The outputs are running hot, only way to do that is to conduct a bunch of current. That is no mystery. We already know those emitter resistors will show excess current, the transistors are hot.
You shorted across Q16 and they went cold. That is not surprising either, since Q16 controls the whole thing. That experiment told us that the outputs are controllable. And by extension, so are the drivers.
Testing by junction drops is a gross exercise, it finds BAD parts. But it won;t find weak or marginal parts, say a transistor that has lost its gain.
Is arctic silver conductive? I sure wouldn't bathe that transistor in it then. The stuff will find its way to the legs.
You have not yet replaced Q16? Why not? It is the central element controlling the output idle current. Don;t have one? Swap it with the good channel. If it is at fault, now the good channel will run hot. And the bad channel will cool down. If it makes no difference, at least you know both bias transistors are OK. And you'd have narrowed it down to something about those resistors around Q16.
You shorted across Q16 and they went cold. That is not surprising either, since Q16 controls the whole thing. That experiment told us that the outputs are controllable. And by extension, so are the drivers.
Testing by junction drops is a gross exercise, it finds BAD parts. But it won;t find weak or marginal parts, say a transistor that has lost its gain.
Is arctic silver conductive? I sure wouldn't bathe that transistor in it then. The stuff will find its way to the legs.
You have not yet replaced Q16? Why not? It is the central element controlling the output idle current. Don;t have one? Swap it with the good channel. If it is at fault, now the good channel will run hot. And the bad channel will cool down. If it makes no difference, at least you know both bias transistors are OK. And you'd have narrowed it down to something about those resistors around Q16.
Thanks for the suggestions.
I didn't want to go unsoldering and soldering without knowing in advance which part was bad. Also, this particular transistor is nearly obsolete, so it's tough finding an exact replacement. The resistors around Q16 are all within spec.
I would prefer to just order a dozen or so of Q16, or equivalents, and pick one based on Hfe, which I can measure. Does anyone know of a good replacement for a 2SD1857? It's in an ATV package.
Arctic Silver is a highly thermally conductive paste used to power transistors. It does not conduct electricity.
Last edited:
I would like opinions on changing R248 (47 ohm) to something larger. In the sim, if I change it to say 100ohms, then changes in the Vbe multiplier are far less dramatic to the outputs. I would have to change something in the Vbe multiplier circuit also of course. I experimented with changing R254 (with R248 at the original 47 ohms) to simulate a trimmer resistor there and small changes in that resistor's value has a dramatic change on the amount of current flowing in the outputs, whereas if R248 is increased, then changes in R254 do not have such a dramatic effect. I was hoping that it would enhance stability a little bit. Power output does not appear to be affected.
Obviously, increasing R248 reduces the current in the drivers and increases the current in the outputs, so I'd have to reduce bias to do this, but I don't have experience with designing power amps so I'm asking for opinions please.
Obviously, increasing R248 reduces the current in the drivers and increases the current in the outputs, so I'd have to reduce bias to do this, but I don't have experience with designing power amps so I'm asking for opinions please.
I think the 2SD1857 could be replaced with a TIP29, but I have to check to see if there's a way to bond it to the heat sink. The hfe is lower for the TIP29 though.
B&D looks like they might have a FEW 2SD1857. They also suggest 2SC4027 as a sub.
www.bdent.com
Mouser sells the 2SC4027 as well.
http://www.onsemi.com/pub_link/Collateral/EN2262D-D.PDF
And you could call Fender and ask if they still have them in their parts department. If so, then any Fender dealer/supplier would be able to get them for you.
AT some point in any repair or diagnostic, you get to the "gotta break eggs to make an omelet" stage. We can;t go through life afraid to solder a few parts. WHy spend tons of time trying to come up with re-engineered circuits and other changes rather than a five minute part swap that answers the big question?
You are proposing redesiging the driver circuit to add "stability", but then also having to redesign the bias circuit to compensate for the driver changes. But the whole time this repair has been unfolding, the other channel has motored along just fine, stable as can be, as it has the last several years.
Just my own opinion, but I'd turn off that sim.
www.bdent.com
Mouser sells the 2SC4027 as well.
http://www.onsemi.com/pub_link/Collateral/EN2262D-D.PDF
And you could call Fender and ask if they still have them in their parts department. If so, then any Fender dealer/supplier would be able to get them for you.
AT some point in any repair or diagnostic, you get to the "gotta break eggs to make an omelet" stage. We can;t go through life afraid to solder a few parts. WHy spend tons of time trying to come up with re-engineered circuits and other changes rather than a five minute part swap that answers the big question?
