Minimum hfe

You want the worst of the worst ?

2. The only transistors that have a gain less than 20 are like high voltage sw itchers. But then you might just be using them. So let's figure even less, like 5.

but thet is for ONE of them. Now if you are using a power transtor for the power, and a driver to make a Darlington, figure an absolute minimum of 20 for the driver, especially in its operationg range, and then 5 for the seco nd transistor. That is 100 gain.

Feed it with voltage, not currnt because the gain curves are not linear. In other words you stick 100uA into the base and get 10mA copllector current, that does NOT mean you can count on getting 20mA collector current from pu tting 200uA into the base. It does not work that way except for some really "good" transistors. From what you say, this is not what you'll be using.

What are you doong with this ? A little detail mighht get you some really g ood advice here. These old folks are the ones they throw money at to keep u s ahead of Uganda in the technology race...

400?

--
John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 

I meanno insult by that. But I see how people talk about the lack of new en gineers who are competent, which is what keeps these guys working into thei r seventies and shit. Smae thing with a few other fields. I know a process engineer still getting work at 67, and you have any idea how the demand has dropped for that in this country ? They just cannot find peopel to do it a nymore. The industry is lucky the demand is gone, otherwise their phone wou ld never stop ringing - with complaints or whatever, and other problems. Th ey just don't get it anymore.

I don't think it's knowable, but if your circuit will work with 50 for the small one and 10 for the big one, you're probably safe.

At least until someone starts using transistors from some pile of specific equipment with esoteric stuff in it...

--
Tim Wescott 
Wescott Design Services 

switchers. But then you might just be using them. So let's figure even less , like 5.

e power, and a driver to make a Darlington, figure an absolute minimum of 2

0 for the driver, especially in its operationg range, and then 5 for the se cond transistor. That is 100 gain.

In other words you stick 100uA into the base and get 10mA copllector curren t, that does NOT mean you can count on getting 20mA collector current from putting 200uA into the base. It does not work that way except for some real ly "good" transistors. From what you say, this is not what you'll be using.

good advice here. These old folks are the ones they throw money at to keep us ahead of Uganda in the technology race...

Its a very minimal class A audio amp, putting out 0.25 - 1w, depending on p su V and LS R. I'm trying to work out if I can get away with nothing more t han the darling pair, or more current gain is needed, and what range of usa ble biasing resistors I can use.

My best guess at min hfe (95%) was 100 for the driver, 15 or 20 for the pow er tr. But I'm rusty these days, and cant remember how much it drops near t he rail. Of course some power trs can give just 3, but I'm accepting 5% of assembled circuits failing to meet spec or work ok. I also can't remember t ypical rf tranny gain.

A 3 stage version of a darlington could solve the low gain issue if it work ed. Would it though? Never tried it, I suspect current leakage might be a p roblem. 3 stage sziklai even better - but npns are far more numerous.

NT

It'll work, but be limited, R_in would be pretty low. It would be a lot more versatile if I could have a lot more gain. Specifying a list of ok transistors is unrealistic though.

Yup. God know what you get from 1970 soviet geraniums, but 5% product failures is acceptable.

NT

I wouldn't count on any better than 35 for each unit but seeing that you are doling a Darlington the out come is going to be up there around at least 500 or more after you over come the other problems and that is the fact that you'll have a large drop from the rail if you do emitter follower or if you do common emitter you get high saturation voltage problems.

you'll lose at least 1.8 volts off the top.

If that is going to be a problem then instead of a Darlington how about a inverter drive config and do common emitter?

Jamie

stage where I need to know the minimum hfe we can count on with 95% confidence.

The darlington pair consists of a small random npn followed by a 2nd tr that must dissipate 1w in hot weather with a small crude heatsink. We can eliminate high voltage trs from the picture.

A few sanity questions... How many of these are gonna be built? Are they built by people who have the equipment and knowledge to identify some random (s)crap transistor and get it wired in correctly? Do you not care about the maximum leakage for the pair? Where's the energy to run it come from?

As for your transistor, I'd build test equipment into the design. "Short this to that. If the output is between A and B, you're good to go." If the builder has no way to tell A from B, you're pissing upwind.

Here in the first world, in first class industrial environments, it's not uncommon for tunnel vision to focus experienced engineers on one tiny aspect of the project and completely miss the point of the exercise.

Depending on how "third world" your destination is, the money you saved by not using a real part and class B might be overshadowed by the energy cost to run it in the first hour.

Sometimes, high-tech is cheaper than scrounge-tech.

The driver tr's c goes to the +supply rather than to the 2nd collector to eliminate that problem. Which I suppose makes it not quite a darlington any more.

What sort of invertor did you have in mind?

NT

A Lot

if not they wont be able to build it. It only takes 2 resistors and a LED to identify transistor type/pinout

No, 0.1A would be acceptable so no problem. Bias is self adjusting.

:) its mains power only for this one, for obvious reasons

I like that idea. I think it can be done not by adding parts, but by powering up before its fully built.

I hear you. The points are:

1. to build saleable goods from junk
2. to build as many as possible where parts supply is limited
3. for people to learn elec eng skills along the way

Minimising part count is very desirable. I'm not short of more standard amplifiers.

I do understand class A :) Analog amplifiers are my thing.

NT

The circuit should be designed such that it'll work at the minimum combined hfe that can exist.

