The even distance distribution rule is not exactly true.
Some 15 years ago an audio mag in the German reich made thermal pictures of a number of power amplifiers.
One of them was the Threshold SA4 model, with different pictures during warmup. The heatsinks of the SA4 showed a rather un-even thermal distribution at the steady-state situation.
Now, if you consider the length of a heatsink as a solid bar and place an output device as a centered heatsource along the bar it is easy to estimate the temperature rise at different points along the bar by setting up a differential equation. By placing the output device three times at a different position and calculating the temperature distribution the total distribution can be estimated by superimposing the three calculations.
The optimal outcome differs from cutting a heatsink in 3 even parts and placing each output device in the middle of 1 heatsink.
At the time when desktop calculation power became available i made a finite elements model of 2 heatsinks: 8"x4" and 12"x6"
A different number of output devices were placed as node heat sources in the model and the computer calculated the temperature distribution and from that the efficiency of the heatsink. By shifting the position of the output devices the total efficiency of the heatsink was raised or lowered.
Nowadays it is common practice to do thermal calculations with finite elements software, posting this seems ancient history.
For small heatsinks it is practical to "cut" it in even parts and place the output devices in the middle, the width of an output device will leave only a small edge of the 17%, accurate it is not and certainly not general.
Some 15 years ago an audio mag in the German reich made thermal pictures of a number of power amplifiers.
One of them was the Threshold SA4 model, with different pictures during warmup. The heatsinks of the SA4 showed a rather un-even thermal distribution at the steady-state situation.
Now, if you consider the length of a heatsink as a solid bar and place an output device as a centered heatsource along the bar it is easy to estimate the temperature rise at different points along the bar by setting up a differential equation. By placing the output device three times at a different position and calculating the temperature distribution the total distribution can be estimated by superimposing the three calculations.
The optimal outcome differs from cutting a heatsink in 3 even parts and placing each output device in the middle of 1 heatsink.
At the time when desktop calculation power became available i made a finite elements model of 2 heatsinks: 8"x4" and 12"x6"
A different number of output devices were placed as node heat sources in the model and the computer calculated the temperature distribution and from that the efficiency of the heatsink. By shifting the position of the output devices the total efficiency of the heatsink was raised or lowered.
Nowadays it is common practice to do thermal calculations with finite elements software, posting this seems ancient history.
For small heatsinks it is practical to "cut" it in even parts and place the output devices in the middle, the width of an output device will leave only a small edge of the 17%, accurate it is not and certainly not general.
sorry to be such slowpoke
Hi,
No big update here just yet...my house is slowly returning to normal, I have almost all the boxes of electronics stuff down from the attic, but only about half unpacked, and my bench is not yet functional...
I have located the boxes of krell stuff, so I think it will be possible to 'refresh' my memory soon.
Getting married is a big deal when you do it at the house, especially on New Years Eve...I hope to have normality restored before the first anniversary...
Stuart
Hi,
No big update here just yet...my house is slowly returning to normal, I have almost all the boxes of electronics stuff down from the attic, but only about half unpacked, and my bench is not yet functional...
I have located the boxes of krell stuff, so I think it will be possible to 'refresh' my memory soon.
Getting married is a big deal when you do it at the house, especially on New Years Eve...I hope to have normality restored before the first anniversary...
Stuart
emitter resistors...
I have a different opinion about emitter resistors...
If the number of output devices is changed by a lot, ie doubled, I'd increase the emitter resistors by the same proportion, perhaps a little less, but increase them... This will keep the bias voltage across the resistors more constant. The damping factor will be largely unaffected, the NFB and paralleled resistors will tend to make the increased value unimportant, plus the output transistor current sharing is enhanced and the outputs are better protected in the event of catastrophe...
HTH
Stuart
I have a different opinion about emitter resistors...
If the number of output devices is changed by a lot, ie doubled, I'd increase the emitter resistors by the same proportion, perhaps a little less, but increase them... This will keep the bias voltage across the resistors more constant. The damping factor will be largely unaffected, the NFB and paralleled resistors will tend to make the increased value unimportant, plus the output transistor current sharing is enhanced and the outputs are better protected in the event of catastrophe...
HTH
Stuart
HI
I would like to ask for help from those who purchase transformer from the Audiohobby guy .
I would order one to the Krell clone ,but I think is only avaleble in US.
At least when I tried to order it ask me for the (5 digit) US postal code for the shipping cost .
I wrote a couple email to the guy about the the tranie but no answer .
If some one has his phone number or know how I contact with him , or how I order one from Canada please let me know . There is no answer to my email.
I need a toroid very badly .
I even able to pay more for it in case jut to finnish my Krell clone finaly.
Cheers
I would like to ask for help from those who purchase transformer from the Audiohobby guy .
