Repair Line6 Spider III Guitar Amp

[obiwanjacobi]

I am trying to repair a Line6 Spider III (30 watt) guitar amp (schematic is attached).

It is a digital 'frontend' with an analog power amp (TDA2050).

When I started the whole digital part lacked a 3.3V power rail and so I replaced the LM317T.
That resulted in a 3.1V rail but the digital part still didn't start up.
I scoped out the MASTER_CLK and it did wobble a bit but was not making a full 0V-3.3V swing.
The MASTER_RESET remained low. I think U3 (is an MCU) will only release the reset when the clock is stable?

Anyway my question is what could make the 3.3V rail be so low (3.1V) and does that have an effect on the oscillator circuit not working properly.
- or what (other things) could cause the oscillator circuit to fail.

Another thing I noticed is that the power amp IC (TDA2050) is getting hot quickly when running the amp with all volumes at 0 and loading the output with a head phone.
Could a malfunctioning TDA2050 cause problems on the digital section? (I can not see it would)

[Vicus]

Something maybe is loading the rail.

[pcprogrammer]

Tricky one.

The master clock is not super high frequency (12MHz) and not being full swing should not matter that much as long as it reaches the specs of the MCU and DSP clock input. The DSP is from the 56000 series, bit of an oldie.

Did you look into the data sheet of that DSP?

The master reset not being released can have several causes. Best test might be to see if the FLASH (U4) is being addressed. Easiest with a logic analyzer (cheap 8 bit 24MHz one will do). That one can contain the code for the DSP, but it might also be that the MCU uses it for its own code to. Depends on what type it is of course, but did not see it in the schematic.

Might be a 8051 based MCU.

There are quite a lot of components loading the 3.3V rail, which makes it hard to figure out what might be the problem.

Do you have some good pictures of the main board?

[obiwanjacobi]

Something maybe is loading the rail.

Yes, that was my initial thought as well.

I desoldered the two main supply smoothing caps to see if that would bring up the rail voltage - it did not.
The rest is all SMD and I lack the tools to work that effectively (I only have a soldering iron).
(I do have some experience soldering SMD but not so much desoldering -without destroying the part)

The master clock is not super high frequency (12MHz) and not being full swing should not matter that much as long as it reaches the specs of the MCU and DSP clock input.

Yeah - but now it sort of oscillates in a +/-500 mV range somewhere around the 2V level... So no way any logic IC is gonna see that as a clock.

Did you look into the data sheet of that DSP?

No. I figured it did not matter much as long as there is no clock and the reset stays active...

The master reset not being released can have several causes. Best test might be to see if the FLASH (U4) is being addressed. Easiest with a logic analyzer (cheap 8 bit 24MHz one will do). That one can contain the code for the DSP, but it might also be that the MCU uses it for its own code to. Depends on what type it is of course, but did not see it in the schematic.

I interpreted it as the MCU being fed the board clock and having startup code that releases the master_reset (that's how I would implement it)...
I'm pretty sure nothing is accessing the FLASH (U4) when there is no clock and reset is active - but I haven't measured it. It's all SMD so I'd have to solder on some wires to be able to measure that.

Do you have some good pictures of the main board?

No - I couldn't find any online - so I will make some (when I get home).
What do you hope to learn from the photo's?

BTW: I also desoldered the power amp IC (TDA2050) that was getting warm without any load. I figured I could tackle that problem later (I have some in stock).

[pcprogrammer]

Maybe there is a part number on the MCU that can provide some insight. And pictures are always nice. 

But with the master clock not being up to par, you are right that not much will work.

Did you scope the power supply voltage itself. Is it stable at this 3.1V or is there a large ripple on it?

I noticed in the schematic that it is possible to desolder a jumper that brings the master clock to the MCU. Does the clock improve if you take this jumper out?

The MCU might be able to start up with an internal clock and have a boot loader that needs to load the main code from the flash first, and then check if the master clock is operational and release the master reset if so. There are many scenarios to think of, but without knowing the actual parts used it is just guessing.

Is there a way to easily separate the analog amplifier part from the digital part, supply and signal wise. That way the problems could be investigated separately.

