Roister Pb-150 Subwoofer repaired, how to prevent from failing again?

[kaboom]

Re: Roister Pb-150 Subwoofer repaired, how to prevent from failing again?

OK, my turn.

Upgrade the drivers, Q47/48. Final gain drops at high output current; driver not only needs to be robust, but also keep its gain up so the previous stages stay linear.
The C5200/A1943 should be driven with something like a MJE15032/33, 15030/31 , or 15028/29 comp pair.
The 028/029 will work since this amp runs ~51v rails; use what's easy to get.
Make sure you transpose the leads; TO-220 is different vs TO-92.
And be sure to put NPN->NPN & PNP->PNP.
These won't need heatsinking beyond the tab.Your PCB is silkscreened for bigger drivers- look close, they cheaped out.

Don't worry about TO-126 for drivers- go right to TO-220.

TO-92 or TO-92(L) is absurd- amazing it lasted this long!
Q45/46 are DC detect monitors, one for each polarity. These kill drive to relay driver when triggered. Pretty low duty, so conductive/corrosive glue or borderline/"second" parts likely.

Q49/50 should be replaced with BD139/140. Make sure NPN->NPN & PNP->PNP and you transpose the leads! Otherwise, you're sending +/-51v into the opamps!


Change the two 47u/25v, across the zeners for the +/-15 regs, to 100u/25v. Make the reference for the reg better vs extra filtering "after the fact."
Change C67 to 47u/50v, 105c, PW, HE, FC, etc. if you haven't already.
Change C66 to 22u/50v, also PW, HE, FC, etc.
If the main filters have been in this long, they'll keep formed at the higher voltage.
Electros are initially formed at a bit more than rated voltage; yours will likely tolerate 54-56v.
The toroidal transformer can/should keep the rails up better than an E/I.

Should make an easy 150w/4ohms, especially with 230 vs 220 in.
Possibly 180-200w peaks if "pushed," but there's only one set of finals; push it only with an external fan on that heatsink.
Don't worry about opamp "rolling," 4558s are just fine in a sub, and besides, those are JRC!


Good luck and "play it loud!"

[goodpsusearch]

Re: Roister Pb-150 Subwoofer repaired, how to prevent from failing again?

Ok, this will be a really long post!

Nice!


No surprise there.
Looking at the magnet of the woofer driver, it has a pretty chunky magnet. That's pretty much the #1 thing needed for a good woofer to go down very low in frequency. The other thing is the box, of course. If tuned properly, it should be able to hit 25-30 Hz easily... which is what would give it that earthquake rumbling sound when it needs to go very low.

According to Greek audiophile forums, this company uses Vifa speakers, I don't know if they are considered good or not. Could you tell from the pics if the woofer is magnetically shielded?

As for your Roister PB-150 circuit, I tried to do some calculations, but it doesn't quite add up. Assuming stable 50V DC comes out before R61 and the relay pulls down R61 on the other side to 24V level, then R61 will have approximately 26V across it, dissipating a little over 2 Watts! But I don't think that's the case, though, or you would have seen a charred resistor. I'm thinking more along the lines that perhaps the relay coil is getting a bit "over-driven" - possibly up to 30V. This should leave only 20V across R61, dropping P_d to 1.2 Watts... which honestly is still a lot. So perhaps it could be over-driven even further? Or perhaps the 10 uF capacitor isn't enough to provide a steady 50V DC, and that further lowers the overall DC voltage? Whichever way I look at it, though, it's honestly a mystery/mess to me. There really ought to be better ways to design that circuit and not use many/more/pricier parts.

The relay is HLS-4453(18F) 240VAC 5A 30VDC DC24V

I think I found the datasheet:
https://cdn.badcaps-static.com/pdfs/...f770a29e5e.pdf
Alternate link:


Coil resistance Ω ( ± 10% ) 650

So, according to voltage divider, we will have approximately 33.163V dropped on the relay coil and 16.837V dropped on the 330 ohm resistor!

