Marshall Mode 4 (mf-350) IC's blowing up

[R.G.]

From everything I have seen, the TDA729x chips look nice on paper, but have some interesting quirks that make them difficult to make reliable. IMHO they're electronically fragile. Which is a shame, because they're otherwise a nice combination of features, with higher power and voltage capability than the LM3886 family. The LM3886 is stable and sturdy, but won't get into the voltage and power territory the TDA729x chips claim.

But then, they work.

Sorry - I know that's no help, but I think it's realistic.

[Mars Amp Repair]

Mod to lengthen the life of the TDA7293 IC's

I'm not certain if anyone has posted this bulletin from Line 6 before, but her is their retrofix for make the TDA
s more hearty. Hope that helps. I never actually performed this mod, so I don't have any history of success or failure. from their description of why they're doing this, it seems plausible as a remedy for premature failure. glen

[R.G.]

Wow. It amazes me that with so much electronic history available to us, any manufacturer ever leaves off catch diodes on the output of a solid state amp. Amazing. Utterly amazing.

That's much like issuing a tech bulletin saying "OK, we got caught on that one. Put in the $0.05 parts to do it right.".

[J M Fahey]

Thanks for posting.
Funny that they explicitly recognize (obviously based on field use, not datashet specs or Lab work) that TDA7293:
a) does not drive 4 ohms reliably, so no "100 Watt" capability . 70/80 into 8 at best.
b) does not reliably stand voltage needed to comply with 70/80 into 8 rating anyway.
So, in a nutshell, very little beyond what LM3886 delivers.
Which does not surprise me either, packed into what amounts to 2 TO220 cases side by side.
I usually make 100W up amps, because my niche is people who actually play live regularly, and that power can play at ease along any drummer (the guy who *really* sets the volume reference on stage), I'm considering making 60/70W combo amps, because that sells a lot, and can get TDA7294 for peanuts, but I guess I'll go discrete instead, I have a huge stash of IRF640 MosFets.
Yes, 2 x TO220 packs, but I can separate them 2 inches for much better heat spreading and think they will be much more reliable.
And PCB will only have 5/6 extra parts, still simple enough.

[J M Fahey]

@ RG:
what I still don't get is that TDA729x chips claim to have output MosFets, which by construction (like it or not) *do* have built-in reverse diodes.
In fact, that's specified and analyzed in MOSFET datasheets , so you take them in consideration when you design switching PSU and the like.
That's why adding external diodes *should* be redundant.
In fact, in my regular TIP142/147 amps I don't use external diodes either, because they also have built-in ones.
Mystery.
I'm still not too sure about Line 6 bulletin, they may be chasing ghosts and adding these diodes needlessly when in fact the chip is underrated for the task.
Just a speculation, of course.

[Steve Conner]

Discrete MOSFETs have a diode from drain to source. But when they are integrated onto a chip, the diode goes from drain to "substrate" which is connected to Lord knows what else inside the chip.

I have experience with gate driver ICs that have MOSFET output stages but need external Schottky protection diodes to prevent them latching up and self-destructing. The datasheet even warns of latchup and specifies a maximum reverse current not to be exceeded. I killed several before adding the diodes.

[Enzo]

Well then it sounds like a good idea. I STILL won;t take in any MF350s for repair though.

[davman]

Thanks teemuk, very useful. I have edited your drawing to show Master/Slaves & their respective motherboard connectors.

[davman]

Just a few discussion points since my last blog (11/sept/14).

Someone should make an aftermarket power board for these things that replaces the modules with a bunch of good old-fashioned TO3s. I'd do it if I thought it would sell enough to make it worthwhile.

I don't actually understand what makes them so temperamental. As other posters said, two of the ICs are driven by the preamp, and the rest are just power followers. Theoretically, you should be able to remove all but the two master modules and it should still work, as long as you don't connect a low impedance load and drive it hard. If you reconnected the speaker unbridged, it should work with only one module.

But as we all know, theory is the same as practice, in theory. My gut feeling would be that the modules go into push-pull parasitic oscillations under some circumstances, and blow each other out.

........................................................ ........................................................ ..................................


a) In the interest of Science (& Steve Conner's comment, 4/4/09), I decided to remove both Slave modules (by unplugging the module connectors to CON3 & CON4) - because I too could not see any problem doing this (in theory).

RESULT: Output Ov dc & good signal into crystal earpiece (with 47K in series). Then 8 ohm monitor speaker added & was tested for several minutes at moderate volume - no problems with sound. IC's running cool.

b) I then reconnected one of the Slave modules - same procedure - and with similar results. So appeared to be stable enough at moderate levels with only 2 Masters & one Slave in circuit.