You are proposing redesiging the driver circuit to add "stability", but then also having to redesign the bias circuit to compensate for the driver changes. But the whole time this repair has been unfolding, the other channel has motored along just fine, stable as can be, as it has the last several years.
Just my own opinion, but I'd turn off that sim.
Thanks for your help.
So, you think it's just the 2SD1857 being either bad, or on the far side of the normal range, that's the problem? Should a transistor for use as a Vbe multiplier have high hfe or low? I'd prefer having a dozen or so and select one that is either high or low instead of just buying one from a dealer in the hopes that it would solve this problem. I did a simple diode test on it and it tests fine, so how can I say it's a bad part?
I thought it was suggested that I put a trimmer in the bias circuit?
I'm waiting on parts before doing anything more with it.
I really haven't been looking at the sim lately.
Hmmm, I wonder what I'm going to do with 49 of the 2SD1857 since the minimum order is 50? Do you need any? <g>
So, you think it's just the 2SD1857 being either bad, or on the far side of the normal range, that's the problem? Should a transistor for use as a Vbe multiplier have high hfe or low? I'd prefer having a dozen or so and select one that is either high or low instead of just buying one from a dealer in the hopes that it would solve this problem. I did a simple diode test on it and it tests fine, so how can I say it's a bad part?
I thought it was suggested that I put a trimmer in the bias circuit?
I'm waiting on parts before doing anything more with it.
I really haven't been looking at the sim lately.
Hmmm, I wonder what I'm going to do with 49 of the 2SD1857 since the minimum order is 50? Do you need any? <g>
Last edited:
Hi,
Swap any suspect parts between the channels. Look for any thermal
coupling differences between the channels. Ascertain why one channel
is OK, the other channel apparently not. Compare good to bad, e.g.
is Vbias different etc...
rgds, sreten.
Swap any suspect parts between the channels. Look for any thermal
coupling differences between the channels. Ascertain why one channel
is OK, the other channel apparently not. Compare good to bad, e.g.
is Vbias different etc...
rgds, sreten.
IMHO, the idling current is probably wrong because the replacement transistors were a bit different to the originals. That's why most amps have a trimmer to start with - to allow for differences between transistors.
Sounds like there's nothing wrong with Q16. Easiest solution is just to put in a trimpot and adjust the idling current to what you want.
Sounds like there's nothing wrong with Q16. Easiest solution is just to put in a trimpot and adjust the idling current to what you want.
B&D wants a minimum oprder of 50 for the original type. The sub they suggest has a minimum order of 1 piece.
Mouser has a 1 piece mimimum on the sub, but wants 10 days to ship them.
Call Fender monday, they may have them too. Their stock number for it is 0551448003
I have run a Fender authorized repair facility for 25 years, and I can;t recall ever having to select a bias transistor. The circuits are just not that critical in design. They make these things to be built by the thousands. They won;t be taking the time or expense to select parts for this application, and in fact they don't want to have to adjust anything either. SO we grab a part from the bin, solder it in, and the amp works. If their design worked out to be so touchy, they wouldn;t send it out that way, they'd redesign it.
Making the bias circuit adjustable is one strategy. It is one way to treat the symptom without finding what the actual problem is. But would probably work. You are concerned that small changes in some resistor make large changes in idle current. OK, but what is the range of useful variation in that resistor? Go ahead and sim it. If it is 825 ohms stock, we might put a 1k trimmer in place of it. But if we find that really, the only useful setting of that 1k trimmer are between 810 ohms and 920 ohms. Then instead of using a 1k trimmer, mount a 800 ohm resistor and a 100 ohm trim in series. Now the tweaking is less touchy, and you have limited the lower end to 800 ohms and the upper end to 900. And that without redesigning the drivers.
I made some suggestions, others made alternative suggestions. It isn;t rocket surgery, try either way.
DO I think it is Q16? It is my prime suspect. I am in the business of solving these problems. I have the need to repair the amp in a timely fashion. I like to be able to analyze things and get a darn good idea what is needed before I tear into the parts. But customers are paying me a dollar a minute. So if I think a part may be bad, I change it. How many dollar minutes should I spend trying to decide if a 75 cent part is edgy? I realize the home repair operation has no such constraints. But my job is to find out what is wrong. SO if it all points to Q16, and I have an identical working amp beside me, I am going to swap those parts in a heartbeat, and voila, I'll know - it is something about that part or not. And that also tells you whether you need to order one or not.
Mouser has a 1 piece mimimum on the sub, but wants 10 days to ship them.