This means 1.69Watts of heat dissipated on the relay coil. I heard from a tech here in Greece that he received another PB-150 that had, among other parts, a failed relay. This could explain why the relay failed on him.

The power dissipation on the 330 ohm resistor would be 0,86W.

I should have replaced the resistor with bigger one...

That aside, you may want to glue some of the larger caps down (or ones with very long leads), because the vibration from the subwoofer may eventually damage their seals on the bottom. Probably won't happen in many years if at all, since you used Japanese caps... but generally, it's a good idea to glue down any heavy components in a subwoofer amp, due to the vibration. Regular household/bathroom silicone should be OK to use if kept away from metal pins and parts, due to being slightly corrosive while it cures. For places where you have to glue over metal parts, there are non-corrosive silicones for electronics... or just plain hot glue.

Very interesting and totally makes sense. I will try to keep that in mind for future repairs, especially when the item is expected to be exposed to vibrations.

I use 2 parts epoxy glue. It needs more than 24 hours to totally dry but it is strong enough for my needs. I also have cyanoacrylate / super glue but it is not suitable for this type of jobs.

That's what I was thinking as well. If it was a P_d issue, we likely would have seen it in the form of darkened PCB around the transistor... unless it all happens very quickly during the On/Off cycle of the amplifier while the speaker protection is engaged / relay coil dis-engaged.

However, I'm also not quite sure how it can be a voltage issue.
Let's assume worst case here: either the 2SA1943 or 2SC5200 is fully conducting (but not both at the same time) when the output stage is getting biased. This would be / could be sending either full +50V or -50V to one side of R92. But Since R92 forms a voltage divider with R102, that should limit the voltage to about 1.786V (without the transistors in the circuit.) With Q45 Emitter terminal grounded and Base connected to other side of R92, though, really R92 becomes more of a current-limiting resistor for the Base of Q45... and that current should be less than 1.85 mA... which again, I don't think would be enough to cause damage. And the Collectors of the Q45 and Q46 should also be fairly protected, since the 47 uF 63V cap that sets the delay for when the relay turns On, will be discharged. Therefore at power up, Q45 and Q46 Collectors should be at 0V. Perhaps only if the Base voltage of Q44 is somehow allowed to float above 50V, this could then be an issue, given that the Emitter of Q46 is tied to ground with a 150 KOhm resistor (and also a diode or Zener diode to the Base of Q55? )

So I was thinking of another possibility. The PCB construction (particularly the "random" appearance/selection of the resistors) reminds me of ultra-cheap pre-made PCBs/circuits from China (and elsewhere), perhaps the manufacturer just bought the cheapest parts online... so possibly some counterfeit transistors? Now, the outputs may be genuine (or at least decent quality parts), but perhaps these smaller TO-92 transistor were not. That could explain the random failure of Q45 and Q46 without any discoloration around them.

Time should tell, though. Hopefully the amp will stay fixed. If it does, I would honestly just suspect dodgy/fake/counterfeit transistors from China were the culprit.

Q45/46 are DC detect monitors, one for each polarity. These kill drive to relay driver when triggered. Pretty low duty, so conductive/corrosive glue or borderline/"second" parts likely.

The previous owner told me that he left the subwoofer on standby and went to work. A worried neighbor called him on his cellphone and told him there is a terribly loud noise coming from his apartment. Then he returned home to find the sub making the noise and he unplugged it. Then he placed an ad to sell it for repair/parts for 40 euros and this is how I got it. So, maybe after Q56 failed, the protection circuit activated to disconnect woofer from amp and either the protection circuit immediately failed resulting on the terrible noise, or the protection circuit stayed too long on this state (protection ON) and this somehow wore and damaged Q45 and Q46 and then there was the noise that worried his neighbors.

It's quite impressive that the woofer survived.



I should also mention that the heatsink of the amp appears non conductive at first but if you rub the multimeter lead on it, then it gets conductive. The reason I mention this is that when I disassembled the unit for the first time, I noticed that the thermal pad of one of the 2 main output transistors 1943 / 5200 was incorrectly seated, leaving the shiny metal back of the transistor directly touch the heatsink. At this point I am just making wild assumptions though.

