Using an AD584 for a 10.0 Volt 1.0 Amp DC Supply

[eevjohn]

Hi all,

I built a DC Supply to give 2.5, 5.0, 10.0 Volts.

I used an AD584JH with a Darlington current boost as per the classic AD Application Notes.
The AD suggested bypass device 2N6040 is a little harder to get, so I used a BDX54 and TIP2955, as a triple Darlington to safely generate over 1.0 amps with ease, and no loading on the 10mA output of the AD584.
For a DC 15V supply I used a 7815 and TIP2955 bypass device to generate 15 V at 1.5 amp without any voltage sag.
I run the whole unit off an external 20V DC 5 amp DC supply.

I came to grief because, although I had 0.1 MFD ceramic bypass caps everywhere, the AD584 was still oscillating at about 250 KHz.
The AD584 was ok at low current outputs of 10mA or less, but operated in a funny mode for higher currents.

Solution was to put a 220MFD to ground on the DC supply to the AD584 on pin 8, which is actually the base of the Darlington chain.
As per the AD Notes, this may not help in all cases; it depends on layout and the loop gain of AD584 and boost Darlington.
Experiment, but have a CRO on the AD584 output.

Now it all works like a charm.

[SiliconWizard]

OK, thanks for sharing your experience. A small attached schematic would have been nice to illustrate your project.
 

[eevjohn]

Here you go.
Roughie cct.

[Ysjoelfir]

Nice project, looks quite simmilar to the thing I built with the AD584 some time ago. I like the 584, pretty versatile thing and not to expensive.
Would you mind showing an inside photograph? I would be interested how you built it.

[Dave]

How exactly is this supposed to provide an accurate current? A reference is not supposed to change its value with respect to load resistance (within reasonable limits).

[001]

How exactly is this supposed to provide an accurate current? A reference is not supposed to change its value with respect to load resistance (within reasonable limits).

Yea

It is not constant current source
It is linear supply only

[eevjohn]

Hi there,

It is a voltage source with very low source resistance.
It is very similar to normal voltage regulators, except its internal resistance is very very low, and the output voltage is maintained accurately by the AD584 (and bypass Darlington).

This unit gives about 3.0 amps before the output voltage falls a few millivolts, even on 10.0 Volts, but some if not all of that is in lead and internal interconnect resistance.
It was designed with reserve capacity.
The unit was only designed for a max of 1.0 amp, which allows for use of normal rating internal switches and "3 amp hookup wire".
There are some overall design constraints, mainly to stay within the current spec of the AD584 of 10 mA max, and to have a very low source resistance power supply which stays above 14.5 volts for the AD584.
Output noise is restrained by 0.1 caps.
The 10 mA AD584 output is "boosted" by the high Hfe of the Darlington string to have ample reserve.
Analog Devices App Notes specified a bypass 2N6040 with a Hfe of 2500 average, which is why I have used a triple Darlington.

[SiliconWizard]

It is a precision voltage source with very low source resistance, so a precise and selected 10 ohm resistor (in the box) will give an accurate 1.0 amp to an ammeter connected to "current" terminals.

Yeah, I suppose the "current" terminals are mainly made to serve as references to check or calibrate a multimeter, not to serve as a general-purpose current source. Of course, always make sure the ammeter's shunt resistor in the range you're using is negligible (within the accuracy you're targetting) compared to your own reference series resistors.

[eevjohn]

Very useful device.

[Dave]

It is a precision voltage source with very low source resistance, so a precise and selected 10 ohm resistor (in the box) will give an accurate 1.0 amp to an ammeter connected to "current" terminals.

A random meter will have let's say 300mR resistance on the current terminals. 10V / 10.3R = 971mA
Hardly an accurate 1.0A source.

You need a regulation loop that adjusts the output voltage so the current flowing through the circuit is constant. Look at the circuit of the electronic loads everyone is building on this forum, except you want to optimize it for long-term stability instead of speed.

[eevjohn]

I have today modified all my posts in this thread to remove words "Precision", "1%", "Calibrator", and a few attachments which are now superfluous.
It is still a useful circuit, and the original intent of the thread is still valid, ie, be cautious about instability.

[Ysjoelfir]

I have today modified all my posts in this thread to remove words "Precision", "1%", "Calibrator", and a few attachments which are now superfluous.
It is still a useful circuit, and the original intent of the thread is still valid, ie, be cautious about instability.

That seems a bit childish to me.
The comments others here are giving may sound a bit harsh, but hey, we're in an engineering forum.
I would allow myself to assume, that most engineers aren't the outgoing, overly kind type of person but more the fact oriented, down to earch ones who like to present a meaning, bach that up with facts and that without giving too much attention to sounding especially nice.
This can sometimes lead to the impression that people are being rough and unfriendly, but most people here are just neutral while posting without any special attitude.

I personally like the project but I know about the shortcomings of this kind of build (experienced them myself, as I said, I build something quite simmilar some time ago), so I keep that in mind and like to know about the oppinions and tips others give me to improve my designs.