I didn't mean the same PHYSICAL chip, but the same TYPE of chip... I guess I need to be really specific on this one. Yes, my board revision may be different than yours, but I would imagine they are very much similiar. I haven't checked my hexfet's yet. I might not get to stuffing the board until later this week, not sure yet.
What differences have you noticed between the rev. 6 and rev. 7?
Cap C11's marking is 104, as stated in the table I added at the top of the schematic, along with the rest of the components.
Jer
What differences have you noticed between the rev. 6 and rev. 7?
Cap C11's marking is 104, as stated in the table I added at the top of the schematic, along with the rest of the components.
Jer
That is very cool, as long as we are fixed on the components values, all of us will be able to fix this weak module.
I have also noticed that you have fixed the reference designators for all components, something that i didnt take care of at all.
did you figured that something else was wrong on the main amplifier board ?
I do have a new revison of the sevice manual and there is a lot of reworks needed to be done. Those reworks dont apears in any of the schematics posted on this site.
I will send it to you.
Cheers.
I have also noticed that you have fixed the reference designators for all components, something that i didnt take care of at all.
did you figured that something else was wrong on the main amplifier board ?
I do have a new revison of the sevice manual and there is a lot of reworks needed to be done. Those reworks dont apears in any of the schematics posted on this site.
I will send it to you.
Cheers.
Hi,
Your board may be different but I see the scimps are still using 6 volts for gate drive.
I believe C3 would be 47p, not 4.7p, so it's not that immense a difference, just sufficient to speed it up a little, likely to compensate from the delay induced by level shift through the MPSA42, I would think this value was selected empirically.
Hey, rev6, rev7, how different can they be, they both fried
Looking forward to seeing more about this, even though it may be an exercise in futility, I find it highly interesting.
Your board may be different but I see the scimps are still using 6 volts for gate drive.
I believe C3 would be 47p, not 4.7p, so it's not that immense a difference, just sufficient to speed it up a little, likely to compensate from the delay induced by level shift through the MPSA42, I would think this value was selected empirically.
Hey, rev6, rev7, how different can they be, they both fried
Looking forward to seeing more about this, even though it may be an exercise in futility, I find it highly interesting.
Hello, and yes, C3 is 47, just like in the schematic.
The datasheet on the IRL640 states that it has a very low voltage requirement for the gate, in fact, the test circuits for the device only use 5V. It isn't like the IRF640, that needs double the voltage! So although the circuit seems inadequate for the HEXFET, it is in fact, about right.
I almost wonder if PPL been thinking that the output transistors were IRF640's... they aren't a compatible replacement for IRL640's.
The datasheet on the IRL640 states that it has a very low voltage requirement for the gate, in fact, the test circuits for the device only use 5V. It isn't like the IRF640, that needs double the voltage! So although the circuit seems inadequate for the HEXFET, it is in fact, about right.
I almost wonder if PPL been thinking that the output transistors were IRF640's... they aren't a compatible replacement for IRL640's.
In response to Pat's question regarding the 47 marked capacitor, it's always been my understanding that any 2 digit number on a cap is always that number in pico... so 47 is supposed to be the same as a 470... 47 pico. Since there is a bit of question on this one cap, I will try to get it tested to confirm. I will keep y'all updated on that cap.
Jer
Jer
Yeah reading cap values from markings is confusing as hell because there's no standard method of marking them for any given cap, chances of a screw up are likely, ability to measure them would be great, wish I had it. The other reason is because while looking for a nice decent clear correct reference on this, I haven't found one that didn't screw something up.
Two examples of screw ups I found on what are the nicer sites I saw:
1. "So 100 would be 10pF multiplied by zero i.e. 10pF."
He got the right answer, just screwed up the instructions.
2. HM, well, these guys don't let you copy and paste from their site, guess they want recognition for their hard work, so here it is:
http://www.radio-electronics.com/info/data/capacitor/capacitor_codes.php
"It can be seen from the tables below"... the tables are above.."that a capacitor with the marking 104J has a value of 10 000pF and has a tolerance of +-5%"
At least their tables were right, doesn't serve to teach anyone well when their "worked example" is wrong.
