Yamaha EMX2000 repair

[Phil Allison]

Hi,

had one of these powered mixers on the bench this week - but without a clear fault report! The unit belongs to a church and was brought in by a regular customer, who did not use it personally.

The main problem proved to be similar, intermittent faults in both power amps - with very low and distorted (ie half sine wave) output when one or both were in fault condition. Both amps could work for periods of a hour or more between episodes of the fault condition.

The amp PCB did not respond to impacts or wiggling any multi-pin connectors and there are no rail fuses. So there was no option but to remove the heatsink and unbolt all 16 flat-pak devices to release the PCB.

This revealed a surprising situation - the solder on each leg of every TO3P device was visibly cracked, no need for a magnifying lens to see them either. Some devices almost fell of the PCB when it was released. Strangely, no other solder joints on the PCB showed any problems at all.

The unit is maybe 15 years old and looked well treated, so WTF was going on ?

The best hypothesis I have is the heatsink is the main culprit - it is a single piece of extrusion, about 20cm square, with raised, flat surfaces on opposite sides for two rows of TO3Ps bolted into threaded holes. The PCB itself is secured by 4 bolts fitted into stand-offs in a square pattern.

So, whenever the aluminium gets hot it expands and this pulls on all the flat-paks, tensioning the legs of each device against the solder joints. Do this thousands of times and fatigue cracks appear.

Not really sure if the solder was Pb free or not ( there is no marking on the unit ) but they all did have that dull look of the evil Pb fee stuff.

Anyone here seen the same thing before ?

BTW

The power amps use a really odd circuit topology called "EEEngine" by Yamaha - with six power Darlington devices, two of them operating as self oscillating, switching current sources.


..... Phil

 

According to

http://www.yamahaproaudio.com/global/en/training_support/selftraining/technology/eeengine.jsp

They are effectively using transistors as active filters for switching regulators, pretty much like class D.

High frequency and current are not nice to solder. Some TV manufacturers used eyelets at the connections to the horizontal (lie) output transistor for this reason. The transformer as well.

I've seen that topology before, may have been Yamaha, or some other brand. I don't know if they have it patented or not, or if it is patentable or worth patenting. Crown apparently patented their grounded bridge, how enforceable it is is another story. I think it would be pretty hard to prosecute.

I see this one is actually DC coupled, but there may have been no other way to make it work. Not sure.

 

[Phil Allison]

jurb...@gmail.com wrote:

According to

http://www.yamahaproaudio.com/global/en/training_support/selftraining/technology/eeengine.jsp

They are effectively using transistors as active filters for
switching regulators, pretty much like class D.

** A better description is that the class B output stage includes signal tracking supplies for each rail employing switching, buck regulators.

The attraction of this example is the extreme simplicity and low parts count - in part made possible by the use of Sanken power Darlingtons rated at 70MHz.

http://www.semicon.sanken-ele.co.jp/sk_content/2sd2560_ds_en.pdf


..... Phil

 

>"* A better description is that the class B output stage includes >signal tracking supplies for each rail employing switching, buck >regulators. "

The way I see it, the main thing it accomplishes is to get the filter out of the output line.

>"The attraction of this example is the extreme simplicity and low >parts count - in part made possible by the use of Sanken power >Darlingtons rated at 70MHz. "

What kind of frequency does that thing actually switch at ? Or didn't you check. No biggie if you didn't, I probably wouldn't.

Were your bad connections on the outputs themselves or the regulators ? If on the outputs, your theory bout the thermal expansion is probably right, but if on the regulators I think the high frequency had an effect on it.

However, of late it is starting to look like the HF has less of a deleterious effect on the unleaded solder than it did on leaded solder, though that is only supposition at this time, I don't have enough data. Not sure, but it is possible that even with all its faults, unleaded solder has a lower resistance than leaded, I don't feel like looking it up right now but that might explain it. In this newer stuff I am seeing good connections in the SMPS but bad one like at the display driver. And I get some strange ones, like in a microphone input that don't even respond to resoldering and I have to install jumpers. I'll get to the bottom of that one day but for now I just get the job out.

By the way, do you use unleaded solder on repairs ? I don't. I remember there was a time when I used Tektronix 2 % silver bearing solder on certain critical spots. You didn't see them but we had the Magnavox (before Phillips bought them) T995 chassis TVs that were actually a fire hazard because of the yoke connections. I used it on them.

But someone did ask on a forum and I said no, why would I use something that doesn't work as well ? That law is for manufacturers. I got a nice big pound roll of 63/37 that'll probably last me a lifetime.

Another question - you said you had to unscrew all the transistors, was there no way to remove the heatsink with the board attached ? Or did they make it so you can't do that ? I swear sometimes I wonder how they manage to even build these things. I mean, if it is that cumbersome to take apart, how the hell did they achieve a decent production rate putting it together ? Since labor is such a "huge" cost they say, why not make it easy ?

 

[Phil Allison]

jurb...@gmail.com wrote:

"* A better description is that the class B output stage includes >signal tracking supplies for each rail employing switching, buck >regulators. "

The way I see it, the main thing it accomplishes is to get the filter
out of the output line.

** But it is not a class D amp in any case, just a basic class B design with a a pair of "helpers" to eliminate most of the heat.

What kind of frequency does that thing actually switch at ?

** About 200kHz, the pos and neg versions are not very symmetrical.

You can see some residual riding on the output sine wave, along with minor crossover notches cos the output pair run zero bias.

BTW You need a decent analogue scope to see any of this, forget using any recent, low cost DSOs with LCD screens which totally blur detail.

Were your bad connections on the outputs themselves or the regulators ?

