I agree there are no absolute rules regards how hot things get and I certainly followed the conservatice approach.
This also makes excelent reading as well and goes into the various factors to be considered http://sound.westhost.com/heatsinks.htm
The aim ultimately is maintaing a good operating temperature for the Mosfets (OK stating the obvious here!). But what is the maximim or optimal operating temp for the mosfets we are using in the Mini-A? As per the article the temp range the transistor operates at varies device to device and the effect of the heat varies also. It has been said somewhere in this forum that the mosfets (in the Mini-A at least) operate better at lower temps and this seems to be born out by the article.
I also note in the article the effect of the type of insulator used and how the transistor is mounted makes a large difference. It reads that typical sil-pad insulators have a thermal resistance around 1C/watt thus if you are running at 25watt/device there is a 25C differential between heatsink and fet (eg my sinks are running at 40C at the moment and my fets are at 65C). So if you use thin mica insulators rather than sil-pad your heats sinks can effectively run 10-15C hotter for the same mosfet temp.
I have loked at the datasheet for the IRF240 and fail to understand the operating temp limits. If anybody understands this I would dearly like know the best operating range for these devices.
Cheers
This also makes excelent reading as well and goes into the various factors to be considered http://sound.westhost.com/heatsinks.htm
The aim ultimately is maintaing a good operating temperature for the Mosfets (OK stating the obvious here!). But what is the maximim or optimal operating temp for the mosfets we are using in the Mini-A? As per the article the temp range the transistor operates at varies device to device and the effect of the heat varies also. It has been said somewhere in this forum that the mosfets (in the Mini-A at least) operate better at lower temps and this seems to be born out by the article.
I also note in the article the effect of the type of insulator used and how the transistor is mounted makes a large difference. It reads that typical sil-pad insulators have a thermal resistance around 1C/watt thus if you are running at 25watt/device there is a 25C differential between heatsink and fet (eg my sinks are running at 40C at the moment and my fets are at 65C). So if you use thin mica insulators rather than sil-pad your heats sinks can effectively run 10-15C hotter for the same mosfet temp.
I have loked at the datasheet for the IRF240 and fail to understand the operating temp limits. If anybody understands this I would dearly like know the best operating range for these devices.
Cheers
Let it run hot, but have spare transistors ready for ten years down the road. I think if you're wondering if it might be too hot, then it's probably OK (although something one has to learn to accept compared with other cooler-running amps).
To me, hanging a fan on >afterward< seems an admission of defeat or an indicator that the builder didn't know what they were doing. A fan should be designed-in from the beginning using a different style of heat sink (like a chimney or tunnel) and take advantage of the possible space savings and styling options this opens up (like NP's A75, for example.)
To me, hanging a fan on >afterward< seems an admission of defeat or an indicator that the builder didn't know what they were doing. A fan should be designed-in from the beginning using a different style of heat sink (like a chimney or tunnel) and take advantage of the possible space savings and styling options this opens up (like NP's A75, for example.)
http://www.enjoythemusic.com/magazine/manufacture/0808/
And that is for a commercial product where long life and low maintenance is almost as important as sonic performance. It is probably worth noting that the IRF-240 (these are fairly typical) has a maximum junction temperature of 150degC.
If you read the Zenv6 article Nelson talks about 4A bias and 40V rails giving stunning performance, but his heatsinks not coping with the thermal load. I'd suggest this is the key - keeping the heatsink temperature to a level which you decide is acceptable at the bias current and rail voltage which you want to run the amp.
If it sounds good and the heatsinks aren't getting much higher than 60degC then that appears to be a nice compromise.
I'm using a pair of Aleph 30 monoblocks built on the brian gt boards. They were running at 52degC in a 22degC room last night. Unfortunately the hot summers mean it's not realistic to bump the bias or rails higher than the present levels.
And that is for a commercial product where long life and low maintenance is almost as important as sonic performance. It is probably worth noting that the IRF-240 (these are fairly typical) has a maximum junction temperature of 150degC.
If you read the Zenv6 article Nelson talks about 4A bias and 40V rails giving stunning performance, but his heatsinks not coping with the thermal load. I'd suggest this is the key - keeping the heatsink temperature to a level which you decide is acceptable at the bias current and rail voltage which you want to run the amp.
If it sounds good and the heatsinks aren't getting much higher than 60degC then that appears to be a nice compromise.
