Transistor tester

LM2594 simple switchers, right. (Lysdexia strikes again.) The caps are polymer aluminums, and there are a bunch of bypasses strewn about. The coils are Murata toroids, and there's a piece of copper-clad Kapton soldered over the whole SMPS section, so the supplies are surprisingly clean. (I wouldn't use them barefoot like that in a real instrument, but this thing has a really small bandwidth.) I've become sort of fond of running gizmos off a laptop power brick--you have to use a rail splitter or a SMPS to get split supplies, but on the plus side, the ground loops pretty much go away, and there's always enough supply current available.

Quite right. Those are in the test fixture--most of the transistors I've tried start oscillating someplace between 10 uA and 100 uA, which is a nuisance. Fortunately the effect is far from subtle--the log conformance error goes from e.g. 14 mV to 3V in one step.

A bead or two, plus a well-placed finger, makes them work better, but it would be nice to have some properly-engineered test fixtures. I haven't thought enough about how to make the same fixture have low enough capacitance to work well at 100 pA while still keeping the transistors stable at 10 mA. Suggestions are very welcome.

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 USA 
+1 845 480 2058 

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

You are one of the few (as in, exactly one) physicist that I've known who can design circuits. DGMS about chemists.

This was designed by chemists, to run in UHV. The opamps run at about

140C, so they added resistors to kill gate leakage. I figure they gave up over 30 dB s/n.

Funny, modern electronics was largely invented by physicists, at the MIT RadLab. Different days.

It isn't just scientists that fall on their faces in that sort of fashion. I've been transferring this relatively simple grating spectrometer to a contract engineering company. Their in-house EE redid the front end (which was fine) but didn't calculate the noise floor (not so fine). He also left out the antialiasing filter on the ADC. They got in a consultant to redesign it.

This is an example of a larger problem that I see all the time--so many folks refuse to do noise and error budgets, even though their livelihood is riding on the outcome. Weird.

Soft solder in UHV???

Physicist circuit designers aren't that rare--I've known several good ones. Jan Hall and Paul Horowitz are probably the best.

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 USA 
+1 845 480 2058 

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

Fact is, the minority of EE graduates are decent circuit designers. It's a weird skill.

[...]

In this case Phil can lowpass the heck out of it so that any xx kHz noise is literally gone. I use switchers a lot in noise-critical circuits, even in radios. More efficient, and we all have to be at least a little green :-)

--
Regards, Joerg 

http://www.analogconsultants.com/

The nice thing is that they are very cheap and usually rated for 100% duty-cycle at a reasonable power level, many remain relatively cool at

50% load.

That's good. I had situations where I noticed a slight noise increase, only to discover that the FM radio has miraculously gone "out".

One really nice method to calm things down are multi-hole ferrite plates. They are made with 9 holes, 15 holes, and so on, for D-Sub connectors. There is no low that says that all holes must be used. Although with the current administration one never knows ...

--
Regards, Joerg 

http://www.analogconsultants.com/

Several kinds of them. Collect the entire set by buying enough equipment!

At some point, drafters were ordered to stop, I think. (I was trained at Tek as an electronics drafter around 1980 or so.)

Jon

Got any doodle pics? Here's one:

Do you think they drew that blue scope waveform by hand, graticule and all?

The originals would be valuable, like old Disney cells.

HP was staid and boring, schematics and the gear itself.

hil > >> might plug in his transistors. The elements are recognisable enoug h - I've > >> put together logging and anti-logging circuits (though not re cently) - but > >> it's not the most intelligible circuit diagram I've ever seen.

ics. > >> It's not a particularly horrible example - I been more peeved by neater

in pulsed > >> laser spectroscopy? Review of Scientific Instruments, 67

3763-4 (1996)

I've got opinions - some negative, some positive - about lots of subjects. I figure that I'm entitled to have negative feelings about a circuit diagra m that purports to be for a transistor tester where I can't immediately see where the transistor/transistors are intended to plug in.

A clue would have been helpful ... I'm not interested enough to spend an ho ur crawling around the circuit working out exactly where they'd have to go, bearing in mind that Phil's handwriting was unclear, and the image contras t sub-optimal.

I've got my own horror stories about chemists and electronics, but they don 't suffer from the "electronics is just a sub-set of physics" hubris which afflicts too many physicists - though not Phil Hobbs. The problem with his circuit diagram was just that it was insufficiently intelligible.

g

But if it worked, would they have cared? As a graduate student, I was propo sing to get a stable voltage by sinking a bunch of ordinary zeners in an ic e bath. My boss sent me over to the Electronic Engineering Department to ha ve my design reviewed, where it was suggested that a reference diode would be a more elegant solution, so I revised the design. From what I now now ab out ice baths, the ice bath would have done a much better job ...

With a lot of help from the UK, and Germany.Hans Ferdinand Mayer - a German physicist - told the British a lot of what the Germans were doing in 1939, and Reginald Victor Jones passed on quite a bit of it to MIT RadLab, along with some British contributions like the cavity magnetron.

Of course the bulk of the work was done by regular electronic engineers lik e Alan Dower Blumlein and his team at EMI Central Research and the people w orking on radar, most of whom seem to have ended up at Royal Radar Establis hment when it was put together in 1953.