** It was ALL the devices on the heatsink - including a pair of 1amp, 15V regulators.

By the way, do you use unleaded solder on repairs ? I don't.

** I have some 99% tin solder - but never use it. The stuff is just awful.

Another question - you said you had to unscrew all the transistors,
was there no way to remove the heatsink with the board attached ?
Or did they make it so you can't do that ?

** The power amp module was easy enough to get out and then you access the underside of the PCB by unbolting all the power devices. A common enough assembly method these days and one Yamaha uses in a few models.

I was just trying to avoid doing that until I was sure there was a fault to be found there. It was amazing that with so many cracked joints, the PCB did NOT respond to hard impacts.


..... Phil

 

>"** But it is not a class D amp in any case, just a basic class B >design with a a pair of "helpers" to eliminate most of the heat. "

Class B or AB ? Things can be class B, they, according to their publications speak of "musicality" or some such IIRC. Class B is not very warming to that, though my Phase Linear does run the actual outputs in class B. It is just that there is enough feedback and low ohm resistors from the drivers that make it sound good at low levels. If it is class B there will be no bias regulator transistors or anything to do with idle bias. Is that so ? The Phase Linear you set the bias for the drivers and they are actually the outputs up to a certain level. Is it like that or what ?

>"** About 200kHz, the pos and neg versions are not very >symmetrical. "

And you said something about self oscillating ? That sounds like a real bitch to engineer but I suppose it can be done. But from what I know about that topology is the collector voltage just stays a few volts higher than the output voltage.

Actually in the past there were transistors that did not like that. An example would be the NEC N-965 VHS, which you probably have not seen. they had transistors for the reset the had 5 volts on the collector and then wanted 5 volts on the emitter so the put 5 volts on the base. They were a common failure. I determined that the problem was that the transistor could not maintain hfe with that low of a Vce. I replaced them with 2SD612Ks which were bias transistors and therefore made to work at low voltages. Never failed. Others had modifications running wires up to a heat sink, drilling hoes and putting the collectors to the 12 volt source. Fuck all that, but that is why I got the big bucks.

So in a way I wonder if these class B outputs might not fail the same way. With the ones in the reset of the N-965 they seemed to fail with an open base. They read infinity every which way on the meter. Could have been an open emitter, no way to tell.

>"** It was ALL the devices on the heatsink - including a pair of >1amp, 15V regulators"

Well that pretty much clinches it that it was heat. Remember when STKs had the loops on the pins ? Obviously they were to eliminate that mechanical stress. I guess shit like that is too much trouble now. But hey, they paid the bill and you mad some money I hope. If nothing ever broke, we would be broke.


>"** The power amp module was easy enough to get out and then you >access the underside of the PCB by unbolting all the power >devices. A common enough assembly method these days and one Yamaha >uses in a few models."

So, after all that do you clean it and put new heatsink compound on ? (also known as bird shit) And then if there are insulators it is even worse. In fact I had a problem with that with a Yamaha, the insulators were big, one for the whole bank of transistors and it was stuck, no way to get it apart without damage. I was looking for a solution to that but quit that job and left them to their own device. It might be different in your country but here people like us are like gold. I know in Europe people know things and can fix things, but not in the US. They can't even change a flat tire.

>"I was just trying to avoid doing that until I was sure there was >a fault to be found there. It was amazing that with so many >cracked joints, the PCB did NOT respond to hard impacts. "

I am pretty goddamn good but I am not sure what to tell you about this. It must be that the mechanical was so strong that even a hard smack would not affect it. Maybe if you had pressed down on the PC board right near the heatsink really hard, not hitting but applying pressure, while in fault condition of course, you may have been able to tell without taking the whole damn thing apart.

that reminds me I got a bunch of amps that have a channel cutting out and it is as if it is a bad connection, and AFTER the volume. I did prelims on them and thought it was the speaker relay(s) but found a couple which is wasn't. I see no bad joints but it has to be one, and I got it on either channel, we got 53 of these things. Not great amps, hundred a channel but can handle four ohms so two hundred a channel. HA, I AM factory service for these things and I have to operate like a boneyard.

They all got a defect in the mechanical design. The bias transistor has no thing forcing it to make contact with the heat sink. Well we know what happens then. And the factory put heat sink compound on it so that means the design wants it. It was a SNAFU between the electronics end ans the mechanics end. But if we sell them like that they are all coming back and they will have to be refunded when, not if, they fry out. They do not want that. I had an idea with a wire clip, under a screw of the output to hold it down. I prefer that because it is plastic and you got no worries about the UL or anything but the boss was asking questions, like why are all these sitting here so I am considering using a paper clip. That is a FACTORY modification !

I have seen worse.

 

[Phil Allison]

jurb

"** But it is not a class D amp in any case, just a basic class B
design with a a pair of "helpers" to eliminate most of the heat. "

Class B or AB ?

** It just happens to be class B = zero bias current in the output devices.

The full manual and schem is available here:

http://elektrotanya.com/yamaha_emx-2000_sm.pdf/download.html

All outputs are Darlingtons, some forward bias voltage is provided set by two diodes and a resistor.

"** It was ALL the devices on the heatsink - including a pair of
1amp, 15V regulators"

Well that pretty much clinches it that it was heat.

** Yep.

"** The power amp module was easy enough to get out and then you
access the underside of the PCB by unbolting all the power devices.
A common enough assembly method these days and one Yamaha uses in
a few models."


So, after all that do you clean it and put new heatsink compound on ?

** The module has long strips of insulating material - the grey coloured silicone, re-inforced with fibreglass threads kind. I was able to re-use them with no problem, cos none of the TO3P devices dissipate really high wattages.

BTW: I think the stuff is a poor heat conductor, compared to mica and white grease.


...... Phil