What is the tolerance of the 0r47 Re?
add that tolerance to the variation on the 15mVre and you may find that your bias currents could be anywhere, or very close. One simply doesn't know if one doesn't hand select the Re to better than 1% and preferably <0.2%
Kap,
the KSA100 is specified as a 100W ClassA amplifier. That 100W is delivered into an 8r0 load.
If you want/need/require a 200W ClassA amplifier for a 4r0 load then you should be looking at 200W into 4r0 ClassA amplifiers.
There is some evidence that lower emitter resistor values do result in lower output distortion. If this holds true for ClassA operation, then reducing the Re from 1r0 to 0r47 may give a small benefit. Keep in mind that with 4pairs of output devices the effective re was 0r25 and will become 0r118 for the proposed change. That is quite low for a +-52Vdc supply. Maybe Krell knew a thing or two after developing and building dozens of high power amplifiers.
Cordell has written a lot about this Re value and supply Vdc and Thermal Stability relationship.
the KSA100 is specified as a 100W ClassA amplifier. That 100W is delivered into an 8r0 load.
If you want/need/require a 200W ClassA amplifier for a 4r0 load then you should be looking at 200W into 4r0 ClassA amplifiers.
There is some evidence that lower emitter resistor values do result in lower output distortion. If this holds true for ClassA operation, then reducing the Re from 1r0 to 0r47 may give a small benefit. Keep in mind that with 4pairs of output devices the effective re was 0r25 and will become 0r118 for the proposed change. That is quite low for a +-52Vdc supply. Maybe Krell knew a thing or two after developing and building dozens of high power amplifiers.
Cordell has written a lot about this Re value and supply Vdc and Thermal Stability relationship.
"The more you learn the more you must learn."
Thanks AndrewT, Stuart Easson gave me a bunch of 0R67--he has a ton of them, just send a private message to him--he said the same, but said I could experiment. I will read Cordell just to understand a new subject involved here--"Thermal Stability relationship."
Thanks for the replies!
@AVWERK: Think that is indeed a good way to test the amplifier, when this beast finally works instead of:
:-D. I will measure carefully how much mV difference there is between the devices. Luckily production processes of trannies have improved a lot compared to the 80's. Sometimes people even say matching is not necessary. If I don’t make a mistake Onsemi, which produces a lot of former Motorola products (like the MJ-trannies) have developped a new production process which makes the specs even more accurate. I thought it was something like a more accurate laser to make the dies, but I don't remember exactely what their improvement was.
Did Krell actually matched the trannies? For instance in my AudioAnalyse the MJ15003/4’s are from the same batch, but that's it. They are not specificly matched. There is about 4 mV spread in bias, emitter resistors are .47 ohm.
@Andrew: Aha, so the KSA is indeed designed fully class A into 8 ohm. My speakers are 4 ohm and have a 3.2 ohm dip (B&W DM604). Thus no full class A disco here
Only kiddin, at 4 ohm the amplifier has for me more than enough class A power to make the neighbours mad haha.
Hmm, Krell used 1 ohm emitter resistors in the original. That is why I bought also 1 ohm 1% emitter resistors. I can’t measure THD, but it would be a nice experiment. I have bought 'Designing Power Amplifiers' from Mr. Cordell a few weeks ago, I will certainly read it.
By the way, made again some progress. I've found original 2SA968 and 2SC2238 Dean! Came across a batch from a little webshop in Holland. I've bought a few pairs, so I have some replacements when I make a stupid mistake.
I have mounted them to the boards and replaced the unrelieable ISC's:
I also received the speaker terminals, don't mind the mess on the floor ;-)
Drilled some holes in the cabinet:
Resulting in:
@AVWERK: Think that is indeed a good way to test the amplifier, when this beast finally works instead of:
An externally hosted image should be here but it was not working when we last tested it.
:-D. I will measure carefully how much mV difference there is between the devices. Luckily production processes of trannies have improved a lot compared to the 80's. Sometimes people even say matching is not necessary. If I don’t make a mistake Onsemi, which produces a lot of former Motorola products (like the MJ-trannies) have developped a new production process which makes the specs even more accurate. I thought it was something like a more accurate laser to make the dies, but I don't remember exactely what their improvement was.
