Didn't check on printer yet. The printer is at the office, I'm on the road. I was planning to check the TO-220AB transistor case dimensions according to the pcb printout.
I used the exact pcb image from this thread. I just put them over each other using PaintShop.
They should be of the same dimensions as the original unless some image resolution trick mixed things up.
I used the exact pcb image from this thread. I just put them over each other using PaintShop.
They should be of the same dimensions as the original unless some image resolution trick mixed things up.
Yeah they did 
I printed at 1200 dpi on the works laser printer and it's an a4 page in size!! So i copied into word, used the original schematic diagram (printed out of adobe) and that was 10.3cm by 7.1cm and works out that most of the parts are pretty much spot on, so resized the image in word and done!
The layouts were excellent though, thanks for them!
Aaron
I printed at 1200 dpi on the works laser printer and it's an a4 page in size!! So i copied into word, used the original schematic diagram (printed out of adobe) and that was 10.3cm by 7.1cm and works out that most of the parts are pretty much spot on, so resized the image in word
and done!The layouts were excellent though, thanks for them!
Aaron
Fuses?
With 2KW of output devices on a 50W stereo amp a circuit breaker in the primary is all that is needed.
BGW also did this on their big amps. The 365W stereo unit had 5KW worth of outputs and only one switch/circuit breaker on the front panel.
Class A amplifiers are much less sensitive to power supply bypass caps as the current drawn is much more on a constant level. Look at the designs from Pass as well.
With 2KW of output devices on a 50W stereo amp a circuit breaker in the primary is all that is needed.
BGW also did this on their big amps. The 365W stereo unit had 5KW worth of outputs and only one switch/circuit breaker on the front panel.
Class A amplifiers are much less sensitive to power supply bypass caps as the current drawn is much more on a constant level. Look at the designs from Pass as well.
"BGW also did this on their big amps. The 365W stereo unit had 5KW worth of outputs and only one switch/circuit breaker on the front panel."
BGW also had an ingenious SCR Crow Bar circuit that in effect shorted the main supply out and triped the breaker if there was any DC, or hf oscillation at the output. I always much admired Brian Wachners designs. They were pretty good amps back in those days.
Mark
BGW also had an ingenious SCR Crow Bar circuit that in effect shorted the main supply out and triped the breaker if there was any DC, or hf oscillation at the output. I always much admired Brian Wachners designs. They were pretty good amps back in those days.
Mark
The last version of the schematic, ver 1.4 has some parts at the output of the amp not included on the output PCB:
D201, D202: Are 1N4007 OK? Soldered directly to the ouput PCB?
R205: is 1/4W ok?
C201: What type, MKP, Mica?
R206 and L201: They are usually a resistor surrounded by wire. Is 1W, carbon type resistor ok? What wire gauge and how many turn, 5 turns maybe?
Will this snubber R206,L201, reduce the damping factor too much?
D201, D202: Are 1N4007 OK? Soldered directly to the ouput PCB?
R205: is 1/4W ok?
C201: What type, MKP, Mica?
R206 and L201: They are usually a resistor surrounded by wire. Is 1W, carbon type resistor ok? What wire gauge and how many turn, 5 turns maybe?
Will this snubber R206,L201, reduce the damping factor too much?
R124 and biasing the beast...
R124 should definitely be about 4.7k, it allows the VBe mult to adjust from ~1.1 to ~4.3v, enough for the original KSA50's needs, and probably for ours...
...those of us making a KSA50 clone need ~2.2amps total bias, ie 1.1a of bias through each output pair, we need ~0.65v for the driver, ~0.65v for the output and 0.55v to push 1.1 amps through Re, so normally the bias voltage will be ~3.7v...2*(~0.65+~0.65+0.55). Decreasing r125 from it's nominal 800ohms would allow for higher levels of bias voltage, but flames are a possible side effect.
If you plan on more bias, like NUTTTR, for class A into 4 ohms, you can just add more output pairs with their resistors, each pair will simply add another 1.1a to the output stage current, and the amp can drive more current before it leaves class A...3.3A gives class A to 87w into 4ohms, 4.4A gets us to 150w...for a 4ohm target, you can't do better, the original amp doesn't have the voltage swing to push much more current (8.8A peak) into 4ohms...unless you have added more volts that is...
The driver and output stage form a darlington with a gain that could, under 'bad' conditions, drop to perhaps 500 (50x10), so if we have 100ma available from the VBE/VA (and I think we do) our biasing assumptions should be OK upto ~50amps peaks in the load...More outputs will diminish Ic beta droop in the outputs, but may encourage it in the drivers, so this is a balancing act...I think doubling the number of outputs is safe, possibly tripling. More than that could make the situation worse. the outputs multiply the beta droop of the driver...but this 'problem' should only occur driving really low impedance loads, the < 1 ohm type...
Fixing this by increasing the current capability of the VBE, adding another driver set etc is what krell seems to have done later...perhaps this additional complexity explains why some people think their amps started sounding 'worse'...
