Constant-current LEDs?

s/years/decades/

Slowman's been designing one simple oscillator for a couple of decades now.

Here's what Doc suggested. Put two diodes in series from ground to the base of a NPN transistor with a 120 ohm resistor in the emitter and a 10K resistor from base to +v. Insert LED in series with the collector. It's 4 parts, but you could eliminate the two diodes with a zener and make it 3 parts.

BTW, I got your book on good friday. I've read the forward part and am now into the main text.

.

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

ers say it is obsolete and unavailable...

for

e s

can

use

h circuits until he gets something that seems to work. Typically it takes h im a fortnight to get a "new" product. Real design takes longer, and he can 't recognise the process when somebody else does it, and can't imagine that it might go on without customers to buy the product.

The particularly complex and slow project that I had in mind got pushed by a boss and a project manager in the hope of getting it to an industry get-t ogether that we couldn't - as it turned out - make.

The project manager didn't give up trying as early as he should have done, and pulled tricks like shipping "completed" circuit diagrams off to printed circuit layout before we'd had a chance to do a deliberate design review, and shipped the printed circuit layouts off to the printed circuit shops eq ually urgently.

We had "complete" and loaded boards early enough that if they had worked fi rst time we could have got to the get-together. They didn't work.

One irritating failure was that a 6-layer board carrying 800NMHz clock sign als had been made with the layers in the wrong order - my manufacturing not es were quite explicit about the layer order, but some clown at the printed circuit house got worried about board warping and cleared a change in stac k-up with somebody back a Cambridge Instruments who didn't have clue about what the board was supposed to do.

The fault that got up my nose was in the twos-complement arithmetic being c arried out in 100k ECL digital signal processing.

When we down-shifted negative numbers (effectively dividing them by some mu ltiple of two) we should have made sure that the new high-order bits create d by the down shift were all "one". The engineer who did the detail design of the board didn't bother. I probably would have caught it at the design r eview, if we'd had one, but the rogue project manager was in too much of hu rry for such time-wasting niceties.

When the software guys finally tracked down that particular error - in the sense that the shift and add wasn't working right - I cut and linked a bodg e that fixed it (for small shifts only) in less than a day (and apologised to the junior engineer responsible for board for going over his head).

John Larkin likes to think that I was "academic" and some bosses did compla in that I was too academic, but I really did work with complicated systems, and getting them right does take time. Sometimes academic input helped, bu t it was never a big part of the job.

--
Bill Sloman, Sydney

George Herold is a good guy, and doesn't do rhetorical questions of that kind.

If you were half as clever as you seem to think you are, you'd know that.

Because you are a dim and unpleasant prick, you want to see it as a rhetorical question, and thus an insult, but George doesn't do that kind of stuff.

--
Bill Sloman, Sydney

Larkin lacks a sense of humour, and lacks the capacity to realise that taking somebody literally can be a way of sending them up.

--
Bill Sloman, Syndey

te:

e:

pliers say it is obsolete and unavailable...

OL for

the

Idss

te

ou can

ver use

with circuits until he gets something that seems to work. Typically it take s him a fortnight to get a "new" product. Real design takes longer, and he can't recognise the process when somebody else does it, and can't imagine t hat it might go on without customers to buy the product.

Jamie can't do the web searches that find my patents and my published paper s.

I didn't publish a lot of the stuff I designed (most of it was commercial a nd copyright) but there's enough of it out there that anybody brighter than Jamie (about 95% of the population) would hesitate to claim that I couldn' t design electronics.

--
Bill Sloman, Sydney

e

ircuits until he gets something that seems to work. Typically it takes him a fortnight to get a "new" product. Real design takes longer, and he can't recognise the process when somebody else does it, and can't imagine that it might go on without customers to buy the product.

That would be a fairly sophisticated set of claims for a child to make. Tab b needs to choose his insults more carefully - repeating yourself doesn't a lways suggest that you have self-consistent world view, but rather - as her e - can suggest that you have a restricted vocabulary

--
Bill Sloman, Sydney

I'm not a communist, and neither is Obama. "Krw" needs to learn a new insul t, not that he has ever shown any capacity to learn or change.

