Now there's an interesting question.
450 watt dissipation means maybe 10 to 20 devices, or 5 to
10 devices in parallel off each rail.
If the 450 watts is with a 25 or 30 deg temperature rise, then
we can talk as few as 10 total, although 16 is more like it.
You can build very large Alephs with stability, since the dominant
pole is the capacitance of the devices themselves, but as the
number of parallel devices piles up, the Cgd starts getting out
of hand, and the distortion at high frequencies gets worse.
I recommend maximal sinking, minimal devices, and maximal
current through the input diff pair and the npn transistor used to
control the current source. If you are using the original Mosfet
input circuits, that means 20 mA per input device to drive the
capacitance.
All very do-able.
450 watt dissipation means maybe 10 to 20 devices, or 5 to
10 devices in parallel off each rail.
If the 450 watts is with a 25 or 30 deg temperature rise, then
we can talk as few as 10 total, although 16 is more like it.
You can build very large Alephs with stability, since the dominant
pole is the capacitance of the devices themselves, but as the
number of parallel devices piles up, the Cgd starts getting out
of hand, and the distortion at high frequencies gets worse.
I recommend maximal sinking, minimal devices, and maximal
current through the input diff pair and the npn transistor used to
control the current source. If you are using the original Mosfet
input circuits, that means 20 mA per input device to drive the
capacitance.
All very do-able.
Thank you for the input Nelson. Based off the Aleph Power and Dissipation cheat sheet attached I was able to dissipate close to 500w with each block without concern, between 400-450 would be a little more safe, not sure if that sheet is 100% accurate though. Right now I have an Aleph with 4 outputs and 0.16R source resistors running at +/-20v rails which comes out to 7.3 amps, and it sounds great. I can only listen to it for an hour though because the coils in the supply start to smell and it gives me a headache; with a low speed window fan I can listen as long as I want, but its noisy. Probably need Solen's 10ga or 12ga coils to make it work reliably and of course a few more fets.
Another question: (based off the attached sheet) How am I getting 148C junction temperatures with a fet thats dissipating 21watts?
-john
edit-edit: I think I can figure out the input diff pair current. By removing R19 (ref Aleph2), is this how we achieve max current for the npn controlling the current source?
Another question: (based off the attached sheet) How am I getting 148C junction temperatures with a fet thats dissipating 21watts?
-john
edit-edit: I think I can figure out the input diff pair current. By removing R19 (ref Aleph2), is this how we achieve max current for the npn controlling the current source?
I take the opportunity to ask for few advice:
I posted this project long time ago...and obviously it will take time since i'm gathering experience on amps by doing other things and first i'll make the Aleph p1.7
This project is pleasently abicious (aleph AX).
Since Nelson mentioned that the Cgate could be an issue,and rises up by paralleling output devices causing therefore bad high frequency responce.
I am actually planning on buildin g an AX version with 8 devices per quadrant (32 devices per channel).
Bias set up to 1.3A per device; dissipation on each IRF 38W.
Tot dissipation 1200W per monoblock (600W per side)
By using a 0.05deg/W the temperature should rises up 30deg.
Current set up on the front end by the CCS is 27mA (so 13.5mA per each 9610 in the front end)
RMS power
100W@8ohm
200W@4ohm
400W@2ohm
First question:
- Nelson mentioned that the current on the input 9610s should be as high as possible....but...does this current fix the output DC offset? does it? So it cannot be arbitrary and not any value can be chosen, correct?
- Assuming that i can handle the heat with this heatinks, aren't the output devices gonna be to many that they will cause a bad high frequency response?
- Is there any general advice on the project that you guys can give me?
any hints/tip is very welcome at this point.
I posted this project long time ago...and obviously it will take time since i'm gathering experience on amps by doing other things and first i'll make the Aleph p1.7
This project is pleasently abicious (aleph AX).
Since Nelson mentioned that the Cgate could be an issue,and rises up by paralleling output devices causing therefore bad high frequency responce.
I am actually planning on buildin g an AX version with 8 devices per quadrant (32 devices per channel).
Bias set up to 1.3A per device; dissipation on each IRF 38W.
Tot dissipation 1200W per monoblock (600W per side)
By using a 0.05deg/W the temperature should rises up 30deg.
Current set up on the front end by the CCS is 27mA (so 13.5mA per each 9610 in the front end)
RMS power
100W@8ohm
200W@4ohm
400W@2ohm
First question:
- Nelson mentioned that the current on the input 9610s should be as high as possible....but...does this current fix the output DC offset? does it? So it cannot be arbitrary and not any value can be chosen, correct?
- Assuming that i can handle the heat with this heatinks, aren't the output devices gonna be to many that they will cause a bad high frequency response?