I'm using a pair of Aleph 30 monoblocks built on the brian gt boards. They were running at 52degC in a 22degC room last night. Unfortunately the hot summers mean it's not realistic to bump the bias or rails higher than the present levels.
ok:
voltage across the resistors is an average of .375V (range 0.360 to 0.395). divided by the resistance is 1.97 which to me is fairly close to 2A - voltage is 12V so each mosfet is putting out 24W, or 48W per side.
Do those figures add up?
This is the detail on the heatsinks:
* HEAT SINK, 150MM, 0.75°C/W
* Thermal resistance:0.75°C/W
* Mounting type:Adhesive Mount
* Length:150mm
* Height, external:25mm
* Width, external:200mm
* Depth, external:150mm
* Length / Height, external:25mm
* Material:Aluminium
* Surface finish:Black Anodised
* Fixing style:Glue bond
What do you guys think? the consensus seems to be that cooler would be better, but I should get away with it. The actual heat off the amp isn't an issue, just whether I am doing harm running it like this. FWIW, I think I have the IRFP140N (red stickers from group buy)
Fran
voltage across the resistors is an average of .375V (range 0.360 to 0.395). divided by the resistance is 1.97 which to me is fairly close to 2A - voltage is 12V so each mosfet is putting out 24W, or 48W per side.
Do those figures add up?
This is the detail on the heatsinks:
* HEAT SINK, 150MM, 0.75°C/W
* Thermal resistance:0.75°C/W
* Mounting type:Adhesive Mount
* Length:150mm
* Height, external:25mm
* Width, external:200mm
* Depth, external:150mm
* Length / Height, external:25mm
* Material:Aluminium
* Surface finish:Black Anodised
* Fixing style:Glue bond
What do you guys think? the consensus seems to be that cooler would be better, but I should get away with it. The actual heat off the amp isn't an issue, just whether I am doing harm running it like this. FWIW, I think I have the IRFP140N (red stickers from group buy)
Fran
Thanks Spzzzzkt. I saw the max junction temp in the datasheet but I was unsure how that would relate to the actual temp you would read on the IRF240 body.
Looks like I have been rather conservative with my heatsinks but I had 6 of them anyway and ended up using 4. I think clearly that 6 would not have worked in the A75 project I bought them for years ago.
On a related subject I have a question around rail voltage and bias current. Having read that 25-30watts/mosfet is a good range to aim for I ended up with 23v and 1.12amps. Question is - is it better to have higer voltage rail or higher bias current in achieving the 25-30 watts? Are there different advantages to each assuming an 8 ohm load?
I kinda understand that higher bias current relates to how much class A the amp runs - am I right and if so what is the relationship betweem the bias current and number of watts in class A?
Cheers
Looks like I have been rather conservative with my heatsinks but I had 6 of them anyway and ended up using 4. I think clearly that 6 would not have worked in the A75 project I bought them for years ago.
On a related subject I have a question around rail voltage and bias current. Having read that 25-30watts/mosfet is a good range to aim for I ended up with 23v and 1.12amps. Question is - is it better to have higer voltage rail or higher bias current in achieving the 25-30 watts? Are there different advantages to each assuming an 8 ohm load?
I kinda understand that higher bias current relates to how much class A the amp runs - am I right and if so what is the relationship betweem the bias current and number of watts in class A?
Cheers
In regards to Pass designs you'll do better reading what Nelson has to say. Rod Elliot says of the Zen amps "Having looked at the original and many of the "improvements" currently on the web, I did a few tests of my own and frankly, found the amp lacking in the fidelity department. Hi-Fi this most certainly is not. " Not exactly a person who's advice I'd look to on optimal operating conditions for a Pass design
Have a look at the Pass article I linked to in my last post. NP explains very clearly that it is bias current which is the most important factor in Class A operation:
Given that I=V/R raising rail voltages will increase the bias. Once you have your rails set it's usually easier to swap resistors.
Heat sink temperatures are a limiting factor not a determinate of performance in themselves. The more heat dissipation you build into the amp the higher the bias current you can run and the lower distortion you'll get.
cheers
Paul
Thanks Paul. Yes I do take what I read on ESP with a grain of salt but what he says about heatsinks seems sound (it was on DIYAUDIO I read about the fets performing better when cooler).
I have been to that link you provided before but I will read through it again!
Just one point tho, you said "Given that I=V/R raising rail voltages will increase the bias". However I have raised the rail voltage a couple of times (reducing value od CRC resistor in PS) and each time the voltage accross the source resistor got a little smaller (same source resistor value .33) thus the bias current got lower. Effectively changing the rail voltage did not raise the dissipation which stayed on 25watt. I assumed I would need to adjust R13 upwards to increase the bias? Must say it did confuse me!