Call Fender monday, they may have them too. Their stock number for it is 0551448003
I have run a Fender authorized repair facility for 25 years, and I can;t recall ever having to select a bias transistor. The circuits are just not that critical in design. They make these things to be built by the thousands. They won;t be taking the time or expense to select parts for this application, and in fact they don't want to have to adjust anything either. SO we grab a part from the bin, solder it in, and the amp works. If their design worked out to be so touchy, they wouldn;t send it out that way, they'd redesign it.
Making the bias circuit adjustable is one strategy. It is one way to treat the symptom without finding what the actual problem is. But would probably work. You are concerned that small changes in some resistor make large changes in idle current. OK, but what is the range of useful variation in that resistor? Go ahead and sim it. If it is 825 ohms stock, we might put a 1k trimmer in place of it. But if we find that really, the only useful setting of that 1k trimmer are between 810 ohms and 920 ohms. Then instead of using a 1k trimmer, mount a 800 ohm resistor and a 100 ohm trim in series. Now the tweaking is less touchy, and you have limited the lower end to 800 ohms and the upper end to 900. And that without redesigning the drivers.
I made some suggestions, others made alternative suggestions. It isn;t rocket surgery, try either way.
DO I think it is Q16? It is my prime suspect. I am in the business of solving these problems. I have the need to repair the amp in a timely fashion. I like to be able to analyze things and get a darn good idea what is needed before I tear into the parts. But customers are paying me a dollar a minute. So if I think a part may be bad, I change it. How many dollar minutes should I spend trying to decide if a 75 cent part is edgy? I realize the home repair operation has no such constraints. But my job is to find out what is wrong. SO if it all points to Q16, and I have an identical working amp beside me, I am going to swap those parts in a heartbeat, and voila, I'll know - it is something about that part or not. And that also tells you whether you need to order one or not.
You should have done it days ago - I can't believe you're still posting about it?.
If adjusting it allows you to set the bias correctly, then it's proved the Vbe trasnsistor isn't faulty, and that (as we've suggested all along) the 'fault' is almost certainly the wrong output transistors fitted.
The Hfe of the Vbe multiplier makes sod all difference, it's not run in that kind of mode (don't forget - a transistor doesn't have a specific 'Hfe', it has one within a specified WIDE range).
I have to order the parts. It takes a week or so to get here. Nothing good is available locally.
The output transistors are the correct types, according to the printing on them.
Yeah, I forgot to compare the voltages between the amps. I've been busy with other things. Thanks for the reminder.
They are different individuals but exactly the correct output transistors according to the printing on them.
The trimmer should be here soon, so that's something I'll try.
Everyone doesn't seem to realize what a hassle it is to keep removing and replacing the main circuit board. There's always a good chance of damaging something getting the board in and out. I don't want to do it unless I know what needs to be done.
The numbers don't matter - with fixed biasing they need to be EXACTLY the same as the original ones. This is why amps have adjustable bias.
Presumably the originals are specially selected for Fender, to avoid any setting up during manufacture?. I've come across similar cases occasionally over the decades, it always seems a stupid idea to me.
Just solder an extra fixed resistor across the top of the 2.43K one - try a 10K for a start and see what that reduces the bias to. Then change it up or down depending what the result is - by paralleling the existing resistor you can only reduce the bias, so you can't 'blow it up'.
I ordered 2 of the 2SD1857 from a parts dealer. So, it'll be another 4 days or so before I can do anything with this amp.
No, the originals were not specially selected, according to other people here. Nothing is matched or selected.
It's too difficult to solder on the board with it in the case. I have to remove the board, do what ever I want, and then put it back. It's a freaking hassle and I'd rather know EXACTLY what's wrong before doing that again.
One thing that's been a real problem for me lately is that my right arm and hand are messed up from spending too much time on computers. I have constant dull pain on the right side going all the way down to my hand. I work from home on the computer, and I've been working almost nonstop since my wife went on vacation in Sweden a few weeks ago. So, I'm having trouble manipulating small objects with my right hand, which makes working on this amp twice as difficult as usual. Holding a soldering iron steady is another challenge for me. I leave for Sweden in a week, so I don't know if much will get done. Thanks everyone for your kind advice and patience with me.
Also, I have never repaired an amp like this before. I don't have very much experience working on solid state gear, and I still don't understand a lot of the stuff in this amp. I have the basics, so maybe that's enough to repair it.
Also, I have never repaired an amp like this before. I don't have very much experience working on solid state gear, and I still don't understand a lot of the stuff in this amp. I have the basics, so maybe that's enough to repair it.
Last edited:
- Status
- Not open for further replies.