[pcprogrammer]

I looked at the manual you provided again and in the parts list there is the part number of the used MCU. It is a NXP 8051 based MCU. https://www.nxp.com/docs/en/data-sheet/P89LPC933_934_935_936.pdf

The DSP is also from NXP.
Datasheet: https://www.nxp.com/docs/en/data-sheet/DSP56364.pdf
User guide: https://www.nxp.com/docs/en/user-guide/DSP56364UM.pdf

A better look at the jumpers shows that the MCU can either use the master clock as input or generate it. The different models will use these different configurations. So you have to check the jumpers to see how it is using the master clock.

The MCU can run from an internal clock. This is configured with flash configuration. The Port 3 pins can be used for plain IO or a XTAL or clock input and or output.

The DSP seems to need just a bit more then the 3.1V to be in range of its operational characteristics. So it is necessary to find the cause of the a bit to low supply. Not easy though since there is no direct short. Maybe some component is heating up a bit more than others?

Good luck with the search.

[obiwanjacobi]

Did you scope the power supply voltage itself. Is it stable at this 3.1V or is there a large ripple on it?

Yes. No ripple to speak off - I do not remember the actual values.

I noticed in the schematic that it is possible to desolder a jumper that brings the master clock to the MCU. Does the clock improve if you take this jumper out?

Will try that but I think the low rail voltage is problem #1.

Is there a way to easily separate the analog amplifier part from the digital part, supply and signal wise. That way the problems could be investigated separately.

The power supply for the power amp looks good and stable. It is powered from separate windings of the transformer. I desoldered the power amp IC so I think we can ignore the analog section for now - unless I am wrong in my analysis/assumptions.

No parts are getting (Very) hot. I think there was one that was getting slightly warm, but that didn't raise any alarm bells with me.

I want to try to unhook the ICs from the 3.3V rail. Is there a quick and easy way to do that short of desoldering the entire chip?
I was thinking of lifting the VCC pin, but I am concerned that the positive voltages on the other pins may cause problems...(or still power the chip and thereform load down the rail)?

[pcprogrammer]

I want to try to unhook the ICs from the 3.3V rail. Is there a quick and easy way to do that short of desoldering the entire chip?
I was thinking of lifting the VCC pin, but I am concerned that the positive voltages on the other pins may cause problems...(or still power the chip and thereform load down the rail)?

Just lifting the VCC pin won't cut it. Like you wrote, the positive voltages on other pins might supply the chip and it might even damage things.

Can you check the current being drawn after the LM317T. This might give an idea of what you are looking for that is problematic.

[obiwanjacobi]

Can you check the current being drawn after the LM317T. This might give an idea of what you are looking for that is problematic.

There is no easy way to check that - besides from cutting traces...
It's not getting warm in any case.

Could the resistors (R63/64) that set the voltage have drifted out of value?

[pcprogrammer]

Might be an option, but not likely. Maybe the specification spread of the LM317 makes it a bit low.

A way to measure the current is to take of the LM317 and solder it via wires with a current meter in series. Keep the wires short or add some small decoupling caps near the LM317 to avoid oscillation.

The above might also be an option to check the current in some of the IC's. Lift the power pin and add a current meter in series.

The MCU has a wider supply voltage range than the DSP, so another test can be to cut the master clock to the MCU and feed in a proper clock from a XTAL oscillator or a function generator. See if the master reset gets lifted and LED's light up.

A cause of it not working might be that the MCU lost it's programming. Maybe it is possible to read it in circuit with a suited programmer.

[obiwanjacobi]

Might be an option, but not likely. Maybe the specification spread of the LM317 makes it a bit low.

I already replaced the LM317T. Before that there was no rail at all. At the time I thought it would be an easy fix 

[obiwanjacobi]

I've ordered a (cheapo) hot air station and will -after some practicing- attempt to remove the SMD chips to find what drains the 3.3V rail...

[pcprogrammer]

I've ordered a (cheapo) hot air station and will -after some practicing- attempt to remove the SMD chips to find what drains the 3.3V rail...

Watch some good youtube videos on how to do this. Plenty of flux and keep moving the nozzle over the pins is what does it for me, but with IC's that have buried ground pads it is more difficult and the IC might not survive.

[obiwanjacobi]

Yeah, I've been watching some YT videos already. Looks logical. I do have some experience soldering SMD, just not desoldering with hot air. I will practice on some old PCBs before trying the real thing. I don't think there are any ICs on there with ground pads...

I also got some more flux coming - the hot air station has arrived. Will start playing with it this weekend.