Hi Geezer,
that important correction is not an improvement. It is an attempt to correct an error in the PCB track layout. As far as we know the same error has been avoided in the pinkmouse PCB. One advantage the earlier PCB had was the two stage low bias/high bias (summer/winter bias) setting. Something to consider if you draw up your own.
Jacco,
If it makes your answer any easier then Yes you can assume that the gain is the same for each output transistor and only the Vbe is different. Now having answered the Variable Re question go back and explain what would be done if gain and Vbe were different.
that important correction is not an improvement. It is an attempt to correct an error in the PCB track layout. As far as we know the same error has been avoided in the pinkmouse PCB. One advantage the earlier PCB had was the two stage low bias/high bias (summer/winter bias) setting. Something to consider if you draw up your own.
Jacco,
If it makes your answer any easier then Yes you can assume that the gain is the same for each output transistor and only the Vbe is different. Now having answered the Variable Re question go back and explain what would be done if gain and Vbe were different.
AndrewT said:
Actually, Jan's boards do have protection circuitry that was absent from the original (hmm.... the 300 series components and Rb and Rv), and I think the Geeze doesn't want to sully his circuit with the extra traces (though realize that they can be skipped/unused, and most everyone has skipped putting these extra components into their project) in his quest for the real macarana. Bravo! Living on the edge.
So I guess you email pinkmouse and ask for a pdf. But if I were you and I knew how to make a PCB in the shed I would use Jan's and simply "white out" the extra traces and edit the black traces for the resistors (to fix the small errors). Then you are good to go as Jan's is the original circuit if you leave out these extra items (well I'm pretty sure it is).
Actually the Pinkmouse board conforms more to the original KSA-50 MK-2 than Jan's board does. The Pinkmouse board is also easier to deal with and properly heatsink the driver devices than is Jan's. The only thing about the Pinkmouse board that differs at all is the pinout for the use of the MPS series small signal devices instead of the "Jap" devices. The "Jap" devices can however easily be used as well just by bending the leads properly. Jan's board is definately the easier one to make if you are doing the boards your self!!
Mark
Howdy all,
My goodness, such a lot of good advice to do this different than what I have in mind. Was thinking to replicate this little amp as much like the original as is possible for an ole guy to recapture his younger days. The easy way would be to just hunt around for a good used one and be done with it. So easy ain't exactly what I had in mind, but to actually make it with my own hands.
Seems there's lots of really smart and clever folks out there, but can't find one who can figure out if a simple pdf of the superior pinkmouse board exists. Have not found such a thing after all the reading. Maybe Mr. P just don't have it or this is just way too much trouble. Must be other folks who missed out on the group buy or prefer to roll their own that could make use of this little request.
But I do thank you all kindly for the generous help that's been offered.
A public appeal to Mr. pinkmouse:
If publishing a pdf of your pretty board is possible, would it be way too much effort to share it with these nice folks here?
Many thanks and best regards,
Elvin
My goodness, such a lot of good advice to do this different than what I have in mind. Was thinking to replicate this little amp as much like the original as is possible for an ole guy to recapture his younger days. The easy way would be to just hunt around for a good used one and be done with it. So easy ain't exactly what I had in mind, but to actually make it with my own hands.
Seems there's lots of really smart and clever folks out there, but can't find one who can figure out if a simple pdf of the superior pinkmouse board exists. Have not found such a thing after all the reading. Maybe Mr. P just don't have it or this is just way too much trouble. Must be other folks who missed out on the group buy or prefer to roll their own that could make use of this little request.
But I do thank you all kindly for the generous help that's been offered.
A public appeal to Mr. pinkmouse:
If publishing a pdf of your pretty board is possible, would it be way too much effort to share it with these nice folks here?
Many thanks and best regards,
Elvin
Your tale has touched my heart! 
There is no reason why you can't have my layout, In fact, I thought it was already here on the thread somewhere. All you will need to do is ignore the two output boards at the top, and roll your own T03 ones. I am at work right now, but I will post a copy tomorrow.
:edit: And it may well be that Mark has a few boards left over after the group buy, IIRC, he was going to get extras just for people like yourself that missed the first order.
There is no reason why you can't have my layout, In fact, I thought it was already here on the thread somewhere. All you will need to do is ignore the two output boards at the top, and roll your own T03 ones. I am at work right now, but I will post a copy tomorrow.
:edit: And it may well be that Mark has a few boards left over after the group buy, IIRC, he was going to get extras just for people like yourself that missed the first order.
Measurements
OK, here are some measurements of my Krell Clone, whipped them out this morning-- they are in the "low" bias setting (about 20WPC Class A). For those of you who don't know my setup-- 4 output pairs per channel at 37.5V rails (actually more like 35 V rails with this running)
All pics are 10V/div and ground is the center horiz line on the graticule. I've got some 100W wirewound dale power resistors about 5 ohms and in parallel about 3 ohms.
I think all except for one have the positive rail (same scale and origin) shown along with the output waveform. You see my DC rails dip to about 35 volts rather than the 37.5 I am supposed to be getting. In fact they run at about 35 to 36 all the time, so I wouldn't call the dip at clipping much of a dip at all. Then again, I wonder where my 2 volts went?
