Repairing subwoofer auto shutting off after a while (Velodyne DLF3750S)

[seed_87]

Hey guys,
(Canada, 120VAC, 60Hz)

Got my hands on a Velodyne sub that shuts off after just a while of use. Trying to fix it.

The issue is that it shuts off after a while of use and afterwards you just hear a relay clicking away sort of it trying to reset itself, unsuccesfully.

I don't know this thing's condition so I opened it up and I'm testing everything I can think of. I see three PCBs, a signal/input handling/preamp board, a power board and what I suspect is an amplifier board enclosed inside a metal box. This thing passed the visual and smell tests so I decided before I start probing and testing things I'd see how well the grounding is overall.

A bit more info (and do correct me if I'm wrong). The IEC plug that plugs into this thing has no earth connection, just Line and Neutral, so this makes this a (Class 2?) PSU. This supported by the fact that I see suppression X and Y caps in the form of:


Namely referring here to the two Y caps going from Line to chassis to Neutral, forming some kind of "earth" there. This earth then goes out from the PCB with a cable bolted to the back plate if the sub.


So I got my DMM set to measure AC voltage, and I do this between the metal black plate and earth, real actual earth from the wall socket. I'm measuring this while the sub is plugged in and it's power switch (which interrupts the LINE cable, neutral goes in directly) turned on.

I'm measuring ~60VAC, I turn the subwoofer's switch off, measure again, ~120VAC! What?

So now I'm thinking: I've handled the sub's back plate back when it was completely assembled, turned on, off, plugged in, not plugged in. Didn't get a single shock, not even a tingle, I didn't die. So these voltages, while they are high, must be very low current.

I tested this. With the sub plugged in, switch OFF. I measure~120VAC open voltage, 0.41mA short circuit current.
Sub plugged in, switch ON. I measure~61VAC open voltage, 0.21mA short circuit current.

Is this normal?
Are these Y caps dying (Marked as 472M, 4.7nF)? They both measure ~8.5nF with my meter.
Where are these high voltages (120V!) coming from? (Especially when the switch is off so only Neutral is connected to the PCB, LINE is interrupted by the switch and the plug isn't inverted or anything, although I don't see how that would affect anything as the primary side looks pretty symmetrical)

Here's some pics of the PSU board and the AC connections:

(Excuse the shitty graphics, I've flipped/rotated the board pics so that they match)

PSU Top:


PSU Bottom:


AC connections:

[seed_87]

Just found more pics of the internals of same or alike velodyne subs. Posting these here just for reference, these are NOT photos taken of my particular sub!

https://imgur.com/a/kvXYBjg
https://imgur.com/a/8VDtIyV

Although now that I see those burnt boards, I suspect I might see something alike when I'm finally able to pry open the metal box that contains the amplifier board. It kinda sorta smells a tiny bit smokey, but It is so faint I'm really not sure till I open it.

Trying to pry the thing open now, will post pics when I do.

[seed_87]

Managed to open the metallic box which contained the amplifier board within.

Looks like a capacitor blew up in there.

Any ideas?

[thm_w]

If you clean the carbon deposits off the board, is there any designator near that burned component that will give you a better idea of what it is?

472 is 4.7nF yeah, sounds fine.

[MathWizard]

so what was the cap attacted to in circuit ? What are those big inductors for if it's an audio amp board, or is it just the low frequency subwoofer reasons, blocking higher freq's ?

[seed_87]

And the story continues!

So I de soldered one of the big inductors (L1) from the board to allow me to see better, clean the board and maybe draw a schematic.

L1 inductor removed:



So I notice a couple of things:

There's an SMD component that blew up in half, I highlighted the pieces in RED.
Highlighted in GREEN is where the Capacitor used to be.
Highlighted in BLUE is what appears to be a freewheeling diode for the Relay.

Went ahead and cleaned the board a bit. Here's how that looks:


Here's that SMD resistor that cracked in half next to the capacitor that immolated itself:

If you clean the carbon deposits off the board, is there any designator near that burned component that will give you a better idea of what it is?

