Small doodad; charge CR2032?

Hey y'all.

I just put the question in an image with the circuit diagram, so hey.



Any ideas?

See attached a more simple alarm circuit. No need to recharge the battery, stand-by current = 0
You need only a PNP small signal transistor, one diode and two resistors. :-D

The original test circuit checks and latches a dropout condition on 12V only. The suggested one does not. If you want a cheap and dirty solution, charge the 3V cell with the 5V pin using 10~20mA max current with an LED indicator for current. 3V Coin cell is maybe ok within 10% or 2.7~3.3 and n.g. outside that. so use 5V source on Molex, with AlGaAs Red 5mm LED (anode to +5) then 1.6V drop to series resistor between 3V & 3.4V drop to 10mA or 400mV/10mA= 40Ω consider higher like 68Ω for margin. Then it will charge when PS is on but passes and share power with battery when PSU 12V only fails and run off Batt only when both fail. So this way you get both trickle charge and sense o both V+ for 12V and 5V on Molex. Note any Voltage on 5V line that is > -5V or >+10V may burn out LED but WTH its cheap & dirty.

so connect superbright RED or Yellow (Amber) 5mm to +5 and cathode to 68ohm resistor to CR2032 ( not all are rechargeable)

SunnySkyguy said:

The original test circuit checks and latches a dropout condition on 12V only

Click to expand...

The original test circuit checks and latches a dropout condition for any voltage > 0.7-0.8V if we change the BC857 for a NPN bipolar transistor. I guess that circuit was never tested in advance. :-D

For a more precise detection in the proposed circuit we may add a 7V5 - 9V1 zener diode. However, if a power supply fails, the +12V power supply simply turn to zero, so we get the same result with or without the zener diode…

yes mister_rf I agree,

but I suggest orignal design is adequate for Go-NoGo test of 12V with momentary switch to enable alarm and battery with latch alarm to test for transient dropouts like loose crimp connectors on Molex. Eudoxus, if my assumption is correct, use the orignal design with my suggestion for adding LED and 68Ω to charge 3V cell from 5V detection.

But if you dont care about intermittent dropouts and prefer to measure adequate voltage consider Mr Rf's circuit on left with zener series drop. If you want both, that is easy to combine.

I admit that I forgot the actual transistor #, so I just found one that looked like it. Teach me to be more careful with model #s! I just asked one of our engineers to give me some parts that would work and he gave them to me (he also pointed out a mistake that would have rendered it inoperative ). I'll be back in the office tomorrow briefly, I'll grab the real # then. The circuit does in fact work, except that I don't have a speaker that will work in it. XD All we have are old motherboard speakers and they all require a modulated signal, not just flat DC. However I tested with a DMM and it shows voltage across the speaker when it's supposed to, and doesn't when it isn't supposed to. So it works.

I like the idea of having an LED to show +5V activity while also charging the battery. Good idea. As for frying the diode, if we've got a sustained spike to +10V on the +5V rail a burned diode is the least of our concerns.

90%+ of PSU fails or protection activations will drop +12V (and usually the other rails as well) to ~0V. 50% of remaining failure modes will at least drop +5V to ~0V. That's 95% of possible situations covered; good enough for something cobbled together out of spare parts.


Anyway, going to Guanghua market tomorrow evening and buying a speaker and any other parts I need. So make any more suggestions quickly.

choose 9~24Vdc or so piezo device not a magnetic speaker.

Could you sketch out that battery charge circuit?

Also consult with Engineers on preload and Cap to filter false trigger. Most SMPS need a 10% pre-load for stable operation.

---------- Post added at 18:28 ---------- Previous post was at 18:27 ----------

no sorry thats too simple for you to read my explanation again and hard for me to sketch...
connect superbright RED or Yellow (Amber) 5mm to +5 and cathode to 68ohm resistor to CR2032 ++
+5V ---|>|---/\/\/\--- +3V on coin cell
. . . . . LED

Our SMPS have voltage inside ATX specs between 0% and 110% load except in extreme crossload situations on some cheaper models. And voltage rise-time in the microseconds. Only time I should have "false" trigger is when in standby mode, since I'm not monitoring +5VSB. But a beep during the brief standby tests shouldn't bug me.

