Ugh I don't know how many of you have seen this, but it seems like somebody has made an upgraded PCB of that blue Chinese one we saw; a MK2 model. His component choices are rather sane, but he used the IRF510/9510 output devices. Don't know why he didn't at least go for 610's, they're better for this application. Layout is very compact but it does sort of remind one of Elliott's S500 subwoofer amplifier 
http://www.diyaudio.com/forums/showthread.php?s=&postid=1025358#post1025358
That means that if he is willing to share the Gerbers and the components are easy enough to find my boards are only good for those who want the classic look
Anyway, here is some feedback on the temperatures of the MK2. Very impressive indeed. At 660mA bias per output device (i.e. 660mV across each resistor; 110W class-A into 8ohms), the main heatsink measures 60degrees, the output transistors about 75degrees, the drivers about 46degrees (on the transistors themselves no less), the predrivers just around 40, and the MOSFET's were basically at ambiant. The drivers are therefore more than just perfectly happy with the heatsink provided.
Funnily enough the main heatsink doesn't get nearly as warm as the Mk1 - I suspect it may be due to the bottom mounted fans, and the fans being more effective than the old top-mounted plastic ones. The larger driver heatsink also clearly paid off.
With the transformers wired for 240V mains (it gets 230V), the DC rails are between 49-50V.
All in all a great piece of apparatus and a definite improvement over Mk1. Mk1 owners, get a decent Papst fan for better performance and lower noise. Bottom mounting will be possible with some ingenuity too. All those clicky relays also provide some peace of mind, and they do a damn good job - when the sparks flew for a while during the voltage changeover they quickly saved the day
I'm going to do a full set of measurements (THD vs freq, THD vs amplitude, frequency response, SNR, DR, 1kHz spectra etc) and repeat them after I've ripped out the boards and replaced the caps with Black Gate NX's.
http://www.diyaudio.com/forums/showthread.php?s=&postid=1025358#post1025358
That means that if he is willing to share the Gerbers and the components are easy enough to find my boards are only good for those who want the classic look
Anyway, here is some feedback on the temperatures of the MK2. Very impressive indeed. At 660mA bias per output device (i.e. 660mV across each resistor; 110W class-A into 8ohms), the main heatsink measures 60degrees, the output transistors about 75degrees, the drivers about 46degrees (on the transistors themselves no less), the predrivers just around 40, and the MOSFET's were basically at ambiant. The drivers are therefore more than just perfectly happy with the heatsink provided.
Funnily enough the main heatsink doesn't get nearly as warm as the Mk1 - I suspect it may be due to the bottom mounted fans, and the fans being more effective than the old top-mounted plastic ones. The larger driver heatsink also clearly paid off.
With the transformers wired for 240V mains (it gets 230V), the DC rails are between 49-50V.
All in all a great piece of apparatus and a definite improvement over Mk1. Mk1 owners, get a decent Papst fan for better performance and lower noise. Bottom mounting will be possible with some ingenuity too. All those clicky relays also provide some peace of mind, and they do a damn good job - when the sparks flew for a while during the voltage changeover they quickly saved the day
I'm going to do a full set of measurements (THD vs freq, THD vs amplitude, frequency response, SNR, DR, 1kHz spectra etc) and repeat them after I've ripped out the boards and replaced the caps with Black Gate NX's.
Unexpected Gremlims
Gentlemen
Last night we wanted to do a quick distortion test on a KSA100 mk1 and hooked it up to the test equipment. At switch on there was a brief crackling sound inside the box then a puff of smoke followed by a burning smell...
Needless to say it was a simple resistor that blew but still a interesting feeling when you see smoke coming out of your amplifier....even worse when you cannot find a screwdriver when you need one!
Jozua
Gentlemen
Last night we wanted to do a quick distortion test on a KSA100 mk1 and hooked it up to the test equipment. At switch on there was a brief crackling sound inside the box then a puff of smoke followed by a burning smell...
Needless to say it was a simple resistor that blew but still a interesting feeling when you see smoke coming out of your amplifier....even worse when you cannot find a screwdriver when you need one!
Jozua
Those boards are only good if you go passive cooling.
+250 watts of dissipation will need big MF heatsinks, even if efficient later Krell heatsink types can be traced.