Did Krell actually matched the trannies? For instance in my AudioAnalyse the MJ15003/4’s are from the same batch, but that's it. They are not specificly matched. There is about 4 mV spread in bias, emitter resistors are .47 ohm.
@Andrew: Aha, so the KSA is indeed designed fully class A into 8 ohm. My speakers are 4 ohm and have a 3.2 ohm dip (B&W DM604). Thus no full class A disco here
Only kiddin, at 4 ohm the amplifier has for me more than enough class A power to make the neighbours mad haha.Hmm, Krell used 1 ohm emitter resistors in the original. That is why I bought also 1 ohm 1% emitter resistors. I can’t measure THD, but it would be a nice experiment. I have bought 'Designing Power Amplifiers' from Mr. Cordell a few weeks ago, I will certainly read it.
By the way, made again some progress. I've found original 2SA968 and 2SC2238 Dean
! Came across a batch from a little webshop in Holland. I've bought a few pairs, so I have some replacements when I make a stupid mistake. I have mounted them to the boards and replaced the unrelieable ISC's:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I also received the speaker terminals, don't mind the mess on the floor ;-)
An externally hosted image should be here but it was not working when we last tested it.
Drilled some holes in the cabinet:
An externally hosted image should be here but it was not working when we last tested it.
Resulting in:
An externally hosted image should be here but it was not working when we last tested it.
Hi Nick,
Thanks!! I have not tested yet, there is still a lot of work to be done. Most parts are mounted to the cabinet, but I have to do a lot of wiring before I can test the amp. I think I start two weeks from now with wiring the PSU, mounting the caps to the cabinet and add little current and increasing voltage to the big caps slowly so they can slowly form their oxidelayer back (they haven't been used for at least 3 years). The PCB design is in my opinion thusfar sound, but I can't say for sure. Planning is to finish the amp somewhere in August when I have vacation
Thanks
!! I have not tested yet, there is still a lot of work to be done. Most parts are mounted to the cabinet, but I have to do a lot of wiring before I can test the amp. I think I start two weeks from now with wiring the PSU, mounting the caps to the cabinet and add little current and increasing voltage to the big caps slowly so they can slowly form their oxidelayer back (they haven't been used for at least 3 years). The PCB design is in my opinion thusfar sound, but I can't say for sure. Planning is to finish the amp somewhere in August when I have vacation
Last edited:
At 1ohm I don,t think they matched anything, but at .15 they better
When you do bring it up, monitor for high frequency oscillation and if that happens shut down immediately!!
Put a trap across the speaker outputs to make it work (.1uf and 5ohm Zobel)
Doing a Bode plot and staggering poles is another whole issue on bandwidth and stability that a simple trap will fix for now.
Regards
David
When you do bring it up, monitor for high frequency oscillation and if that happens shut down immediately!!
Put a trap across the speaker outputs to make it work (.1uf and 5ohm Zobel)
Doing a Bode plot and staggering poles is another whole issue on bandwidth and stability that a simple trap will fix for now.
Regards
David
That is exactely what I wanted to ask next David ! It takes a few weeks before I can finally fire her up, but I better be prepaired. I am curious how to test the amplifier incrementally. Is there a sort of checklist? Ofcourse I will check bias, voltages across the caps, DC offset and so on, but what is sane to check further?
I've made a large dummyload with big 2 ohm resistors. I've mounted 4 of them in series on a tunnel cooler per channel, so in total there are 8 resistors. It can dissipate at least 200W @8ohm per channel. I added one 1uF capacitor per channel parallel to the resistors to approach the capacitative character of a loudspeaker a bit more. Is addition of this capacitor sane? An oscilloscope (with two channels, so I can compare both channels) and two AC RMS mV meters (one per channel) are attached to the dummyload so I can measure accurately how much output and what kind of sinewave comes out of this 'beast' when I inject a signal. I use my computer as tone generator by the way. I have also a variac so I can make the voltage incrementally higher.
Will I see a bad sine wave (when I inject 1000Hz or so) at the oscilloscope when there are HF oscillations? And do I have to measure oscillation across the speaker terminals or at the driver boards? Do I need specialised equipment to make a Bode diagram? Or is it sufficient to do a tone sweep while monitoring dips in output power?