It is my understanding as well that when it only shows two digits, then you just take that as the given value in pico farades
Your table is perfectly correct until you get to the bottom.
In order to get 4.7 out of 47 you have to multiply by 10^-1, as you know, and it does make sense to do because it's simply the next logical step. But the simple and confusing fact is, this is where they break with convention. There is no multiplier given, so as a rule, it doesn't get multiplied, you just have to take it as face value and that it is in pico.
So, you're perfectly correct to say that 470=47X10^0=47, but, it is the same thing as simply having it say 47... and they say there's no black magic in electronics.
There's three whole pages in the student manual of Art of Electronics dedicated to helping one stand a chance at reading caps properly, even with that aaand the internet at my hand, I find caps that just leave me scratching my head. At least resistors have their act together.
Anyway, back to the good stuff.
Jeremy can you comment on how poor the drivers power supplies are in your version? Find us some burnt parts Nice looking schematic and dissection as well. I think it would be neat to see you guys get that working half decent, somehow, accomplish what JBL and their ilk could not.
Regards
Two examples of screw ups I found on what are the nicer sites I saw:
1. "So 100 would be 10pF multiplied by zero i.e. 10pF."
He got the right answer, just screwed up the instructions.
2. HM, well, these guys don't let you copy and paste from their site, guess they want recognition for their hard work, so here it is:
http://www.radio-electronics.com/info/data/capacitor/capacitor_codes.php
"It can be seen from the tables below"... the tables are above.."that a capacitor with the marking 104J has a value of 10 000pF and has a tolerance of +-5%"
At least their tables were right, doesn't serve to teach anyone well when their "worked example" is wrong.
It is my understanding as well that when it only shows two digits, then you just take that as the given value in pico farades
Your table is perfectly correct until you get to the bottom.
In order to get 4.7 out of 47 you have to multiply by 10^-1, as you know, and it does make sense to do because it's simply the next logical step. But the simple and confusing fact is, this is where they break with convention. There is no multiplier given, so as a rule, it doesn't get multiplied, you just have to take it as face value and that it is in pico.
So, you're perfectly correct to say that 470=47X10^0=47, but, it is the same thing as simply having it say 47... and they say there's no black magic in electronics.
There's three whole pages in the student manual of Art of Electronics dedicated to helping one stand a chance at reading caps properly, even with that aaand the internet at my hand, I find caps that just leave me scratching my head. At least resistors have their act together.
Anyway, back to the good stuff.
Jeremy can you comment on how poor the drivers power supplies are in your version? Find us some burnt parts
Nice looking schematic and dissection as well. I think it would be neat to see you guys get that working half decent, somehow, accomplish what JBL and their ilk could not.Regards
well, in a nutshell, since there are no zeners or other regulators in the amp module, i would say that it relies solely on the external circuit for regulation. I haven't looked into the outside circuit much yet since I isolated the problem with the sub. Also in the same nutshell, from what I do recall looking at, it is strictly unregulated, direct dc straight from the bridge rect. from the step-up transformer. I could be wrong, I only skimmed through the board prior to removing the amplifier unit.
The plan is for me to go through the devices tomorrow. I strongly suspect the op amp is what actually failed, initially, since that came out in 2 parts. BUT, an op amp isn't going to blow my 2A fuse either. There is some arc damage evident between the source and drain of M2 (Q2 on the schematic). With the exception of the ICs, all the semiconductors and resistors test good, and the caps show no signs of mesurable short or blistering, but I haven't tested them extensively, yet.
I honestly believe that the circuit is adequate, considering it's application. In anything but a subwoofer, you would notice the difference in the unregulated power supply rails to the driver HEXFETs. Subs can vary in their power as much as 10% and have no real noticable effect to the ear.
The biggest downfall to this poor design is the heat dissipation....in this encapsulated and virtually closed design for a heatsink, there is almost no thermal efficency. I'm seriously considering adding a small cooling fan to a custom heatsink assembly with a higher thermal conductivity. Most powered subs rely on the movement of the air from the cone to cool the unit, and as I've read througout the net, these JBLs most often fail when in standby.
