LED reference current source

It's got aerogel in it. I don't have a lot of room so it's pretty nice. A 2" thick slab of styrofoam might be even better if the space is available.

I guess you can nest ovens and tweak heat flows until it doesn't matter anymore, but you still have to live with hysteresis unless you have uninterrupted power. Unlike your nice crystal, a lot of parts are not happiest at 60°C or 75°C.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

weak

itivity will be

AL

wo

citance current

nd the

ves the stable

stor makes for a

ng fast enough/low

d an LM4140.

better reference

h being a

iest Way Of

ur is hotly

t Flame War Award.

nce.

of you want.

own

rward

s of

n of

ently, and

uite

an

on. =A0It

tance or

ects

ive

timer.

rs,

ike

And you wanted to stick with Imogen "Suggs"?

didn't think much of Helena Bonham Carter, but didn't bother mentioning Imelda Staunton at all - she's a reliable actress, but not exactly memorable.

The Royal Shakespeare Company's "Midsummer Night's Dream" directed by Peter Brook went around the world in the early 1970's. I saw it in Southampton (of all places). Alan Howard was already desperately bored with the Theseus/Oberon part even then and played it very strangely.

%281970%29

-- Bill Sloman, Nijmegen

around

In the stuff i have seen or worked with it was a dashed line box with a label. It seems to get the message across. Should i go look up applicable symbols in IEEE 315?

?-)

I took apart the OCXO from an old HP Nixie tube counter that arrived horribly smashed. They used something a lot like that for the insulation.

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
845-480-2058

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

What's it blown with, xenon? Good foam all has about the thermal conductivity of air, i.e. 0.026 W/m/K at room temperature.

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
845-480-2058

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

I have a 5245L, a 5248 and these: Hewlett-Packard 5223 Electronic Counter Hewlett-Packard 5233 Electronic Counter Hewlett-Packard 5325B Universal Frequency Counter Hewlett-Packard 5532A Frequency Counter

Only the 5245L has a good timebase, and I want to repair it. I first used one in '73 in Alaska.

is some of the equpiment that I need to check out, after over five years in storage.

Actually it is only two more parts. And the BJT can be a cheapie since capacitance wouldn't be an issue.

As Phil indicated, the bandgap is a noisier solution.

Depletion fet ISS is not very accurate. The impedance would probably be higher, but the actual current would not be as well controlled.

Why? You can have a solid metal external box which will pretty much eliminates any thermal gradients, good (foam) insulation around the OCXO itself, so the thermal time constant can be long - an hour would be nice, but you'd probably have to add thermal mass to get that - then a good PID controller with a thermistor right on the OCXO which should stabilise the OCXO temperature to ten or twenty microdegrees. A platinum resistance sensor would be even nicer, but you'd have to AC- excite it to get down to the microdegree level.

People have been doing that well for forty years now, with well insulated objects. Two nested crummy ovens aren't going to be as good as one good one.

-- Bill Sloman, Nijmegen

.

Then you can use a Peltier junction, but that degrades your thermal time constant - the thermal resistance across the Peltier junction isn't all that high and the exhaust side of the junction has to be at room temperature (or pretty close).

-- Bill Sloman, Nijmegen

It was a 5245L iirc, with the good timebase. The idiot seller shipped it in two layers of bubble wrap inside a single cardboard box. It arrived with both handles broken off, the faceplate cracked, and gross internal damage.

It was interesting taking it apart though--to control the Nixies, they used neon bulbs shining on CdSe photoresistors. I kept a few of the modules.

And if you'd like to get rid of one of those Boonton 92s when you're done, I'd be interested.

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
845-480-2058

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

I'm charging a 5 pF cap at 5 mA, to get a 1 v/ns ramp, so I do need a low capacitance PNP. A BCX71 has wonderful beta, but about 6 pF at the collector. A serious GHz PNP might have a beta of 20. The MMBTH81 is a workable compromise.

The depletion fet would reduce the power supply sensitivity some, which is currently my biggest error. I'll just have to specify a stable +5, I guess.

