Did I just toast my Hafler XL-280? Need help/advice!

polyshades

Member
2016-04-30 12:02 pm
I am pretty new to working on power amplifiers. I bought a used Hafler XL-280 off ebay and measured the DC offset. The right offset was -100mv and the left was about 10mv. Of course I opened it up anyway trying to "fix it" and set my multimeter to measure amps. The bias was around 290ma-300ma which seemed ok though the DC offset wouldn't budge from -100mv when adjusting the bias trimpot. I figured I probably would need to replace/match the transistors to get the dc offset lower.

Being a huge idiot, and it being 4 AM, I then proceeded to connect the multimeter to the speaker outputs thinking I was checking DC volts, but forgot it was still hooked up as an ammeter. Poof! Both speaker fuses blew (2A). I replaced the fuses but the damage was done. The right channel output now shows something like 38 volts and the left channel shows something like 16 volts. It doesn't seem to pass signal from the inputs.

Dammit. :mad:

Is this at all repairable? I'm not sure how much permanent damage I did, especially to the transformer and/or mosfets, and no idea how to begin assessing the situation. At this point, I'm ready to sell the damn thing for parts and pretend the whole thing never happened...but if I just toasted the transistors then I might have a chance...

Any advice or help is appreciated...thanks!
 

polyshades

Member
2016-04-30 12:02 pm
Ah, I guess I misunderstood my friend who took a look at it before I did. I thought there was a relationship between bias and DC offset, since he said something like "the amplifier won't bias" when I asked him about adjusting the DC offset. So the only way to adjust offset is to replace parts/match transistors?

The internal power supply fuses F1 and F2 don't appear to be blown.
 

Trollet

Member
Paid Member
2015-07-20 9:35 pm
Westcoast
I bought those mosses from a retiring music instrument shop/repair man.
They did also manufacture PA-amps in the eighties but I guess there was no success :(
I bought a lot of spare parts and a bunch of parts from their manufacturing.
They are all dated to mid eighties and no fakes :)

About the Hafler - These mosfets are rather sensitive especially to exeded gate voltage.
If they are blown they usually short between pins so you probably can do a easy check.


Best regards Figge.
 

Trollet

Member
Paid Member
2015-07-20 9:35 pm
Westcoast
I've been looking to the schematic from Hafler homepage.
The component values are'nt printed on it, but thats less inportnant.

I would check D1-D6 and the drivers first. Then the outputs.
Then turn it on and measure the voltages in VAS and driverstage.
The fact that F1 and F2 are still alive talks against that the mosfets are gone I think. But there are many much moore skilled technicans on this forum than me so...:shy:

When it was shorted it propably pulled current, or a spike, "backwards" in some way.
Measuring without knowing the values means that you have to do some calculating and comparing (balance between halves) and see if your measurements seems resonable.

By the way - what are the railvoltages B+, B++ etc.

Figge.
 

polyshades

Member
2016-04-30 12:02 pm
So I poked around a bit more. First thing is doing a visual inspection I noticed the legs of diodes D3 and D4 had some soft black stuff around them, maybe a sign of diode failure? It was very soft though and flaked off. I briefly tested conductivity around the diodes while in circuit, then tried to remove them from the board to test properly but could not for the life of me manage to desolder and pull them out. Is there some trick to pulling out components on old two sided boards?

Testing DC offset again it measured about 16volts with nothing on the inputs. if I short the inputs, the DC offset goes to zero, which I guess makes sense if it is DC coupled. Contrary to my original post, the unit does pass a test signal, however it is very noisy and the offset remains high at 16 volts. For instance, sending a 1kHz square wave of 1 volt amplitude in, I get a pretty gnarly looking 22 volt amplitude square wave coming out.

In the Hafler schematic towards the end, it appears to show some points on the board to test the voltage levels. Other than the rail voltages, none of the other voltages I get match up. The mosfets don't seem to short between pins as you mention so crossing my fingers they are ok...

Rail voltages:

B- : -66v
B+: 66 v
B--: -75.5v
B++: 75.5v

Once I figure out how to pull out components, looks like I'll just need to systematically check through them one by one...unless I can find a better way to narrow down the problem.

Thanks for all the help/pointers so far. Trying to stay motivated rather than bang my head against the wall...
 
Last edited:

Trollet

Member
Paid Member
2015-07-20 9:35 pm
Westcoast
Hi.
It seems to be some kind of problem in the inputstage.
If the output goes to zero when grounding input it looks to me that there's imbalance between the LTP's. Checking drop over the resistors may give you a clue.
There may be a good idea to check Jfets and Transistors in IPS.

There's no need to resolder the diodes if they measure OK in circuit. If they are dead it will show quickly on diodetest.
If the outputs were blown the rail fuses most certain also will leave for next life.

+/- 75 volts rails is serious stuff ;)

Figge.
 

Trollet

Member
Paid Member
2015-07-20 9:35 pm
Westcoast
I just measured a LSJ74.
With diode test it had 0,753 volts drop source - drain and 0,02 volts gate drain. Gate don't conduct other way if I remember right.
So it shall be somewhere in that area.
The drop over R11 - R14 shows where the current flows. V/R gives the current.