Putting such a board on a fan duct means quite a long heatsink tunnel, also implies placing it horizontal.
Separate boards are more flexibel, enable a compact construction, and the front end can be kept away from the heatsink if so desired.
Pierre's board design looks more attractive to me, with a combined device type output board it would be perfect.
+250 watts of dissipation will need big MF heatsinks, even if efficient later Krell heatsink types can be traced.
Putting such a board on a fan duct means quite a long heatsink tunnel, also implies placing it horizontal.
Separate boards are more flexibel, enable a compact construction, and the front end can be kept away from the heatsink if so desired.
Pierre's board design looks more attractive to me, with a combined device type output board it would be perfect.
A few comments after having just spent a few hours trying to restore the old MK2's boards:
Krell really should've gone through the trouble to clean up their solder flux. There's so much of the stuff on the bottom that it looks like Maple syrup. Although the frequencies are low, it's not good for crosstalk.
The mystery cap next to the feedback electrolytics is a 20pF, in parallel with the 20k feedback resistor (running from the output to the LTP inverting input).
I hope the cap just reads very funny, since the 62pF's have the letters "620" printed on it. It may perhaps be 62x10^0 = 62, but I'll still measure one up to be on the safe side.
I think I should mention too that I didn't connect the mounting holes to anything for protection sake. On the original these holes were connected to the respective voltage rails for improved juice to the output stage on which the standoffs were mounted, but I decided against it since there's a wire running to it anyway. Those who really want the juice should run an extra wire or use a ring lug to connect the standoffs to the rails again.
Krell really should've gone through the trouble to clean up their solder flux. There's so much of the stuff on the bottom that it looks like Maple syrup. Although the frequencies are low, it's not good for crosstalk.
The mystery cap next to the feedback electrolytics is a 20pF, in parallel with the 20k feedback resistor (running from the output to the LTP inverting input).
I hope the cap just reads very funny, since the 62pF's have the letters "620" printed on it. It may perhaps be 62x10^0 = 62, but I'll still measure one up to be on the safe side.
I think I should mention too that I didn't connect the mounting holes to anything for protection sake. On the original these holes were connected to the respective voltage rails for improved juice to the output stage on which the standoffs were mounted, but I decided against it since there's a wire running to it anyway. Those who really want the juice should run an extra wire or use a ring lug to connect the standoffs to the rails again.
OK, after receiving a few emails about various ideas for the KMA boards it seems like everybody wants support for something else at varying degrees of complexity. Trying to accommodate everybody will lead to a very bulky and inefficient layout.
The board I'm designing for myself is long and narrow, specifically to fit in a KSA100 MK1 in the area between the PSU caps and the heatsinks, that will be upgraded with the Mk2 clone boards. In the centre of the board the soft-start relay and resistor will be, and the regulator sections at the sides. The outermost cap on each side will have a 20VA toroidal with dual secondaries (47 and 42VAC) each. The two respective sides of the board will have two regulators for each channel's LTP and driver pair. The 52V driver will have a two stage regulator (65 to 58 to 52V) and the LTP a three stage one (58 to 52 to 45 to 39). All will be based more or less on the KMA ones, which are standard zener-followers, with a few improvements and SMD components in some of the non-critical areas to save space.
Since I doubt that this board would be very popular with most people here, those wanting KMA support should either use stripboard or somebody should design a more generic version with various degrees of complexity to keep everybody happy.
The board I'm designing for myself is long and narrow, specifically to fit in a KSA100 MK1 in the area between the PSU caps and the heatsinks, that will be upgraded with the Mk2 clone boards. In the centre of the board the soft-start relay and resistor will be, and the regulator sections at the sides. The outermost cap on each side will have a 20VA toroidal with dual secondaries (47 and 42VAC) each. The two respective sides of the board will have two regulators for each channel's LTP and driver pair. The 52V driver will have a two stage regulator (65 to 58 to 52V) and the LTP a three stage one (58 to 52 to 45 to 39). All will be based more or less on the KMA ones, which are standard zener-followers, with a few improvements and SMD components in some of the non-critical areas to save space.