--

John Larkin                  Highland Technology Inc
www.highlandtechnology.com   jlarkin at highlandtechnology dot com   

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME  analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

It becomes a second-order lowpass on thermal transients. The inner oven sees almost constant heat flow to the world.

You can have a solid metal external box which will pretty much

How about two good ones?

The best, most expensive XOs are SC-cut crystals in double ovens.

Why AC on the RTD? Chopper opamps have nV/degC offsets, and even that is inside the box.

--

John Larkin                  Highland Technology Inc
www.highlandtechnology.com   jlarkin at highlandtechnology dot com   

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME  analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

I actually don't remember doing that, but once you found the vending machine, N,D,D,D would get you out.

ne.

Should be unnecessary.

History. Customers buy what they've always bought.

Thermocouple voltage on the RTD leads. They should be stable inside a well thermostatted box, but micro-degree control with an RTD would mean that very small voltage shifts would matter.

-- Bill Sloman, Nijmegen

,

hane.

And of course, 1/f noise. Micro-degree temperature controllers are actually limited by the Johnson noise in the sensor resistance.

With AC excitation, you can amplify everything and rely on the demodulator to get rid of the 1/f components.

-- Bill Sloman, Nijmegen

urethane.

There are chopamps nowadays that don't exhibit 1/f noise, e.g. the OPA378, which I've used in laser cavity lockers with excellent performance (10**-10 frequency stability) down to time scales of a few days. They effectively do all that demodulating stuff inside. 1/f noise in the RTD itself is due to conductivity fluctuations, and so won't be improved by AC techniques. Of course their flatband noise is worse, about 20 nV in 1 Hz, so it's a tradeoff. However, a 1-k RTD with a couple of volts' excitation will give you about 600 uV per kelvin, so

20 nV is equivalent to about 33 microkelvins per hertz, which is fine for lots of applications.

Putting all the sensing and loop control electronics inside the controlled volume (except the power driver) gets rid of most of the thermocouple worries, especially in a two-shield system. It's sort of the thermal equivalent of using an op amp plus a voltage reference--if you bias the reference from the op amp output, the reference regulates its own bias current.

The other reason for two-shield thermal control is to get rid of transients, especially spatially nonuniform ones. Making a shield thick doesn't speed up thermal diffusion at all, so making it small and its environment approximately isothermal are how you win. Putting a few zones in the outer box can help a lot in more difficult situations, e.g. where there's variable local dissipation in small confines. I did that recently in a downhole version of the laser locker--six zones in the outer shield, one in the inner.

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
845-480-2058

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

The RTD, the bridge resistors, and the opamp would all be inside the inner oven. All non-RTD tempcos just change the loop gain a tiny amount.

Here's one oven-ish thing I did.

There's a home-made manganin current shunt glued to the block, under the PC board. Minco heater on the bottom, thermistor surface-mount on top. I used "shunt" manganin, whose parabolic TC curve is flat at the operating temperature.

It goes in a metal box with some styrafoam, which seemed to help a little in this case. I did a little XO like this, too. There must be some pictures around here somewhere...

The most interesting problem was avoiding magnetic pickup and eddy current effects in the block.

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation

The loop has some bandwidth either way, and there's Johnson noise in that bandwidth. Modulation does get rid of 1/f noise in the opamp, but a chopper amp doesn't have 1/f noise.

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation

Practically necessary items such as the electrical conductors and robust mounting hardware pump disturbances into the controlled zone. If you can heat sink them to an external, mostly constant, temperature then the inner zone(s) can be much better controlled, particularly dynamically. It can be improved maybe an order of magnitude with more exotic materials, but a nested oven can give at least two orders of magnitude improvement easily. Reducing an external variation from (say) -20C to + 50C to 1mK or 100uK is about 5-6 orders of magnitude, if you need 10uK that's 7 orders of magntude.

A simple-simon PID controller will in theory eliminate all static error at the sensor, at least as far as the controller knows, but will have a constant offset if you ramp an external temperature disturbance. More sophisticated types of controller can do better, of course, but there are other problems.