Upgrade the drivers, Q47/48. Final gain drops at high output current; driver not only needs to be robust, but also keep its gain up so the previous stages stay linear.
The C5200/A1943 should be driven with something like a MJE15032/33, 15030/31 , or 15028/29 comp pair.
The 028/029 will work since this amp runs ~51v rails; use what's easy to get.
Make sure you transpose the leads; TO-220 is different vs TO-92.
And be sure to put NPN->NPN & PNP->PNP.
These won't need heatsinking beyond the tab.Your PCB is silkscreened for bigger drivers- look close, they cheaped out.

Don't worry about TO-126 for drivers- go right to TO-220.

TO-92 or TO-92(L) is absurd- amazing it lasted this long!

Perfect, Thank you for your recommendations, now I have to go shopping on my local electronics store

Btw, you are perfectly right about the PCB supporting bigger Q47 Q48 components, but if you look closely you will see that there is marking on the PCB that clearly says 647 and 667.

Q49/50 should be replaced with BD139/140. Make sure NPN->NPN & PNP->PNP and you transpose the leads! Otherwise, you're sending +/-51v into the opamps!

more to add to the basket, thanks

Change the two 47u/25v, across the zeners for the +/-15 regs, to 100u/25v. Make the reference for the reg better vs extra filtering "after the fact."
Change C67 to 47u/50v, 105c, PW, HE, FC, etc. if you haven't already.
Change C66 to 22u/50v, also PW, HE, FC, etc.

Could you please explain what exactly C66 does in the circuit?

If the main filters have been in this long, they'll keep formed at the higher voltage.
Electros are initially formed at a bit more than rated voltage; yours will likely tolerate 54-56v.

This is really good news!

The toroidal transformer can/should keep the rails up better than an E/I.

Should make an easy 150w/4ohms, especially with 230 vs 220 in.
Possibly 180-200w peaks if "pushed," but there's only one set of finals; push it only with an external fan on that heatsink.
Don't worry about opamp "rolling," 4558s are just fine in a sub, and besides, those are JRC!


Good luck and "play it loud!"

If I remember correct, I measured woofer impedance with multimeter and it was 6ohm. Of course the real impedance will vary depending on frequency.

Well, I had it running at unbearable high levels of bass and the backplate+heatsink got very hot!



Thank you both for your advice again! This was great fun and learning experience

[kaboom]

Re: Roister Pb-150 Subwoofer repaired, how to prevent from failing again?

According to Greek audiophile forums, this company uses Vifa speakers, I don't know if they are considered good or not. Could you tell from the pics if the woofer is magnetically shielded?

They seem to be better than average. That driver isn't shielded, there's no "can" around the magnet.

I should also mention that the heatsink of the amp appears non conductive at first but if you rub the multimeter lead on it, then it gets conductive.

Probably just the probe scratching thru the anodizing.

Perfect, Thank you for your recommendations, now I have to go shopping on my local electronics store

You're welcome...

Btw, you are perfectly right about the PCB supporting bigger Q47 Q48 components, but if you look closely you will see that there is marking on the PCB that clearly says 647 and 667.

If I find datasheets for those, I'll upload them. It's possible they've got the same pinout as TO-220- some of those TO-92Ls were the same. But never assume!

Could you please explain what exactly C66 does in the circuit?

That dropper resistor is shunted by that cap. Looks like it supplies a high initial current for when the relay pulls in, then C66 charges and only the resistor powers it. They're probably trying to reduce dissipation. May not even need C66 if C65 is 47u or greater.

If I remember correct, I measured woofer impedance with multimeter and it was 6ohm. Of course the real impedance will vary depending on frequency.

True. d:

Have fun & don't blow anything up!

[momaka]

Re: Roister Pb-150 Subwoofer repaired, how to prevent from failing again?