Since I doubt that this board would be very popular with most people here, those wanting KMA support should either use stripboard or somebody should design a more generic version with various degrees of complexity to keep everybody happy.
Dont complicate this project, we want the boards now!
Thats exactly the right thing to do as Mark suggested in post #390: a separate tread in "Power Supply Design" We could for example call it KSA power project or what ever.
PWatts: If you dont have done it, add some jumper pads (or socket pads) for the option to split the LTP power rails, driver stage etc.. for a separate stabilized multi power supply (-board) for those who want it . I dont think this will change the complexity of the board from what it is right now? But dont think of it, if this will demand a re-design of the board.
Right now I think the prototype board should be built for testing and evaluating, or....what next? Keep on going PWatts
Regards
Thats exactly the right thing to do as Mark suggested in post #390: a separate tread in "Power Supply Design" We could for example call it KSA power project or what ever. PWatts: If you dont have done it, add some jumper pads (or socket pads) for the option to split the LTP power rails, driver stage etc.. for a separate stabilized multi power supply (-board) for those who want it . I dont think this will change the complexity of the board from what it is right now? But dont think of it, if this will demand a re-design of the board
.Right now I think the prototype board should be built for testing and evaluating, or....what next? Keep on going PWatts
Regards
Flodstroem, Mark has submitted the prototype boards for manufacture yesterday after I mailed him the final Gerbers. Unless the prototype goes up in smoke or something serious is wrong, my part is (for the time being) is done so I'm not letting the KMA stuff get in the way of the Mk2 boards.
And yes, there is indeed a jumper to separate the LTP and driver power, each with their own happy variety of decoupling caps.
I'll try to take a load of pictures and gather as much information of both the Mk1 and Mk2 I currently have for those who might be inteerested to recreate them as faithfully as possible.
About the KMA stages, it really isn't very complex except that those wanting a simple one will have to use a much larger board than they need to accommodate all the functions.
Besides the normal stuff such as caps, resistors, diodes etc., it may also be good to make it a "controller board" to house a delayed startup for the output stage - a well regulated supply for the LPT and drivers will start up slower than the output stage, and may lead to all kinds of nasty sounds on the speakers until it settles, especially if the regulated rails do not start up simultaneously. The circuit I'm going to build simulates a hefty 30 seconds to reach full voltage although it will be usable quite earlier.
A simple timing circuit on the board will be useful to keep everything together, and a variety of footprints for lots of relay sizes can be used. This is a purer option than to put a relay in series with the speaker outputs. While using relays you can just as well add a second one for the soft-start - the thermister in the MK1 got way too hot. If going for a compact layout all of it can be squeezed in a rather tight area, but since I'm going to use SMD stuff for most due to very constained space I doubt if my design will be helpful.
However, depending on my time I'll try to take spectral measurements with my Audio Precision of the noise levels of the various supplies (unregulated, regulated with different component values and stages etc.). I can tell you now however that the pure DC from the main bus will have a massive 50Hz component and harmonics. On the output of the Mk1 it was larger than the 1kHz input signal's harmonics, and was modulated in healthy doses around the 1kHz fundamental as well. If that's the way it looks on the amplifier output, it will be much worse on the DC rails due to the PSRR of the amp. Seen in that light, even a simple RC network for the drivers and LTP will already do a lot of good.
Not intending to start a massive discussion, but it was interesting to measure that the good old LM317 if properly implemented has a substantially lower noise floor than many of the modern stuff (excluding Linear Technology but their high price tag indicates it). The newer REG1117 from TI is very popular but in fact it performs worse; like 20dB worse. And batteries.. well unless you use a massive capacitor bank and some with a much larger Ah rating than what you figured you would need, they often performed worse than properly mains-filtered and regulated supplies. Under no-load they perform much better, but the moment they are loaded the performance rapidly drops with the mains regulated ones remaining far more stable. Moral of the story: only use batteries if you need ground isolation or if it's for very low current applications; use a lot of them and a large low-ESR capacitor bank too.
BTW the Omron relays used in the original Mk2 is still in production and are damn good with four large contacts wired in parallel and produce a hefty clunk when switched. They are $25 dollars at RS Components, meaning you can probably get them for around $10 at normal vendors.