I want to make really sure that everyting is safe and sound when I are going to use the amplifier, so the more test, the better If there are more suggestions of testing the amplifier, please post them, I am really curious
! It takes a few weeks before I can finally fire her up, but I better be prepaired. I am curious how to test the amplifier incrementally. Is there a sort of checklist? Ofcourse I will check bias, voltages across the caps, DC offset and so on, but what is sane to check further?I've made a large dummyload with big 2 ohm resistors. I've mounted 4 of them in series on a tunnel cooler per channel, so in total there are 8 resistors. It can dissipate at least 200W @8ohm per channel. I added one 1uF capacitor per channel parallel to the resistors to approach the capacitative character of a loudspeaker a bit more. Is addition of this capacitor sane? An oscilloscope (with two channels, so I can compare both channels) and two AC RMS mV meters (one per channel) are attached to the dummyload so I can measure accurately how much output and what kind of sinewave comes out of this 'beast' when I inject a signal. I use my computer as tone generator by the way. I have also a variac so I can make the voltage incrementally higher.
Will I see a bad sine wave (when I inject 1000Hz or so) at the oscilloscope when there are HF oscillations? And do I have to measure oscillation across the speaker terminals or at the driver boards? Do I need specialised equipment to make a Bode diagram? Or is it sufficient to do a tone sweep while monitoring dips in output power?
I want to make really sure that everyting is safe and sound when I are going to use the amplifier, so the more test, the better
If there are more suggestions of testing the amplifier, please post them, I am really curious
Last edited:
More likely that all the answers have been given before on this Forum, some many times.
Hi Guys,
Haha maybe David. But summer or not, replies or not Kaplaars is still @work ;-) !
Hmm, I've searched but could not find a checklist Andrew. Hope I did not overlooked a topic that explained it. Maybe I have to ask my questions again on the Dutch equivalent of this DIY-forum, or maybe someone can repeat the explanation because I am pretty curious. The basis of gaining knowledge is repeating knowledge so if someone wants to repeat, please do
I red a few things about the zobel network. Great suggestion from AVWERK. It seems that this network will prevent instability due to the sometimes capacitive character of the loudspeakers. I hate smoking, I hate it even more if my Krell will smoke, so I have to prevent instability!
@Rolandong; that is exactely how I want to mount the fans So push-pull principle. Noise will be far more less when I use two fans because noise is reduced quadratically with fanspeed. Have heard a lot of, sometimes very different, opinions about how to mount the cooling tunnels. I think I leave them the way they are. Please dont think I am stubborn, but I think this will be in my case the best solution. The tunnels are capable to dissipate at least 200W per channel with fans off (tested it with resistors mounted to them). But I am a little confused by now, because some people told me that I should use AC fans and not DC-fans because DC-fans would add distortion. Is there anything against DC-fans if I use brushless fans?
Have a little update, very busy at the moment so not very much progress. Lots of hours @lab (I now know everything about alcohol dehydrogenases... yay). But ontopic again Mr. Kaplaars; I've finished the PSU!!! I added very carefully current to the capacitors with a little stabilised PSU. I did this so the caps could reform their oxide layer back. I added a 1,2K resistor in series with the capacitor to limit current and added 25VDC to the caps. It was really cool to measure that the leakage current caused by the caps became smaller and smaller and the capacitors thus got better and better. After an hour the leakage current was only 3uA (it was 50uA). Very very good for electrolytic capacitors, but hey we don't want anything less than very very good.
Setup:
When I measured that the capacitors were OK I mounted the capacitors to the chassis. I bought realy cool braided hose to cover the cables with. Is is really nice if I mag say so Everytime when I search the internet for parts I am astonished by what there is to buy; there are so much nice parts around.
Mounted AC-cable:
I also replaced the relays from the speaker protection. I've replaced them by Omron G2RL-1's. They should be a little bit better than the ones which came with the kit, and they were not realy expensive, so it was worth a try.
And how the amp is right now:
To be continued....
Haha maybe David. But summer or not, replies or not Kaplaars is still @work ;-) !
Hmm, I've searched but could not find a checklist Andrew. Hope I did not overlooked a topic that explained it. Maybe I have to ask my questions again on the Dutch equivalent of this DIY-forum, or maybe someone can repeat the explanation because I am pretty curious. The basis of gaining knowledge is repeating knowledge so if someone wants to repeat, please do
I red a few things about the zobel network. Great suggestion from AVWERK. It seems that this network will prevent instability due to the sometimes capacitive character of the loudspeakers. I hate smoking, I hate it even more if my Krell will smoke, so I have to prevent instability!