- Is there any general advice on the project that you guys can give me?
any hints/tip is very welcome at this point.
MEGA-amp said:
The semiconductor is going to experience something like 1 to 2
deg C per watt, so you would get a junction temp like that if
your heat sink is over 100 deg, which seems to be the case on
your sheet.
Regarding the npn current, you are looking to reduce the
resistor values all around that part of the circuit to get the
current (and thus the speed) up.
Stefanoo said:
Let's remember that by the standards of most of the guys over
in Solid State, the distortion of the Alephs is too much anyway.
The 9610's are power transistors - you can crank them right on
up as long as you heat sink them. As you do, the Drain resistance
on them needs to go down proportionately, and as it does, you
will see the open loop gain decline and the bandwidth increase.
Could be quite good. Remember not to use the IR parts on the
9610's.
Hi Mega Amp,
just have look at the cheat sheet. You´ve used 0,2°K/W heatsinks wich will give a 100°C temerature rise over ambient (for 511watts). So at 25°C the heatsinks will be 125°
The 21watts per fet will raise the junction a further 21*0,83° plus the thermal resistance between device and heatsink.
So if you want to have precise resiults you have to feed the sheet with the right data.........
William (who always seems to be a bit irritated when people doubt his spreadsheet)
just have look at the cheat sheet. You´ve used 0,2°K/W heatsinks wich will give a 100°C temerature rise over ambient (for 511watts). So at 25°C the heatsinks will be 125°
The 21watts per fet will raise the junction a further 21*0,83° plus the thermal resistance between device and heatsink.
So if you want to have precise resiults you have to feed the sheet
with the right data.........William (who always seems to be a bit irritated when people doubt his spreadsheet)
Ah yes, I'm still trying to figure the proper heatsink parameters to make it accurate. I knew my data wasn't completely correct. But I do have two massive heatsinks and plan on attaching a cold plate to the very top of both. This would allow me to make single chassis for both channels, but the PS will have to be separate of course. That sheet is very helpful for someone like myself who isn't into formulas, so I thank you.
cold plate
-john
cold plate
-john
Attachments
Around here the plans change on a minute by minute basis. That is what I have my Aleph attached to currently. My initial Aleph experiments were with those big boxes I built a while ago. By themselves those sinks are much better off with forced cooling. I was dissipating 700 watts on one of those with a high speed fan on it, and it was cool to the touch. By placing a cold pate on the top of any sink, aren't we effectively raising the efficiency of sink exponentially? With those big sinks, I would probably still need a fan, but I can run it slowly and quietly, couple that with a cold plate on top and I could dissipate as much as I want, I would imagine.
edit: 5 fins/inch
edit: 5 fins/inch
Which one would that be ?
The lowest "theoretical" die temperature would be 50-55C, depending on the ambient temperature.
A heat exchanger can only function with a temperature difference, therefore the cool plate will be hotter than ambient.
By controlling the waterflow the cool plate temperature can be tuned, just regulate the waterpump speed.
The lowest "theoretical" die temperature would be 50-55C, depending on the ambient temperature.
A heat exchanger can only function with a temperature difference, therefore the cool plate will be hotter than ambient.
By controlling the waterflow the cool plate temperature can be tuned, just regulate the waterpump speed.
So I bias to wherever I want to, adjust water pump and/or radiator fan speed accordingly making sure that the fets are well within their SOA? How much bias is proper for 16 fets/channel? Probably best to configure the trafos for 240v mains.
cold plate :
Performance @ 1 GPM: 0.0110 (spec sheet doesn't say what unit its in, i'm assuming its heatsink thermal resistance).
If Rthh-a = 0.0110
then @ 3 GPM: .0110/3 = .0037 (?)
---------
441W into 2ohms!!! I'm gonna have to change my name to MEGA-speakers!
cold plate :
Performance @ 1 GPM: 0.0110 (spec sheet doesn't say what unit its in, i'm assuming its heatsink thermal resistance).
If Rthh-a = 0.0110
then @ 3 GPM: .0110/3 = .0037 (?)
---------
441W into 2ohms!!!
I'm gonna have to change my name to MEGA-speakers!
Nelson Pass said:
Let's remember that by the standards of most of the guys over
in Solid State, the distortion of the Alephs is too much anyway.
The 9610's are power transistors - you can crank them right on
up as long as you heat sink them. As you do, the Drain resistance
on them needs to go down proportionately, and as it does, you
will see the open loop gain decline and the bandwidth increase.
Could be quite good. Remember not to use the IR parts on the
9610's.
thanks Nelson for your advice.
I'll be trying to get the FC part.
Actually only the IRF part seems available i'll try to see if i can get it somewhere.
Thanks again.
Best,
Stefano.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.