I wasn't objecting to John Larkin having customers - John Larkin was misrea ding what I wrote in a way that made him feel better - but rather suggestin g that people who don't have as many customers as John can still be interes ted in designing stuff, which I contend to be something John Larkin finds d ifficult to believe.

--
Bill Sloman, sydney

:

iers say it is obsolete and unavailable...

for

he

ss

can

r use

th circuits until he gets something that seems to work. Typically it takes him a fortnight to get a "new" product. Real design takes longer, and he ca n't recognise the process when somebody else does it, and can't imagine tha t it might go on without customers to buy the product.

.

It's not a "simple" oscillator. And it's designed, and the design has been stable for about six months now - I need to turn it into a printed circuit layout and get it built, which is mainly a matter of firing up my Linux par tition and plugging the circuit into KiCad. I've made a start, but don't se em to be making much progress at the moment.

And the design may go back to something I did for Cambridge Instruments in

1987, but the current design lost it's inductors a few years ago, and took it's current form around June last year.

I've fooled around with other projects in the meantime, but this one does l ook more promising than most, though it does seem to exist in a customer-fr ee space.

Build a better low-distortion oscillator, and the world will beat a path to your door? Probably not.

--
Bill Sloman, Sydney

matched temperature

ement, it's not a big problem.

with delta-T is

uns away when its dissipation gets to roughly

Alas, I goofed the calculation. John Larkin's pessimistic thermal couplin g estimate seems borne out, too. Just to check, I looked up 2N3425, a metal-cased dual tra nsistor of yesteryear, and it is nearly as bad as the modern ones (250 C/W going by the spec sheet ). Monolithic multiple transistors, on a common metal paddle (in a DIP package) presumably are bet ter, but it looks like modern plastic dual transistors are all thermally misdesigned. For a milliwatt they might make an OK mirror, but deciwatts won't work.

f you wanted a good 5 mA mirror, I wonder how much it would cost to get a B eO table on a D2PAK-5 (TO-263-5) base, with the bases of the two transistors on the comm on tab?

Its maximum frequency will be measured in decades per cycle...

Actually it is set up for a nominal operating frequency of 15.9kHz, or 100,000 radians per second, to keep the calculations easy.

You seem to be competing with krw in finding ways to be rude that don't have much to do with reality - krw has an excuse, in that he seems to be completely incapable of processing new information, but you used to be able to do better.

--
Bill Sloman, Sydney

Not an insult, a reprimand.

Still smarter than you.

Dan

You may want to think so.

You seem to want to think so.

You do seem to be further into self-delusion than most, though John Larkin may be a marginally worse case.

--
Bill Sloman, Sydney

Yup. There's some excuse for the 6-pin ones, but it's true even of the ones sold as current mirrors, e.g. the BCV61/62. If you look at the BCV62 datasheet, on P2 it has a spec for this

Thermal coupling of transistor T1 and T2 @V_CE = Maximum current of thermal stability of I_C1 5 mA TYP

So Infineon is even more pessimistic than I am.

For a milliwatt

A lot. It's considerably simpler and cheaper to use a Wilson-style mirror. (I like the switcheroo Wilson, which has much higher Zout--it actually contains three separate feedback loops, using only four transistors.

The cascoded mirrors force V_CE to match between the two bottom transistors, which are what determines the mirror ratio.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

ers/April_Fool_Lim_2.JPG

say it is obsolete and unavailable...

ositive VGS, the output characteristic showing over 5mA at VGS 1.0V for the 1.6mA IDSS part. How come you don't make use of that feature in your April Fool's circuits...

IDSS of the LND150 is unworkable without parts and effort, you need a deple tion FET with MIN IDSS in excess of your current source. Ckt below is adapt ation of precision current source by Harris from an ALD white paper, sing c heapest crap I cold find on Mouser. The bss126 is a 600V part IDSS min 7mA and -2.7

I take being compared to John Larkin as a compliment.

And still richer than you.

Dan

And there's always opamps.

This might work, sometimes:

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

My PNP booster will work with the LND150. And there are higher Idss depletion fets. We use DN2530, Idss 200 mA min. Adding a source resistor will drop the current as needed. 2 parts for a decent current limiter.

Accuracy of the Supertex parts is plenty good to regulate the current to an LED.