Help w/Thule Spirit hi fi amp repair please

[Martin B]

Hi all! First post for me. Been watching Dave's vids for some years and browsed this forum occasionally. I'm not an engineer or electronics repair pro, my background is in hobby retro computer repairs and projects. Now I'm a bit out of my depth with my Thule Spirit IA150B stereo integrated hi fi amp that went boom the other day (literally boom like a movie explosion through the speakers) and then it behaved very strangely and i could hear a relay switching on/off rapidly, sounded like the standby switch relay. I was playing at low volume at the time.

Thinking the problem might be in the power amp section, I hooked the pre out connectors up to another power amp, and tested it. But after about 10 mins it happened again.

Thule are long gone, but I found the service manual (which I have attached to this post). Some of the electrolytic caps were bulging, so I replaced all of them. I also discovered that the circuit board has been burned to a crisp in the area around a set of voltage regulators (LM317LM and LM337LM, REG 1 and 2 on the schematic) where there are some very hot resistors that have been bodged in place, on the underside of the board, in parallel with the visible resistors (R336 and R371). These resistors are not visible on the attached pictures as they are on the underside of the board. Is it possible they were added during production due to lack of resistors of correct value? Could it be that those resistors have been heating the board for so long that the board has been slowly burned?

Anyway I changed all the SMD caps that looked zinged in the area and reattached the voltage regulators to their traces. Unfortunately, my repairs did not solve the problem, so after 10 minutes of use the amp still produces a loud boom and the relay goes nuts.

I am, of course, suspecting the voltage regulators may be faulty. Heat damage? Any help as to the next steps greatly appreciated. Thanks!

[terminus]

Hi Martin,

Welcome to the club!

Those big TH resistors are needed to burn excessive voltage to reduce the power the LM3x7s are dissipating. It is difficult to say what it was, bad design or increased current draw on the output on the ±5 and ±15 rails by some short somewhere. I think you should start by measuring the voltage on the output of the regulators. Also it would be interesting to check what is the voltage drop on the TH resistors to calculate the current on the rails.

wbr

[Martin B]

Thanks, Terminus!

I made some measurements. I only have 10 min windows to measure, then It goes bonkers and I have to switch it of and let it cool down for 10 mins.

Voltage drop across R366 is 12.3 V (specified to be a 100R resistor, there are two resistors in parallel, one bodged on the underside, I can't suss out the values). These resistors are connected to the input on REG1.
Voltage drop across R371 is 12.4 V (spec is 100R, also two in parallel). These resistors are connected to the input on REG2.
Voltage drop across R382 is 16.4 V (spec is 300R, no bodge resistor). This resistor is connected to the input on REG5.
Voltage drop across R387 is 16.8 V (spec is 300R, no bodge resistor). This resistor is connected to the input on REG6.

Voltage across Vin-Vout on REG1 (LM137LM in worst heat damaged area) is -2.9 V
Voltage across Vin-Vout on REG2 (LM337LM in worst heat damaged area) is 4.2 V
Voltage across Vin-Vout on REG5 (LM137LM in less heat damaged area) is -5.5 V
Voltage across Vin-Vout on REG6 (LM337LM in less heat damaged area) is 6.0 V

I tried to measure between Vref and +15/-15 V, and got +13.7/-12.7 V (not used to this type of circuit). Also I noticed that when the amp started acting up, voltage fluctuated wildly on REG1 and 2, but seemingly not on REG 5 and 6.

[Martin B]

I have put some LM137LMs and LM337LMs on order just to be on the safe side.
Any other advice would be greatly appreciated.

[floobydust]

I see ten voltage regulator IC's  and kinda need to narrow down which one is sagging.
The power supply is odd, they seem to make an artificial ground for each channel with R60,R61,R68,R69 and R160,R1612 etc. They can do this because the amplifier is bridged so the loudspeaker return currents are not to GND - it's actually a 4 channel amplifier.

So I would check the balance of the GND point, the voltage across C29, C30 and C31, C32 should be around 20V each?
If the speaker thumps and the protection relay trips, then I would suspect REG7-REG10 +/-26VDC which are for the power amp. The voltage-doubler capacitors are high stress (ripple current) and generally don't have a long life. C23, C24, C123, C124 100uF 35V so I would suspect them. If you can test them, or carefully look at the four 26V rails and see how they are doing. It's best to measure voltages from GND to various parts - assuming the artificial ground is working, it's balanced.