Sine Clipping into 5 ohms, 2KHz.
Sine Clipping into 5 ohms, 40Hz.
Wanted to show these, as the mid frequency clipping behavior is not symmetric. Just as you enter clipping the negative signal distorts about 5 volts up.. and then as you increase the input the output distorts about 10 volts up. Right were the vertical line is on the neg phase. So where the negative signal was undistorted prior to clipping it is now distorted-- rather than simply becoming flat. Not with the positive- it simply gets flat. And not so at clipping into 40 Hz. I can get 50 V pk-pk undistorted into 5 ohms, but when it clips the distortion points move up to what you see here.
I am able to get 25 volts (50 peak to peak) into 5 ohms, actually almost 30 (60 pk-pk) as shown in the 40 Hz pic. I think this at least 78 watts at 5 ohms.
Sine Clipping into 3 ohms, 2KHz.
Sine Clean output into 3 ohms, 2KHz.
Sine Clean output into 3 ohms, 40Hz.
So you see 25 volts (50 peak to peak) undistorted into 3 ohms. V^2/R = 208 Watts at 3 ohms? Also 8.3 amps.
Did not want to try this into less ohms due to the lack of protection circuitry and me wanting to actually listen to this amp before blowing it up.
Square wave
Square into 5 ohms, 20KHz No DC Rail Shown.
The following pic is not as good but its super cool to see the positive rail and where it is having trouble maintaining the voltage, charging and discharging.
Square into 5 ohms, 20KHz With Pos Rail.
Looks like this will put out at least 8 amps into 3 ohms, no sign of letup but I'm just not ready to try 2 and 1 ohm yet.
One other thing, when I've tested amps to clipping like this in the past, at major power levels you can actually hear the test tones eminating from the inside of the amp somewhere. With this clone at the max output into a resistor its completely silent, which I suppose is a good thing.
Anyway, please comment away about the measurement. Good/bad? Is this thing working right? Do these show what one would expect or if not what's different?
OK, here are some measurements of my Krell Clone, whipped them out this morning-- they are in the "low" bias setting (about 20WPC Class A). For those of you who don't know my setup-- 4 output pairs per channel at 37.5V rails (actually more like 35 V rails with this running)
All pics are 10V/div and ground is the center horiz line on the graticule. I've got some 100W wirewound dale power resistors about 5 ohms and in parallel about 3 ohms.
I think all except for one have the positive rail (same scale and origin) shown along with the output waveform. You see my DC rails dip to about 35 volts rather than the 37.5 I am supposed to be getting. In fact they run at about 35 to 36 all the time, so I wouldn't call the dip at clipping much of a dip at all. Then again, I wonder where my 2 volts went?
Sine Clipping into 5 ohms, 2KHz.
Sine Clipping into 5 ohms, 40Hz.
Wanted to show these, as the mid frequency clipping behavior is not symmetric. Just as you enter clipping the negative signal distorts about 5 volts up.. and then as you increase the input the output distorts about 10 volts up. Right were the vertical line is on the neg phase. So where the negative signal was undistorted prior to clipping it is now distorted-- rather than simply becoming flat. Not with the positive- it simply gets flat. And not so at clipping into 40 Hz. I can get 50 V pk-pk undistorted into 5 ohms, but when it clips the distortion points move up to what you see here.
I am able to get 25 volts (50 peak to peak) into 5 ohms, actually almost 30 (60 pk-pk) as shown in the 40 Hz pic. I think this at least 78 watts at 5 ohms.
Sine Clipping into 3 ohms, 2KHz.
Sine Clean output into 3 ohms, 2KHz.
Sine Clean output into 3 ohms, 40Hz.
So you see 25 volts (50 peak to peak) undistorted into 3 ohms. V^2/R = 208 Watts at 3 ohms? Also 8.3 amps.
Did not want to try this into less ohms due to the lack of protection circuitry and me wanting to actually listen to this amp before blowing it up.
Square wave
Square into 5 ohms, 20KHz No DC Rail Shown.
The following pic is not as good but its super cool to see the positive rail and where it is having trouble maintaining the voltage, charging and discharging.
Square into 5 ohms, 20KHz With Pos Rail.
Looks like this will put out at least 8 amps into 3 ohms, no sign of letup but I'm just not ready to try 2 and 1 ohm yet.
One other thing, when I've tested amps to clipping like this in the past, at major power levels you can actually hear the test tones eminating from the inside of the amp somewhere. With this clone at the max output into a resistor its completely silent, which I suppose is a good thing.
Anyway, please comment away about the measurement. Good/bad? Is this thing working right? Do these show what one would expect or if not what's different?
Howdy pinkmouse,
Wooo........Haaaa...........
You are a prince and must be a really good guy to help out like this.
My most generous thanks for all your hard work and superior design.
Buying the boards would be quite OK, 'cepting for the fact that when you make stuff yourself, you get the responsibility of fixin' it yourself.
My 16 year old grandson will be helping/learning during this creation and these are good life lessons also. My granpa was a cabinet maker when folks used real solid wood. He taught me that when folks looked at your new table or chair and said, "Nice piece, did you make it yourself?", that meant you did it poorly. They should have said, "Where did you buy it?"
