neychi said:
Neychi,
Very interesting, it may be a pain in the *** to replace components....
But in any case I suggest that this type of discussion could be done in a new or appropriate thread since the Krells will have a regular PCB.
You are a very welcomed contributor but rather new to this forums so let me say that we avoid discussing out of topic matters. Already the ML stuff was rather out of topic but accepted since it came when talking about star grounds and bus bar materials.
We will be glad if you join us building the Krells!!! Sign in on the Krell WIKI to get the PCB’s and enjoy the construction and this nice thread.
I’m not a moderator but deem appropriate to make this comment.
bus bars ...
I prefer using stainless steel for all my hardware nowadays (including telescopes). Set and forget. It has marginally lesser conductivity than brass or copper, but most importantly will STAY the same long after we are all dead. And few miliohms here or there are easily compensated by using larger bars.
One thing I don't recommend is having the recifiers on bus bars close to main capacitors. In class A amp they get HOT. And I mean HOT ! I was surprised how hot they get. Last thing you want is to heat the caps more than they need. I keep the rectifiers on a separate plate (black anodized aluminium) away from capacitors. Even with fairly large area available (and this plate is connected to the chasis via thick aluminium bars) it is still almost as hot as heathsinks.
Also note how signal star ground connection is offset from the power ground. It is much easier to fight ground loops this way than if you bunch up all of the ground connection together.
You can also see 8 gauge OFC wires (rated 60A) I used to feed the power to the output stage. Important if you want to squeeze those last few amps you need for 1KW+ testing on 1 ohm loads
Bratislav
I prefer using stainless steel for all my hardware nowadays (including telescopes). Set and forget. It has marginally lesser conductivity than brass or copper, but most importantly will STAY the same long after we are all dead. And few miliohms here or there are easily compensated by using larger bars.
One thing I don't recommend is having the recifiers on bus bars close to main capacitors. In class A amp they get HOT. And I mean HOT ! I was surprised how hot they get. Last thing you want is to heat the caps more than they need. I keep the rectifiers on a separate plate (black anodized aluminium) away from capacitors. Even with fairly large area available (and this plate is connected to the chasis via thick aluminium bars) it is still almost as hot as heathsinks.
Also note how signal star ground connection is offset from the power ground. It is much easier to fight ground loops this way than if you bunch up all of the ground connection together.
You can also see 8 gauge OFC wires (rated 60A) I used to feed the power to the output stage. Important if you want to squeeze those last few amps you need for 1KW+ testing on 1 ohm loads
Bratislav
Attachments
rabstg said:
... and this is where it gets interesting. It's not always straightforward to connect chassis ground (earth) and two separate signal/power grounds without introducing ground loops.
When you have huge ripple currents pulsing away at big caps' centerpoints ... everything gets complicated that extra bit.
Re: bus bars ...
Nice setup show on the pic.
For me, I may stress, the theoretical concept of a bus bar is one that has equal potential at any point, so installing two point star ground over it is irrelevant or the same as having only one.
If I want equal potential at any point on the bus bar, cross section and material is of prime importance. So copper (or silver if I could afford it) is my first choice always. As second choice, alu or brass, but never any kind of steel which is considered a bad conductor. Of course stainless could be the material of choice when you only consider corrosion and not electrical properties.
My two cents.
Bratislav said:I prefer using stainless steel for all my hardware nowadays (including telescopes). Set and forget. It has marginally lesser conductivity than brass or copper, but most importantly will STAY the same long after we are all dead. And few miliohms here or there are easily compensated by using larger bars.
One thing I don't recommend is having the recifiers on bus bars close to main capacitors. In class A amp they get HOT. And I mean HOT ! I was surprised how hot they get. Last thing you want is to heat the caps more than they need. I keep the rectifiers on a separate plate (black anodized aluminium) away from capacitors. Even with fairly large area available (and this plate is connected to the chasis via thick aluminium bars) it is still almost as hot as heathsinks.
Also note how signal star ground connection is offset from the power ground. It is much easier to fight ground loops this way than if you bunch up all of the ground connection together.
You can also see 8 gauge OFC wires (rated 60A) I used to feed the power to the output stage. Important if you want to squeeze those last few amps you need for 1KW+ testing on 1 ohm loads
Nice setup show on the pic.
For me, I may stress, the theoretical concept of a bus bar is one that has equal potential at any point, so installing two point star ground over it is irrelevant or the same as having only one.
If I want equal potential at any point on the bus bar, cross section and material is of prime importance. So copper (or silver if I could afford it) is my first choice always. As second choice, alu or brass, but never any kind of steel which is considered a bad conductor. Of course stainless could be the material of choice when you only consider corrosion and not electrical properties.
My two cents.
Hi neychi
Mine is a custom made heatsink. It's milled out of a solid piece of aluminium. All walls are 3mm thick and the length is 75mm.
From what I gathered here is that all transistors should be mounted on the same heatsink for better tracking.
This is a question for the expert cloners. (Mark, Al, K-amps where are you.....)
Cheers
Mine is a custom made heatsink. It's milled out of a solid piece of aluminium. All walls are 3mm thick and the length is 75mm.
From what I gathered here is that all transistors should be mounted on the same heatsink for better tracking.
This is a question for the expert cloners. (Mark, Al, K-amps where are you.....)
Cheers
Attachments
thermal tracking...
The only transistor that has to thermally track anything much is q111. It has to be at a temp which is going to change in the same way as the output stage. This can be achieved by bolting it to the same heatsink as the outputs, or by bolting it to the heatsink for the drivers, whichever is most convenient. There is no specific reason why the drivers need to be on any heatsink with anything else....
If the driver heatsink option is used, the heatsink should be chosen to give temps that approx. track the output stage. It would be detrimental to the whole thermal compensation process to put q111 on an overly large driver heatsink that keeps it cool, if the outputs are cooking...
Stuart
The only transistor that has to thermally track anything much is q111. It has to be at a temp which is going to change in the same way as the output stage. This can be achieved by bolting it to the same heatsink as the outputs, or by bolting it to the heatsink for the drivers, whichever is most convenient. There is no specific reason why the drivers need to be on any heatsink with anything else....
If the driver heatsink option is used, the heatsink should be chosen to give temps that approx. track the output stage. It would be detrimental to the whole thermal compensation process to put q111 on an overly large driver heatsink that keeps it cool, if the outputs are cooking...
Stuart
Hi Neychi,
re offer in post 2795; Is this the board that Stuart, Mark & others are building just now?
The Vbe multiplier (q111) must be on the output heatsink.
The drivers can be mounted on separate heatsink/s, but the heatsink/s should be sized so that the driver case temp (Tc) is similar to the output case temp.
You can predict this if you know the driver Iq, +-Vrails, Rth mounting, Rthsink, Tambient (internal).
Here is another suggestion for the Wiki; how to size the driver heatsinks.
re offer in post 2795; Is this the board that Stuart, Mark & others are building just now?
The Vbe multiplier (q111) must be on the output heatsink.
The drivers can be mounted on separate heatsink/s, but the heatsink/s should be sized so that the driver case temp (Tc) is similar to the output case temp.
You can predict this if you know the driver Iq, +-Vrails, Rth mounting, Rthsink, Tambient (internal).
Here is another suggestion for the Wiki; how to size the driver heatsinks.