is.

HA!

John

"1418"

John

--- Since the beginning.

Early-on you asked for some numbers and I provided them and showed my work, from which it should have been obvious that I was using Vf(max), since I called it that, as I recall.

If you wanted to contest the numbers and you wanted to use values other than mine, then you should have said so instead of just presenting a number.

---

--- Problem with that is that you didn't "say so and do the math", you just did the math using different quantities in order to muddy the water.

---

--- Since a DC coil relay won't usually drop out until at about 25% of the rated coil voltage you can get better power gain that way too.

---

--- Oh, for chrissakes, John, grow up. How long are you going to whine about those early numbers?

---

---

Here's the full thing:

"Well, let's see...

Using their 400 mA 300mW absolute maximum spec, it appears the output voltage drop across the part will be:

P 0.3W E = --- = ------ = 0.75 volt I 0.4A

Subtracting that from the 100V limit for the part means that the load will see 99.25 volts, and the gain equation will look like this:

Eload * Iload 99.25V * 0.4A PG = --------------- = ---------------- ~ 1418 " Eled * Iled 1.4V * 0.02A

"1418" is the integer part of the decimal number and since there is no fractional part (1418.XXXX would be a four decimal point number) it's really not even a "1 decimal place" number.

JF

Yeah. I started doing math on slide rules, and there's some mental knack, which I can't really describe, that one gets for tweaking the exponent as a result of the basic mul or div results. It helps to work in "engineering notation", namely express everything in powers of three. "20 n over 5 p is 4 k", like that.

Sometimes somebody will say something like "is the pcb capacitance under that pad enough to make that opamp unstable?" Usually things like this can be guestimated on the spot, to an order of magnitude, and maybe dismissed from further consideration. It's sort of a sport, but it comes in very handy.

I find myself automatically checking numbers in newspaper articles and such. The mistake level here is huge.

John

I strongly object to your misrepresentation of my position.

"Couldn't get it right" implies that I tried to get it right and failed. That's not true. I was very clear in stating that I consider your switching the conversation from your obvious errors concerning latching electromechanical relays to a discussion of non-latching solid-state relays to be a red herring. I was very clear in stating that I refuse to play that game.

When you accused me of agreeing with John Fields I was very clear about neither agreeing nor disagreeing when I haven't bothered to check his work myself. I did say (and it's still true) that he makes few errors, his errors tend to be minor, and he quickly admits errors and corrects them, while you make huge glaring errors and use every cheap debating trick in the book rather than admit to any error. Based upon that, if I had to choose, I would choose JF as being more likely to be correct, but that is *not* the same thing as saying I replicated his work and confirm it to be error-free. If JF pointed at a flock of sheep and told me that they are all white, I would say that he is likely to be correct but that I can only attest to the fact that the sides facing me are white.

Also for the record:

It is not true that a latching relay has an infinite power gain.

It is not true that the energy gain of a relay is the same as the power gain of a relay.

You cannot calculate power gain using joules in. Joules are a unit of energy and can only be used to calculate energy gain, not power gain.

You cannot calculate any sort of gain using joules in and joules per second out.

It is not true that most relay datasheets rate contacts in VA.

It is never true that any relay datasheet rates contacts under steady-state conditions in VA.

It is not true that most engineers fail to differentiate between infinite and unbounded.

Did I mention that it is not true that a latching relay has an infinite power gain?