I would order one to the Krell clone ,but I think is only avaleble in US.
At least when I tried to order it ask me for the (5 digit) US postal code for the shipping cost .
I wrote a couple email to the guy about the the tranie but no answer .
If some one has his phone number or know how I contact with him , or how I order one from Canada please let me know . There is no answer to my email.
I need a toroid very badly .
I even able to pay more for it in case jut to finnish my Krell clone finaly.
Cheers
gaborbela said:
Want a 1kVA Plitron? 33-0-33 dual secs? I got one but will use the 2kVA's instead. PLease note: With 33v secs, you get rails of about 43vdc loaded which is good for a 70 watt amp.... email me if intersted.
HI
Thanks for your offer , I realy want that transformer from the Audiohobby guy.
I live in Toronto , usualy I order tranie from Plitron but their quallity is so so . I'm not to satisfied with , I already purchased fom Plitron 6 toroids for diferent DIY projects ..
Your transformer is 33-0-33v ,I dont know if I have to modifie the circuit for the higher voltage .
Please let me know how much you want for it , in case....
You know when I order from Plitron I go and pick it up so I can save the shipping ,I dont know if worth to buy yours from US.
I want that tranie from the Audiohobby because I know the quallity of that ..
Cheers
Thanks for your offer , I realy want that transformer from the Audiohobby guy.
I live in Toronto , usualy I order tranie from Plitron but their quallity is so so . I'm not to satisfied with , I already purchased fom Plitron 6 toroids for diferent DIY projects ..
Your transformer is 33-0-33v ,I dont know if I have to modifie the circuit for the higher voltage .
Please let me know how much you want for it , in case....
You know when I order from Plitron I go and pick it up so I can save the shipping ,I dont know if worth to buy yours from US.
I want that tranie from the Audiohobby because I know the quallity of that ..
Cheers
HI
I didn't said Plitron transformer is so so because is yours , I said because I had bad experience with it .
I use Plitron to because it is here near by the commpany .
But to comper with that wich I'm intrested there is a quallity diference beetwen them and I'm not sure if I could use yours with out changing the resistors ect. on the pc board.
Sorry about if you feel uppset , I realy didn't want to do that . Instead I'm very greatfule you guys to gave these opurtunity so we can have such a good amplifier for that price .
I just can say thanks .
Regards
I didn't said Plitron transformer is so so because is yours , I said because I had bad experience with it .
I use Plitron to because it is here near by the commpany .
But to comper with that wich I'm intrested there is a quallity diference beetwen them and I'm not sure if I could use yours with out changing the resistors ect. on the pc board.
Sorry about if you feel uppset , I realy didn't want to do that . Instead I'm very greatfule you guys to gave these opurtunity so we can have such a good amplifier for that price .
I just can say thanks .
Regards
🙂 I am not upset at all.... I was only trying to help because you seemed a little desperate and wanted something fast. Don't worry abot me, I will find some use for it or I can always sell it for $300 on ebay 😉
As far as circuit mods, no you wont need any changes except that you'd need larger heatsinks to dissipate the extra heat. Just look at Mark, he went from 200mV bias to 400mV bias and his system had no issues.
Purely as far as voltages go, I'd reckon you could go up to +/-55vdc to 60 vdc before you'd need to change values in the PCB. namely the input diff and probably the excess heat in the 27v zener regulated portion.
Anyway good luck!
As far as circuit mods, no you wont need any changes except that you'd need larger heatsinks to dissipate the extra heat. Just look at Mark, he went from 200mV bias to 400mV bias and his system had no issues.
Purely as far as voltages go, I'd reckon you could go up to +/-55vdc to 60 vdc before you'd need to change values in the PCB. namely the input diff and probably the excess heat in the 27v zener regulated portion.
Anyway good luck!
Mark A. Gulbrandsen said:
Hi Mark,
When you say better, how do you mean? I used a single 30-0-30 500vA in the amp with Jan's boards and two 28-0-28 400vA trannies in the amp with Al's boards and I can't hear any difference between them. What is it that you notice as an improvement?
Thanks, Terry
Ok.. Since no one answered I just pulled the trigger and bought some of these cases.. They will be easy to just drop in my parts once ALL the GB kits are gone.
NO I HAVE NOT stuffed my boards... I am not that rude.. I will get the last 2 out this week and then start mine.
http://www.prestostore.com/cgi-bin/store.pl?ref=mrothacher&ct=58800
NO I HAVE NOT stuffed my boards... I am not that rude.. I will get the last 2 out this week and then start mine.
http://www.prestostore.com/cgi-bin/store.pl?ref=mrothacher&ct=58800
rabstg said:
Troy, Those cases would work well for a Gainclone, but for the KSA they might just be a tad aneamic.... heatsink wise. However if you went Monoblock, they could work....