According to Greek audiophile forums, this company uses Vifa speakers, I don't know if they are considered good or not. ...
The previous owner told me that he left the subwoofer on standby and went to work. A worried neighbor called him on his cellphone and told him there is a terribly loud noise coming from his apartment. Then he returned home to find the sub making the noise and he unplugged it. Then he placed an ad to sell it for repair/parts for 40 euros and this is how I got it.

That's a pretty cool story!
If the speaker driver survived all day making loud noises that the neighbors noticed and called... yeah, I'd say it's a "decent" speaker driver, at the very least.
Like I said, the magnet and construction looks "chunky", which is a rather good thing in this case.

So, maybe after Q56 failed, the protection circuit activated to disconnect woofer from amp and either the protection circuit immediately failed resulting on the terrible noise, or the protection circuit stayed too long on this state (protection ON) and this somehow wore and damaged Q45 and Q46 and then there was the noise that worried his neighbors.

Hard to pinpoint exactly who killed what, but I don't think Q45/Q46 staying turned On for too long (protection On) should have killed either of them. After all, they are driving the Base of Q44 through a 10k resistor, so it's not like they can get overloaded with current from this. And on their input, R92 is 27k, so again a bit too high to let too much current through. So I'd still chuck it to dodgy transistors.

It's quite impressive that the woofer survived.

I should also mention that the heatsink of the amp appears non conductive at first but if you rub the multimeter lead on it, then it gets conductive. The reason I mention this is that when I disassembled the unit for the first time, I noticed that the thermal pad of one of the 2 main output transistors 1943 / 5200 was incorrectly seated, leaving the shiny metal back of the transistor directly touch the heatsink. At this point I am just making wild assumptions though.

Good catch there!
Well, even if that wasn't the problem, it could become one. So good that you eliminated this too.

All in all, this looks like a decent sub, but just poorly/hastily assembled/designed in some parts and possibly with some dodgy components. But the "frame" is there for a good subwoofer, so it's definitely a worthwhile repair.

Coil resistance Ω ( ± 10% ) 650

So, according to voltage divider, we will have approximately 33.163V dropped on the relay coil and 16.837V dropped on the 330 ohm resistor!

This means 1.69Watts of heat dissipated on the relay coil. I heard from a tech here in Greece that he received another PB-150 that had, among other parts, a failed relay. This could explain why the relay failed on him.

Could sure be that. Or maybe more failed transistors in the protection circuit "buzzing" the relay On and Off constantly until it failed?
But yeah, 1.7 Watts of heat is quite a bit.

On that note, I do wonder if a higher resistance for R61 can be used - something like 680 Ohms. That would actually increase the heat dissipation slightly on R61, but should reduce the current going through (and power dissipated) by the relay coil to under 1 Watt... so you'd end up saving around 0.5 Watts. Or perhaps go even higher resistance for even lower voltage on the relay's coil. Of course to make sure it turns On, you might have to increase the capacitance of C66, as suggested by kaboom. Then you may be able to realize even further power reduction.

I use 2 parts epoxy glue. It needs more than 24 hours to totally dry but it is strong enough for my needs. I also have cyanoacrylate / super glue but it is not suitable for this type of jobs.

Yes, CA (super glue) generally isn't great for gluing parts that will see lots of vibration. Flexible glues as better for such task... though epoxy will probably be OK as well - just hope in the future you won't have to remove whatever you glue with it. Probably will be a bit of a paint to remove afterwards. And hot glue may also not be idea if the components inside the box get very hot... though if they do get hot enough to melt the hot glue, that's another issue in itself. Hence why I like using hot glue on electronics - if it runs so hot that it melts the hot glue, it's time to go back to the drawing board on the circuit design.

Thank you both for your advice again! This was great fun and learning experience

You're welcome.

Don't worry about TO-126 for drivers- go right to TO-220.

This comment really cracked me up.

That driver isn't shielded, there's no "can" around the magnet.

Or a 2nd (and weaker) magnet in reverse on the back of the woofer.

I'm guessing this was designed after the CRT era, so not as much regard for shielding... though I've also seen plenty of 90's gear with unshielded speakers.