Thule Spirit IA 150B

[HRmeteohub]

A friend of mine asked me to take a look at his Thule Spirit amp. The amp presented with distortion and the speaker protection circuit was clicking the relays often.

Outside, design is ... Danish simplicity from 1999. Inside, the schematic is very interesting, power amp entrance is basic but then the design gets really symmetrical, and not needing true grounding tap on the transformer means that parts in the power supply need to be really balanced. If anything goes wrong, whole switching part with lot of preamps get asymetrical voltages. Guess what...

Danish were a bit stingy and used some not-so-top-of-the-line capacitors and soldering tech. Also, the use of 100mA LM317/337 in some cramped spaces, while dissipating quite a lot of power(up to 0.9W) resulted in some burn marks on the board. Finally, as far as I am concerned, using 1458 op-amp was a bit old fashioned even then - better op-amps were available, more appropriate for this really expensive amp (sorry, there are no 1458 in sound path).

Quick look at the voltages revealed severe imbalance on REG1 and REG2, but not one part was visibly bad. Then again... one (out of four) 10 000uF filter cap had small smoke tail under it, not visible before de-soldering. So... I removed all electrolytic caps, both through-hole and SMD ones. About half of them did not had rated capacity and several were basically resistors.

I am waiting for new, 5000 and 10 000 hours, 105°C capacitors from Panasonic, Wurth and Vishay. I also ordered all new regulators, just in case (200mA versions). I have also re-soldered all power transistors, connectors, basically all large joints, and cleansed the board.

So far, this was easy fix. I'll see how it goes from here. In the mean time, you can take a look at the pictures, and perhaps you can give me an advice.

How can I improve cooling of regulators (marked with circles) in order to prolong the life of this nice amp? TO220 are so close to filter caps that I can't fit any cooler, and SMD reg's are so cramped, even the small coolers for RaspberryPI can't fit.



























View attachment thule_audio_ia150b.pdf

 

[caution]

How can I improve cooling of regulators (marked with circles) in order to prolong the life of this nice amp? TO220 are so close to filter caps that I can't fit any cooler, and SMD reg's are so cramped, even the small coolers for RaspberryPI can't fit.

Those heatsinks are HUGE, why didn't they attach the 337's to them? They're so close! Maybe you could add a small heatsink on the SMD regs. Even better, the top of that is vented, so mounting tiny fans inside, on the hot spots, could be enough. I once fixed an overheating display chip for a spectrum analyzer this way.

 

[HRmeteohub]

Yes, my thoughts exactly. There is plenty of space inside, they could do number of things with these regulators.
Now I have to move the regulators away from board or they will carbonize the board - and than the amp will be broken beyond repair. I am seriously thinking about buying dozen of LM317/117 in TO220P , mounting them on separate heatsink, and soldering wires to the main board. That could give more years to the board and capacitors. But, it will look ugly as hell.

The other solution is fan(s). However, in my experience, fans always fail after some time, and than we are back at the same problem. No, I guess I'll have to move the regulators away from circuit board, they have almost fried it anyway. There is no point in risking it.

 

[HRmeteohub]

I have two ideas on how to solve the overheated-regulators-problem.

First idea is ugly, bad but robust.
I will make a bracket between left and right cooler, bolted on the transistors, over the board and under the upper plate. The regulators will be mounted on the bracket and soldered using short leads down on the board. It is not going to look pretty, but it will be stable and the leads are going to be reasonably short. I feel like a vandal, destroying the nice layout of this amp with this, but...
Second idea is more elegant but not so robust.
I will put small aluminium coolers (made for Raspberry Pi) on regulators (both TO220 and SMD), glued with double-sided thermal tape. I am having second thoughts about this, because I am afraid what will happen one day when these cooler(s) fall down onto the board.

I would appreciate any thoughts you may have on this.

Edit: There is maybe one more way. Perhaps I could replace SMD regulators with TO-92 versions?

 

[HRmeteohub]

Update...

I got some TO-220 regulators, removed the old SMD ones from the board and cleansed the board. I also made some angular iron boards to mount the regulators on it (on top of the image13), but they look rather heavy. Therefore, perhaps using a thinner iron strip and mount it between left and right heatsink.

The capacitors (only through-hole, SMD's are still in bag) are mounted despite the fact that they should go in last. This choice makes my moving around more cumbersome and riskier, but It is necessary in order to evaluate the space available for modifications.

It looks like .. this:


... and my plan for the support structure ... I guess it could be mounted using a screw that holds one power transistor to the main heatsink...


The board under removed SMD regulators sustained a bit of heat damage, but it is still functional:

 

[HRmeteohub]

Brackets are in place, I started populating SMD capacitors...



Next time - regulators and bunch of wires to the main board (3 per regulator, 8 regulators = 24 ouch!).

 

[HRmeteohub]

Next step - finally found the perforated strip that fits to this amp:


... the old regs are out, new await. It is rather funny to see that metal part of TO-220 can be 50% thinner than it was 20 years ago. I hope that this week will all caps and reg will find the way to the board.

