Hi,
I have a H&K Attax 100 on the bench blowing +/-45V rails fuses, the two output transistors with C-E shorted. (MJ4035/MJ4032). Schematic attached.
One of the connectors connecting the transformer to the PCB was broken, there was a bad contact.
Those output transistors seems to be hard to find. I checked a cross reference table, MJ11015/MJ11016 seems to be a potential replacement and easy to find.
I am not overly familiar with solid state amplifiers, could you confirm that those two transistors are a viable replacement ?
Also, I checked the bridge rectifier, filter caps, T5, T6, T7, T8, D31, D32, D33 and D34, no shorts, no broken solder. What should I check next ?
Thanks
I have a H&K Attax 100 on the bench blowing +/-45V rails fuses, the two output transistors with C-E shorted. (MJ4035/MJ4032). Schematic attached.
One of the connectors connecting the transformer to the PCB was broken, there was a bad contact.
Those output transistors seems to be hard to find. I checked a cross reference table, MJ11015/MJ11016 seems to be a potential replacement and easy to find.
I am not overly familiar with solid state amplifiers, could you confirm that those two transistors are a viable replacement ?
Also, I checked the bridge rectifier, filter caps, T5, T6, T7, T8, D31, D32, D33 and D34, no shorts, no broken solder. What should I check next ?
Thanks
MJ11015/16 from Motorola (obsolete) have an actual SOA chart as opposed to the MJ4032/35 from mospec, so in that respect they are a better part. Darlington, 120 vceo instead of 100 should be a fit. The Ic at 120 v is 100 ma, so you can't actually use them at 120 v Vce.
That soa is 0.3 A @ 90 volts and 4 A at 45 v so I wouldn't repair this amp without a major redesign. I definitely wouldn't guarantee it to a customer. Single TO3 output pair is generally reliable at +-35 v and 50-70 w/ch so unless the transformer has major resistance limitations this design is hugely optimistic. Plus darlington output designs have the reputation of being much less reliable than designs with separate driver transistors with separate heatsinks. See comments by Sakis who ran a repair shop in Greece catering to the beach bar trade. Life of this design with an infinite heat sink maybe weeks? If ever run more than an hour @ 100 w/ch. Dual output transistors would extend the life some, or changing the transformer to +-35. But darlington output SS amps were avoided by major vendors Peavey, Crown, QSC, Yamaha, that had a reputation to protect. Sakis named a popular Greek brand that used darlingtons that went bankrupt in the early 80's.
That soa is 0.3 A @ 90 volts and 4 A at 45 v so I wouldn't repair this amp without a major redesign. I definitely wouldn't guarantee it to a customer. Single TO3 output pair is generally reliable at +-35 v and 50-70 w/ch so unless the transformer has major resistance limitations this design is hugely optimistic. Plus darlington output designs have the reputation of being much less reliable than designs with separate driver transistors with separate heatsinks. See comments by Sakis who ran a repair shop in Greece catering to the beach bar trade. Life of this design with an infinite heat sink maybe weeks? If ever run more than an hour @ 100 w/ch. Dual output transistors would extend the life some, or changing the transformer to +-35. But darlington output SS amps were avoided by major vendors Peavey, Crown, QSC, Yamaha, that had a reputation to protect. Sakis named a popular Greek brand that used darlingtons that went bankrupt in the early 80's.
Last edited:
If you can find them from reliable suppliers (no EBay, etc. but, say, Mouser, Digikey, Farnell, etc.)) yes, they are "better" in all aspects than originals: 120V, 30A, 200W instead of 100V, 15A , 150W
EDIT: as of Darlingtons in modern Guitar amps: Fender, Laney, Hugher & Kettner, Ampeg, Fishman Loudbox, Marshall (plus yours truly) use plastic TIP142/147 by the truckload, they are proven robust after a couple decades or more.
EDIT: as of Darlingtons in modern Guitar amps: Fender, Laney, Hugher & Kettner, Ampeg, Fishman Loudbox, Marshall (plus yours truly) use plastic TIP142/147 by the truckload, they are proven robust after a couple decades or more.
Last edited:
If T6, T7, T8 and their associated resistors/diodes are ok, then it won’t blow straight up if new outputs are installed. Still best to start it up on a dim bulb, because the bias might run high and need to be trimmed down. When higher power outputs are used, vbe often runs lower so Ic goes up. No bias pot so you have to do it the hard way. Might be ok, but you might end up with a 200 milliamp standing bias, too. Dim bulb will at least let you know before a new set of outputs dies. If anything else is busted it won’t be fatal and you can address if necessary.