@Rolandong; that is exactely how I want to mount the fans
So push-pull principle. Noise will be far more less when I use two fans because noise is reduced quadratically with fanspeed. Have heard a lot of, sometimes very different, opinions about how to mount the cooling tunnels. I think I leave them the way they are. Please dont think I am stubborn, but I think this will be in my case the best solution. The tunnels are capable to dissipate at least 200W per channel with fans off (tested it with resistors mounted to them). But I am a little confused by now, because some people told me that I should use AC fans and not DC-fans because DC-fans would add distortion. Is there anything against DC-fans if I use brushless fans?Have a little update, very busy at the moment so not very much progress. Lots of hours @lab (I now know everything about alcohol dehydrogenases... yay). But ontopic again Mr. Kaplaars; I've finished the PSU!!! I added very carefully current to the capacitors with a little stabilised PSU. I did this so the caps could reform their oxide layer back. I added a 1,2K resistor in series with the capacitor to limit current and added 25VDC to the caps. It was really cool to measure that the leakage current caused by the caps became smaller and smaller and the capacitors thus got better and better. After an hour the leakage current was only 3uA (it was 50uA). Very very good for electrolytic capacitors, but hey we don't want anything less than very very good.
Setup:
An externally hosted image should be here but it was not working when we last tested it.
When I measured that the capacitors were OK I mounted the capacitors to the chassis. I bought realy cool braided hose to cover the cables with. Is is really nice if I mag say so
Everytime when I search the internet for parts I am astonished by what there is to buy; there are so much nice parts around.Mounted AC-cable:
An externally hosted image should be here but it was not working when we last tested it.
I also replaced the relays from the speaker protection. I've replaced them by Omron G2RL-1's. They should be a little bit better than the ones which came with the kit, and they were not realy expensive, so it was worth a try.
And how the amp is right now:
An externally hosted image should be here but it was not working when we last tested it.
To be continued....
Last edited:
Thanks Jozua! That is a concern which I share with you. I try to keep the build as modular as possible, so that in case of failure the amplifier is relatively easy to disassemble. That is why I used for example cable clips at the side pannel instead of tiewraps and faston plugs instead of soldering the wires directly to the bridge rectifiers. From the other hand I have to assemble the tunnel coolers with a lot of bolts, they weigh a lot. So dissassembling the tunnel coolers will not be an easy task.
I realy hope so! I have never heard the original Krell KSA-100. Would be fun to compare it to an original KSA-100.
I realy hope so! I have never heard the original Krell KSA-100. Would be fun to compare it to an original KSA-100.
Kaplaars
I have the KSA 100 mk1 and mk 2. Subjectively the mk2 is more mellow wheras the mk 1 seems more dynamic in my system.
With my KSA 50 clone top notch resistors (Ricken)were used and subjectively I seem to imagine it having slightly more detail.
When I find some good looking heatsinks, or a KSA 100 box or a spare set of heat tunnels my clone 100 will proceed.
Consider yourself very lucky with your costing. This is a very expensive amp to build.
Regards
J
I have the KSA 100 mk1 and mk 2. Subjectively the mk2 is more mellow wheras the mk 1 seems more dynamic in my system.
With my KSA 50 clone top notch resistors (Ricken)were used and subjectively I seem to imagine it having slightly more detail.
When I find some good looking heatsinks, or a KSA 100 box or a spare set of heat tunnels my clone 100 will proceed.
Consider yourself very lucky with your costing. This is a very expensive amp to build.
Regards
J
The AC vs DC fan debate is just that a debate. I used both styles because of the ..debate...
Having built 6 Krell monoblocks that use both types, I can not reliably tell a difference switched in or out that it does anything negative to the soundstage or on the scope for that matter.
The zobel is cheap insurance. Bedini never used them thinking I assume it does something negative to the sound. I smoked a 150 mk2 without anything connected to the outputs while changing components and then the phone rang....magic smoke from the distance? It oscillated itself out of exsistance
When you do power up you can use the light bulb method. I haven,t done that but it is excellent.