What I don't like about this design

i) The gain-defining resistors to the sensitive 'virtual earth I/P' of the Master modules are too far away (on the motherboard, several inches away from the modules). C1 helps (across TDA pins), but apparently not enough!
The corresponding wires to CON6 & CON7 then act as aerials & are more likely to pick up HF transients, which can be considerable when higher powers are used.
Perhaps this is why the IC manufacturer shows them positioned close to the chip (pin2) in their datasheet layout!

ii) No supply rail fuses, hence blown module tracks.

iii) Connectors to modules seem rather fragile - I always inspect them for tarnish & clean the pins with fibre pen, & sockets with alcohol.


Question: I wonder why the Zobel network on the speaker O/P is 100n/10ohm on the Mode4, but 100n/2ohm on the datasheet, for the same configuration?

I have added some more info on Teemuk's simplified diagram, also corrected component value error on it.

[Jazz P Bass]

Has anyone ever noticed that there are two datasheets for the TDA7293.
1 is from 1999 & the other is from 2003.

What is odd is that in Slave Mode the +IN & -IN are tied to -Vss on the 2003.
Whereas they are tied to Ground on the 1999 sheet.

Is this the root cause of the preponderance of failures.
Hmmm.

Link: Custom PCB for DIY electronics: Failure in the official datasheet of TDA7293

[davman]

Datasheet differences

Has anyone ever noticed that there are two datasheets for the TDA7293.
1 is from 1999 & the other is from 2003.

What is odd is that in Slave Mode the +IN & -IN are tied to -Vss on the 2003.
Whereas they are tied to Ground on the 1999 sheet.

Is this the root cause of the preponderance of failures.
Hmmm.

Link: Custom PCB for DIY electronics: Failure in the official datasheet of TDA7293

Thanks JPB for highlighting this. It may well be an 'improvement'.
This Amp was designed around the 1999 datasheet application circuit, so does not incorporate the latest changes.

I contacted a Field Applications Engineer from the chip manufacturer, who said it would be difficult to find the relevant info to explain the change - however, any TDA7293 purchased after the the latest (2003) datasheet should be wired to the application circuit recommended there.

In other words, the 2 Slave TDA7293's must now have +IN & -IN wired to -Vss.

I have done this mod on a recent repair (where the chips lasted 12 months after the last blowout) & all seemed well.

I will show pictures of the mod soon.

[ILAC]

Enzo, this ones for you if you dont mind. Been reading through many of these postings and i still have a question. Fixed one of my fans and before i put everything back together in regard to the fan,heatsink, module for that one side i decided to turn the amp on to see if the fan would kick on with me holding the plastic casing on the fan mount. The fan worked but in the process the heatsink touched the chassis. Ya i know...Stupid me! No spark at that time fan kept running no smoke from IC's. Shut the amp down. Proceeded to screw the fan back to the chassis and the first screw that hit the chassis sparked a little. Once all back together sure enough the IC's fried when i turn it on. The preamp still works fine through the headphone jack. Replaced all 4 IC's after about a month of the amp sitting. I did not discharge the caps since i thought that after a month they would have discharged, maybe im wrong. New chips blew instantly. My question is do you feel that when the heatsink touched the chassis it fried something on the main board that caused the chips to blow the 2nd time or should i have discharged the caps even though it sat for a while? The preamp works so im thinking the main board is OK, except maybe for the mute circuit? or would the preamp not have sound through the headphone jack if the mute circuit was bad? My understanding is that the mute circuit effects the speaker output. Will the amp still power on with new chips without frying them if the mute circuit is damaged or will it just have no sound? The caps are all good on the small module boards just the IC's were bad. I would love to hear you opinion before i hook up the ribbon cables to the module boards. New IC's in, caps are discharged, afraid to hook up the ribbon cables. Although i have many spare IC's i dont want to go through that again. Thanks in advance for your thoughts. I do apologize for the long winded post, just wanted to make sure all details were included.

[Jazz P Bass]

The metal plate on the back of the TDA ic's is connected to the minus power supply.
So, yeah, you should check that the power supply is good after being shorted to the chassis.
I would also check that there is no continuity between the ic heatsink & the chassis before powering up the unit.

The 'mute' pin is a slowly charging voltage that tells the TDA's to unmute.
I believe 3.5 Vdc is the unmute threshold.

If the ic's are new and the capacitors are discharged, you should be good to go.

[Enzo]

I can't really speculate what may have happened when two things touched that ought not to. We just soldier on as though we are just now seeing the amp. the mute circuit is a secondary protection. As a rule it doesn;t blow things up.


I avoid that model, and I think I have been consistent that I do not take them in for repair at my shop.

[g1]

Agree with JazzP that you have to really check the insulation of the heatsink. I think something in the fan mounting may be messed up. When you said there was a spark when installing a screw, that was a warning.
Make sure the caps are discharged before connecting modules, and make sure the heatsinks are completely insulated before applying power.