Look up "suicide bias" >:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 

ill be built from random electronic scrap. There's a darlington pair class A

ail as it can. Peak tr current is anything upto 1A.

ruggling to answer that question.

Your requirements and mine are rather different. If I go with 95% working w ell on 2 transistors, more income is made than 100% working right with 3 tr ansistors. I've got more conservative designs, this one's about optimising return where resources are scarce. And its a circuit that should be buildab le for decades.

Bias is fed from the power tr's collector, I'm sure that's stable enough.

I've been looking at some of the circuits on your site. Some interesting th ings there.

cheers, NT

Is this close to what you're working on? I forgot but I think it has something to do with head phone amp? This will operate from a wide range of voltages.

Version 4 SHEET 1 880 680 WIRE -240 -32 -480 -32 WIRE -144 -32 -240 -32 WIRE 0 -32 -144 -32 WIRE 144 -32 0 -32 WIRE -144 -16 -144 -32 WIRE -480 64 -480 48 WIRE -240 80 -240 -32 WIRE 144 80 144 -32 WIRE 144 80 128 80 WIRE 240 80 144 80 WIRE -560 128 -704 128 WIRE -480 128 -496 128 WIRE -144 128 -144 64 WIRE 64 128 -144 128 WIRE -704 176 -704 128 WIRE -480 176 -480 128 WIRE -144 192 -144 128 WIRE 240 192 240 160 WIRE 416 192 240 192 WIRE 0 208 0 -32 WIRE -240 240 -240 160 WIRE -240 240 -336 240 WIRE -208 240 -240 240 WIRE 128 240 128 176 WIRE 176 240 128 240 WIRE -240 288 -240 240 WIRE -704 304 -704 256 WIRE -336 320 -336 304 WIRE -144 320 -144 288 WIRE -80 320 -144 320 WIRE 0 320 0 304 WIRE 0 320 -80 320 WIRE 240 320 240 288 WIRE -80 336 -80 320 WIRE 64 368 64 256 WIRE 192 368 64 368 WIRE 240 368 192 368 WIRE 416 368 416 192 WIRE 416 368 320 368 WIRE -240 400 -240 368 WIRE -480 480 -480 256 WIRE 192 480 192 368 WIRE 192 480 -480 480 FLAG 240 320 0 FLAG 128 320 0 FLAG -480 64 0 FLAG -704 304 0 FLAG -80 416 0 FLAG 64 448 0 FLAG -336 320 0 FLAG -240 400 0 SYMBOL ind 224 64 R0 SYMATTR InstName L1

SYMATTR SpiceLine Rser=32 SYMBOL npn 176 192 R0 SYMATTR InstName Q1 SYMATTR Value 2N2222 SYMBOL npn 64 80 R0 SYMATTR InstName Q2 SYMATTR Value 2N3904 SYMBOL res 112 224 R0 SYMATTR InstName R1 SYMATTR Value 100k SYMBOL voltage -480 -48 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 6 SYMBOL Misc\\signal -704 160 R0 WINDOW 123 0 0 Left 2 WINDOW 39 24 132 Left 2 SYMATTR InstName V2 SYMATTR Value SINE(0 .01 100hz) SYMBOL npn -208 192 R0 WINDOW 0 63 20 Left 2 WINDOW 3 46 47 Left 2 SYMATTR InstName Q4 SYMATTR Value 2N3904 SYMBOL res -160 -32 R0 SYMATTR InstName R2 SYMATTR Value 56k SYMBOL cap -496 112 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C2

SYMBOL npn 64 208 M0 SYMATTR InstName Q3 SYMATTR Value 2N3904 SYMBOL res -96 320 R0 SYMATTR InstName R4 SYMATTR Value 1k SYMBOL res 48 352 R0 SYMATTR InstName R5 SYMATTR Value 50k SYMBOL res 336 352 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R6 SYMATTR Value 100k SYMBOL res -496 160 R0 SYMATTR InstName R7 SYMATTR Value 100 SYMBOL res -256 64 R0 SYMATTR InstName R10 SYMATTR Value 47k SYMBOL cap -352 240 R0 SYMATTR InstName C4

SYMBOL res -256 272 R0 SYMATTR InstName R9 SYMATTR Value 9k TEXT -512 -88 Left 2 !.tran 0 1 0 10u startup TEXT 200 56 Left 2 ;Head Budds?

Jamie

(snip disaster)

But not very well over a range of frequencies. Input impedance is about

160 ohms at 5kHz. Gain goes from 40 to 80 dB from 100Hz to 5kHz and it begins clipping badly somewhere between.

snip spice

Well... I've just installed ngspice on this linux pc, but its not recognising an .asc file, and I don't know how to run it :/.

Here we go... so far:

+----+---------+---o +5v | | | | LS | | | | +------ ( ---+ | | | | | R1 | | _|_ | _|c | ___ C2 2000 uF

-------| |-----+-----(_) | | Cin | |e _|c | in R2 +--(_) tr2 | | |e |

---------------+------------+---------+---o 0v

Its a small amp for mains powered pc speakers.

NT

I'd be interested to see a jpg, can't open spice stuff atm. I knew win98 well, but the workings of these linuxes are a bit alien to me.

NT

Install LTSpice and you should be able to see it.

How about having them build one sample with each "batch" of parts. If that's good, build the rest? Perhaps a different BOM for the test article to throw in some margin?

Where would I get that, thingified to work on this system? Nothing I've installed on this runs.

NT