For that reason I believe it isn't the poor design of the unit, rather, but an unclean signal getting to the unit causing it to stay "ON" but not enough to move the cone and cool. What JBL might have done to prevent this is have a better incomming signal filter to prevent a false "ON" situation.
Makes sense to me! (although I have been wrong before)
Jer
The plan is for me to go through the devices tomorrow. I strongly suspect the op amp is what actually failed, initially, since that came out in 2 parts. BUT, an op amp isn't going to blow my 2A fuse either. There is some arc damage evident between the source and drain of M2 (Q2 on the schematic). With the exception of the ICs, all the semiconductors and resistors test good, and the caps show no signs of mesurable short or blistering, but I haven't tested them extensively, yet.
I honestly believe that the circuit is adequate, considering it's application. In anything but a subwoofer, you would notice the difference in the unregulated power supply rails to the driver HEXFETs. Subs can vary in their power as much as 10% and have no real noticable effect to the ear.
The biggest downfall to this poor design is the heat dissipation....in this encapsulated and virtually closed design for a heatsink, there is almost no thermal efficency. I'm seriously considering adding a small cooling fan to a custom heatsink assembly with a higher thermal conductivity. Most powered subs rely on the movement of the air from the cone to cool the unit, and as I've read througout the net, these JBLs most often fail when in standby.
For that reason I believe it isn't the poor design of the unit, rather, but an unclean signal getting to the unit causing it to stay "ON" but not enough to move the cone and cool. What JBL might have done to prevent this is have a better incomming signal filter to prevent a false "ON" situation.
Makes sense to me! (although I have been wrong before)
Jer
Hi, if you look closer to the schematic is sent you, there is plenty of dc filtration here and there, as well as a low pass and high pass filter too. C32 is right berore the amp module to filter any dc.
So in your module there was IRL 640? mines were defenitively IRF's. I dont find any data sheets for the IRL's...
I would say what make them fry is the absence of proper voltage regulation, and that the gate drivers are powered thru theirs very limit at +6v. If i ever work on that again, my first mods will to add some zener at the vcc of both chips to clip it at anything more than 5.6v.
The schematic that you posted, wich is mine but modified, differs very slightly from what i did, and i still believe that i did it right, and that we dont have the same exact boards. At least, part of my job is to reverse the competition products, to see if they have copied my (our) design
So in your module there was IRL 640? mines were defenitively IRF's. I dont find any data sheets for the IRL's...
I would say what make them fry is the absence of proper voltage regulation, and that the gate drivers are powered thru theirs very limit at +6v. If i ever work on that again, my first mods will to add some zener at the vcc of both chips to clip it at anything more than 5.6v.
The schematic that you posted, wich is mine but modified, differs very slightly from what i did, and i still believe that i did it right, and that we dont have the same exact boards. At least, part of my job is to reverse the competition products, to see if they have copied my (our) design
Hello,
IRL640? I'll tell ya what the difference is.... ready for this?
IR(L)...L= logic level. Vth is 1 to 2 volts. That'll go a long way to keep the mosfets cooler.
If you just type "IRL640" into google you'll find the data sheet no problem.
http://www.alldatasheet.co.kr/datasheet-pdf/view/IRF/IRL640.html
It's not so much that you might hear some difference in power by poor regulation, it is as Pat said that they're running the inverters a little too high, they used to expect them to do the job of a 12V driver, which would be hard on the mosfets. In this case they have an easier load being logic level , but they do have a max voltage and 6 is reaaaal close, it is well beyond their "recommended" max voltage.
If their supply is as you say, a small power line surge of say 10%, not uncommon, would be enough to flat line them.
Instead of a fan, maybe you can find / make some decent heatsinks for it.
Just leaving the epoxy junk out and leaving it in free air might be a big difference in that regard too.
I bet your op amps would have been fine, they're very robust devices from my experience.
Anyway, keep us posted, something is wrong with it
Regards
Chris
IRL640? I'll tell ya what the difference is.... ready for this?
IR(L)...L= logic level. Vth is 1 to 2 volts. That'll go a long way to keep the mosfets cooler.
