Hi jacco,
Talking of Kuhlkorpers, some have arrived. Do you have the dimensions of the power boards? I'll see if they fit. Yeah, I know I should have done that bit the other way round.
Regards,
Chris
Talking of Kuhlkorpers, some have arrived. Do you have the dimensions of the power boards? I'll see if they fit. Yeah, I know I should have done that bit the other way round.
Regards,
Chris
Some progress has been made on the heatsink attachment problems. If you go to this address: Extrusion Profiles with Clip Attachments - Max Clips
you will see that clips can be retro-fitted using screws instead of using machine made countersunk attachment points. This method will allow a degree of Mosfet movement under the clip whilst fully aligning the attached pcb which I feel should be a major bonus.
The MAX07G is 12mm wide and applies 11.2lbs whilst the MAX08G is 18mm wide and applies 16.9lbs.
I'm going for it.
Regards,
Chris
you will see that clips can be retro-fitted using screws instead of using machine made countersunk attachment points. This method will allow a degree of Mosfet movement under the clip whilst fully aligning the attached pcb which I feel should be a major bonus.
The MAX07G is 12mm wide and applies 11.2lbs whilst the MAX08G is 18mm wide and applies 16.9lbs.
I'm going for it.
Regards,
Chris
the screw on clips you have linked to are an almost perfect DIY solution for mounting power devices.
Yes you do. Mosfets's back plate is Drain (as well as middle leg). You need to provide electrical insulation and at the same time the best possible thermal conductivity. That is achieved by placing mica insulators between MOSFET and heat sink and applying thermal compound-paste on the both sides of the mica insulator. There are also Kapton pads that do not require thermal paste. You could also buy kit all together with mica, washers, bolt and nut.
Check at the bottom of the page.
Kapton pads are a thin sheet of flat plastic.
They too need thermal paste on both sides.
Similarly all hard surfaced thermal conductors need thermal paste at the interfaces.
Thermal pads, that conform to the device and heatsink, do not need thermal paste. There is some literature that suggests that adding thermal paste hinders heat flow.
They too need thermal paste on both sides.
Similarly all hard surfaced thermal conductors need thermal paste at the interfaces.
Thermal pads, that conform to the device and heatsink, do not need thermal paste. There is some literature that suggests that adding thermal paste hinders heat flow.
I hope those clips are very strong and keeps a firm tight pressure on the devices
remember we are approaching the limits of cooling, temperature stability, etc
or else it might be possible to adjust pressure with copper plate
The manufacturer makes the point that a screw-in method directly through the hole in the semiconductor case (as this hole is up one end of the device) can cause part of the casing to ride up, away from the heatsink. The other advantage to the clip is that it rests against the plastic and shouldn't need electrically insulating, whereas a conventional screw would need an insulating collar.
I see on Mouser Electronics they have an insulator called Thermasil. Is this Kapton or Mica does anyone know. I see also the insulator kit AR2 is referring to but I can't see whether it is the same material.
Regards,
Chris
I see on Mouser Electronics they have an insulator called Thermasil. Is this Kapton or Mica does anyone know. I see also the insulator kit AR2 is referring to but I can't see whether it is the same material.
Regards,
Chris
Incidentally, I couldn't find anyone selling the Keystone CL-60 Thermistor in the UK but I have bought, what I think is an identical part, under the GE Sensing/Thermometrics brand called the CL-60, which is rated at 5A and 10ohms. Its part number from Farnell is 1653469.
Regards,
Chris
Regards,
Chris
Another bonus feature of the clips is that they allow the use of a #6 screw which is much easier for those of us who are metal working challenged to tap.
Thermalsil is a silicone rubber isolator. Good: no need for goop Other: higher thermal resistance than mica, .9C/W vs. ~.7 with appropriate grease. If you are pushing it go for mica.
I think the issue with grease causing higher thermal resistance comes if you use too much. You want just enough to fill the surface irregularities, not so much that it squeezes out all over the place when you tighten the mounting screw(s).
The GE CL-60 is what you want.
Thermalsil is a silicone rubber isolator. Good: no need for goop Other: higher thermal resistance than mica, .9C/W vs. ~.7 with appropriate grease. If you are pushing it go for mica.
I think the issue with grease causing higher thermal resistance comes if you use too much. You want just enough to fill the surface irregularities, not so much that it squeezes out all over the place when you tighten the mounting screw(s).
The GE CL-60 is what you want.
It is also worth mentioning that Nelson used CL60 in various locations for various purposes. It is used in line to AC feed to transformer, which provides sort of slow start to the transformer, and I found out sometimes eliminates the mechanical broom of transformer. He also use it to separate starpoint ground from chassis ground.
that is partly true
but mostly related to the medium power driver type devices
plastic casing high power devices are cast with built in insulation for the mounting screw
a thick big planed washer will help some on the problem with the offset position of the hole
and a tooth spring/lock washer
I dont know how much the ordinary scew mount will do for heat transfer from device casing to heatsink
but sure, some of the thicker and softie insulators could play tricks that are not so obvious
with the screw hole close to top of device, a softy insulator could be unevenly pressured, theoretically
but there are other ways to place the pressure in the middle
question is what happens to the casing
the device casing is reinforced around hole, and is meant to have the load pressure there
on the other hand, edges are probably equally strong
Attachments
Just received a Mouser delivery containing Thermasil insulators which I think I'll use so I don't need to use any heat transfer compound, rather than the mica insulators.
