Maker Pro
Maker Pro

431 shunt regulator help

Version 4
SHEET 1 880 680
WIRE 192 16 -288 16
WIRE 368 16 192 16
WIRE 192 64 192 16
WIRE 368 144 368 16
WIRE 192 160 192 144
WIRE 336 160 192 160
WIRE 544 176 400 176
WIRE -288 192 -288 16
WIRE -128 192 -288 192
WIRE 336 192 64 192
WIRE -288 256 -288 192
WIRE 192 272 192 160
WIRE 256 272 192 272
WIRE 400 272 336 272
WIRE 544 272 544 176
WIRE 544 272 480 272
WIRE -128 288 -176 288
WIRE 192 384 192 272
WIRE 416 384 192 384
WIRE 544 384 544 272
WIRE 544 384 480 384
WIRE 192 448 192 384
WIRE -288 560 -288 336
WIRE -176 560 -176 288
WIRE -176 560 -288 560
WIRE 192 560 192 528
WIRE 192 560 -176 560
WIRE 368 560 368 208
WIRE 368 560 192 560
WIRE -288 592 -288 560
FLAG -288 592 0
SYMBOL References\\LT1021-5 -32 240 R0
SYMATTR InstName U1
SYMBOL Opamps\\LT1006A 368 112 R0
SYMATTR InstName U2
SYMBOL voltage -288 240 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 12
SYMBOL res 176 48 R0
SYMATTR InstName R1
SYMATTR Value 100k
SYMATTR SpiceLine tol=0,1
SYMBOL res 176 432 R0
SYMATTR InstName R2
SYMATTR Value 100k
SYMATTR SpiceLine tol=0.1
SYMBOL res 496 256 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R3
SYMATTR Value 124k
SYMATTR SpiceLine tol=0.1%
SYMBOL res 352 256 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R4
SYMATTR Value 1k
SYMATTR SpiceLine tol=1
SYMBOL cap 480 368 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C1
SYMATTR Value 0.1µ
The LT1021-5 is a ridiculous over-kill for the application, but it is
in the LTSpice library while the cheaper LM4040C50FTA  I mentioned
earlier isn't.

I've also left out the 100nF bypass capacitor that even the LT1006
single supply op amp should have.

Almost any other low off-set single supply op amp that could survive
the peak supply voltage would do as well as the (elderly) LT1006.

I've followed Jim Thompson in figuring that the OP wants to digitise
the supply voltage range from 10V (which would give 5V at the output
of the LT1006 in this circuit) to 14V (which would give 0V at the
output of the LT1006, if the LT1006 could pull right down to the 0V
rail, which it can't, since it can only pull 40uA down to about 50mV).

The 0.1% resistors are E96 values and available ex-stock from Farnell,
some of them as single parts (though for many you have to buy five at
once, typically for more than a buck apiece).

Sorry about the typo in the LTSpice circuit diagram - not 10k
resistors, but 100k resistors (corrected above).

The biggest soucre of uncertainy on the output is the tolerance on the
voltage reference - the LM4040C50FTA
offers +/-0.5% on 5V, or +/- 25mV, which shows up as +/-56.25mV on the
output, equivalent to +/-45mV on the 10V to 14V range of the supply
voltage to be digitised. The LT1021-5 would be a factor of ten better,
but does cost about ten times as much (at around $10), and the
contribution from the tolerances on the. 0.1% resistors would swamp
that.
 
B

Bill Sloman

Jan 1, 1970
0
John Larkin said:
I'm just finishing up the "frame compiler" firmware part of the most
kick-ass digital delay generator anybody has ever made.

I know that your marketeers insists that you make these claims.
What is actually going to be the difference between this one and all
other the ones you've made the same claim about in the past?
I think I'm still getting better at electronic design. But even if I
plateau one
of these years, I'll still be designing kick-ass electronics for a
good while.

And you aren't even content to be useless.

Worse still, I'm frustrated. The last digitial delay generator for which
I was responsible pre-dated the MC100E195.

I got to design one some ten years later where I could design it in
and I worked out how I could cope with its nasty temperature
sensitivity, but the potential customer ran out of graduate students
before we could start building it, leaving me with a couple of
hundred pages of A4 sheets of Orcad circuit design, and no evidence
of how well it might have worked in practice.

Now Motorola are selling the MC100EP195, which would have let
me get away with a 100MHz clock (though 500MHz crystal
clocks were essentially off-the-shelf items, at a price, using chemically
thinned crystals and offering sub-picosecond jitter).
 
K

krw

Jan 1, 1970
0
I have no "marketeers" and when anybody insists on anything in my
company, it's me. Except when they order Hawaiian pizza, which is
repulsive, and I an powerless to stop thaem.


Two differences: one can load a new set of delays and widths and queue
them into the hardware, and they get loaded without trashing any delay
cycles. Most DDGs abort the current timing cycle if you reprogram
them, which can be bad news. Even better, one can pre-program a
sequence of thousands of shots, and it will load a new set of timings
every time it's triggered. This is cool for things like stroboscopic
photography or radar target simulation. The idea of reprogramming a
delay generator while it's being triggered at a high rate is
inherently tricky.