Hi Googler,
the 0.45inch referred to the root thickness of the 5inch high fins not the baseplate.
If the base plate is 0.3inch thick then 10 times thickness =3inch.
This tells you that the distance to any edge from a transistor centre should be less than 3inch to allow the sink to achieve reasonable efficiency similar to the manufacturer's spec sheet.
For your cut down sink the maximum dimension is diagonally = 7.8inch. Therefore for a single device the corners of the sink will run cooler because the 7.8/2=3.9inch exceeds 10*thickness.
Put on two devices and the area of sink that lies outside your 3inch radius becomes very small so efficiency is higher than using a single device. Similarly increasing the spread of the devices improves the overall efficiency because more of the sink is running at a temperature close to the Tc of the devices. It's all quite logical.
If and when Jacco gives you some real advice on how to layout your devices for good/acceptable heatsink efficiency then compare the methods/strategy and go for either.
the 0.45inch referred to the root thickness of the 5inch high fins not the baseplate.
If the base plate is 0.3inch thick then 10 times thickness =3inch.
This tells you that the distance to any edge from a transistor centre should be less than 3inch to allow the sink to achieve reasonable efficiency similar to the manufacturer's spec sheet.
For your cut down sink the maximum dimension is diagonally = 7.8inch. Therefore for a single device the corners of the sink will run cooler because the 7.8/2=3.9inch exceeds 10*thickness.
Put on two devices and the area of sink that lies outside your 3inch radius becomes very small so efficiency is higher than using a single device. Similarly increasing the spread of the devices improves the overall efficiency because more of the sink is running at a temperature close to the Tc of the devices. It's all quite logical.
If and when Jacco gives you some real advice on how to layout your devices for good/acceptable heatsink efficiency then compare the methods/strategy and go for either.
Speading the devices like you posted is the most practical.
Shifting devices more to the outer edges only gives an improvement on very large heatsinks. My post was not an advice, merely thought it was interesting to mention.
Shifting devices more to the outer edges only gives an improvement on very large heatsinks. My post was not an advice, merely thought it was interesting to mention.
K-amps said:
Aaaah man, certainly NOT what I wanted to hear after buying 4 of them.
Well I guess I am building mono blocks again now. That's what I really wanted but didn't want the mess of building 2 enclosures..
I have much larger heat sinks I could change these out with... But if I do go mono blocks I will have alot of very pretty(in my eyes) heat sinks for sale...
I know man, they looked real good....
with 3 pairs per channel and 38v rails, you are dissipating over 135 watts. Those are rated at 100w for 55c... so you are's kinda close, but if you want to get the benefits that Mark got by over biasing, then definitely go Monoblocks.
with 3 pairs per channel and 38v rails, you are dissipating over 135 watts. Those are rated at 100w for 55c... so you are's kinda close, but if you want to get the benefits that Mark got by over biasing, then definitely go Monoblocks.
Andrew and Jacco - Thank you both... I think I have a handle on the transistor placement now. I am thinking 6 OP pairs in a pyramid arrangment 3-2-1.
I measured the base plate last night and it was .5inch - making the fins 4.5inches. I didn't measure the fin thickness, but I suspect they are ~.15inch. The thickness of the fins are equal at base and tip. Given your previous math on root thickness, the fin shape is not the most efficient.
The fin ridges are sinusoidal with each ridge having another smaller and higher frequency sine wave on it, so once I measure them I should get a good idea of surface area.
Thanks again!
~Brad
I measured the base plate last night and it was .5inch - making the fins 4.5inches. I didn't measure the fin thickness, but I suspect they are ~.15inch. The thickness of the fins are equal at base and tip. Given your previous math on root thickness, the fin shape is not the most efficient.
The fin ridges are sinusoidal with each ridge having another smaller and higher frequency sine wave on it, so once I measure them I should get a good idea of surface area.
Thanks again!
~Brad
Hi,
a side effect of those very high fins is that they are cooler to the touch at the edges.
Can you avoid touching the hotter parts by placing a cover over them but air spaced to allow circulation?
Remember to keep the bottom edge up off the floor to allow cold air in at the hottest part.
a side effect of those very high fins is that they are cooler to the touch at the edges.
Can you avoid touching the hotter parts by placing a cover over them but air spaced to allow circulation?
Remember to keep the bottom edge up off the floor to allow cold air in at the hottest part.
pcb
Hi guys,
does anybody paln to produce pcb and sell it? Where can I find a list with suggested components?
I have two toroids 35-0-35V with 650VA each, do you think they can do the job?
KEPha
Hi guys,
does anybody paln to produce pcb and sell it? Where can I find a list with suggested components?
I have two toroids 35-0-35V with 650VA each, do you think they can do the job?
KEPha
