Squelching U4 when in shutdown

[Grumpy_Geoff]

After a period of many hours in shutdown/deep standby mode, our U4 'squelches'. I assume this is from the AlwaysON block being powered, but that's just me guessing.
After being in shutdown for say, 6-8 hours, the squelch is intermittent, something like a short burst every 30-60 secs. If it's been a longer shutdown period, say after 9-10 hours, then the squelching is pretty constant.
Of course, if I power off via the rear switch then there's no squelch.

Has anyone else observed this?


Cheers,
Geoff

[prl]

No, but I'll have a listen to my test U4 after it's been in shutdown overnight.

[MrQuade]

Are you describing coil whine?
The fan does still have a quiet whine in standby, but it is much improved.

Intermittent noises sound like a bit of a worry though...I will have to have a listen to see if I hear that too.

[peteru]

It's the fan. Depending on where the blades of the fan "park" you can get any scenario ranging from silent to buzzing. There is no (practical) known workaround for that, short of leaving the unit on and the fan spinning.

The impractical workaround is to use a thin plastic stick to move the fan blades slightly to a silent position.

[peteru]

It has occurred to me that it may be possible to eliminate the buzzing by putting a diode or two in series with the fan. It may be worth a try. To test, add enough LEDs in series, such that they do not light up when the fan is supposed to be powered off, but do light up when the fan is supposed to spin.

Be aware that this will reduce the fan operating voltage, so it will spin slower.

[BigBen]

This has been discussed further over here

Sounds like you would need at least 2 diodes like the 1N4004

Is this right?

Code: Select all

  + Fan -
  |     |
  -     |
diode   |
  +     |
  |     |
  -     |
diode   |
  +     |
  |     |
  + U4  -

Could you still use a capacitor like this?

Code: Select all

  +  Fan  -
  |       |
  -       |
diode     |
  +       |
  |       |
  |—+cap-—|
  +  U4   -

Now edited to be (hopefully) correct

[sub3R]

Ben, if I’m reading your drawing correctly you have the diodes around the wrong way, which means no current will flow. See my drawing below. The capacitor is correct in your drawing.

Beyonwiz-U4-fan-control-diodes_TC21.jpg (29.83 KiB) Viewed 1632 times

[prl]

Ben, if I’m reading your drawing correctly you have the diodes around the wrong way, which means no current will flow.

That's what I wondered, too. My understanding is that the idea is to make use of the forward-bias voltage drop of the diodes (about 0.6V each), so that no current will flow if the fan control output voltage is less than about 1.2V.

Is that fan control output smoothed, though, or is it simply the direct PWM output of the fan driver? If the latter, will the diodes help much?

1N4004s have had a spectacularly long product lifetime. They are also very much overspecified for the purpose: they're 1A sustained forward current & 400V reverse breakdown. But I assume they're still one of the most common and cheapest power diodes around.

[sub3R]

My understanding is that the idea is to make use of the forward-bias voltage drop of the diodes (about 0.6V each), so that no current will flow if the fan control output voltage is less than about 1.2V.

That was my understanding from peteru’s untested theory.

Is that fan control output smoothed, though, or is it simply the direct PWM output of the fan driver? If the latter, will the diodes help much?

I don’t know the answer to any of that. A bit of experimenting is the only way to find out for certain. Peteru may know about any smoothing in the cct though.

The 1N4004s are definitely an overkill, but I suggested any of the 1N series to Ben because they can be picked up cheaply just about anywhere. I hate to think how long ago I was playing around with 1N4004 diodes.

[Paul_oz53]

This has been discussed further over here

Sounds like you would need at least 2 diodes like the 1N4004

Is this right?

Code: Select all

  + Fan -
  |     |
  +     |
diode   |
  -     |
  |     |
  +     |
diode   |
  -     |
  |     |
  + U4  -

Could you still use a capacitor like this?

Code: Select all

  +  Fan  -
  |       |
  +       |
diode     |
  -       |
  |       |
  |—+cap-—|
  +  U4   -

I think yes, the combination may work better than diodes or capacitor alone. Ideally, the diodes should have a forward drop that exceeds the residual DC voltage measured on the capacitor. Polarity matters, as noted above.

Peteru suggested using LEDS, which have a high forward voltage drop - typically 1.8V (red) to 3.3v (blue) whereas the 1n4004 is around 0.6-0.7v. The higher the voltage drop, the fewer diodes needed.