Q-tips and IPA can gently clean off the flux on the board.

[Martin B]

Thanks, Floobydust.

Yes. There are ten regulators. I can always replace them all.

I am used to circuits with common ground, so it was hard for me to figure out where to measure against. There are both balanced and unbalanced outputs, maybe they have separate channels?

I suspected the power amp stage too, however when connecting the pre outs to an external power amp thus bypassing (?) the power amp, the problem still persisted. Therefore I suspect the problem is before the power amp stage. Also the display pulsates dimly and shows 0 volume when the noise kicks in, so I think it's maybe the standby switch relay I hear going bananas. I think maybe the only thing the standby relay does in this amp is shut off the outputs and the display, leaving the amps on. The amp does get very hot even in standby.

As mentioned earlier the electrolytics have all been replaced, including the caps you mention. So they're certainly all good. There is no flux residue on the board, the discoloration you see on the pictures is all heat damage.

[floobydust]

As long as you know the power supply is very tricky and they used many Vregs.
With this kind of problem, you have to take a guess and then work forward or backwards. If the LED display flickers, I would look at its 5V power.

U22/U21 LM1972 feeds the balanced outputs as well as the power amp inputs. It's all DC-coupled, so any +/-5.5V imbalance troubles will go straight to the outputs. But you say no problem with sound when using the balanced outs connected to another power amp? You aren't bypassing the power amp though. It's still there getting signal and the protect relay will trip and chatter if there is a large DC offset coming in.

I'd start by measuring the voltage regs to see which rail is having problems, keeping in mind ground is not ground. I see REG4 makes another artificial ground for DGND. So the regulators interact which makes it more fun. LED display power comes from +5V DIG so I would look at REG3 and REG4.

Standby is supposed to shut off bias current to the power amps (i.e. Q36, Q136) so it should be cool. I wonder if U2, U3 are powered ok.

[terminus]

What I would do first is to disconnect the power stage fuses S1, S2, S101, S102 and 26V rails - S5,6,7,8.
C23,24,123,124 as Floobydust already mentioned can easily go wrong and need to be checked.

Check the voltage on the +-20VL, +-20VR the ground should be in the middle. if not, check REG4 as it seems to keep the GND in place from -20VL rail. If not, start removing the input resistors on REG1,2,3,5,6 and checking the voltages.

If symmetry and output voltages on REGs 1-6 are correct then you can try inserting S5,6,7,8 and check the 26v rails symmetry.

That's the algorithm I would follow.
wbr

[Martin B]

Thanks again, guys!
Just to clear two things up:
1) As mentioned twice previously, every electrolytic cap on the board is brand spanking new. Replacing them was the first thing I did. This did not change anything, the problem did not go away.
2) NOT connecting any speakers directly to this amp, but feeding the pre out signal from this amp to another power amp, does NOT change anything either. There is a loud boom (max volume), the standby
 relays start switching on/off rapidly, the display starts flickering and shows 0% volume. This ONLY happens once the amp gets hot (about 10 min), and if you power off and on again before it has cooled down, it happens again much sooner. Only allowing the amp to cool down before switching it back on buys you another 10 mins in which to make measurements before it malfunctions again.

So I measured across the other Vregs too:
Reg 7: -1.25 V
Reg 8: 9.64 V
Reg 3: -1.24 V
Reg 4: 6.85 V
Reg 10: 9.52 V
I couldn't get to REG9 as I would have to disassemble the amp to get at it from the reverse side.

I then disconnected fuses S5, 6, 7, 8 in turn. At first I thought this made the amp stable, but when I put them all back it was still stable. So it may take some time for the problem to show up again...
I couldn't disconnect S1, 2, 101, 102 are soldered in place and I couldn't get around to checking them tonight as I would have to disassemble the amp again for that.

[terminus]

There is not much sense measuring voltages like you did last time. You need to find which rail sags and what pulls the artificial ground. It can be done only by disconnecting loads.

wbr