Realistically, why not pot the whole circuit in some propriatary mixture.
epoxy, marble chips, sand, carbon fiber, etc. Leave the heatsink attachment surface protruding 1/32" at the surface of the block. Make a heatsink out of a 1/2" thick plate minimum. Bond this to the potted blockwith epoxy with the plate in contact with the chip , filling in the gap around the chip at the same time. Of course the chip and heatsink need grease between them. Now you have bonded disimilar materials together- even better damping
Then suspend the block inside the cabinet using lengths of monofiliment fishing line under tension. You can adjust the tension using a guitar string tuner knob.
Make sure the solid core wire going in and out has a loop in it to
allow movement and avoid (unlikely) fatigue breakage. Stranded wire might add noise when it flexes?
Another idea is just trim cut the epoxy/aluminum and buff so it is a clean solid polished block. Don't even put it in a case The only trick is how to embed the in/out connecters into the block. How
about blocks soldered to each wire and when you embed the circuit leave the blocks partially protruding. Then they get cut off flush with the surface during the trimming/polishing operation. Now drill and tap the block for a screw to secure the outputs, power, etc.
Wait..... instead of a potting mixture, use the clear resin that they use for those paperweights with bugs in them. (available at hobby shops) Then you end up with the visible circuit bonded to the aluminum plate, all polished up. OK I feel better now, thanks........
Or don't do all this- it's just what I plan to do....
epoxy, marble chips, sand, carbon fiber, etc. Leave the heatsink attachment surface protruding 1/32" at the surface of the block. Make a heatsink out of a 1/2" thick plate minimum. Bond this to the potted blockwith epoxy with the plate in contact with the chip , filling in the gap around the chip at the same time. Of course the chip and heatsink need grease between them. Now you have bonded disimilar materials together- even better damping
Then suspend the block inside the cabinet using lengths of monofiliment fishing line under tension. You can adjust the tension using a guitar string tuner knob.
Make sure the solid core wire going in and out has a loop in it to
allow movement and avoid (unlikely) fatigue breakage. Stranded wire might add noise when it flexes?
Another idea is just trim cut the epoxy/aluminum and buff so it is a clean solid polished block. Don't even put it in a case The only trick is how to embed the in/out connecters into the block. How
about blocks soldered to each wire and when you embed the circuit leave the blocks partially protruding. Then they get cut off flush with the surface during the trimming/polishing operation. Now drill and tap the block for a screw to secure the outputs, power, etc.
Wait..... instead of a potting mixture, use the clear resin that they use for those paperweights with bugs in them. (available at hobby shops) Then you end up with the visible circuit bonded to the aluminum plate, all polished up. OK I feel better now, thanks........
Or don't do all this- it's just what I plan to do....
Peter
You might also use natural oil for insulation and thermal transport (instead of all those poisonous xformer oils). I once heard that some natural oil (I vaguely remember it was peanut oil) was used during WWII for high-voltage capacitors in RADAR devices. So its dielectrical properties shouldn't be that bad at all !
But I fear that its heat transfer properties will be much much worse than those of aluminium for instance.
Regards
Charles
You might also use natural oil for insulation and thermal transport (instead of all those poisonous xformer oils). I once heard that some natural oil (I vaguely remember it was peanut oil) was used during WWII for high-voltage capacitors in RADAR devices. So its dielectrical properties shouldn't be that bad at all !
But I fear that its heat transfer properties will be much much worse than those of aluminium for instance.
Regards
Charles
I work for a motor manufacturer and I know that there are submerged motors (pods) that are oil filled, and that are higher voltage fed that an amp, with inverters that have a dv/dt in the order of infinity and beyond, so the dielectric capability is proven. Unfortunately, we don't build them and can't help you with a reference for the oil.
What I can suggest you is to use an internal blower to toss the oil bath. Don't laugh, please. The micro turbulences and waves can act like a resonance barrier with very good results. You only have to tune it to the desired Fs and Q, (I think the speed &power control of the motorfan will do) that is the one that causes less problems to the chip. This concept is studied by me in another field (nothing to do with electronics) and maybe has some use for you. I hope so.
What I can suggest you is to use an internal blower to toss the oil bath. Don't laugh, please. The micro turbulences and waves can act like a resonance barrier with very good results. You only have to tune it to the desired Fs and Q, (I think the speed &power control of the motorfan will do) that is the one that causes less problems to the chip. This concept is studied by me in another field (nothing to do with electronics) and maybe has some use for you. I hope so.
I once found a couple of dielectric properties tables while i was playing with high voltage; and it's true; some oils are quite good isolators (with dielectric constants comparable to porcelain, 5-10). Water (distilled, deionized one of coz) is a one of the best known electrical insulator, with a dielectric constant of 75-80.
Whatever you choose to use, i'd make sure to have a *tight* seal on the can, so not only it doesn't leak at all but also it doesn't get dirty.
Whatever you choose to use, i'd make sure to have a *tight* seal on the can, so not only it doesn't leak at all but also it doesn't get dirty.
Hi Peter
I had a look into an older ARRL handbook. There you could find the construction of an RF dummy load. There is also a discussion about availability/properties of xformer oils and mineral oil.
They claim that the product "Regal Oil R&O No.46" by TEXACO (sold in 5 gallon cans as "turbine oil" for servicing purposes) was in fact a relabelled heat transfer oil as it is used in some xformers. It seems to be the only heat transfer oil available in "small" quantities.
Regards
Charles
BTW: I have seen huge inverters in a lab intended for railway locomotives, using oil cooling. I have seen some electronic containers that were ripped open by boiling oil when "things got interesting"*.
*copyright by NP
I had a look into an older ARRL handbook. There you could find the construction of an RF dummy load. There is also a discussion about availability/properties of xformer oils and mineral oil.
They claim that the product "Regal Oil R&O No.46" by TEXACO (sold in 5 gallon cans as "turbine oil" for servicing purposes) was in fact a relabelled heat transfer oil as it is used in some xformers. It seems to be the only heat transfer oil available in "small" quantities.
Regards
Charles
BTW: I have seen huge inverters in a lab intended for railway locomotives, using oil cooling. I have seen some electronic containers that were ripped open by boiling oil when "things got interesting"*.
*copyright by NP
Peter ,
Water has about 25 times higher thermal conductivity than air, but in autoconvective cooling as in this case we can expect the thermal resistance junction-ambient to decrease maybe a factor 10. (a engineers guestimate)
Take the Rtja value for LM3875 is 43 C/W, we land on about 4,3 C/W. With a powerloss of 15 W => delta T = 65 C. Add this to the thermal resistance of the rest of the equipment where heat is dissipated.
As for oils, forget it. They usually have a much higher viscosity so the Rtja will not decrease 10 times, and if it only decreases 5 times you land on a temperature rise of about 130 C from chip junction to liquid. Perhaps the 3-M stuff could do it, Fluoriert.
Anyway, you could probably coat the whole smackarang in epoxy, or even mold it into expoxy to protect it. What you lose in form of higher Rtja by more material, you probably gain by increasing surface in contact with cooling liquid. plus you can use deionized water, whcih perhaps has to be replaced from time to time because it dissolves impurities .
BTW , water hardly stops vibrations.
/rickard
Water has about 25 times higher thermal conductivity than air, but in autoconvective cooling as in this case we can expect the thermal resistance junction-ambient to decrease maybe a factor 10. (a engineers guestimate)
Take the Rtja value for LM3875 is 43 C/W, we land on about 4,3 C/W. With a powerloss of 15 W => delta T = 65 C. Add this to the thermal resistance of the rest of the equipment where heat is dissipated.
As for oils, forget it. They usually have a much higher viscosity so the Rtja will not decrease 10 times, and if it only decreases 5 times you land on a temperature rise of about 130 C from chip junction to liquid. Perhaps the 3-M stuff could do it, Fluoriert.
Anyway, you could probably coat the whole smackarang in epoxy, or even mold it into expoxy to protect it. What you lose in form of higher Rtja by more material, you probably gain by increasing surface in contact with cooling liquid. plus you can use deionized water, whcih perhaps has to be replaced from time to time because it dissolves impurities .
BTW , water hardly stops vibrations.
/rickard
water dampens vibrations better than metall, yes, but hardly stops them.
Immersion an object in water is a very very bad way of damping vibrations. Compare it to an ultrasonic bath, or an "utltrasonic agitator , a rod vibrating with ultrasonic frequencey.
It is partly affected by different temperatures , since the speed of sound alters with water temperature, thus "bending" the sound waves , and make it possible to reflect the sound.
The point is that you should put different layers of substance , which all have a very wide dispersion of speed on sound.
This leads to that instead of transfering sound (vibrations) between the different layers , it is reflected out.
And oil is such a lousy heat transfer media without agitation, i e natural convection that I hardly thinkt that it will work. The dampening factor is probably higher, due to higher viscosity and lower speed of sound, but this also means that the convective heat transfer factor is reduced dramatically.
/rickard
Immersion an object in water is a very very bad way of damping vibrations. Compare it to an ultrasonic bath, or an "utltrasonic agitator , a rod vibrating with ultrasonic frequencey.
It is partly affected by different temperatures , since the speed of sound alters with water temperature, thus "bending" the sound waves , and make it possible to reflect the sound.
The point is that you should put different layers of substance , which all have a very wide dispersion of speed on sound.
This leads to that instead of transfering sound (vibrations) between the different layers , it is reflected out.
