AD584J: strange behavior

[siealex]

Recently I purchased a reference module with AD584J from Aliexpress. 2.5 V, 5 V, 7.5 V work perfectly, but 10 V behaves strangely. If I set the jumper to 2.5 V, then switch it immediately to 10 V, the output voltage drops down to 2.1..2.2 V, then slowly (within 10..30 seconds) crawls up to 4..5 V and then jumps to 10.00 V (screenshot 1, 5 s/div!).
To get 10 V instantly, I need to set 5 V (or 7.5 V), then switch to 10 V (screenshot 2, also 5 s/div). Is it normal for this chip? The module is powered off a 12 V 0.25 A transformer (not switching) power supply.

[Andreas]

The module is powered off a 12 V 0.25 A transformer (not switching) power supply.

Hello,

Stabilized 12 V or unregulated?
The data sheet says nominal 15 V. (minimum VOut + 2.5V)

with best regards

Andreas

[siealex]

Stabilized, in fact slightly higher than 12 V (~12.5 V).

[magic]

Is the board fitted with a noise reduction capacitor between CAP and V_BG? Or an unusually large output capacitor? Maybe something is wrong with those caps?

Anything weird at the STROBE pin?

I don't have this chip to test. But...

sudden removal of the 10V-2.5V jumper inserts the internal /4 divider into the feedback path and causes 2.5V tap voltage to dip below 1V if output is currently at 2.5V. This may have an effect of reducing bias on Q3,Q4 and slowing down the regulation loop, which may perhaps take a moment to recover, particularly if the CAP node is abnormally loaded by a defective capacitor, board leakage or whatever.

[siealex]

The board is fitted with three capacitors: one electrolytic 470 μF 25 V at the input (connected to the input socket via a 1N4007), one 0.1 μF connected to the first one in parallel and another 0.1 μF connected to pins 6 and 7. The strobe pin (5) is not connected anywhere.

[siealex]

sudden removal of the 10V-2.5V jumper inserts the internal /4 divider into the feedback path and causes 2.5V tap voltage to dip below 1V

Yes, the output voltage dips for a very short time, but then returns to 2.5 V and slowly (20 seconds!) crawls up to approx. 4 V, then suddenly jumps to 10 V.

which may perhaps take a moment to recover

20 seconds? And why does it jump to 10 V immediately if I touch (even slightly) the 5 V or 7.5 V pins with the jumper?

[iMo]

AD584J from Aliexpress

The question is what chip you have really got..

[magic]

AD584J from Aliexpress

The question is what chip you have really got..

I doubt there are any pin compatible substitutes.
Maybe if those boards start to sell in millions on AliExpress somebody will make a clone in China

sudden removal of the 10V-2.5V jumper inserts the internal /4 divider into the feedback path and causes 2.5V tap voltage to dip below 1V

Yes, the output voltage dips for a very short time, but then returns to 2.5 V and slowly (20 seconds!) crawls up to approx. 4 V, then suddenly jumps to 10 V.

Not the output voltage but the 2.5V pin voltage. It dips below 1V, at which point Q5 can't feed base current into Q3,Q4 at all. It frankly looks like the regulation loop completely shuts down and only base current of the Q13,Q14 darlington charges the noise filter capacitor. This may take a while.

And why does it jump to 10 V immediately if I touch (even slightly) the 5 V or 7.5 V pins with the jumper?

Doing so increases 2.5V pin voltage.
It also applies a positive voltage step to V_BG, which is AC coupled into CAP, which is maybe enough positive feedback to unlock the chip. Note that the output is simply the CAP voltage buffered by a few emitter followers.

Try without the CAP-BG capacitor. You probably don't care about noise above 100Hz anyway.

[iMo]

Also doublecheck the pcb has been cleanup , sometimes they do not clean the pcb and the residual resin/flux between the pads may cause parasitic leakages..

[TizianoHV]

Could be an issue with one of the caps values:

See https://www.markhennessy.co.uk/ad584_references/ (Unit #2):

"But note the capacitor between pins 6 and 7. The datasheet advises a maximum value here of 100n. This needs a closer look..."

"I changed the capacitor from 330n to 100n. This shortened the time the unit spent drifting, but it didn't reduce the occurance of the slow drift. So I changed the capacitor to 10n, and this meant that the "drift period" was reduced to less than half a second."

[siealex]

"But note the capacitor between pins 6 and 7. The datasheet advises a maximum value here of 100n. This needs a closer look..."

On my board it's 100n. Tomorrow I'll try to find a smaller one.

[siealex]

The question is what chip you have really got..

Markings on the chip:
▶️ AD
584JH
9338△
The chip is in a metal 8-pin circular case.

[magic]

For checking DMMs this capacitor isn't necessary or useful at all. A smaller value is even more useless than a larger one, as shown in the datasheet.

Sufficiently bad leakage current from those pins will affect accuracy, but I don't know how bad is bad enough.
I have previously had bad experience with MAX6350 and X7R noise filtering capacitor

Markings can be erased and changed. But if it quacks like AD584 it should be AD584; besides Maxim's second source there seem to be no other chips with exactly the same function. Probably harvested from e-waste.

[siealex]

Replacing that capacitor by a 10 nF almost solved the issue, now the switch time is less than 0.5 s.