Many, many thanks for your kindness.
Elvin
Wooo........Haaaa...........
You are a prince and must be a really good guy to help out like this.
My most generous thanks for all your hard work and superior design.
Buying the boards would be quite OK, 'cepting for the fact that when you make stuff yourself, you get the responsibility of fixin' it yourself.
My 16 year old grandson will be helping/learning during this creation and these are good life lessons also. My granpa was a cabinet maker when folks used real solid wood. He taught me that when folks looked at your new table or chair and said, "Nice piece, did you make it yourself?", that meant you did it poorly. They should have said, "Where did you buy it?"
Many, many thanks for your kindness.
Elvin
Re: Measurements
You are forgetting your 0.499 ohms resistors, Mr Green.
Divided by 4 is 0.125 Ohms.
With a 5 Ohms load, the actual load on the powersupply is 5.125 Ohms.
37.5 divided 5.125 times 5 = 37.5 * 40 / 41~ 36.6 volts, the emitter resistors make the output voltage drop a point.
Pushing 7 amps into your dummy makes the powersupply drop a volt.
Is that bad ?
Unsymmetrical clipping: no good.
lgreen said:
You are forgetting your 0.499 ohms resistors, Mr Green.
Divided by 4 is 0.125 Ohms.
With a 5 Ohms load, the actual load on the powersupply is 5.125 Ohms.
37.5 divided 5.125 times 5 = 37.5 * 40 / 41~ 36.6 volts, the emitter resistors make the output voltage drop a point.
Pushing 7 amps into your dummy makes the powersupply drop a volt.
Is that bad ?
Unsymmetrical clipping: no good.
Hi Lgreen,
the clipping pics seem to show distinct sticking of the signal when coming out of negative clipping and just perceptable sticking coming out of positive clipping.
To all,
Is this discrepancy normal in complementary topology?
Is there a method of juggling the circuit parameters to eliminate the sticking? or to improve the negative to the same standard as the positive?
the clipping pics seem to show distinct sticking of the signal when coming out of negative clipping and just perceptable sticking coming out of positive clipping.
To all,
Is this discrepancy normal in complementary topology?
Is there a method of juggling the circuit parameters to eliminate the sticking? or to improve the negative to the same standard as the positive?
Re: Vbe mismatch in outputs
Many seem to see base to emitter of a transistor as a diode with a floating voltage drop.
Different base resistors could make output voltage equal for different Vbe drop transistors, for a given input voltage.
But, with equal Hfe values, with input voltage going higher or lower the output voltage difference between devices would become negative if input voltage moves in one direction, positive if input voltage goes the other way.
Thus, creating non-linearity between 2 devices, making matters far worse.
To me this resembles excessive feedback design, huge deviation corrected by huge feedback, out comes a bucket full of nasty harmonics.
Without different value base resistors, with equal Hfe values and differing Vbe drop, there would be a difference in output voltage all the time.
That difference however is constant, if Vbe behavior is equal for several devices.
Suppose base resistors are equal, but a diode is placed between base resistor and base, diodes with differing diode drop to equalis(/z)e Vbe drop differences.
Output voltage would be equal and linear.
But then the diodes had to be selected instead of selecting output devices.
My 2 Pence, i had the same thought at the time when i had no means(funds) of selecting
AndrewT said:
Many seem to see base to emitter of a transistor as a diode with a floating voltage drop.
Different base resistors could make output voltage equal for different Vbe drop transistors, for a given input voltage.
But, with equal Hfe values, with input voltage going higher or lower the output voltage difference between devices would become negative if input voltage moves in one direction, positive if input voltage goes the other way.
Thus, creating non-linearity between 2 devices, making matters far worse.
To me this resembles excessive feedback design, huge deviation corrected by huge feedback, out comes a bucket full of nasty harmonics.
Without different value base resistors, with equal Hfe values and differing Vbe drop, there would be a difference in output voltage all the time.
That difference however is constant, if Vbe behavior is equal for several devices.
Suppose base resistors are equal, but a diode is placed between base resistor and base, diodes with differing diode drop to equalis(/z)e Vbe drop differences.
Output voltage would be equal and linear.
But then the diodes had to be selected instead of selecting output devices.
My 2 Pence, i had the same thought at the time when i had no means(funds) of selecting
AndrewT said:
The sticking problem can be reduced if the input stages were allowed to clip before the output stage. Because of the large charge storage effects of the output transistors, they don't come out of clipping as fast as the smaller transistors, so it's better if the output stage doesn't clip.
But you are right as always in that its better if some other people check to see if they get the same results.
Mark A. Gulbrandsen said:
Other than using MJ21193/94 output devices, 47pF for C105 and C106, large 5W zeners, and not quite exact resistor values (the ones I used are shown here ), my setup is just like Jan's part list, here) minus the output board and minus the protection devices. I may have used some AVX multi layer film caps (yellow caps on the board) to replace a few Wima's (Mouser ran out of Wima's when I was ordering), but I was able to locate and use all the specified transistors.