The Max Clips have also arrived and look fine but at what point on the heatsink to attach them? The heatsinks I'm looking at are 16.5cmX20.0cm so is the ideal position in the middle or doesn't it really matter if they conduct efficiently? I was going to place the Mosfets near the bottom and attach the pcb to the base plate which seemed a logical approach.
I see I've ordered the wrong thermistors which are 5A and 5ohm rather than 10. Twas ever thus for us manufacturers!
Regards,
Chris
The Max Clips have also arrived and look fine but at what point on the heatsink to attach them? The heatsinks I'm looking at are 16.5cmX20.0cm so is the ideal position in the middle or doesn't it really matter if they conduct efficiently? I was going to place the Mosfets near the bottom and attach the pcb to the base plate which seemed a logical approach.
I see I've ordered the wrong thermistors which are 5A and 5ohm rather than 10. Twas ever thus for us manufacturers!
Regards,
Chris
try to find heatsink manufacturers' data or info sheet on optimum locations.
For a single row across the width the height from the bottom to the centres of the interface should be ~40% of the heatsink height.
If you have 2devices then locate them at 25% in from each edge.
If you have 3devices then the middle is 50% in from the edge and the outers are at 17% from each edge.
For 4devices the spacing is decided by dividing the heatsink into 4 equal vertical slices and placing each device in the middle of it's slice, i.e. @ 12.5% & 37.5% from the edges.
This rule also fits the earlier examples and for many devices in a single row.
If you have two horizontal rows then the lower row should be ~20% up from the bottom, the upper row ~60% up from the bottom.
For a single row across the width the height from the bottom to the centres of the interface should be ~40% of the heatsink height.
If you have 2devices then locate them at 25% in from each edge.
If you have 3devices then the middle is 50% in from the edge and the outers are at 17% from each edge.
For 4devices the spacing is decided by dividing the heatsink into 4 equal vertical slices and placing each device in the middle of it's slice, i.e. @ 12.5% & 37.5% from the edges.
This rule also fits the earlier examples and for many devices in a single row.
If you have two horizontal rows then the lower row should be ~20% up from the bottom, the upper row ~60% up from the bottom.
BA2 cascode front end
Well the new cacode boards are back and built up (new Layout) but with 55volts +/- the dissipation of R208 and R209 is around 1 watt which is hotter than I like. When the output boards get hooked up and biased the supply will load down to around 50 to 52 volts. My question is can I double the value of R208 and R209 to around 1500 ohm each? This will keep the current approx. the same as original design. I don't quite understand the boot strap function of R208 and R209 so I am not sure what the results of the change will do. Any help would be appreciated.
Well the new cacode boards are back and built up (new Layout) but with 55volts +/- the dissipation of R208 and R209 is around 1 watt which is hotter than I like. When the output boards get hooked up and biased the supply will load down to around 50 to 52 volts. My question is can I double the value of R208 and R209 to around 1500 ohm each? This will keep the current approx. the same as original design. I don't quite understand the boot strap function of R208 and R209 so I am not sure what the results of the change will do. Any help would be appreciated.
BA2 cascode front end
Changed resistors R208 and R209 as per post 416 and increased the size of heat sink on Q203, everything seems to running ok on bench test.The Amp can swing 80 volts (peak to peak) into 4 ohms load. With out any visiable distortion on scope. Will run bias around 150ma per fet as the heat sink got a little to hot at 250ma, so it looks like 150 to 170 ma is about all the sinks can handle safetly. Now to get the stuff into the chassis will post pictures when get it in the box. By the way the amp will swing 92 volts (peak to peak ) into 10 ohms.
Changed resistors R208 and R209 as per post 416 and increased the size of heat sink on Q203, everything seems to running ok on bench test.The Amp can swing 80 volts (peak to peak) into 4 ohms load. With out any visiable distortion on scope. Will run bias around 150ma per fet as the heat sink got a little to hot at 250ma, so it looks like 150 to 170 ma is about all the sinks can handle safetly. Now to get the stuff into the chassis will post pictures when get it in the box. By the way the amp will swing 92 volts (peak to peak ) into 10 ohms.
Have you increased R208, 209 because you've used a higher voltage from the power supply?
Components from Jack have arrived safely so all I need are the pcbs and a bit of get up and go.
I have some 10mH chokes here (wire wound ferrite). Does anyone have a view on using one of these instead of the four parallel R1 to R4 resistors on the power supply.
Regards,
Chris
Components from Jack have arrived safely so all I need are the pcbs and a bit of get up and go.
I have some 10mH chokes here (wire wound ferrite). Does anyone have a view on using one of these instead of the four parallel R1 to R4 resistors on the power supply.
Regards,
Chris
Yes with the 750 ohm and +/- 55 volt supply (no load) the resistors were pushing 1.0 watt (calculated) even though they were 3 watt resistors they were getting hotter than I preferred. Also the IRF 610 driver was pushing around 1.5 watts, which was also more than I liked. (Even though this is a burning Amp).