Cheers
I have been to that link you provided before but I will read through it again!
Just one point tho, you said "Given that I=V/R raising rail voltages will increase the bias". However I have raised the rail voltage a couple of times (reducing value od CRC resistor in PS) and each time the voltage accross the source resistor got a little smaller (same source resistor value .33) thus the bias current got lower. Effectively changing the rail voltage did not raise the dissipation which stayed on 25watt. I assumed I would need to adjust R13 upwards to increase the bias? Must say it did confuse me!
Cheers
Cool, (OK, not so cool!)
So the overall thermal resistance is 1.1degC/W junction to case and about the same case to heatsink (using sil-pad). Thus mosfet running at 25watts the junction is around 55C greater than the heatsink temp. With a max junction temp of 150C the heatsink can get pretty darned hot!
Thats good to know.
So the overall thermal resistance is 1.1degC/W junction to case and about the same case to heatsink (using sil-pad). Thus mosfet running at 25watts the junction is around 55C greater than the heatsink temp. With a max junction temp of 150C the heatsink can get pretty darned hot!
Thats good to know.
If I do go the route of trying to add some fans - and right now its a big if, am I better blowing into the case or out of the case? Seems they both should be the same, but what do I know??!
I have 2 small 2" fans that I could mount easily enough, but I want to hook them up to the bench supply first and try them out for noise. I think if I drop the feed voltage just a little they should get quieter. I also have a 4" and a couple of 5" fans but they would be much harder to accomodate in the case.
Still sounds really great tonight, this is quite the revelation for me.. never thought this amp would sound so good. Like, it sounded good, but not top notch, ie I would be keeping the valves going, fine for a second system etc. But not since this upping the bias. Bowls me over completely. Really a big jump.
What needs to happen now is for some of you with mini-alephs to up the bias and see what you hear. This should be replicated and tested. If you want to get the same value as me, with 12V rails, you need to parallel a 0.33 with the 0.47 that was stock.
(Its interesting that a value of 25-30W is desireable for the mosfets - and thats what I am just at now. Makes me wonder what it would sound like if the supply was 14V instead of 12 (although when I measured it tonight it was 13V, which would give me 26W per device)
Fran
I have 2 small 2" fans that I could mount easily enough, but I want to hook them up to the bench supply first and try them out for noise. I think if I drop the feed voltage just a little they should get quieter. I also have a 4" and a couple of 5" fans but they would be much harder to accomodate in the case.
Still sounds really great tonight, this is quite the revelation for me.. never thought this amp would sound so good. Like, it sounded good, but not top notch, ie I would be keeping the valves going, fine for a second system etc. But not since this upping the bias. Bowls me over completely. Really a big jump.
What needs to happen now is for some of you with mini-alephs to up the bias and see what you hear. This should be replicated and tested. If you want to get the same value as me, with 12V rails, you need to parallel a 0.33 with the 0.47 that was stock.
(Its interesting that a value of 25-30W is desireable for the mosfets - and thats what I am just at now. Makes me wonder what it would sound like if the supply was 14V instead of 12 (although when I measured it tonight it was 13V, which would give me 26W per device)
Fran
Chris
yes you are right - my bad. The rail voltage doesn't effect the bias, just the dissipation.
http://domainedupossible.free.fr/pagemus/ALEPH5/Aleph-Power.xls
cheers
Paul
yes you are right - my bad. The rail voltage doesn't effect the bias, just the dissipation.
http://domainedupossible.free.fr/pagemus/ALEPH5/Aleph-Power.xls
cheers
Paul
"If I do go the route of trying to add some fans - and right now its a big if, am I better blowing into the case or out of the case? Seems they both should be the same, but what do I know??!"
Fans exhausting hot air fail sooner due to the higher operating temperatures. But we are talking slight differences here.
Fans exhausting hot air fail sooner due to the higher operating temperatures. But we are talking slight differences here.
Just on the question of which is better, higher bias current or rail voltage I will post this quote from Grollins:
"...The thing about reducing the rail voltage is that distortion will rise rapidly; I was pushing it a bit by building the circuit with 15V rails. The problem being that the MOSFETs' Gate capacitances increase below roughly 25V or so. That, in turn, makes the MOSFETs harder to drive, particularly at high frequencies. As a direct consequence, distortion increases.
