Hi all,
I have removed the extra 15pF cap//R15. That appeared to be moving in the wrong direction (maybe R+C here?)
I increased the Cdom caps from 69pF (47//22) to 94pF (47//47).
Nearer success.
5Vpp 10kHz square wave nearly square, slightly rounded corners, the overshoot has gone.
Ripple non existant with 100nF//8r.
Damped ripple just evident with 680nF//8r.
Damped ripple about 2cycles <+-2% with 1uF//8r.
I can attach the 1uF to the spkr terms or to the the audio ground or to the output before the Thiele to audio ground. It seems to be stable in all locations now.
The Cdom are two pair of 47pF NP0 ceramic.
I've tried a few more voltages ranging from 5Vpp to 100Vpp. At this highest test voltage, I can just hear a tiny whistle from the amp and ripple has gone up to about +-5%.
Now that I know that a slightly reactive load will not send the amp into oscillation I can try listening.
Any feedback on this HF measuring procedure?
I have removed the extra 15pF cap//R15. That appeared to be moving in the wrong direction (maybe R+C here?)
I increased the Cdom caps from 69pF (47//22) to 94pF (47//47).
Nearer success.
5Vpp 10kHz square wave nearly square, slightly rounded corners, the overshoot has gone.
Ripple non existant with 100nF//8r.
Damped ripple just evident with 680nF//8r.
Damped ripple about 2cycles <+-2% with 1uF//8r.
I can attach the 1uF to the spkr terms or to the the audio ground or to the output before the Thiele to audio ground. It seems to be stable in all locations now.
The Cdom are two pair of 47pF NP0 ceramic.
I've tried a few more voltages ranging from 5Vpp to 100Vpp. At this highest test voltage, I can just hear a tiny whistle from the amp and ripple has gone up to about +-5%.
Now that I know that a slightly reactive load will not send the amp into oscillation I can try listening.
Any feedback on this HF measuring procedure?
Hi Jozua,
Eventually I will be doing the same thing... going fanless. Will be striping out my Aleph 2's and installing KSA-100 in those chassis. I used the same heat sinks that the original KSA-80 has so even extreme dissipation isn't a problem. I am also going to add 4 more output devices as I have 6 of those heat sinks per monoblock.
What transformer did you end up getting?
Mark
Somewhere in Agoura Hills, CA at DTS
Eventually I will be doing the same thing... going fanless. Will be striping out my Aleph 2's and installing KSA-100 in those chassis. I used the same heat sinks that the original KSA-80 has so even extreme dissipation isn't a problem. I am also going to add 4 more output devices as I have 6 of those heat sinks per monoblock.
What transformer did you end up getting?
Mark
Somewhere in Agoura Hills, CA at DTS
Fanless
Mark
The transformers you saw in the picture are 1 KVA models that I bought for the AlephX project. I simply used them to give an idea of the scale of the heatinks. The transformers for the 100.2 clone still need to be ordered. At this stage my priority is simply to get the heatsinks sorted out so that the PCB and output devices mounting layout can be finalised.
Personally I dont like the idea of monoblocks but the sheer size of the heatsinks (and their weight alone) makes it impractical. I expect the amp to run to run cool as I am using 3 heatsinks blocks per channel. The AlephX runs quite hot when the cooling area of two blocks is used.
I am possibly not moving as fast as AndrewT but I am following his tweaks with interest. I suspect that once the heasink layout is finalised that this project will start to move faster.
Jozua
Mark
The transformers you saw in the picture are 1 KVA models that I bought for the AlephX project. I simply used them to give an idea of the scale of the heatinks. The transformers for the 100.2 clone still need to be ordered. At this stage my priority is simply to get the heatsinks sorted out so that the PCB and output devices mounting layout can be finalised.
Personally I dont like the idea of monoblocks but the sheer size of the heatsinks (and their weight alone) makes it impractical. I expect the amp to run to run cool as I am using 3 heatsinks blocks per channel. The AlephX runs quite hot when the cooling area of two blocks is used.
I am possibly not moving as fast as AndrewT but I am following his tweaks with interest. I suspect that once the heasink layout is finalised that this project will start to move faster.
Jozua
Hi Jozua,
The sinks in the pic look good for a passive monoblock.
They are quite tall and I see 6 main spines running vertically.
This would be ideal for a 6pair output stage.
A row of 6 upper devices above a row of 6 lower devices and the emitter resistors tying the upper to the lower with a big bussbar for the output line. The Bussbar could be solid core 1.8mm diam or bigger if you feel necessary, or multicore, or extremely fine multi hundred core.
Remember if you go for two rows spaced vertically, that to get the two rows to run at the same temperature, the vertical positions will not be symmetrical nor near the middle. The bottom row will be about 25% up from the bottom and the upper row will be about 60% up from the bottom. Try to find Wakefield's paper on this, because my recollection is a bit hazy.
What's wrong with Monoblock? I find they are easier to get hum free and lighter and simpler to assemble. But your 12 sinks could be placed on both sides to create a giant KSA50 style (the passive one).
The sinks in the pic look good for a passive monoblock.
They are quite tall and I see 6 main spines running vertically.
This would be ideal for a 6pair output stage.
A row of 6 upper devices above a row of 6 lower devices and the emitter resistors tying the upper to the lower with a big bussbar for the output line. The Bussbar could be solid core 1.8mm diam or bigger if you feel necessary, or multicore, or extremely fine multi hundred core.