 

[caution]

That's exactly what I was envisioning, nice work so far!

 

[Superduper]

I know nothing specific about that amp so I'm just going to comment on what I see. It takes a lot of heat to scorch a pcb. Before you go through all that heat remediation upgrades, I would have suggested that you replace all the bad components first (resistors, capacitors) and then power it up to see if (1) it works normally again, and then (2) check the temperatures of the regulation circuits to see if they are reasonable. When some components go bad, it could cause some circuits to oscillate which will add a lot of heat. Furthermore, if the caps are blown or become resistors or shorted completely, it obviously would cause increased current that will roast the regulators. In short, if you resolved the amplifier issues, your heat issues might have also resolved.

If on the other hand, the amp was designed to roast like purgatory, then it was just a poor design and I would drop it like a hot potato. I'm not particularly liking the idea of attaching a bracket to TWO of the output transistors, especially if the regulators are expected to generate a lot of heat. A spring snap on heatsink might be all you need. On the other hand, if you feel you must attach those regulators to that heat sink, then a far better choice might have been to drill and tap the top edge of the heatsink. This will allow any heat to sink directly to the aluminum chunk instead of through the transistor, if that's what I'm envisioning. If the output transistors are in parallel like I suspect, then heating up 2 of the transistors more than the rest could cause them to conduct more than the rest and I think you know where I'm going with this. Or maybe the idea here is that the bracket is intended merely to suspend the regulators to located them off of the board and not to transfer heat to the heat sink. Ok, except now the regulators will be near the top of the board where it's going to be hottest and will it cause issues with something stacked on top, or will the regulators blow due to excess heat seeing as instead of air flowing up off the regulators, they are now up against the top panel/shield will reflects heat back on it? Maybe consider mounting angle bracket against the heat sink and use thicker piece of aluminum bridge to actually function as a heat sink.

Then when I look at the PCB, and I see traces underneath that look scorched too. Seems like what I mentioned before about the circuits drawing too much current, hopefully due to malfunction that will be corrected. Because if it normally works that hot, then the traces probably aren't even designed large enough to carry the expected current in which case, it's just a bad design and probably I'd consider it a hot potato..... get rid of it, lol.

Oh well, I probably don't even know what I'm talking about haha.

 

[HRmeteohub]

@ caution - Thank you sir!
@Superduper - You are reading my mind.

I have found one blog (in german) that deals with similar problems with this amp. The solutions were, kind of similar.

May I add that this amp was working for about 15 years in rather warm climate in room without air-conditioning. Up to 95°F/35°C - for months at the time.

 

[caution]

It's amazing what a little air flow can do for cooling! As PCB designer I sometimes work on systems without any airflow, liquid cooling only, and we have to make sure that even the tiny (<1W) chips are under the coldplate or they can overheat.

 

[JVC Floyd]

Not to sound like an expert but most good kick ass amps have built in cooling fans and in most cases more than one. Now it takes extra noise cancelling circuitry to kill the electrical noise from the fan motors running.

 

[HRmeteohub]

@ cautiom - sir, you could not be more right. My experience is the same. Even several holes through the board would have done nice things for this board. Now that the thing is half-roasted, I believe this is the best bet of giving this amp another 20 years of life.

@JVC -- well, this is not a kick-ass amp, this is high end amp, on the border of esoteric tech. Very clean design, therefore - no fans are allowed. I do feel like a vandal, destroying the artistic beauty of this tech. Perhaps, one could say that I am doing it in a good way, in order to preserve a piece of quality tech with some care and even a reverence to original design. I could have done better, probably...

Anyway.... an update:
Four replaced regs are connected, four remain. All caps are replaced.

 

[HRmeteohub]

All parts are replaced, double checked. I used shink-wrap as stress release for cables, and after all is tested I intend to fix the wires to smd pads with some glue (hot glue or something like that).
Next - the testing.

 

[HRmeteohub]

Finally, first sound of the restored amp!

First testing is done with small speakers, although this amp is very safe design. I have measured the temperature of regulators, and they are running at 44°C (111°F). Power resistors run even hotter, above 60°C (140°F).

I have to say, I like the fact that the amp works, but I am sad that I had to place a rail above the board. However, I can't say that any other solution would be better for this amp-





https://youtu.be/v6ebimWeCWw


What have I done:
- celansed the amp of the dust and debris.
- removed SMD LM317/337 regulators 1-4, 7-10, replaced them with TO-220 versions, placed on perforated rail above the board.
- replaced SMD capacitors (3,3uF, 35V, Panasonic, 5000 h, 105°C)
- replaced rail filtering capacitors (100uF, 63V, Wurth, 5000 h, 105°C)
- replaced filter capacitors (10 000 uF, 63V, Wishay, 10 000 h, 105°C)
- regulators are mounted at the rail using thermaly conductive isolators.

And now, the sweet part - testing for at least a week. I just hope this SARS-Cov-2 pandemic will not prevent me getting it back to the owner (after I am done)!

Edit (2020-03-22): It did. The amp is now in storage until better times come. Drži se Zagrebe!