MJ11015/6 should be ok. I’ve analyzed the SOA protection - it limits at 3 amps at the zero crossing and allows about 6.5 amps at clipping. But it does require a 200 watt transistor for safety. 100 watt like TIP142/7 is a little outside it’s comfort zone. They are the most rugged of all TIPs though. 11015/6 will likely become unobtainium soon, if not already.
MJ11015/6 should be ok. I’ve analyzed the SOA protection - it limits at 3 amps at the zero crossing and allows about 6.5 amps at clipping. But it does require a 200 watt transistor for safety. 100 watt like TIP142/7 is a little outside it’s comfort zone. They are the most rugged of all TIPs though. 11015/6 will likely become unobtainium soon, if not already.
Thanks for your inputs,
MJ11015/6 are still available at most of the reputable resellers, and marked as active at Onsemi.
I will try it, it will be the opportunity to try the limiting bulb, never had to do it so far.
I was considering ditching the whole power amp and replace it with a class D module, but I thought that the original power amp could play a role in the sound ? There is negative feedback through R146/C85, does it change the frequency response of the amp ?
What about the damping factor of this amp compared to a class D amplifier ?
As for the bias, what should be the idle current ? Can I achieve variable bias by replacing one of the resistors around T6 by a trimpot ?
Hypothetically, would it be possible to use mosfets instead of BJT in this topology, without too much modifications ? Or does it require a totally different topology ?
Also, I'm searching for articles or books analyzing in details (with calculations) the different topologies of solid state power amps. I've read Audio Power Amplifier Design by Douglas Self, but it only scratch the surface I think.
MJ11015/6 are still available at most of the reputable resellers, and marked as active at Onsemi.
I will try it, it will be the opportunity to try the limiting bulb, never had to do it so far.
I was considering ditching the whole power amp and replace it with a class D module, but I thought that the original power amp could play a role in the sound ? There is negative feedback through R146/C85, does it change the frequency response of the amp ?
What about the damping factor of this amp compared to a class D amplifier ?
As for the bias, what should be the idle current ? Can I achieve variable bias by replacing one of the resistors around T6 by a trimpot ?
Hypothetically, would it be possible to use mosfets instead of BJT in this topology, without too much modifications ? Or does it require a totally different topology ?
Also, I'm searching for articles or books analyzing in details (with calculations) the different topologies of solid state power amps. I've read Audio Power Amplifier Design by Douglas Self, but it only scratch the surface I think.
Class D amps are basically designed for Hi Fi (duh!!) so usable but dull/flat for Guitar use.
The original one has mixed feedback to kill damping and please Guitar players better.
In any case, I guess you just have this single one and the idea is to make it playable again, so guess it does not merit much as far as deep redesigning, , so try to repair it in the most practical way .
As is, a very good sounding amplifier, and very well built (by Guitar amp standards that is)
In a mid sized Guitar combo comparison, its little brother Attax 40 came out as the most "tube sounding", yet it was the only one not using a tube, go figure. All others were Valvestate type, with a single 12AX7.
The original one has mixed feedback to kill damping and please Guitar players better.
In any case, I guess you just have this single one and the idea is to make it playable again, so guess it does not merit much as far as deep redesigning, , so try to repair it in the most practical way .
As is, a very good sounding amplifier, and very well built (by Guitar amp standards that is)
In a mid sized Guitar combo comparison, its little brother Attax 40 came out as the most "tube sounding", yet it was the only one not using a tube, go figure. All others were Valvestate type, with a single 12AX7.
If the idle bias comes out anything less than 10 or 20 mA, just leave it. If it comes out high replace R145 with a series combo of 6k8 and a 2.5k or 5k pot. Measure voltage across the emitter resistor with no speaker connected. It may turn out very low, or zero - sometimes the designer intended to run fully class B. If that’s the case it won’t be very sensitive to transistor type changes and shouldn’t require adjustment at all. I’m not super familiar with the BD679’s vbe/ic curve other than it’s a darlington, but the resistor ratio looks like bias would be toward the low side.
The most straightforward fix is to put in MJ11015/6. Don’t think that MJ11032/3 are any “better”. They are the next size up, BUT— SOA above 30 volts VCE is exactly the same. You will not have any reliability gain and they are stupid expensive now. The 2N equivalents of your original types do show up on the surplus market from time to time. And yes, usually they are real (literally no market for them now and they just want to move them). Yeah, I buy them, but not for amplifiers running off +/- 45 volts.