Remember if you go for two rows spaced vertically, that to get the two rows to run at the same temperature, the vertical positions will not be symmetrical nor near the middle. The bottom row will be about 25% up from the bottom and the upper row will be about 60% up from the bottom. Try to find Wakefield's paper on this, because my recollection is a bit hazy.
What's wrong with Monoblock? I find they are easier to get hum free and lighter and simpler to assemble. But your 12 sinks could be placed on both sides to create a giant KSA50 style (the passive one).
Heatsinks saga
AndrewT
Your idea of a busbar sounds nice and easy but Pierre and I need to discuss the exact layout. If need be only two heatinks blocks might be used with a lower class A rating and then using the 3rd block for the Pinkmouse 50 clone.
If possible I want to follow the KSA50 concept of keeping the PCB wires to the out put devices as short as possible. Based on my hookup experiences with the 100.2 I now suspect that the very short hookup wires of the KSA50 is a major factor in making it sound so good.
Jozua
AndrewT
Your idea of a busbar sounds nice and easy but Pierre and I need to discuss the exact layout. If need be only two heatinks blocks might be used with a lower class A rating and then using the 3rd block for the Pinkmouse 50 clone.
If possible I want to follow the KSA50 concept of keeping the PCB wires to the out put devices as short as possible. Based on my hookup experiences with the 100.2 I now suspect that the very short hookup wires of the KSA50 is a major factor in making it sound so good.
Jozua
Hookup setup
AndrewT
I am the last person to give a technical opinion. At this stage I am simply working on my subjective gut feeling. Pierre is most likely better qualified to comment. Strictly speaking lab measurements are required. That is why I need to get my heatsinks sorted out so that one can experiment and measure.
At this stage I am very keen to read about your initial sound impressions.
Jozua
AndrewT
I am the last person to give a technical opinion. At this stage I am simply working on my subjective gut feeling. Pierre is most likely better qualified to comment. Strictly speaking lab measurements are required. That is why I need to get my heatsinks sorted out so that one can experiment and measure.
At this stage I am very keen to read about your initial sound impressions.
Jozua
Re: Hookup setup
Have been listening to one channel for three hours this morning. Pretty mediocre. no sparkle, no get up and go, there's treble and mid and bass, but it does not impress. very non descript. I hope that is not what high accuracy sounds like.
I suspect that Cdom @ 94pF is killing the sound. But do keep in mind that Cdom is still NP0 ceramic.
listened for an hour last night before I went out.Jozua said:
Have been listening to one channel for three hours this morning. Pretty mediocre. no sparkle, no get up and go, there's treble and mid and bass, but it does not impress. very non descript. I hope that is not what high accuracy sounds like.
I suspect that Cdom @ 94pF is killing the sound. But do keep in mind that Cdom is still NP0 ceramic.
I toured the QSC facility yesterday morning. They have an absolutely amazing production line... a good part of it is automated. One thing that interested me was the spring steel clip they used to fasten the TO-247 output devices to the heatsinks. Pretty nifty... they say it aloows the devices to breathe....
Mark
Mark
They won't allow photos on the tours. I think there is a video tour on their web site though.The amazing thing on the line was the Panasonic hi speed parts insertion machine. Stuffs an entire board in one minute! Makes a heck of a racket too. I was allowed to take photos at Christie Digital the day before however...
Mark
Mark
Speaking of spring steel clips, I was looking for off-the-shelf product like this. I've seen heat sinks that are made to accommodate clips. But what I was looking for is a bar with a line of clips that could be screwed on to a flat surface of a heat sink. This way, any holes tapped into the heat sink would be well away from the transistor heat plate, which would avoid any issues with burrs from tapping holes for transistor mounting screws, or a lifting of the flat heat sink if the transistor mounting screws were over torqued.
I realize that clips may not give the PSI as screw through the transistor, but it could probably be adequate if designed correctly.
Does anyone know of any available clip product like this?
I realize that clips may not give the PSI as screw through the transistor, but it could probably be adequate if designed correctly.
Does anyone know of any available clip product like this?
Hey guys,
just want to confirm.
1. Are R38,R39,R40 and R41 (75 ohms) 0.25 watt as per the 12/2006 BOM? Some photos show them as 2 or 3 watt.
2. What post has the latest silk screen? My gold boards do not have the components printed on them.
3. Is there a circuit showing the optional caps and jumpers?
Thanks
just want to confirm.
1. Are R38,R39,R40 and R41 (75 ohms) 0.25 watt as per the 12/2006 BOM? Some photos show them as 2 or 3 watt.
2. What post has the latest silk screen? My gold boards do not have the components printed on them.
3. Is there a circuit showing the optional caps and jumpers?
Thanks
a1.)the driver emitter resistors carry about 18mA. That amounts to less than 25mW. Do the calcs.Harry3 said:
However, have you changed anything down stream. The 75r value is worked back from the voltage across the outputs+their emitter resistors. I set output Re=0r35 and used 6pair of 230W plastic packages. The driver Re becomes 51r to ensure adequate driver current.
a3.)no, I have manually added the jumpers and caps to the standard schematic. And moved one diode pair connection to the correct location.
Hi,
6pairs of 200W To264 devices just about matches the capability of 4pair of 250W To3 devices. Good choice if you go with plastic packaged.
Now go and calculate the voltage drop across the emitter resistors in the two schemes (4pair and 6pair). The 6pair will run with LESS voltage across the output pair bases. This will put LESS voltage across the driver emitters. The drivers should be set up to pass the same bias current as Krell chose, or YOU must have a good reason for changing the driver bias current.
If anyone changes the down stream side after the drivers, the driver emitter resistors must be recalculated if you want to mimic the Krell set up.