If you just type "IRL640" into google you'll find the data sheet no problem.
http://www.alldatasheet.co.kr/datasheet-pdf/view/IRF/IRL640.html
It's not so much that you might hear some difference in power by poor regulation, it is as Pat said that they're running the inverters a little too high, they used to expect them to do the job of a 12V driver, which would be hard on the mosfets. In this case they have an easier load being logic level , but they do have a max voltage and 6 is reaaaal close, it is well beyond their "recommended" max voltage.
If their supply is as you say, a small power line surge of say 10%, not uncommon, would be enough to flat line them.
Instead of a fan, maybe you can find / make some decent heatsinks for it.
Just leaving the epoxy junk out and leaving it in free air might be a big difference in that regard too.
I bet your op amps would have been fine, they're very robust devices from my experience.
Anyway, keep us posted, something is wrong with it
Regards
Chris
Its cool that they have updated the circuit with logic mosfets, but not cool that i have herited maybe the very first units ever produced with non-logic mosfets??
ahaha !!
anyway, its still a piece of crap imho. And they burn anyway with the logic level mosfets...maybe they last a bit longer with them, who knows, mine burned after 2 month, while it was not playing.
lets see what Jer will do...
ahaha !!
anyway, its still a piece of crap imho. And they burn anyway with the logic level mosfets...maybe they last a bit longer with them, who knows, mine burned after 2 month, while it was not playing.
lets see what Jer will do...
Another interesting fact to note with these mosfets is since they are logic level, Vgs MAX is only +-10 volts. Some protection Zeners may have gone a long way towards preventing that little bit of "arcing"..
I think the fets are cooked, maybe the drivers too. Guess it's wait and see time now
Hey, don't feel too bad, just think of how many modules they had to have fried before they were goaded into switching to a logic level fet! I bet it wasn't in the "dozens", thousands?? Hah... I hope so.
I think the fets are cooked, maybe the drivers too. Guess it's wait and see time now
Hey, don't feel too bad, just think of how many modules they had to have fried before they were goaded into switching to a logic level fet! I bet it wasn't in the "dozens", thousands?? Hah... I hope so.
Chris, thanks for the info.....
I'd like to add to that... The 6V rail is actually regulated by a 6.2V zener on the mainboard. There is also a diode in series with each device, so with the diode drop, one isn't doing too bad for the driver ICs. Granted, it can be better! I am thinking about dumping the whole additional zener idea and just resorting to the good ol' standby of a couple lm2940ct's right before the power enters the IC. That would keep the logic level at a true logic level.
My plan is to work a nicely design and modded HS for the HEXFETs and that would give me ample locations for the new 5V VRs as well. A lot of people swear by the zeners, but I do not, for the most part.
Originally I was just going to do a part-for-part replacement, but after talking with ya'll, It does make me kind of tempted to do some mods since it wouldn't take much to make it better than orig. I'm still not sure what my final result is going to be... most likely I will rebuild it to start out with, watch it fry, get all upset, and then redesign it... like I should have in the first place
Aw heck, here is what I say, modify the design so it is a shoe-in replacement, with a lot better stability, patent it, and offer it on the black marketf for a modest profit!
Tell ya this much, watching ebay, it might just be worth it to buy up all those fried subs and with an hour's work, resell it for tripple, hands down. Mine was only $25!
(if there are any JBL spies around here, that should scare em up)
I'd like to add to that... The 6V rail is actually regulated by a 6.2V zener on the mainboard. There is also a diode in series with each device, so with the diode drop, one isn't doing too bad for the driver ICs. Granted, it can be better! I am thinking about dumping the whole additional zener idea and just resorting to the good ol' standby of a couple lm2940ct's right before the power enters the IC. That would keep the logic level at a true logic level.
My plan is to work a nicely design and modded HS for the HEXFETs and that would give me ample locations for the new 5V VRs as well. A lot of people swear by the zeners, but I do not, for the most part.
Originally I was just going to do a part-for-part replacement, but after talking with ya'll, It does make me kind of tempted to do some mods since it wouldn't take much to make it better than orig. I'm still not sure what my final result is going to be... most likely I will rebuild it to start out with, watch it fry, get all upset, and then redesign it... like I should have in the first place
Aw heck, here is what I say, modify the design so it is a shoe-in replacement, with a lot better stability, patent it, and offer it on the black marketf for a modest profit!
