Just picked up a Krell KSA-200B. What a monster!!
Anyways it seems to be running "cool" at about 120F. All the talk states the unit should be running pretty toasty. This is definitely not the case especially when compared to my Threshold 400A i just rebiased to 55C.
I sent an email to Krell about the temperature and the concern about the unit being underbiased. They responded stating 120F sounds about right cause the unit was fairly low bias. Of note I have to say Krell has even great about responding to my several emails.
I found a schematic/service manual for all KSA/KMA models showing appropriate voltages and bias. The manual states all models should be set to 325mV. This raises two questions-
1) does the bias setting add up to 200 WPC class A? I read the article at stereotimes about the KSA-250 and calculating class A watts. If I use that calculation then I'm no where near 200 with 9 output pairs per channel.
2) anybody have experience with this amp to know how hot the heat sinks get?
I plan on opening the amp up to check bias myself but I don't feel like lugging this 150lb beast to the bench quite yet. I wanted to ask around cause there are few threads about bias level on these KSA amps (Did see some info that the KSA-100 had bias set to 500mV).
Thanks for the help
Anyways it seems to be running "cool" at about 120F. All the talk states the unit should be running pretty toasty. This is definitely not the case especially when compared to my Threshold 400A i just rebiased to 55C.
I sent an email to Krell about the temperature and the concern about the unit being underbiased. They responded stating 120F sounds about right cause the unit was fairly low bias. Of note I have to say Krell has even great about responding to my several emails.
I found a schematic/service manual for all KSA/KMA models showing appropriate voltages and bias. The manual states all models should be set to 325mV. This raises two questions-
1) does the bias setting add up to 200 WPC class A? I read the article at stereotimes about the KSA-250 and calculating class A watts. If I use that calculation then I'm no where near 200 with 9 output pairs per channel.
2) anybody have experience with this amp to know how hot the heat sinks get?
I plan on opening the amp up to check bias myself but I don't feel like lugging this 150lb beast to the bench quite yet. I wanted to ask around cause there are few threads about bias level on these KSA amps (Did see some info that the KSA-100 had bias set to 500mV).
Thanks for the help
In of itself, "325mV" is a meaningless term.
325mV is part of an equation that will get you to your answer. Also I'd stay away from any bias procedure where the temperature at some location is the target. The temperature at any particular location will be a function of dissipation and ambient temperature and airflow.
To calculate what each transistor is dissipating, you can sample the idle current through it by measuring the voltage across each transistors emitter resistor, then dividing that voltage by the emitter resistor's value. That will get you the current. To find the power, while idling, of each transistor, multiply each transistor's current by the voltage across the Emitter and Collector.
325(mV)/0.15(Ohms)= 2.1(A), 325(mV)/0.47(Ohms)=691(mA)
If the rails are low-ish, 2.1(A)*40(Volts) = 86(Watts) per device, 691(mA)*40(Volts) = 27(Watts) per device.. If this amp does not have forced cooling, given time, it will exceed 120 deg (F), I believe. I'd check the bias values. Make sure all of the emitter resistors are not open and all have a similar voltage drop across them.
325mV is part of an equation that will get you to your answer. Also I'd stay away from any bias procedure where the temperature at some location is the target. The temperature at any particular location will be a function of dissipation and ambient temperature and airflow.
To calculate what each transistor is dissipating, you can sample the idle current through it by measuring the voltage across each transistors emitter resistor, then dividing that voltage by the emitter resistor's value. That will get you the current. To find the power, while idling, of each transistor, multiply each transistor's current by the voltage across the Emitter and Collector.
325(mV)/0.15(Ohms)= 2.1(A), 325(mV)/0.47(Ohms)=691(mA)
If the rails are low-ish, 2.1(A)*40(Volts) = 86(Watts) per device, 691(mA)*40(Volts) = 27(Watts) per device.. If this amp does not have forced cooling, given time, it will exceed 120 deg (F), I believe. I'd check the bias values. Make sure all of the emitter resistors are not open and all have a similar voltage drop across them.
Last edited:
Thanks for the help. I have attached the service manual and schematic I have obtained.
I agree. I should have been a little more specific. The emitter resistor value is 1 ohm, rail voltage is 77, and number of transistors is 36.
Based on the stated bias (325mV), rail voltage (77V)emitter resistor value (1 ohm), and number of output traunsistors (36), I get this equation:
325(mV)/1(Ohms)= 325(mA)
325(mA)*77(Volts) = ~25(Watts) per device
25(Watts)*36(Devices) = ~900(Watts) consumption at idle
Based on my minimal understanding of class A being extremely inefficient the 200WPC is a little high for the 900(Watts) dissipation.
