Terry I suspect some crosstalk issues here, I have already made new layout and put it back together at the weekend.
Learning process takes some time but hopefully we get to the point.
OS
Thanks a lot!!
Regards
Learning process takes some time but hopefully we get to the point.
OS
Thanks a lot!!
Regards
Hi Borys,
I guess we'll see when you are able to test again. I am puzzled as to what is making the bias jump up so high when adding a load to the speaker out. I think the reason I was able to play music through it at first is because I adjusted the bias with a load attached so I didn't notice that issue until I increased the input volume enough to get it to lock up the drivers. I have been going over the boards and comparing the layout to the schematic and I can't find any errors but something is wrong somewhere. I have a spice file for the slewmaster and I changed the values in it to match your schematic and it runs fine so I'm stumped. I have built three separate boards. One with all the original parts, one with R13, 14 & 22 jumpered and one with those jumpered and KSC3502/A1381 pre-drivers. They all do the same thing. Very strange.
Blessings, Terry
I guess we'll see when you are able to test again. I am puzzled as to what is making the bias jump up so high when adding a load to the speaker out. I think the reason I was able to play music through it at first is because I adjusted the bias with a load attached so I didn't notice that issue until I increased the input volume enough to get it to lock up the drivers. I have been going over the boards and comparing the layout to the schematic and I can't find any errors but something is wrong somewhere. I have a spice file for the slewmaster and I changed the values in it to match your schematic and it runs fine so I'm stumped. I have built three separate boards. One with all the original parts, one with R13, 14 & 22 jumpered and one with those jumpered and KSC3502/A1381 pre-drivers. They all do the same thing. Very strange.
Blessings, Terry
One with all the original parts, one with R13, 14 & 22 jumpered and one with those jumpered and KSC3502/A1381 pre-drivers. They all do the same thing. Very strange.
You mean all these burned?
One didn't burn because I never ran it without the light bulb. It still has the big bias jump with a load attached. I guess what I'm saying is none of the remedies tried have changed that issue.
It has got to be an oscillation you just can't see on the oscilloscope. That will cause cross conduction and your rocketing bias and rampant destruction. Just curious if you tried running the drivers and pre-drivers at a higher current or not?
I did lower R28 to 150R. What current should I shoot for and what changes would I make to get the current?
Terry do you have two boards that are identical? One that works and one that does not work? I got lost on the sequence long ago. Thanks
I'm thinking about the same. That's the only practical reason, why the load, connected to zero-offset output causes the bias jump. Must be something very high frequency (order of 100MHz or something like this).
100mhz+ from fet's...maybe ?
BJT's just don't have the gain @ HF. Much lower freq. oscillations.
(where there is gain). With modern 30+ mhz outputs , you could still
get 20mhz+ oscillations. More typical is 5-10mhz , where there is high
Hfe.
Some Sanken low Cob/high Ft output stages might be exceptions (40+mhz
oscillations). Cob + trace inductance would determine this (L/C).
The graph here -http://cache.freescale.com/files/rf_if/doc/app_note/AN139A.pdf
- (fig. 4) illustrates this .
PS- when I screw up on simulations 😱 , BJT's "squeal away" at <10mhz ,
typically. I've modeled HEXFET circuits that will oscillate at way over 100mhz.
OS
OS
BJT's just don't have the gain @ HF. Much lower freq. oscillations.
(where there is gain). With modern 30+ mhz outputs , you could still
get 20mhz+ oscillations. More typical is 5-10mhz , where there is high
Hfe.
Some Sanken low Cob/high Ft output stages might be exceptions (40+mhz
oscillations). Cob + trace inductance would determine this (L/C).
The graph here -http://cache.freescale.com/files/rf_if/doc/app_note/AN139A.pdf
- (fig. 4) illustrates this .
PS- when I screw up on simulations 😱 , BJT's "squeal away" at <10mhz ,
typically. I've modeled HEXFET circuits that will oscillate at way over 100mhz.
OS
OS
OK so if it is oscillation that is so high, what is causing it and how do we eliminate it? I just remembered that the Slewmaster does have pads for adding compensation to the drivers. I believe I added 82p there on the first boards I built. Seems like someone along the way said they weren't needed but they may still be there on those first two boards. I will try adding some in the morning and see what happens. I suppose lowering R10 and R28 will raise the current on the pre-drivers and drivers too if you think that is necessary. What do you guys think?
Seems odd. So, all is normal until we connect a load (real speaker or dummy load?) then we get an oscillation that burns out the drivers and a set of associated resistors. I'm correct in my understanding so far?
Have a look at your zobel components perhaps? They are supposed to ensure external stability but appear to not be doing their job since it is the external load that sets it off. You might try using 4R7 and 220nF components for the zobel and verify the damping R across the coil, make it 10R if it is not right.
Have a look at your zobel components perhaps? They are supposed to ensure external stability but appear to not be doing their job since it is the external load that sets it off. You might try using 4R7 and 220nF components for the zobel and verify the damping R across the coil, make it 10R if it is not right.
I can try that. It is per the layout in post 632 except I have a 10R soldered inside the coil. The zobel is 10R/100n. I could try adding 10R/100n in parallel with them. I'll try the caps on the drivers first.
My biggest goal is to learn so I don't mind going through the process. The smaller footprint that this board affords is worth the effort IMO.
Blessings, Terry
My biggest goal is to learn so I don't mind going through the process. The smaller footprint that this board affords is worth the effort IMO.
Blessings, Terry
I'll be frank ....
The layout itself might not be "right" for a stable build. The very fact that known
sourcing blows up backs this up. 😱
It might be better to start anew for a board with this form factor.
I'm doing the new 5P , I might consider doing a thinner 2X height 2-3p
OPS , as well.
All my originals (JK tweaked and single sided) work well , but I have
even more ideas for improvements (decoupling , Vbe position ....)
Edit - electrically , it should work (borys schema) .. I have one with similar omissions. I can even
disconnect the zoble ... do anything - it still works.
OS
Last edited:
5 pair of whatever , my main goal is to spread the thermals out across
>220mm + . Better for hot summer "action".
PS - the MT-200 capability is a added perverse "extra" (700W !).
OS
The only thing that stands out to me with that layout is the way the fuse is connected. Any sign of resistance from the fuse is going to cause the drivers to take over as output devices. Maybe have a close look at the fuses and see if they are discolouring the elements?
Separate fuses for drivers,or separate power supplies?
The flaming resistors are the fuses for the drivers. To me it would make sense to fuse the whole output board together so if the output fuse opens the drivers are disabled at the same time. As is the driver current will spike when the rail fuses blow. The input board can be safely powered and fused separately though.