Tell ya this much, watching ebay, it might just be worth it to buy up all those fried subs and with an hour's work, resell it for tripple, hands down. Mine was only $25!
(if there are any JBL spies around here, that should scare em up)
Hi Jeremy,
I'd be tempted to do the same. If you watch it burn you can "see" the exact problem, at least if you're looking at the right spot when it happens. The problem is that isn't likely to happen right away, and if you poke and prode it into showing itself, then it broke/burnt from the poking and proding and not because of the actual circuit's dynamics.
Zeners are good enough, it's just a switch (the driver that is), but the drivers would appreciate a lower value, a regulator IC would work just fine as well. 6V is on the high side but there's a chance the drivers can take it, for awhile, but for how long, any kind of inductive spike would be enough to send them over the edge. You said the regulators were kind of far away from it, seriously, I think they should be alot closer to the driver, in as small a loop as possible with the driver and mosfet. Whichever option you go for, try to keep it close and well bypassed and it should take that possible problem out of the equation. With logic level mosfets dropping it by a volt or so will not hurt anything, they'll still switch fully on.
Something I noticed is, at least in the schematics drawn, there is no bypassing of the driver IC's?? Huge mistake!! It has to pull charge through how much inductance before it gets to the regulator/power supply? If it is there and was just left out for purposes of clarity, great, if it isn't there... seriously add some, right across the hex inverter power pins.
Of course it should go without saying, when you do find whatever is burnt on it, that'll be a strong indicator of the other/main problem area to be concerned with, and we already know it requires some attention as it failed already.
That way, with some luck, it'll be fixed right the first time, at least until another problem area rears its ugly head.
I really think it should be possible to fix it right and have it last.
I think threads like this should be enough to teach JBL a further lesson, I also think if we can get it to work decent where they couldn't, would be a nice twist. I only hope there's not some underlaying critical flaw to the whole design we haven't spotted.
Regards
EDIT:
Just had this other thought, the fact it's no longer in potting might be enough of a difference to mask some problems as well. Let's say the hex inverters are running cooler, they'll probably take a bit more abuse willingly, still, it should be more robust than that.
I'd be tempted to do the same. If you watch it burn you can "see" the exact problem, at least if you're looking at the right spot when it happens. The problem is that isn't likely to happen right away, and if you poke and prode it into showing itself, then it broke/burnt from the poking and proding and not because of the actual circuit's dynamics.
Zeners are good enough, it's just a switch (the driver that is), but the drivers would appreciate a lower value, a regulator IC would work just fine as well. 6V is on the high side but there's a chance the drivers can take it, for awhile, but for how long, any kind of inductive spike would be enough to send them over the edge. You said the regulators were kind of far away from it, seriously, I think they should be alot closer to the driver, in as small a loop as possible with the driver and mosfet. Whichever option you go for, try to keep it close and well bypassed and it should take that possible problem out of the equation. With logic level mosfets dropping it by a volt or so will not hurt anything, they'll still switch fully on.
Something I noticed is, at least in the schematics drawn, there is no bypassing of the driver IC's?? Huge mistake!! It has to pull charge through how much inductance before it gets to the regulator/power supply? If it is there and was just left out for purposes of clarity, great, if it isn't there... seriously add some, right across the hex inverter power pins.
Of course it should go without saying, when you do find whatever is burnt on it, that'll be a strong indicator of the other/main problem area to be concerned with, and we already know it requires some attention as it failed already.
That way, with some luck, it'll be fixed right the first time, at least until another problem area rears its ugly head.
I really think it should be possible to fix it right and have it last.
I think threads like this should be enough to teach JBL a further lesson, I also think if we can get it to work decent where they couldn't, would be a nice twist. I only hope there's not some underlaying critical flaw to the whole design we haven't spotted.
Regards
EDIT:
Just had this other thought, the fact it's no longer in potting might be enough of a difference to mask some problems as well. Let's say the hex inverters are running cooler, they'll probably take a bit more abuse willingly, still, it should be more robust than that.