Looking for a 10mA constant current source.

If you don't need a floating source, you could perhaps use a TL431. Use a 1% 249 ohm resistor from ref to anode, ground the anode, and connect your diode between ref and cathode. Then, use a resistor from Vcc to the cathode to get the current into the ballpark + 10%. You can read the voltage across the diode at the cathode.

The TL431 is temperature compensated.

--
Regards,
  Bob Monsen

If a little knowledge is dangerous, where is the man who has
so much as to be out of danger?
                                  Thomas Henry Huxley, 1877

Most are

tol of 10%?

15 to 30 is a very small range, is this part of 10% or in addition too?

Add a balanced bridge or switch to a constant voltage source and use bipolars to add drive current.

Any parts count limitations?

know.

Sounds like all you need is a single resistor. Add an SSR in series if you want to chop it not too fast.

John

[snip]

Is current FORWARD or REVERSE thru the noise diode? Is 28V supply

+28V or -28V?

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

If the 28 volts is stable to 0.1 volt, and the zener changes by less than 0.1 volt over temp (probably less in real life) that's +-0.2 volts out of 28, or +-0.7 per cent max.

You've got to calibrate the zener's enr anyhow.

John

Hi, I'm looking for a constant current source that must

a) Be floating.

b) Provide 10mA - exact value is not too critical.

c) Have as low a temperature coefficient as possible. Operating range

15-30 deg C.

It is to be used to provide current to an avalanche noise diode. The input voltage will be 28V, the diode will drop about 8 V, so I need to drop 20V or so across a resistor and constant current source. The input voltage will be pretty stable (28 +/- 0.2V), so it does not need to work over a wide range.

I was thinking about an LM334 with the temperature compensation diode added. The only problem is 10mA is on its limit, and might cause significant internal heating, so it will be difficult to keep the temperature compensation diode at the same temperature as the regulator IC. Why the hell NI did not build the diode into the package I do not know.

The 28V input is actually switched, the frequency of which I do not know. But I don't think it will be too high, and so the rise/fall time of the constant current source should not be an issue.

Any better suggestions?

HP / Agilent's 346B noise source

is what I am trying to copy, but a look at the manual (page 36 of 81)

shows a degree of complexity I can't really understand. They use the 28V to make an oscillator, regulate that and use that to drive the constant current source. I can't understand the logic for just not using the 28V.

Thinking about it, there is no need for it to be floating.

Yes, as long as it stays stable. Absolute number is not critcal, but stability is.

The exact current is not critical. It will affect the operation of the diode and change the noise output, but the noise output will be calibrated.

What I don't want is the current changing once the unit has been calibrated, so the current must remain fixed with temperature.

I'm not sure what you mean here. I don't have a constant voltage source

- only +28V pulses, but I can use a regulator internally.

This will be connected to an HP 8970A noise figure meter, which puts out 28 +/- 0.1 V according to the spec. Knowing HP, it will be a lot better than +/- 0.1 V, but I assumed +/- 0.15V as there will be drops across the cable/connectors. Not much I agree at 10mA over a couple of metres.

I guess I should find out how much current I can take from that 28V supply. I need to about 10mA for the diode, but whether the noise meter can source 20 mA, 50 mA or 100mA I do not know.

I was hoping for somthing I can stick on a bit of stripboard no bigger than about 50 x 70 mm. Would rather avoiding having to make a PCB, but it would not be the end of the world. Hence componenets only avialable in SM would not be an ideal choice.

Why would you think this? Remember the cardinal engineering law - KISS.

HP certinaly do not use a simple resistor in their noise sources.

And your point is?

Yes.

Over your given temperature range it won't change squat.

And this means???

How you would get 1% error is beyond me, but if you want to futz around with exotic solutions, have at it.

Jim

>

A resistor is certianly not a bad idea, but I don't think it is accurate enough. HP certinaly do not use a simple resistor in their noise sources.

There are reasons for this.

1) Diode noise output changes with current - both in terms of amount and the frequency spectrum.

2) Diode voltage changes with temperature.

3) The output voltage of the noise meter is speced as 28 +/- 0.1 V.

4) A 10% change in current will typically cause a 0.5dB change in Excess noise ratio (ENR). But the ENR will need to be measured to much greater accuracy than that.

I just looked at Microsemi and they have one at 10mA. But there are no specs for tempearature stability.

You can make a good current reference out of the LM185. The typical performance on temperature stability is 30ppm calculated over the entire operating range, but these parts are optimized for nearly zero tempco in the vicinity of 25oC, and should be very much less over 15o-30o like

3ppm average or less. View in a fixed-width font such as Courier. 28V>--|>|-------+--------------+ | | LM185 | [120] _|_/ | // \\ ------------+ --- | | e | |/ | --| | | |\\ | e c | |/ | +-------| | | |\\ | | c | [270K] | | | ----+ | | | [R] R~120 ohm | | | | | | | - | ^ ND | | +------+-------+ | gnd

I'm not sure if you would say I need it floating or not now. First I thought yes, then no. But I think I was right the first time.

But I don't think the TL431 can be used.

I need to ground (0V) the anode of my noise diode. Anything else can be grounded too if need be. The cathode must be connected to the 28V supply via the current control element.

Would you call that floating or not? I think it does need to be floating.

The data sheet on the TL431 does show a precision current sink, but that would require that I lift the anode lead of my noise diode off of the 0V line. That is not possible

I'm now thinking about perhaps using the TL431 or similar to regulate the 28V +/- 0.1 V to say 24.000 V, then use a resistor in parallel with a thermistor.

Pehaps this is why HP use an oscillator and rectify that. At first I thought it was a bit odd, but I must admit I can't think of anthiing that will be accurate.

I did wonder about measuring the temperature coefficient of the diode, which should be linear over the small range of interest (only around room temperature). Then using a series resitors, with a thermistor fitted somewhere such that the current is kept constant at the two temperature extreams (15 and 30 deg C). Hopefully it would not very too much in between, although I would have to model it to be sure. A bit messy though.

So put two LM334/resistor/diode 5 mA circuits in parallel.

Ed

Why the hell NI did not build the diode into the package I do not know.

The diode is reverse biased and will break down around 10V. The supply is +28V

Thanks. I'll look at that in some detail later, but the spec seems good if that can be achieved.

Yes, if it is not going to compromise performance.

Well HP have been making noise figure meters and noise sources for ages. I think they know a thing or two about them. The HP 436B

is the noise source most suited to the meter I have. The operation manual for that

certainly shows a constant current source (page 36 of the pdf). I suspect they would use a resistor if they felt it adequate.

That is specified (see page 9 in the pdf) for operation over the range 0 to 55 deg C, which is wider than what I need, but note that the spec is only guaranteed at the calibration temperature - *not* over the full range.

These things are quite critical. The room temperature has to be entered into the instrument, as it will affect the amount of noise the attenuator generates on the output (noise power = k T B).

Assuming squat means an insigificant amount, then I would agree with you . Hence it will change a significant amount.

It can be anywhere from 27.9 to 28.1.

Calibration data on a diode will show resolution of 0.01 dB.

I was considering that, athough it will need 3, since they say self heating is a problem above 4mA.

I think you have the device wrong there - the LM185 is a 2 terminal device and you seem to be showing 3. I assume the transistors are both PNP.

Hello Fred. What does the 'R~120 ohm' do?

Thanks.

--
Tony Williams.