Micrel has some ECL delay-line chips. 10 bit resolution and guaranteed
monotonic. Better than the On stuff, and much cheaper.

http://www.micrel.com/_PDF/HBW/sy89296u.pdf


Hmmm, yet another failure. You seem to have a lot of failure stories.


Not Motorola: OnSemi.


The ECL delay-line chips have nasty tempcos that change with
programmed delay. And they are expensive power hogs. But they are true
delay lines, storing multiple edges inside, so have their place.

The CMOS delay-line chips I've used are really bad.

Have you looked at EC2 (Engineered Components Company)? I haven't
used them in ~30 years but they had some 2-decade 10nS ECL delay
lines that were reasonably good. I used tons of their one decade
10nS delay lines (the program got canceled before we needed many of
the .1nS resolution parts).
 
<snip John Fields restatement of the original problem>

If the OP has an opinion on my circuit. I'd be glad to hear it.

Your opinion on the subject is less interesting.
And, unless I've done something dreadfully wrong, your circuit certainly
doesn't provide that.

Nor do you have a source which varies from 0 to 14 volts, and you
haven't even bothered to take the comma out of the tolerance spec for
R1.

Also, there's no run-time associated with the file, so did you even
bother to run it or did you just draw it and make a lot of assumptions?

It's a circuit diagram, not a simulation. Do you find that you have to
simulate this kind of stuff?
 
B

Bill Sloman

Jan 1, 1970
0
In other words, it reflected your inadequate understanding of the problem.

Or so you would like to think.
equally interesting since, if he has eyes in his head, he'll see your
circuit
can't possibly work the way he asked for.

Of course not. For some bizarre reason. he wanted to do the job with
a TL431, which needs a minimum cathode current of 1mA to perform to
specification.
The LM4040C50FTA can get by with 60uA.

It's always nice to check your solution. My sub-concious does it for me,
but it does seem to take it's time.
Besides, as Mr. Larson quite rightly pointed out in an earlier thread,
one should always check one's work before making it public.

And one should confine one's self to posting about stuff where one knows
what one is talking about, which would pretty much silence you.
 
B

Bill Sloman

Jan 1, 1970
0
John Larkin said:
I have no "marketeers" and when anybody insists on anything in my
company, it's me. Except when they order Hawaiian pizza, which is
repulsive, and I an powerless to stop thaem.


Two differences: one can load a new set of delays and widths and queue
them into the hardware, and they get loaded without trashing any delay
cycles. Most DDGs abort the current timing cycle if you reprogram
them, which can be bad news. Even better, one can pre-program a
sequence of thousands of shots, and it will load a new set of timings
every time it's triggered. This is cool for things like stroboscopic
photography or radar target simulation. The idea of reprogramming a
delay generator while it's being triggered at a high rate is
inherently tricky.

It was called double-buffering when we did it for our stroboscopic
electron beam tester back in 1989-91. That machine was intended
to sample as fast as the secondary electron detector would let us,
(which should have been about 25MHz but got stuck at 12.5MHz
for fairly irritating practical reasons) and collected samples at every
possible point on the stroboscopic cycle in arbitrary succession.
Micrel has some ECL delay-line chips. 10 bit resolution and guaranteed
monotonic. Better than the On stuff, and much cheaper.

http://www.micrel.com/_PDF/HBW/sy89296u.pdf

Looks like a me-too of the MC100E196 - which is also guaranteed
monotonic, like the MC100E195. The "better" I might believe if I had
some independent measurements to look at.
Hmmm, yet another failure. You seem to have a lot of failure stories.

Success stories are tediously similar.
Not Motorola: OnSemi.

Same company. It seems to get badge-engineered every few years.
The ECL delay-line chips have nasty tempcos that change with
programmed delay.

As I'm well aware. The solution for my application was going to be
to spend a millisecond or so measuring all 128 delays every few minutes
and adjust the programmed delays accordingly. Other applications
would call for different approaches
And they are expensive power hogs. But they are true
delay lines, storing multiple edges inside, so have their place.

The CMOS delay-line chips I've used are really bad.

Most of the ones that I looked at seemed to be monostables that
marketing wanted to sell as delay lines.
 
B

Bill Sloman

Jan 1, 1970
0
So, it seems to me, had you stayed silent instead of mouthing off,
pretending to know what you were talking about, you'd have been much
better off and would have relieved me of the responsibility of showing
you up for the phony phuck you are.

But you're not that smart, are you?

You'd like to think so.

You also like it if I took your mouthing off seriously enough to work out
what you think you are carrying on about.

I've been foolish enough to waste my time this way in the past, but all I've
ever been able to establish is that once you have get a bee in your little
bonnet, you will keep on posting incoherent nonsense that reduces to the
opinion that you have to be right and I must be crazy to disagree with you.
 
Top