I am surprised by lack of quality test equipment for this production test. Ask designers to help.

What visual notification on fancy testers? LED? or PC monitor?
If LED, then use PD to detect edge of LED on to latch alarm. ha...

We have quality test equipment. Chroma just neglected to add a speaker that beeps when voltage drops to zero unexpectedly. And they charge a fortune for custom modifications. So if I'm busy with non-test related stuff and the unit I'm testing fails, all that happens is that the PSU fan stops, the voltage displayed on the DC load bank changes, and maybe a relay clicks. Maybe a fet blows up, but that's rare with our units and I'm not worried about it. Considering the load banks are cooled by 4kRPM Delta howlers it's unlikely I'd hear the relay, or the fan stopping; and due to the placement of the 3-phase outlet in the room and the other stuff in the way, I can't arrange things so I can see my monitor and the load banks at the same time.

So I need something to beep when a PSU fails or shuts down.

Trust me to not be a complete dunce; just because I don't have much experience designing circuits doesn't mean I don't understand SMPS. I probably know more about some high-level aspects than you do, although I wouldn't try to design one.

IN that case if you have an expensive Turnkey test solution if it does ATE and logging without alarm but has visual then it may be easy to interface to that signal or output to a customer alert port. You can always ask ...

---------- Post added at 18:59 ---------- Previous post was at 18:56 ----------

IN that case if you have an expensive Turnkey test solution if it does ATE and logging without alarm but has visual then it may be easy to interface to that signal or output to a customer alert port. You can always ask ... on facebook ha.. https://www.facebook.com/evver

SunnySkyguy said:

choose 9~24Vdc or so piezo device not a magnetic speaker.

Click to expand...

The piezo buzzer need to choose a low voltage type (typical operating voltage: 1.5~5Vdc), as long this alarm work at 3V power supply.

It costs an even grand just for them to respond to a request like that. Vs. <$10 worth of components, most of which were already lying around. Which one is easier to justify on the budget? Hint: the one so minor it doesn't even get tabulated in.

Also, I asked you to sketch the battery charge circuit because I was thinking it would be putting ~6V through T1 and R1; however I figured out why it won't do that, so no problema. Thanks for the advice in general though!

I do not recommend charging any type of lithium battery, sometimes those may explode… And what’s the point for charging if we get a zero stand-by current circuit? :roll:

Yes, you may add supplementary monitoring LED’s to any of the +5V/+12V using series limiter resistor( 1k to 2.2k).

gotcha no direct tech support. Got Service manuals ? I imagined you were testing many in parallel. Not one at a time.

mister_rf said:

I do not recommend charging any type of lithium battery, sometimes those may explode… And what’s the point for charging if we get a zero stand-by current circuit? :roll:

Yes, you may add supplementary monitoring LED’s to any of the +5V/+12V using series limiter resistor( 1k to 2.2k).

Click to expand...

When +12V fail condition is detected the speaker will draw current from the battery and after a while the battery will be drained. :roll:

---------- Post added at 03:10 ---------- Previous post was at 03:06 ----------

SunnySkyguy said:

gotcha no direct tech support. Got Service manuals ? I imagined you were testing many in parallel. Not one at a time.

Click to expand...

The situation is kind of complex. Multiple machines from multiple vendors. This is something I want for my setups in particular (I've got two). If engineering or RMA think the doodad is useful enough one of the engineers might whip up a full fledged PCB and make a dozen or two (presumably improved; Mk.II). For now it's just me and my test racks.

How long? How many times per week/month? :smile:

ty Mr_Rf my oversight on piezo Voltage.. They often run in 1.5V to 16V or 24Vdc (not as I guessed badly

I remember working at Burroughs Canada HDD production in the early 80's in a office with 50 engineers & techs all with the same bell tone on their phone. in an open office. So I rigged up a 7KHz piezo device with passive parts to detect the ringin voltage (200Vac) on my phone and chirp at a high pitch. No one else even noticed it at same time as bell. But I could hear it bounce off the walls anywhere in the office with my own distinctive ring tone in 1983. with 10Hz modulated 7KHz chirping when bell was active..

Here's a 4KHz one **broken link removed** that is more like a smoke alarm. so tape over the hole or better add RC in series to drop current and alarm level as needed. eg 5~20KΩ // 10uF