As the output of the U4 control source is pulse width modulated, the capacitor should act to smooth the output into a lower DC average value with a ripple component. The aim of the diodes would be to drop the apparent voltage seen by fan to zero when the PWM output has a low pulse width, but I suspect they might not be effective on their own and without the capacitor. This is because the peak pulse voltage remains the nominal voltage. There is still a chance enough energy will break through to trigger the fan noise. [Edit: which is prl's point above.]

If the capacitor voltage is less than 0.7v, one silicon diode would probably work if the ripple is low, which it should be when the fan is ostensibly off. If the residual voltage across the capacitor is less than 1.8V then one red or blue led or two to three silicon diodes would be needed.

The type of diode isn't important, rather it is the forward voltage drop divided into the residual voltage with a safety margin (say 0.2V) that decides how many diodes are needed. Of course, this is more untested theory so practical experiments are needed to see what works.

BTW, my U4 used to make the horrible noise when left in deep standby, after a delay of a few hours from memory - I responded by leaving it in standby.

[sub3R]

Yes, using basic LEDs would save on possibly a string of diodes.

If Ben wanted to try LEDs, just treat the LED like a normal diode when connecting it as per my earlier drawing. To identify the anode & cathode of the LED; the cathode (-ve) lead is next to the flat part of the plastic housing, & the longer lead, if it hasn’t been cut & shortened, is the anode (+ve).

[prl]

You have to make sure that LEDs don't get any reverse voltage applied. They tend to have a pretty low reverse breakdown voltage. It's hard to be sure that that can't happen without knowing the details of the control circuit.

[Paul_oz53]

You have to make sure that LEDs don't get any reverse voltage applied. They tend to have a pretty low reverse breakdown voltage. It's hard to be sure that that can't happen without knowing the details of the control circuit.

Indeed, true.

If peteru's diagram of the fan control circuit in the other thread is correct, reverse polarity should not occur. However, it occurs to me that adding a capacitor may be self-defeating without a bleed resistor.

If the diodes block the discharge path, the capacitor will slowly charge via the 19k base bias resistors to the full V+ rail voltage. To control this, a relatively low-value bleed resistor is needed to prevent voltage rise above the forward conduction point of the diodes. If it's a 12v input and the 19K is correct, a 2.2K 1/4W should limit the voltage to around 1.14V or one LED or two diodes. Compared to the fan impedance, 2.2K should be large enough to not affect normal operation.

This gets more complicated with every iteration!

[sub3R]

You have to make sure that LEDs don't get any reverse voltage applied. They tend to have a pretty low reverse breakdown voltage. It's hard to be sure that that can't happen without knowing the details of the control circuit.

Good point. Typically around 5V.

If peteru's diagram of the fan control circuit in the other thread is correct, reverse polarity should not occur.

I agree, but we don’t know what else is in the cct. As peteru mentioned; ‘a simplified circuit diagram for the fan drive looks something like this (there's a lot more to it, but this should be enough to get the idea)’.

So to be on the safe side, adding a diode, such as a 1N4004, in series with the LED/s in the forward bias direction should protect any LED/s.

The capacitor bleed resistor is a good idea.

[prl]

So to be on the safe side, adding a diode, such as a 1N4004, in series with the LED/s in the forward bias direction should protect any LED/s.

I tried that one time when I wanted to replace an incandescent panel lamp with a white LED, where the panel lamp was fed low voltage AC. It didn't go well, and the LED died (twice IIRC). I eventually found an incandescent bulb that fitted, so I never found out what I did wrong.

[sub3R]

I can’t see why a series diode wouldn’t work, it should limit reverse voltage to the LEDs.

But another option is to put a silicon diode in parallel with all the LEDs in the reverse bias direction.

I think using LEDs should be for testing only, due to their forward current specs. If they worked I would replace the LEDs with silicon diodes to give the same voltage drop. I don’t like the idea of that fan failing in the U4.

Edit: Corrected 1st paragraph.

[BigBen]

I corrected my initial post. Serves me right for trusting a YouTube video.
+ = anode / input
- = cathode / output / striped end of diode

The 1N4004 diodes seem to be the only model of the 7 that Jaycar sells these days. They're extremely small and cheap so I don't think it matters.

[sub3R]

I corrected my initial post. Serves me right for trusting a YouTube video.

Your diagram is correct now. Yes, there is a lot of incorrect information on the internet. In particular on a lot of forums, with this one being the exception.

... They're extremely small and cheap so I don't think it matters.

If only using the 1N4004s without any LEDs, plus adding an electrolytic capacitor, I would go with Paul’s suggestion of a bleed resistor across the capacitor.