I actually worked for one of the UK physicists (1969- 1971) - Alan Butement

who invented the proximity fuse. He also developed the back-pack radio that worked so very badly at Arnhem - a little bit of engineering training migh t not have gone amiss.

Not different enough.

--
Bill Sloman, Sydney

pA

is

has

I did.

e

"E" and "C" are on one sheet, "B" on the other. "B" is on the left of it's sheet, which is a non-intuitive place to put an output. Not user-friendly.

It does make sense once you tell us what you had in mind, but it completely escaped me on first and second inspections. I may be getting stupid in my old age, but I'm probably still above average ...

ger > > > >>>> box!

ter

er - > > 32mW isn't much. Farnell stocks them for $A0.44 in small quantitie s

nt

ably > > grew out of the -3046 and -3096 that started off with RCA.

ell, > > but never found an excuse to use one.

Only the HFA3046 has a differential pair with a common emitter, and that's the only pair in the whole menagerie that has a specified matching. 5mV wor st case for input offset voltage isn't great either.

You could probably buy bare die if you bought enough of them. Getting acces s to a wire bonder might pose the same kind of problem. When I was younger, making thick film hybrids was a kind of cottage industry, and the people t hat did it could bond wires to bare chips and package the result to give yo u a product which could survive humid atmospheres, but surface mount kille d that particular industry.

We probably ought to be grateful that Intersil can make enough money out of the arrays to sell them at all - transistor arrays are very handy for very specific jobs, but they aren't a mass market product and it's comforting t hat there now seem to be enough specialised applications - none taking as m any as a hundred parts a year - to make it worthwhile for Intersil to make a batch from time to time.

they work brilliantly, as long as you put the right little ferrite bead in the right place. ;) "

hem > > oscillating except by sticking about 33R in series with the base

sen > > ferrite bead would do the same job, rather more expensively.

I once got caught when the inductance of a bead was high enough to matter a t a frequency I cared about - I ended up having to put a resistor in series with it to critically damp the resonance. At 100uA that's unlikely to be a problem.

I've done more or less that with my low distortion variant of the Baxandall class-D oscillator. I'm sure that Barry Gilbert does better, but I've not yet been motivated to work out how.

Jim Thompson probably knows - he wasn't clever enough to work it out for hi mself before Barry did, but Barry has published, and Analog Devices have so ld the circuits for a long time now.

The AD734B is good to 80dB

--
Bill Sloman, Sydney

"ding, ding" (sound of bell ringing) Ahh, I kept looking for a |B> on the o ther sheet to connect to the other...

George H.

ger box!

ter than

s have

er - 32mW isn't much. Farnell stocks them for $A0.44 in small quantities

nt part numbers might be HFA-3101, -3102, -3127, -3128, -3134 and and proba bly grew out of the -3046 and -3096 that started off with RCA.

ell, but never found an excuse to use one.

they work brilliantly, as long as you put the right little ferrite bead in the right place. ;) "

hem oscillating except by sticking about 33R in series with the base connec tion, right up against the base tag, but I can see how a well-chosen ferrit e bead would do the same job, rather more expensively.

It sure looks like it. Check the number "5", for example. On the Y-axis the upper dash is shorter than the lower body, on the X-axis it is of same length or maybe even a bit over. It doesn't even look like they used a stencil, more like handwriting. Calligraphy was still taught at schools back then.

I have to object. The HP-3577 and the HP-3585 are both unrivaled in terms of precision and dynamic range. Some of the pulse-echo noise analyses I did with the 3585 are next to impossible with other analyzers without blowing their inputs or at least distorting the results. But, no doodles in the service manuals.

--
Regards, Joerg 

http://www.analogconsultants.com/

Ditto the 8566B and 7000x series spectrum analyzers. HP only learned how to build scopes about 10 years ago, right around the time Tek forgot.

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

:-)

Problem is, Rigol, Instek and a few others learned that as well and now they are eating Tek's and HP's lunches. I would have never thought I'd buy an Asian scope anytime soon, and then a few years ago I did.

--
Regards, Joerg 

http://www.analogconsultants.com/

Do any of these scopes (including Tek) do proper decimation?

I have not looked in detail at it, but it seems they sample at 2G samples/second or whatever and just discard the majority of the measurements at low sweep rates.

With mine (Instek GDS-2204) the sample rate goes down at lower sweep settings. It is from a generation that only has 25k sample memory. The

30% more expensive Tek had a paltry 4k which would have made serious ultrasound pulse-echo tests nearly impossible.

--
Regards, Joerg 

http://www.analogconsultants.com/

I don't like reading dropbox files, so I didn't look at all. But regarding switchers, there is always the philosophical question as to just how large is an inductor. There is the physical size, and then there is the reach of the time variant EMF it produces.

There is a difference between filters and shielding. Filtering is cheap. Shielding isn't.

They don't seem to low-pass filter so the aliasing is uncontrolled.

I appreciate that it's not easy to implement a digital filter at

2G samples/second, but...

Sometime try looking at a square wave at such a frequency that the nominal sampling frequency is an exact multiple of the square wave frequency- at a very low sweep rate.

--
Not true. 

We used an HP scope to good advantage as a piece of depot-level 
maintenance gear for B-52 ECM equipment which we built in the late 
fifties/early sixties, as I recall. 

Back then there was HP's graticule, deposited on the vacuum side of 
the screen, under the phosphor, which got rid of parallax. 

Genius.