Any other fix requires some sort of workaround or kluge to fit parts in there that weren’t supposed to. Depending on what kind of machine tools you have access to it could be very good or very bad. Paralleled TIP142/7, discretes using Real Audio Transistors, MJ15024/5 with separate drivers. Possible, but doing it right is not cheap or easy. Mosfets? Theoretically possible since the amp uses a double bootstrap to get the drive voltage above the rails on BOTH sides. But you would have to zener clamp Vgs and totally re-think the Vbe multiplier. Then retrofit TO-247 case devices on the heat sink, since the only viable parts here are IRFP240/9240. They are the only ones I would trust here, unless you manage to find non-fake laterals. And the TO-3 cased version of the 240 is long gone.
The most straightforward fix is to put in MJ11015/6. Don’t think that MJ11032/3 are any “better”. They are the next size up, BUT— SOA above 30 volts VCE is exactly the same. You will not have any reliability gain and they are stupid expensive now. The 2N equivalents of your original types do show up on the surplus market from time to time. And yes, usually they are real (literally no market for them now and they just want to move them). Yeah, I buy them, but not for amplifiers running off +/- 45 volts.
Any other fix requires some sort of workaround or kluge to fit parts in there that weren’t supposed to. Depending on what kind of machine tools you have access to it could be very good or very bad. Paralleled TIP142/7, discretes using Real Audio Transistors, MJ15024/5 with separate drivers. Possible, but doing it right is not cheap or easy. Mosfets? Theoretically possible since the amp uses a double bootstrap to get the drive voltage above the rails on BOTH sides. But you would have to zener clamp Vgs and totally re-think the Vbe multiplier. Then retrofit TO-247 case devices on the heat sink, since the only viable parts here are IRFP240/9240. They are the only ones I would trust here, unless you manage to find non-fake laterals. And the TO-3 cased version of the 240 is long gone.
Is this possible to do the same with class D ? I've been asking myself this question for a long time, since I played with TPA3116/8 modules.
I also tried the ICEpower AS modules, I suppose those class D amps are "tuned" to be a better fit for instruments amplifiers.
I will try the MJ11015/6, monitor bias and see how it goes. Thanks for the answers, very useful.
Last question, if you had to design a solid state amp for guitar, 50-100W into 8 ohms, what design would you be inspired by?
I am certain it can be done; someday "somebody" will have to take the bull by the horns and try it.
I am not being pressed by that since 50-100W is still ample for Guitar, and that is easy for Class AB amps ,just a couple TIP142/147 or MosFets, and for Bass amps which use >200W RMS, where it´s advantageous, mixed feedback is not really needed (although it still helps if available).
Not sure what that label means.
Perhaps that they take more abuse than usual?
Might be; otherwise I do not see any significant sound difference.
You are asking the wrong guy 😉 , I design from scratch, stating with a blank piece of paper 🙂
That said, "there is a Fender inside every Guitar amp" , even in Marshall or Vox, go figure.
And by extension in all the derivatives, such as Mesa, Soldano, Engl, Diezel, whatever, which are not "true" new designs but endless tweaks on the classics, adding a tube or two for extra gain, slight EQ tweaks, etc.,
In the SS World, unjustly despised Crates are incredibly good, their dynamically varying waveform distortion is to die for, Hugher & Kettner too, and of course Tech 21/Sansamp ; most others not that much, even those adding a tube for flavour (Valvestate, etc.) and failing miserably.
I started making amps in 1969, same year I started Engineering, and in fact Electronics was a "mild" hobby which I could do in an Apartment, compared to my earlier wild Chemistry experiments, model airplane building (and flying), rocket launching with self made Rocket engines, etc. but it grew out of control.
I started , like everybody, cloning Tube Amps, straight from Jack Darr´s "Guitar Amp Servicing" book back pages, so Fender, Gibson and Ampeg it was, first one a Gibson GA5 , their version of Fender Champ, (No Internet way back then) but in 1972 Argentina fell in one of her periodical Crisis and all importing was blocked in its tracks, so NO tubes of any kind, period.
Most makers fell into TV tubes (6DQ6) which were locally made, but by then I was starting to learn about Op Amps, the "universal building block", so I started designing around them, mimicking and cloning what tubes did on their own.