Oh, yeah, the driver transistors, they are going to have Ic at ~50ma, plus ~30ma per output transistor, so you can figure their dissipation, in the clone case, as the voltage rail, 35v * (0.05+0.03+0.03) watts, or 3.85 watts, so lets say 4watts, so if 40 degrees above ambient is OK, you need 40/4 or 10c/watt for each driver transistor. I'd allow for some margin, so 8c/watt sounds fine...somebody else should do this reasoning and math to make sure I'm not out in left field...
Hope that helps
Stuart
R124 should definitely be about 4.7k, it allows the VBe mult to adjust from ~1.1 to ~4.3v, enough for the original KSA50's needs, and probably for ours...
...those of us making a KSA50 clone need ~2.2amps total bias, ie 1.1a of bias through each output pair, we need ~0.65v for the driver, ~0.65v for the output and 0.55v to push 1.1 amps through Re, so normally the bias voltage will be ~3.7v...2*(~0.65+~0.65+0.55). Decreasing r125 from it's nominal 800ohms would allow for higher levels of bias voltage, but flames are a possible side effect.
If you plan on more bias, like NUTTTR, for class A into 4 ohms, you can just add more output pairs with their resistors, each pair will simply add another 1.1a to the output stage current, and the amp can drive more current before it leaves class A...3.3A gives class A to 87w into 4ohms, 4.4A gets us to 150w...for a 4ohm target, you can't do better, the original amp doesn't have the voltage swing to push much more current (8.8A peak) into 4ohms...unless you have added more volts that is...
The driver and output stage form a darlington with a gain that could, under 'bad' conditions, drop to perhaps 500 (50x10), so if we have 100ma available from the VBE/VA (and I think we do) our biasing assumptions should be OK upto ~50amps peaks in the load...More outputs will diminish Ic beta droop in the outputs, but may encourage it in the drivers, so this is a balancing act...I think doubling the number of outputs is safe, possibly tripling. More than that could make the situation worse. the outputs multiply the beta droop of the driver...but this 'problem' should only occur driving really low impedance loads, the < 1 ohm type...
Fixing this by increasing the current capability of the VBE, adding another driver set etc is what krell seems to have done later...perhaps this additional complexity explains why some people think their amps started sounding 'worse'...
Oh, yeah, the driver transistors, they are going to have Ic at ~50ma, plus ~30ma per output transistor, so you can figure their dissipation, in the clone case, as the voltage rail, 35v * (0.05+0.03+0.03) watts, or 3.85 watts, so lets say 4watts, so if 40 degrees above ambient is OK, you need 40/4 or 10c/watt for each driver transistor. I'd allow for some margin, so 8c/watt sounds fine...somebody else should do this reasoning and math to make sure I'm not out in left field...
Hope that helps
Stuart
Resistors values questions:
RB resistor. I found good quality 2.4R, 1% instead of the 2.2R specified. Will it work with 2.4R? It is really close, so I guest it will work.
R122,R123 are 101R, can I use 100R,1% instead?
R102 is 22K2, as you probably know standard 1% resistor value is 22K1, will it do?
Same for R130,R131 specified at 22K. Can I use 22K1,1%?
Thanks in advance.
RB resistor. I found good quality 2.4R, 1% instead of the 2.2R specified. Will it work with 2.4R? It is really close, so I guest it will work.
R122,R123 are 101R, can I use 100R,1% instead?
R102 is 22K2, as you probably know standard 1% resistor value is 22K1, will it do?
Same for R130,R131 specified at 22K. Can I use 22K1,1%?
Thanks in advance.
sounds fine to me...
The resistors you mention are all fine, there really aren't any hyper-critical tolerances here, the pots take care of the final adjustment, so any minor imbalance is removed as part of the final setup...
One thing to bear in mind, using more output transistor pairs makes each transistor use a smaller range of it's forward transfer curve, this makes the inherent non-linearities smaller. To paraphrase John Linsley Hood, a small enough segment of any curve approaches a straight line...Given that more outputs uses more of the forward transfer curve of the drivers too many is going to make the drivers more non-linear...What all this verbiage is getting to is that there are probably an optimum number of outputs for any given load, and driver output combination. I assume Krell found this to be 2 pairs of mj15003/4 for 8ohms, perhaps its 3 pairs for 4ohms, assuming you can measure the output for THD or IMD you should be able to choose the correct number by trial and error...if not, you'll just have to listen.
Can't find my digital camera or I'd take some pictures of piles of heatsinks, capacitors and toroids...
Stuart
The resistors you mention are all fine, there really aren't any hyper-critical tolerances here, the pots take care of the final adjustment, so any minor imbalance is removed as part of the final setup...
One thing to bear in mind, using more output transistor pairs makes each transistor use a smaller range of it's forward transfer curve, this makes the inherent non-linearities smaller. To paraphrase John Linsley Hood, a small enough segment of any curve approaches a straight line...Given that more outputs uses more of the forward transfer curve of the drivers too many is going to make the drivers more non-linear...What all this verbiage is getting to is that there are probably an optimum number of outputs for any given load, and driver output combination. I assume Krell found this to be 2 pairs of mj15003/4 for 8ohms, perhaps its 3 pairs for 4ohms, assuming you can measure the output for THD or IMD you should be able to choose the correct number by trial and error...if not, you'll just have to listen.
Can't find my digital camera or I'd take some pictures of piles of heatsinks, capacitors and toroids...
Stuart