When I power up I ramp my Variac fairly quickly to about 1/3rd to get the trannies to switch while looking at the DVM set in the milivolt
range hung on the outputs. You can place a 2nd DVM across a emitter also.
Then I wait about 3 seconds while sweating like a dog with knees vibrating, nail bitting..the whole package.. A few beers might help here BTW!
If I haven,t had a heart attack yet and can observe the DVM coming down into the milivolt range then I can start slowly increasing the Variac while looking at both DVM,s and adjusting bias along the road ahead..
My variac gets stiff fairly easily if something is wacked and that why light bulb method is a better item to asist.
If you get this far you need to sit with it awhile and keep an eye, actually both eyes on everything.
Regards
David
Having built 6 Krell monoblocks that use both types, I can not reliably tell a difference switched in or out that it does anything negative to the soundstage or on the scope for that matter.
The zobel is cheap insurance. Bedini never used them thinking I assume it does something negative to the sound. I smoked a 150 mk2 without anything connected to the outputs while changing components and then the phone rang....magic smoke from the distance? It oscillated itself out of exsistance
When you do power up you can use the light bulb method. I haven,t done that but it is excellent.
When I power up I ramp my Variac fairly quickly to about 1/3rd to get the trannies to switch while looking at the DVM set in the milivolt
range hung on the outputs. You can place a 2nd DVM across a emitter also.
Then I wait about 3 seconds
while sweating like a dog with knees vibrating, nail bitting..the whole package.. A few beers might help here BTW!If I haven,t had a heart attack yet and can observe the DVM coming down into the milivolt range then I can start slowly increasing the Variac while looking at both DVM,s and adjusting bias along the road ahead..
My variac gets stiff fairly easily if something is wacked and that why light bulb method is a better item to asist.
If you get this far you need to sit with it awhile and keep an eye, actually both eyes on everything.
Regards
David
Haha yeah I was pretty lucky with obtaining parts Jozua. There are so much details that cost money. For example I had to order all wires, terminal rings, cable clips ed for the amp, costed me another 50 euro's! Bought that stuff new which makes the parts far more expensive than second handed parts. But OK it's great quality cable which can withstand 180 degrees of temperature. No beer this week, but cables instead ;-)
WOW 6 monoblocks! How turned your clones out David? Do you have pictures of yours?
Great to hear I can still use the DC fans. I've bought pretty good ones from Nidec (Beta V). They are dead silence, you realy can't hear them. I am thinking of building PWM-control, but.... I've heard PWM-control was not sane because of distortion. Hope that is a fable too!
Great suggestion. I always use the lightbulb method to start old tube radio's (I collect them) which havent been used for years. This way you can see very fast if something is wrong. The mains electrolytic capacitor has this way also an opurtunity to form it's oxide layer a bit slower back.
Hahaha, will some Russian water, or in other words Vodka also do? ;-)
Okay David, will do, I will slowly increase voltage with my variac hoping it does not go kaboom. I've build my variac in a cabinet and added a fuse of 3A together with a VAC- and ampère meter. This fuse has proven it's value, had once a radio which was completely short circuit. Luckily for me the fuse blew which prevented further damage. I will also measure voltage across the outputs and emitters carefully. Do you attach a dummyload and maybe an oscilloscope to the outputs aswell when you are measuring across the outputs?
WOW 6 monoblocks! How turned your clones out David? Do you have pictures of yours?
Great to hear I can still use the DC fans. I've bought pretty good ones from Nidec (Beta V). They are dead silence, you realy can't hear them. I am thinking of building PWM-control, but.... I've heard PWM-control was not sane because of distortion. Hope that is a fable too!
Great suggestion. I always use the lightbulb method to start old tube radio's (I collect them) which havent been used for years. This way you can see very fast if something is wrong. The mains electrolytic capacitor has this way also an opurtunity to form it's oxide layer a bit slower back.
Hahaha, will some Russian water, or in other words Vodka also do? ;-)
Okay David, will do, I will slowly increase voltage with my variac hoping it does not go kaboom. I've build my variac in a cabinet and added a fuse of 3A together with a VAC- and ampère meter. This fuse has proven it's value, had once a radio which was completely short circuit. Luckily for me the fuse blew which prevented further damage. I will also measure voltage across the outputs and emitters carefully. Do you attach a dummyload and maybe an oscilloscope to the outputs aswell when you are measuring across the outputs?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.