Going through my E-waste pile today , I noticed most common
PC PS's (250-300w units) have the perfect 75mm extruded units for
the "slewmonster" (below).
Out of 15 SMPS's , 12 had 75-80mm (5-7C/w estimated) long
"overkill" extrusions. Could be lower than this.
A 300W SMPS at 88-92% efficiency would have 30w waste (heat).
Even as it is fan cooled - this is far more than the slew driver dissipation.
Being so common (PC SMPS's) this would be perfect (and free)
to apply to a build.
PS - with one of these , Driver Re @ 68 (or even 47-56)- no issue. A flat plate
is over 15C/W - 3 X the cooling for nothing !
OS
In light of that, I will ask again. What is the benefit of running the 68R over say a 150R? I'm running 100R and the little plate is about 110F. Will the outputs be happier with the drivers running at higher bias? Why not just go to 100-150R and not worry about the heat?
Thanks, Terry
VSSA alone with output power mosfets has much better performance than your scope images.
Member Keentoken mentioned of some possible "latchup" concerning
the use of large drivers/High Cob .... so I downsized the Re to 68R.
Simulation says all is the same from 68-220R (thd - stability).
But even at 150R , in 40C+ ambient or under-ventilated case ,
why not lower the C/W on the HS ?
PS - I going back to that environment ... I want cool !!
OS
Are you using the 1220/2690 pair at VAS ? That is cutting it close to
Vceo. The 1220 falls short and the 1220A is just enough for 80V rails -
NO derating
.OS
OS is this the same pcb as post 599?I looked , 1.2 -1.3 are the same. I might of tweaked a resistor value
or two. PCB is identical.
(schema below)
Edit - I did update the PCB (either will perform identical).
below 2/3 - allows diode or active transistor saturation clamp.
If you are using 1.2 - It is still SOTA !!
OS
Thimios.
Tiny bit different , option 1/2 is either transistor or diode
saturation clamp. You have a choice - some prefer one or the
other , so I added both. Also has the option of ground or
active reference for LTP cascode.
OS
At +-80V rails?
Square
Will do that tomorrow.
I don't know what the C/W is or the HS. Can you just explain what part number and what value to change to?
Yes, but I'm not married to them. I have KSA1381/KSC3503, MJE340/350 in my junk drawer. Open to suggestions. I just used what Jason suggested.
I don't know what the input filter is. Tell what parts you are referring to.
Thanks, Terry
Terry,
I posted back a short ways to lift or remove C2 to defeat the input filter. PMI and I had discussed a change (lowering) the cut-off frequency some time ago. I went from the original PeeCeeBee cut-off that was quite high at ~1.6MHz to about 10x the audio band at ~220kHz. When testing higher frequency squares this filter will almost certainly be having an impact on what you see and the amplifier may itself be plenty fast. This is why the suggestion to run a test without it.
Also for these tests we usually put the scope probe before the output coil rather than the speaker output spade.
Your VAS devices are the 'A' suffix parts IIRC and are rated for 160V. With 70V rails we still have a little margin. If you wanted to change them to something else then use the 1381/3503 pair.
The C/W of a heat sink is a rating of its efficiency in degrees Celsius per watt of heat applied, so a lower value is always the better heat sink. A 3C/W heat sink being heated by a transistor dissipating 5W will rise in temperature by 15 degrees Celsius above ambient. If ambient is 30 degrees Celsius then the heat sink would rise to 45 degrees Celsius assuming unobstructed free air convection.
I posted back a short ways to lift or remove C2 to defeat the input filter. PMI and I had discussed a change (lowering) the cut-off frequency some time ago. I went from the original PeeCeeBee cut-off that was quite high at ~1.6MHz to about 10x the audio band at ~220kHz. When testing higher frequency squares this filter will almost certainly be having an impact on what you see and the amplifier may itself be plenty fast. This is why the suggestion to run a test without it.
Also for these tests we usually put the scope probe before the output coil rather than the speaker output spade.
Your VAS devices are the 'A' suffix parts IIRC and are rated for 160V. With 70V rails we still have a little margin. If you wanted to change them to something else then use the 1381/3503 pair.
The C/W of a heat sink is a rating of its efficiency in degrees Celsius per watt of heat applied, so a lower value is always the better heat sink. A 3C/W heat sink being heated by a transistor dissipating 5W will rise in temperature by 15 degrees Celsius above ambient. If ambient is 30 degrees Celsius then the heat sink would rise to 45 degrees Celsius assuming unobstructed free air convection.
OK ... no acronyms !
The celsius per watt (c/w) is how many degrees per watt
of dissipation a particular extrusion will rise.
The HS is the heatsink (extrusion).
The final devices in jason's amp before the outputs (if you used jason's
bom /schema)...these were shown as 2sa1220/2sc2690. Nice for small amps.
With even 60-70V rails ... not recommended. At full amplitude , either device
will see nearly the full rail to rail voltage . About 135V for 70V rails.
the 1220/2690 are only rated 120V ... the "a" class of these devices @ 160V.
It is much safer (almost essential), to derate these devices. A >300V device
would be best. On a data sheet , this is specified as "Vceo" (mje340/350 are
300V Vceo ... so are the 1381/3503 pair).
If you are using those low Vceo devices (1220/2690) , that could be the
problem (or part of it).
OS
There isn't the PCB jpg in zip file.I looked , 1.2 -1.3 are the same. I might of tweaked a resistor value
or two. PCB is identical.
(schema below)
Edit - I did update the PCB (either will perform identical).
below 2/3 - allows diode or active transistor saturation clamp.
If you are using 1.2 - It is still SOTA !!
OS
Thimios
Last edited:
Hi Guys,
Thanks for taking the time to clarify for me. I had no idea that the vas devices saw rail to rail voltages. I thought they fed the bases of the pre-drivers so actually passed less than one rail. No matter, I will change them out. Any down side to using the mje parts rather than the ksa/c parts? I only ask because I have a lot more of the mje's.
As for the drivers on the ops, I see now what you meant. I was thinking lower heat with lower bias and you were talking about better heatsinking. So my question is still, does the higher bias on the drivers produce better sound or lower distortion or what? If not, why wouldn't we just lower the bias?
Thanks, Terry
Thanks for taking the time to clarify for me. I had no idea that the vas devices saw rail to rail voltages. I thought they fed the bases of the pre-drivers so actually passed less than one rail. No matter, I will change them out. Any down side to using the mje parts rather than the ksa/c parts? I only ask because I have a lot more of the mje's.
As for the drivers on the ops, I see now what you meant. I was thinking lower heat with lower bias and you were talking about better heatsinking. So my question is still, does the higher bias on the drivers produce better sound or lower distortion or what? If not, why wouldn't we just lower the bias?
Thanks, Terry
I never claimed to be sure of what I'm doing'
. The IPS I'm using was designed to drive a vbe and two MOSFETs. It is just something for me to play around with while I wait for someone to create some Gerbers so I can have some of OS's IPS's made. Jason was kind enough to sell me a pair of his vssa boards so I could play. I'm not sure if others are learning from this, but I am.I populated another pair of slew monster boards yesterday. All I need is a couple caps, a transformer and some heatsinks and I will be ready to make some really good a/b comparisons. Fun stuff!!!
I never claimed to be sure of what I'm doing'
This is what makes this type of thing fun. Not everything is perfect off the hop, though many of us try, but it always offers an opportunity to learn something new.
I think for you Terry, having a 'known good' OPS will allow for lots of exploration while removing some variables at the same time. Keep up the good work!
Last edited: