The package limitation was the main problem with the PA amplifiers. The first solution was the QSC inside-out amplifier, where the output devices was mounted without insulation. That saves 0,2-0,5K/W thermal resistance, which allows to use much smaller heatsink and/or less output devices.
Sajti
Sajti
ok thanks for Information, I will add output devices
but only for general information,..
why offer IXYS Linear Mosfet 960W Power Dissipation for power amplifier
when only about 300W possibile ? This mosfet cost 25 EUR pcs.
In ths case I can buy for 5 EUR heavy Onsemi / Fairchild / Infineon 250V 300-500W Mosfet and add devices
In this case it makes no sense to pay for expensive Linear Mosfet
Datasheets gives the theoretical maximum dissipation (Tcase 25C), which means practically not too much information. Different mounting methods allows different maximum dissipation, which is not part of the catalogue.
Check the IRFP240 datasheet! It gives case to sink thermal resistance (0.24K/W), with no insulation. The proper way to calculate the maximum dissipation is the following:
1, calculate the total junction to heatsink therma resistance:
- junction to case+case to heatsink (case to heatsink is somewhere between 0.25 - 1K/W depended by the insulator, and mounting technology)
2, calculate the junction and heatsink temperature difference:
Maximum junction temperature is 150C. Heatsink temperature depended the heatsink applied. Say 50C, so the difference is 100C
3, Max dissipation is divide the temperature difference by the total thermal resistance.
Sajti
Just a practical calculation:
T(j-c)=0,05K/W (2500W max dissipation)
T(c-s)=0,24K/W (mounting with no insulation)
Maximum possible dissipation:
T(heatsink)=25C Pmax=431W
T(heatsink)=35C Pmax=396W
T(heatsink)=45C Pmax=362W
T(heatsink)=55C Pmax=327W
T(heatsink)=65C Pmax=293W
T(heatsink)=75C Pmax=258W
T(heatsink)=85C Pmax=224W
T(heatsink)=95C Pmax=189W
Sajti
T(j-c)=0,05K/W (2500W max dissipation)
T(c-s)=0,24K/W (mounting with no insulation)
Maximum possible dissipation:
T(heatsink)=25C Pmax=431W
T(heatsink)=35C Pmax=396W
T(heatsink)=45C Pmax=362W
T(heatsink)=55C Pmax=327W
T(heatsink)=65C Pmax=293W
T(heatsink)=75C Pmax=258W
T(heatsink)=85C Pmax=224W
T(heatsink)=95C Pmax=189W
Sajti
Add to that : heatsink temperature is NOT ambient air temperature.
So even if you have a 0.2C/W heatsink which is either:
* HUGE
* heavily air cooled
* water cooled
or 2 or 3 of the above combined.
Now suppose you are dissipating 431W , the first case above with heatsink at 25C and which is already a huge step down from prominently displayed "960W dissipation".
You still need to pull away those 431W being injected into the heatsink so they are carried away using air or liquid.
So you are pushing 431W through heatsink , which has a thermal resistance of 0.2C/W , which means temperature drop will be 431*0.2=86C
Do we agree so far?
Draw the *full* thermal path from junction to air and do the Math yourself.
Now: 25C heatsink temperature means cooling medium (air or liquid) needs to be at ...... 25C-86C=-61C
Yes, you read it right: 61 Degrees Celsius below zero.
As you see, this justifies my beef that "dissipation at 25C case temperature" is the single most misleading spec in a datasheet.
Notice I called it misleading , although it is Mathematically correct ... it just seems to promise huge power that is not achievable in practice, by a huge margin.
So even if you have a 0.2C/W heatsink which is either:
* HUGE
* heavily air cooled
* water cooled
or 2 or 3 of the above combined.
Now suppose you are dissipating 431W , the first case above with heatsink at 25C and which is already a huge step down from prominently displayed "960W dissipation".
You still need to pull away those 431W being injected into the heatsink so they are carried away using air or liquid.
So you are pushing 431W through heatsink , which has a thermal resistance of 0.2C/W , which means temperature drop will be 431*0.2=86C
Do we agree so far?
Draw the *full* thermal path from junction to air and do the Math yourself.
Now: 25C heatsink temperature means cooling medium (air or liquid) needs to be at ...... 25C-86C=-61C
Yes, you read it right: 61 Degrees Celsius below zero.
As you see, this justifies my beef that "dissipation at 25C case temperature" is the single most misleading spec in a datasheet.
Notice I called it misleading , although it is Mathematically correct ... it just seems to promise huge power that is not achievable in practice, by a huge margin.
Last edited:
How comes that ICU puts out 3 times more power than is transferred to the wheels having radiator of resonable size?
One of the smallest cold plates: 2x2" 0.045 deg C per W. The very same 0.43 plus 0.21 yeilds delta T of 128 C at 200W. 200W @ 0.045C/W & 25C at plate surface yeilds 9C between plate surface and coolant meaning 16 C temperature of coolant.
In industry it's mostly assumed 40C at plate surface thus lowering power on avail 176 W and coolant temperature of 32 C.
One of the smallest cold plates: 2x2" 0.045 deg C per W. The very same 0.43 plus 0.21 yeilds delta T of 128 C at 200W. 200W @ 0.045C/W & 25C at plate surface yeilds 9C between plate surface and coolant meaning 16 C temperature of coolant.
In industry it's mostly assumed 40C at plate surface thus lowering power on avail 176 W and coolant temperature of 32 C.
Well,
It has been a LOT of time ago, but reconnecting to what NMOS said, I had a thread about that.
Such IXIS MOSFETS NMOS says exist in SOT-227 package , too.
That package has roughly double the contact area of the TO-264, has 2 M4 mounting screws and is fully insulated.
Vertical MOSFETs for a PA amplifier OP
IXTN90N25L2 - L2™ Series - N Channel Linear Discrete MOSFETs - Littelfuse
Would a pair device like IXTN110N20L2 or IXTN90N25L2 be suitable for the OP's 300W goal?
I mean without exotic cooling , just a 150 x 150 x 75 mm fin (i.e. 0.5K/W) and good forced ventilation?
It seems feasible to me.
It has been a LOT of time ago, but reconnecting to what NMOS said, I had a thread about that.
Such IXIS MOSFETS NMOS says exist in SOT-227 package , too.
That package has roughly double the contact area of the TO-264, has 2 M4 mounting screws and is fully insulated.
Vertical MOSFETs for a PA amplifier OP
IXTN90N25L2 - L2™ Series - N Channel Linear Discrete MOSFETs - Littelfuse
Would a pair device like IXTN110N20L2 or IXTN90N25L2 be suitable for the OP's 300W goal?
I mean without exotic cooling , just a 150 x 150 x 75 mm fin (i.e. 0.5K/W) and good forced ventilation?
It seems feasible to me.
Yes it will work, but its time to make Mosfet amp efficient similar Class D
With one pair Huge IXYS device, you can have easy 2000W 4 Ohm
125 V rail +/-
very cheap to add Class D / EEEngine Rail tracking and you have eartquake mosfet monster amp, with very less heat dissipation
But there is no Future for this technology, I get 2000W 4 OHM with 200 x 80 mm PCB Class D Amp and less components