Starting with independent discovery of mixed feedback, since my cloned Twins had output damping=1 (I had measured it) and that was easy to do applying Operational Amplifier technology to SS amps.
Same with asymmetric clipping, variable duty cycle, lots of intermodulation, etc.
Very advanced in the 1972-1976 period where I developed most of it, and so successful that locally famous Musicians sold their Twins and Marshall all tube heads to buy mine, go figure.
Same as selling Ampeg SVT Classic heads to buy mine, replacing original 2 x 6550 Leslie power amps by mine, etc..
So I suggest you build a Crate Flexwave or an Attax amp, listen to it, and take it as a reference,you´ll be starting with the right foot 🙂.
MJ15015/16 are suitable and available 2N6287/84 also suitable and available.
Im sure you have checked all other transistors on the board.
This amp has simple over current/ short protection T7/T8
and is somewhat common with output transistor faults
these transistors are damaged as well.
Even possible for 2nd gain stage T5 to be damaged.
Since output section damaged check .22 ohm 5 watt emitter resistors
R158/159 make sure not open or have excessive drifting.
likely 10% rated so measured values should be within that range
Very simple fix, and fun simple amplifier to fix.
Class D amplifier sound horrid going into distortion.
much more elaborate preamp design to make
work for guitar amp.
much easier to replace 2 transistors
and run again.
check all transistors and start with light bulb limiter
check for DC offset on output.
no need to change bias.
idle current would be set at a low conservative value
and is very common and intentional for MI amplifiers.
likely around 12 to 20ma.
Im sure you have checked all other transistors on the board.
This amp has simple over current/ short protection T7/T8
and is somewhat common with output transistor faults
these transistors are damaged as well.
Even possible for 2nd gain stage T5 to be damaged.
Since output section damaged check .22 ohm 5 watt emitter resistors
R158/159 make sure not open or have excessive drifting.
likely 10% rated so measured values should be within that range
Very simple fix, and fun simple amplifier to fix.
Class D amplifier sound horrid going into distortion.
much more elaborate preamp design to make
work for guitar amp.
much easier to replace 2 transistors
and run again.
check all transistors and start with light bulb limiter
check for DC offset on output.
no need to change bias.
idle current would be set at a low conservative value
and is very common and intentional for MI amplifiers.
likely around 12 to 20ma.
6487/84 are obsolete - but apparently Digikey still has some dwindling stock. They are some of the types that show up on the surplus market from time to time because no one is using them anymore (at least not in manufacturing quantities). They are closer to the original MJ types, and 11015/6 is the upgrade. NTE still has most of the old TO-3 darlington types - probably take them a while to unload them all too since all they are going to sell is for the odd stereo or guitar amp rebuild. All the old switching applications that used to call for them have long since gone to mosfet.
I was surprised to see MJ11022/11021 still on the ‘active’ list. I thought they were long gone. Full power handling to 50 volts.
Microsemi is still making all the old JAN versions. At ridiculous prices, of course. Since those old darlingtons have been used in all manner of switching applications decades ago, the military still needs to have a source. And that was from when EVERYTHING for military application needed hermetic packaging. Modern stuff is moving away from that, but 30 years ago you didn’t even mention ‘plastic package’.
I was surprised to see MJ11022/11021 still on the ‘active’ list. I thought they were long gone. Full power handling to 50 volts.
Microsemi is still making all the old JAN versions. At ridiculous prices, of course. Since those old darlingtons have been used in all manner of switching applications decades ago, the military still needs to have a source. And that was from when EVERYTHING for military application needed hermetic packaging. Modern stuff is moving away from that, but 30 years ago you didn’t even mention ‘plastic package’.
Last edited:
one thing that worries me (when overdriven ) if the overcurrent transistor T7 trips, the drive current into that node by T5 is not limited , destroying usually T5 and or T7.
a small resistor in the T5 emitter and 2 diodes from rail to base could remedy that.
a small resistor in the T5 emitter and 2 diodes from rail to base could remedy that.
That current is limited - by 139 ohms of resistance (Plus a couple of diode drops). Might not be enough to keep T5 safe for sustained shorts to ground on the output. 45/139 = about a third of an amp. Current limiting T5 itself to say, 20 mA is usually a good idea. But that does require installing an emitter resistor in it.