So the capacitor in GREEN is C27, no idea of it's value or type.
The SMD component is R39, I can read "100"off of it, so a 10 ohm resistor?

so what was the cap attacted to in circuit ? What are those big inductors for if it's an audio amp board, or is it just the low frequency subwoofer reasons, blocking higher freq's ?

The blown cap is C27, which is in parallel with the series formed by the blown resistor R39 in series with a 100n SMD cap C28
Not sure about the inductors, this is a bit outside my area of expertise, I'd be happy to hear from somebody more knowledgeable!

In any case do check out this quick schematic:



I should mention: Back when the sub worked for a bit and then started failing, I could hear the relay closing then opening again, closing for a bit then opening again, multiple times. I haven't checked the rest of the circuit but it could be it's self detecting some failure (hopefully related to the blown bits) activating some protection, then trying to turn on again. There's an IC on this board that's the brains of this operation, an IRS20124s which is a "Digital Audio Driver with Discrete Deadtime and Protection"

According to the datasheet some of its features are: (https://datasheet.octopart.com/IRS20124SPBF-International-Rectifier-datasheet-10898300.pdf)

"Integrated  deadtime  generation  and  bi-directionalover-current  sensing  simplify  design"
"Shutdown  function  protects  devices  from  overload conditions"


Happy to hear your thoughts!

[thm_w]

Does D10 still test ok? That relay is a bit toasted..
R39 should be 10 ohm yeah. If you do replace it, use something a bit bigger. Check C28 is not shorted or has any sort of low resistance.

There is a schematic here with the same RC snubber, but no C in parallel: https://elektrotanya.com/velodyne_cht-8_cht-10_sch.pdf/download.html
So you could leave C27 off.

Maybe the continuous opening and closing of K1 with the L1 caused a spike enough to start heating up R39. But that still seems extreme.

[cncjerry]

That board is going to be a bear to fix.  Once you pull those other rotted green caps off, I think you might find some other damage under there.  An alternative to repairing that board would be to get one of the single channel boards off ebay.  Wondom ? or Sun audio ?, can't remember their name, make decent products, you can spend a lot more money though on better amps.   I use their DSP products and they work well with great customer support.  If you drive that sub with a line level input, then Wondom makes a DSP crossover that would drop in right in front of the amp. Just a thought, but I've tried to repair boards like that and after all that abuse I think you will be fixing it off an on for a while. 

[seed_87]

I'm back, thanks for replying!

Does D10 still test ok? That relay is a bit toasted..

Pulled it out. It has a T4 marking on it, quick Googling suggests that that's an 1N4148. It seems to have gone bad, as It appears to have a Vf = 1.6v. Reverse does read open circuit. I'm replacing this one.
Relay switches on/off just fine. I guess it's just the case that got toasty. Still, replacing it cause it's seen better days.

R39 should be 10 ohm yeah. If you do replace it, use something a bit bigger.

Ok, a bit bigger like ~20 ohm? I'm curious, are you suggesting this based on the fact that you're also suggesting (Based on the schematic that you linked) to leave C27 off?

Check C28 is not shorted or has any sort of low resistance.

Desoldered and checked out of circuit, measures ~101nF both ways. Resistance goes up and up and up until its open circuit. so I guess this one's OK.

There is a schematic here with the same RC snubber, but no C in parallel: https://elektrotanya.com/velodyne_cht-8_cht-10_sch.pdf/download.html
So you could leave C27 off.

Maybe the continuous opening and closing of K1 with the L1 caused a spike enough to start heating up R39. But that still seems extreme.

Nice find! Could it be that the capacitor was just getting bad/hot (I'm assuming it's a tantalum or something that goes bad and boom), damaging D10 so it's now not letting the relay switch on properly, thus exhibiting this on off on off behaviour?

I'm assuming lots of things probably...

That board is going to be a bear to fix.  Once you pull those other rotted green caps off, I think you might find some other damage under there.  An alternative to repairing that board would be to get one of the single channel boards off ebay.  Wondom ? or Sun audio ?, can't remember their name, make decent products, you can spend a lot more money though on better amps.   I use their DSP products and they work well with great customer support.  If you drive that sub with a line level input, then Wondom makes a DSP crossover that would drop in right in front of the amp. Just a thought, but I've tried to repair boards like that and after all that abuse I think you will be fixing it off an on for a while. 