And oil is such a lousy heat transfer media without agitation, i e natural convection that I hardly thinkt that it will work. The dampening factor is probably higher, due to higher viscosity and lower speed of sound, but this also means that the convective heat transfer factor is reduced dramatically.
/rickard
I think that by the time you get through, you can achieve the same results by vibration isolating a normal heatsink. That methoid (sp) would allow one to easly tune the isolators for a specific frequency.
PS: I spent many years working with motors used in underwater remote operated vehicles and we used Floronert.
PPS: Oil cooled amp would be BIG marketing idea though - especially using oil with specific gravity to cancel out negative gravity waves.
PS: I spent many years working with motors used in underwater remote operated vehicles and we used Floronert.
PPS: Oil cooled amp would be BIG marketing idea though - especially using oil with specific gravity to cancel out negative gravity waves.
hot chips
Hi peter,
I recently contacted electrolube here to find the best resin for my resin & small stones chip mounting idea. They came back to me today and in the course of the conversation It came out that either of the resins that they recommend - both of which are heat conductive compounds - have heat transfir properties of about 100 times less than aluminium. that's what they said
That got me thinking a bit...
can anyone out there confirm that these are good ballpark figures
do you have similar data for any of your proposed fluids ?
mike
Hi peter,
I recently contacted electrolube here to find the best resin for my resin & small stones chip mounting idea. They came back to me today and in the course of the conversation It came out that either of the resins that they recommend - both of which are heat conductive compounds - have heat transfir properties of about 100 times less than aluminium. that's what they said
That got me thinking a bit...
can anyone out there confirm that these are good ballpark figures
do you have similar data for any of your proposed fluids ?
mike
for those who think so hard about resistor/capacitor/heatsink sound because of the additional capacitance (change) caused by insulation pads, aren't you going to be concerned about now massive (relatively speaking) amount of capacitance between the leads, the container, and the wires because of the dielectric liquid?
That thing itself is a giant capacitor.
That thing itself is a giant capacitor.
You want something new !?
What about soldering a LM3875 to a heatsink ?
Take a heavy soldering iron and first heat up the solder on a heatsink (takes some time). Then push the LM3875 on the heatsink. Don't know if it works. It should work, I guess. The advantage is a good coupling of heat and a reduction of resonances...
If the heatsink can not be soldered, you could try a thin metal plate (5x5 cm or so). Then attach the plate to the heatsink!
Fedde
What about soldering a LM3875 to a heatsink ?
Take a heavy soldering iron and first heat up the solder on a heatsink (takes some time). Then push the LM3875 on the heatsink. Don't know if it works. It should work, I guess. The advantage is a good coupling of heat and a reduction of resonances...
If the heatsink can not be soldered, you could try a thin metal plate (5x5 cm or so). Then attach the plate to the heatsink!
Fedde
Burning your fuses
Guys, you're burning your fuses with a simple thing.
I would do it this way:
1. Mount everithing (chips, components, etc), be it P-P or in a circuit. Circuit is better for this.
2. Connect all the wires and TEST everything.
3. After tested, drawn everything, holding by the wires, in a can of VARNISH.
4. Let it dry holding by the wires and do it again two more times.
5. Attach a very small heatsink to each chip.
6. Put everything inside the box and THROW THE WATER IN.
That's it.
Well... you have to see if the box doesn't leak.
Guys, you're burning your fuses with a simple thing.
I would do it this way:
1. Mount everithing (chips, components, etc), be it P-P or in a circuit. Circuit is better for this.
2. Connect all the wires and TEST everything.
3. After tested, drawn everything, holding by the wires, in a can of VARNISH.
4. Let it dry holding by the wires and do it again two more times.
5. Attach a very small heatsink to each chip.
6. Put everything inside the box and THROW THE WATER IN.
That's it.
Well... you have to see if the box doesn't leak.
the chip is available in either unpackaged die or wafer form. So get one of those (one wafer could be too much), use the minimal configuration (two resistors plus wire out for two by-passing caps. solder those two smd resistors on, and seal the whole thing with one drop of epoxy.
now, that will get any "shortest signal path" freak to smile,
you may be able to pack a 4-channel gaincloen in a Timex.
now, that will get any "shortest signal path" freak to smile,
you may be able to pack a 4-channel gaincloen in a Timex.
carlosfm said:
This chip deserves a Rolex.
Somebody actually suggested to me a Patek.
Thanks everybody for the ideas. Of course I want to make it as simple as possible and I will definitely try it. One reason for me to pursue that type of approach is building active speaker system, where drivers are mounted in a wall (as it's made of 2x4s, I will have resonances to deal with). The amps, I want to be mounted as close to the drivers as possible (so the speaker wires effect is also eliminated) and that's why the demand for more isolation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.