Those transistors must be darlingtons so they need to be replaced with mj15031 32 drivers and mj15001 4 ? Outputs. If TO3 pkg
Tip125 is a to220 darlington
Tip125 is a to220 darlington
If I was going to replace the darlingtons with discretes that’s exactly what I would use. It would be somewhat of a kluge regardless.
TIP125 isn‘t up to the task, even if it was in a compatible case. TIP142/7 would have a chance of working.
TIP125 isn‘t up to the task, even if it was in a compatible case. TIP142/7 would have a chance of working.
Yes.
It being a combo, if connected only to internal speakers and no chance of external ones, they will be on the edge but survive , I made a few such amps and substitutions.
Not in a head, the slightest short will blow them, even if with "short protection", SOAR is a b*tch.
Modern TO3 transistors such as MJ1502x are best, way better than originals, but of course will need drivers such as mentioned above (mj15031 32), some tracks will need to be cut and a small PCB made to accommodate them.
They may be mounted "in the air" (similar to PCB less chipamps) but vibration (normal in a combo shaken by its own speakers, plus regular transport all over the place) will eventually crack their legs, a simple stamp sized PCB will solve that.
It being a combo, if connected only to internal speakers and no chance of external ones, they will be on the edge but survive , I made a few such amps and substitutions.
Not in a head, the slightest short will blow them, even if with "short protection", SOAR is a b*tch.
Modern TO3 transistors such as MJ1502x are best, way better than originals, but of course will need drivers such as mentioned above (mj15031 32), some tracks will need to be cut and a small PCB made to accommodate them.
They may be mounted "in the air" (similar to PCB less chipamps) but vibration (normal in a combo shaken by its own speakers, plus regular transport all over the place) will eventually crack their legs, a simple stamp sized PCB will solve that.
Or you just use a 150/200 watt T03 Darlington like the original.
Hence MJ11015/16, MJ11033/34 , 2N6284/87
Which have already been discussed
Hence MJ11015/16, MJ11033/34 , 2N6284/87
Which have already been discussed
The key spec for ruggedness is how high in voltage they handle full power. 11022/21 are the best available in this regard, with full power to 50 volts. 11015/6 only go to 30 volts like most all of the others, but the initial roll off is mild and you have useful SOA at 45V. 11033 falls like a stone above 30V (just don’t use them above that as linear amps - actually worse than TIP35’s here). When you exceed the dashed THERMAL limited portion of the SOA all it means is that you are exceeding the maximum rated junction temperature. You can do this for HOURs or WEEKs before failure (Depends how far over, lifetimes go down by a factor or two for each 10C increase). If the music does this for short periods of time (tens of milliseconds at a time) it is just no big deal. Long term reliability has to do with the accumulated average. But exceeding the second breakdown limits has altogether different consequences. Failure is instantaneous when it happens - it is a one time event not an accumulated average. If you look at the pulsed curves (for thermal and S/B) they tell you how LONG you have until you hit the limit. You may only have 5 milliseconds - ONCE - before catastrophic failure. There is margin in the S/B curves - but experience has told me (probably you too) not to push one’s luck.
Yes , very much I like to assume worst case scenario.
Actually very good point.
Thermally knew these T03's were pretty tough.
Just rummaged through datasheets
And yes S/B voltage was a lot lower than expected
on a few of my favorite Darlington's.
As always, real trick is lowering current with multiple devices.
But for doing repairs with Single pair amplifiers pretty common like this.
11022/21 are really holding up in S/B territory
And have learned not to assume S/B is OK
That was a fun exercise rummaging through datasheets.
Good info wg_ski
Actually very good point.
Thermally knew these T03's were pretty tough.
Just rummaged through datasheets
And yes S/B voltage was a lot lower than expected
on a few of my favorite Darlington's.
As always, real trick is lowering current with multiple devices.
But for doing repairs with Single pair amplifiers pretty common like this.
11022/21 are really holding up in S/B territory
And have learned not to assume S/B is OK
That was a fun exercise rummaging through datasheets.
Good info wg_ski
Multiple devices is better if you can do it - but that’s even more of a kluge than putting in TO-3Ps where TO-3s go. Drilling extra holes in the board and running wires isn’t something you want to do with an amp that has a very loud built in speaker. When something shakes loose in a solid state amp bad things can happen. Multiple darlingtons is problematic because current sharing depends on the accuracy of the built-in base-emitter resistors. The tolerance is pretty **** poor. When multiples are used, they should be discretes with a single driver. And one base-emitter resistor (actually base to output, Re inclusive) serving the entire bank.