Thanks for your input! This is definitely plan B: If plan A fails to fix this sub, I'd definitely try using it as a passive sub.

I'll check under those capacitors, I've measured them in circuit and capacitance-wise they are in spec. I'm worried about the fact that they got some heat from the fireworks next door. I've got one of those T7 tester thingies in order to help me sort of gauge ESR on these and other caps, or I might just replace them altogether.

This guy:


I'm hoping to use it as a sort of pass/fail tester for ESR and other things, not exactly expecting much accuracy.

[thm_w]

Ok, a bit bigger like ~20 ohm? I'm curious, are you suggesting this based on the fact that you're also suggesting (Based on the schematic that you linked) to leave C27 off?

Bigger just as in larger physically. The original resistor was 10R 1W, so 10R 2W or something would be good to replace it with IMO. In case the original resistor was getting too hot.

C27 I'm just thinking its not used on their other board so its not critical, also can't imagine its purpose, other than filtering a bit (preventing some thump or turn on noise?).

Nice find! Could it be that the capacitor was just getting bad/hot (I'm assuming it's a tantalum or something that goes bad and boom), damaging D10 so it's now not letting the relay switch on properly, thus exhibiting this on off on off behaviour?

I'm assuming lots of things probably...

Possible. Yeah most things here are assumptions. Until its working, at least somewhat, then you can test them out.

D10 shouldn't prevent switch on if Vr is open as you stated (diode can often fail short). But the Vf reading is definitely bad.

[seed_87]

Bigger just as in larger physically. The original resistor was 10R 1W, so 10R 2W or something would be good to replace it with IMO. In case the original resistor was getting too hot.

So while I'm waiting for some parts I figured I'll just replace D10 and R39 with some comparable parts I have here available, like so:



Wired it *all* back up except I didn't hook up the speaker to the amp board (As I didn't want to kill it potentially?).

Gave it a try and R39 (which is now 1W, while the SMT resistor that was there was 0.5W) started getting very hot and released a bit of it's smoke/soul.

I started to doubt C28 so I replaced that too, same thing R39 super hot.

Hooked up a K probe thermocouple to R39, turned the sub back on for 5 seconds, went from 21C to ~40C.

Did I do wrong by not hooking up the speaker? I figured that board shouldn't just kill itself and should be able to handle having no speaker connected/speaker blown.

So yeah, appreciate any comments here as this is just a tiny update. I think I'll want to reverse engineer more of the board a bit more, I'll try and do so this week!

[Rasz]

Did I do wrong by not hooking up the speaker? I figured that board shouldn't just kill itself and should be able to handle having no speaker connected/speaker blown

no idea, but you could try connecting dummy load in its place to see if it needs to see impedance there in order to work correctly

[thm_w]

Shouldn't kill itself for sure, could be oscillating at a high frequency. Oscilloscope would be best to look at that (carefully).
Either due to other damage on the board, or maybe the board signal input needs to be terminated if its floating.

[Avee]

Hello,
I got the same velodyne, amp module cooked. Wanna confirm, there’s two ceramic capacitors, you labeled 220nf, it’s on the output stage. Are those value correct.
Also, did you fix the module?

[A Hicks]

Here is an actual schematic from Germany

[EHT]

After having tried to repair two Class-D subs like this, especially one which was a Velodyne, I strongly recommend that you give up repairing the amp, and just wire the speaker to an external amp.
The components get push to their limits in all ways - high frequency (class D ~ 100kHz), high current, and high temperature because they are boxed in with no airflow. Then you have a case like yours where there is a multiple component failure, burnt out, you can replace a lot that are clearly damaged and still not have got to the root cause. Also there are deadly high voltage/current in the amp and not isolated from the AC line (thus need a scopemeter or HV differential probe to see waveforms) .
I saw some sub designed to have the amp on the outside of the speaker box with a heatsink, that's a good improvement.