Are there 2sj44 and 2Sk163 in the Circuit ? If any of those are broken it may not be easy to get new ones.
I've heard that they could be replaced by 2 Sj74 and 2SK170 but I guess there are others on this forum that knows this better than me. Diyaudio store sells LSJ and LSK, matched.


Figge.
 

polyshades

Member
2016-04-30 12:02 pm
If any of the output mosfets were actually shorted, wouldn't that mean I wouldn't see AC signal on the speaker terminals? In other words, if I am able to input a 1 volt squarewave and see a 22 volt square that indicates the mosfets are ok, right? And since my rail voltages seem within 1-2 volts of spec the power supply section is probably ok too?

If so then that would seem to at least indicate the problem is isolated to the PC-40 board. But maybe I'm oversimplifying.

I think there may also be a loose joint somewhere. The DC offset is unstable. With no input sometimes it is -67vdc / 1.2 vdc, other times it will be around -34 / -16, then if I apply a mono input to L/R the DC offset sometimes evens out to -16 / -16. Shorting the input doesn't have any effect on offset (I can't repeat the previous test I did, not sure why.)
 

polyshades

Member
2016-04-30 12:02 pm
Turns out I had very old desoldering braid, which was basically dried out and useless. Now I can actually pull components...

All resistors appear to be ok. Diodes appear ok.

With a diode test, P-channel 2SJ72 FET gets around .8 volts drain-gate, .8 volts source-gate, non-conducting when probes reversed. I get around .02 volts drain-source when gate is turned on. I can turn on/off drain-source by touching the gate with DMM probes, which seems to indicate the FET is working.

@ Trollet, not sure your numbers for LSJ74 make sense. I would expect .753 volts was between gate-drain and .02 volts between drain-source. Drain-source seems like it should only ever show open/closed depending on the channel state.
 
Are Q15 and Q16 fine? I would doubt the IPS is gone (unless you blew something while testing). Usually with a shorted output, drivers and outputs cook. I would replace all output transistors and driver transistors, and test the rest. If the diodes test fine, leave them. Do the zeners measure their correct voltage when the amp is on? I think in general, the zeners should be ok if they measure a diode-drop (0.6 V - 0.8 V) anode to cathode (diode test - amp off).

Unfortunately with MOSFETs, you can't just replace a transistor with another with similar voltage and current ratings.

Good luck! I've burned many a transistor. It's just easier to replace them all.

ETA: Sometimes resistors appear to be fine. Just make double-sure. They often burn on the side you can't see.

Oh, and please make sure your power supply caps are discharged before working on the amp. 150 V will give you an unpleasant kick, and can very well blow something while you're testing. I use a 10 W 10 ohm resistor to discharge my caps.
 
Last edited:

polyshades

Member
2016-04-30 12:02 pm
I pulled all the transistors from the board. Q14/Q15 (and all others) pass the diode test. I can't seem to get any readings for hFE for any of them though. I'm not sure if no hFE measurement indicates transistor failure or if my DMM is just unable to measure the hafler ones. However I tested another transistor I had lying around (2N3904) and the DMM did show an hFE around 170.

Is the diode test "good enough" to figure out if a transistor is bad? Or should I be paying more attention to the hFE reading?
 

phase

Member
2004-10-04 11:59 pm
You should probably think about carefully removing the output mosfets and testing them(do a search on that). That will rule out needing to write mr. Qua-Co (epay vendor) a check for $200 for complete matched sets of 6 + 6 new parts.
They basically will toggle on and off by energizing them in a way, then testing to be sure that they conduct in the manner as intended. They can be removed by removing the screws and lightly prying them away from the heat sink.
 

polyshades

Member
2016-04-30 12:02 pm
Fixed it!

Repair steps:
1. Resoldered loose/corroding power supply wires for the right channel PCB. I think this was causing intermittent problems.
2. Replaced R23 47k ohm resistor which had failed open.
3. Replaced R201 5ohm thermistor which goes between input gnd and power board gnd. This is there to reduce hum/noise apparently? This seems to have been the main failure that caused high DC offset in both channels.
4. Since I had the unit open and had extra parts, I added an Ametherm 5 ohm thermistor (inrush current limiter) between the main power fuse and the power switch. This easy mod was suggested in this thread. It seems to reduce the initial power thump when turning on the unit.
5. Adjusted the bias on both channels to ~290mA (manual recommends around 300mA)

I think some of the problems may have happened during shipping, and when I shorted the output it put more thermal stress on the components.

Somehow after all this I actually improved the initial DC offset. It's now -66 mV in the right channel, 14 mV in the left channel. Not perfect, but it's within operating range based on the Hafler manual so I'd rather just use it as-is than spend time and money on any upgrades/fixes right now.

Also during this process, I bought a cheap Yosoo GM328 unit off Amazon, which gave me much more confidence when testing transistors and capacitors. Very useful device!

Thanks for all the help in this forum and to Mr. Qua-Co, who suggested I double check the power supply board components again before attempting to replace transistors. That suggestion lead me to find the R201 problem, a very cheap fix.