... It's just a question of being aware of the tradeoffs before you start making changes.
On the other hand, reducing the output bias will also increase distortion a bit. There's no such thing as a free lunch.
Just for the record, the lowest 'optimum' rail voltage for MOSFETs often turns out to be 25V or so. At that point, the Gate capacitance tends to make a dogleg and begins flattening out. Further increases to the rail voltage tend to produce only incremental reductions in Gate capacitance. ..."
Which is why I decided on a higher rail voltage (23V). But Nelson says in the "leaving class A" article higher bias also reduces distortion. Best of both worlds results in more heat to get rid of. I guess at some point there is an optimal balance between bias and rail voltage for a given heatsink capability.
Cheers
"...The thing about reducing the rail voltage is that distortion will rise rapidly; I was pushing it a bit by building the circuit with 15V rails. The problem being that the MOSFETs' Gate capacitances increase below roughly 25V or so. That, in turn, makes the MOSFETs harder to drive, particularly at high frequencies. As a direct consequence, distortion increases.
... It's just a question of being aware of the tradeoffs before you start making changes.
On the other hand, reducing the output bias will also increase distortion a bit. There's no such thing as a free lunch.
Just for the record, the lowest 'optimum' rail voltage for MOSFETs often turns out to be 25V or so. At that point, the Gate capacitance tends to make a dogleg and begins flattening out. Further increases to the rail voltage tend to produce only incremental reductions in Gate capacitance. ..."
Which is why I decided on a higher rail voltage (23V). But Nelson says in the "leaving class A" article higher bias also reduces distortion. Best of both worlds results in more heat to get rid of. I guess at some point there is an optimal balance between bias and rail voltage for a given heatsink capability.
Cheers
an cooling
Fran,
With just the 1" fins and 0.75*C/W (?) it would be better to use the larger diameter fans (5") and run them slower - best results are obtained by pushing the air directly onto the centre of the sinks opposite the fets (increased thermal transfer) although this is more inconvenient and also requires a larger space between fan and fin.
The more common method is to mount the two sinks (L & R) face-to-face and blow the air quietly down the fins, and using a simple temp control switch, can easily maintain 55*C..
Usually, most of the noise from the super quiet computer fans comes from too tight/ridgid mounting methods (use silicon blocks)or not keeping the blades away from the fins (minimum is 1/4 fan dia) and also need to "round off" the heatsink fin edges.
A lot of pro gear does this and the fans and sinks are cleaned periodically when the amps are serviced - it works very well and they often run 24/7, too.
A most surprising thing is that nearly all computers have appalling heat transfer systems resulting in noisy air cooling despite the quiet fans - you have to go "after market" to shut them down - go figure that one!
Many people "shy-away" from fan cooling but it's been used for years with very little problems and the new super quiet fans make noise a thing of the past with just a little effort.
Fran,
With just the 1" fins and 0.75*C/W (?) it would be better to use the larger diameter fans (5") and run them slower - best results are obtained by pushing the air directly onto the centre of the sinks opposite the fets (increased thermal transfer) although this is more inconvenient and also requires a larger space between fan and fin.
The more common method is to mount the two sinks (L & R) face-to-face and blow the air quietly down the fins, and using a simple temp control switch, can easily maintain 55*C..
Usually, most of the noise from the super quiet computer fans comes from too tight/ridgid mounting methods (use silicon blocks)or not keeping the blades away from the fins (minimum is 1/4 fan dia) and also need to "round off" the heatsink fin edges.
A lot of pro gear does this and the fans and sinks are cleaned periodically when the amps are serviced - it works very well and they often run 24/7, too.
A most surprising thing is that nearly all computers have appalling heat transfer systems resulting in noisy air cooling despite the quiet fans - you have to go "after market" to shut them down - go figure that one!
Many people "shy-away" from fan cooling but it's been used for years with very little problems and the new super quiet fans make noise a thing of the past with just a little effort.
as far as using fans. they push air better then they pull. the most effective method is to arange the sinks as was previously described then place the fan on the bottom letting the fan and natural convection lift the heated air out the top. larger diameter fans generaly make less audible noise. i find that enermax enlobal computer fans are very quiet, plus the are magnetically suspended so the bearing lasts seemingly forever. I boght the last ones at Mcm electronics. ifind nothing wrong with using fans either other than the periodic need to clean out rhe accumilated dust. My deciding factor was when the wife saw how big the new amp would be she freaked ,so fans were the only option to get the new amp out of the shop.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.