And yes, there is indeed a jumper to separate the LTP and driver power, each with their own happy variety of decoupling caps.
I'll try to take a load of pictures and gather as much information of both the Mk1 and Mk2 I currently have for those who might be inteerested to recreate them as faithfully as possible.
About the KMA stages, it really isn't very complex except that those wanting a simple one will have to use a much larger board than they need to accommodate all the functions.
Besides the normal stuff such as caps, resistors, diodes etc., it may also be good to make it a "controller board" to house a delayed startup for the output stage - a well regulated supply for the LPT and drivers will start up slower than the output stage, and may lead to all kinds of nasty sounds on the speakers until it settles, especially if the regulated rails do not start up simultaneously. The circuit I'm going to build simulates a hefty 30 seconds to reach full voltage although it will be usable quite earlier.
A simple timing circuit on the board will be useful to keep everything together, and a variety of footprints for lots of relay sizes can be used. This is a purer option than to put a relay in series with the speaker outputs. While using relays you can just as well add a second one for the soft-start - the thermister in the MK1 got way too hot. If going for a compact layout all of it can be squeezed in a rather tight area, but since I'm going to use SMD stuff for most due to very constained space I doubt if my design will be helpful.
However, depending on my time I'll try to take spectral measurements with my Audio Precision of the noise levels of the various supplies (unregulated, regulated with different component values and stages etc.). I can tell you now however that the pure DC from the main bus will have a massive 50Hz component and harmonics. On the output of the Mk1 it was larger than the 1kHz input signal's harmonics, and was modulated in healthy doses around the 1kHz fundamental as well. If that's the way it looks on the amplifier output, it will be much worse on the DC rails due to the PSRR of the amp. Seen in that light, even a simple RC network for the drivers and LTP will already do a lot of good.
Not intending to start a massive discussion, but it was interesting to measure that the good old LM317 if properly implemented has a substantially lower noise floor than many of the modern stuff (excluding Linear Technology but their high price tag indicates it). The newer REG1117 from TI is very popular but in fact it performs worse; like 20dB worse. And batteries.. well unless you use a massive capacitor bank and some with a much larger Ah rating than what you figured you would need, they often performed worse than properly mains-filtered and regulated supplies. Under no-load they perform much better, but the moment they are loaded the performance rapidly drops with the mains regulated ones remaining far more stable. Moral of the story: only use batteries if you need ground isolation or if it's for very low current applications; use a lot of them and a large low-ESR capacitor bank too.
BTW the Omron relays used in the original Mk2 is still in production and are damn good with four large contacts wired in parallel and produce a hefty clunk when switched. They are $25 dollars at RS Components, meaning you can probably get them for around $10 at normal vendors.
. Guess you PWatts is a little bit excited for the boards function? 
KRELL transformer information:
Hi
I have received an answer from Avel Lindberg, former transformer supplier to Krell Electronics.
Avel have confirmed that they can produce the original "Super Quiet" mk-II transformer
Further, Avel informed me that the difference between mk-I and mk-II was that the mk-II was a "Super-Quiet" version of the mk-I transformer. This lead to an increase in heights of the mk-II steel-can. except from that, all electrical data and flexible lead colors are the same.
Because there is two transformers in a stereo-unit, the original transformer is well suited for to build either stereo or mono block versions.
If there is any interests in this Forum for to order the original mk-II transformer I would like to help with this. I know nothing about price-levels for the moment only a request for quotation kan tell. But if there is somebody who wants to get a quotation: take contact with me.
Regards
Hi
I have received an answer from Avel Lindberg, former transformer supplier to Krell Electronics.
Avel have confirmed that they can produce the original "Super Quiet" mk-II transformer
Further, Avel informed me that the difference between mk-I and mk-II was that the mk-II was a "Super-Quiet" version of the mk-I transformer. This lead to an increase in heights of the mk-II steel-can. except from that, all electrical data and flexible lead colors are the same.
Because there is two transformers in a stereo-unit, the original transformer is well suited for to build either stereo or mono block versions.
If there is any interests in this Forum for to order the original mk-II transformer I would like to help with this. I know nothing about price-levels for the moment only a request for quotation kan tell. But if there is somebody who wants to get a quotation: take contact with me.
Regards