If I use the equation from the review by Atkinson (stereotimes) in regards to calculating true class A watts I get this:
If assuming 200 WPC:
Using equation: sqrt(WPC/2*speaker Ohm)
sqrt(200/2*8) = 3.54A
then using: Amperege/transistor pairs per channel
3.54/9 = 393mA
393mA = 393mV given 1(Ohm) resistor value
Based on that the voltage change across the resistor should be higher than the spec. Of course I am assuming the equation is an accurate estimate.
If it is and I reverse the equation given the 325mV bias I would get only ~137 WPC in class A.
I have left the unit on for several hours and I think the highest it has gone is around 130F. Ultimately I plan on setting the bias (not by temp
) and checking various components. I just find this stuff very interesting and enjoy learning about how these amps work.
I agree. I should have been a little more specific. The emitter resistor value is 1 ohm, rail voltage is 77, and number of transistors is 36.
Based on the stated bias (325mV), rail voltage (77V)emitter resistor value (1 ohm), and number of output traunsistors (36), I get this equation:
325(mV)/1(Ohms)= 325(mA)
325(mA)*77(Volts) = ~25(Watts) per device
25(Watts)*36(Devices) = ~900(Watts) consumption at idle
Based on my minimal understanding of class A being extremely inefficient the 200WPC is a little high for the 900(Watts) dissipation.
If I use the equation from the review by Atkinson (stereotimes) in regards to calculating true class A watts I get this:
If assuming 200 WPC:
Using equation: sqrt(WPC/2*speaker Ohm)
sqrt(200/2*8) = 3.54A
then using: Amperege/transistor pairs per channel
3.54/9 = 393mA
393mA = 393mV given 1(Ohm) resistor value
Based on that the voltage change across the resistor should be higher than the spec. Of course I am assuming the equation is an accurate estimate.
If it is and I reverse the equation given the 325mV bias I would get only ~137 WPC in class A.
I have left the unit on for several hours and I think the highest it has gone is around 130F. Ultimately I plan on setting the bias (not by temp
) and checking various components. I just find this stuff very interesting and enjoy learning about how these amps work.
Yeah, I think that the Class A output may be up to some output power, but not all the way up to 200W into 8 Ohms. That's just a guess on my part.
Do you have any AC line measurement device, like a Kill-A-Watt plug in thingie? You might be able to use something like that, without lugging the whole thing to a bench, to get an idea of current draw and power dissipation.
Thanks for the Atkinson method of determining full Class A. I remember how the Levinson ML-2 would pull >3A from the wall (@120VAC) idling, then make a slight dip in current draw at clipping. That amp was 25 Watts into 8 Ohms, and mono. ....But it did 50 Watts into 4 Ohms, 100 Watts into 2 Ohms all in Class A. That's another aspect of Class A, is the load has an affect of how Class A it is. The lower the load impedance, the higher the bias needs to be to maintain full bias for the full voltage swing, so there may be "grades" of Class A too.
Do you have any AC line measurement device, like a Kill-A-Watt plug in thingie? You might be able to use something like that, without lugging the whole thing to a bench, to get an idea of current draw and power dissipation.
Thanks for the Atkinson method of determining full Class A. I remember how the Levinson ML-2 would pull >3A from the wall (@120VAC) idling, then make a slight dip in current draw at clipping. That amp was 25 Watts into 8 Ohms, and mono. ....But it did 50 Watts into 4 Ohms, 100 Watts into 2 Ohms all in Class A. That's another aspect of Class A, is the load has an affect of how Class A it is. The lower the load impedance, the higher the bias needs to be to maintain full bias for the full voltage swing, so there may be "grades" of Class A too.
I don't have an AC line measurement. Been meaning to get one cause it would be useful for a few amps I have worked on. Previous threads have stated it pulls 14A at idle.
Talking about the doubling down with lower impedance probably gets out of my understanding of class A. I have heard some people say that is the case while I have read a couple articles stating this is not a function of Class but the power supply. Per them Class A should be cut in half as impedance drops. Can't say I know which is correct. I do know that they way Krell adertises they make it sound like these (Class A and doubling down of watts) as two separate things.
I also found this post while diging through DIY which backs up my thoughts. Maybe this person will chime in and provide a little more insight:
"No, it is not.
The KSA-200 is biased at 325mV across the 1R emitter resistors, 325mA per power transistor. (Krell issued number, btw)
The power amp has 3 blocks of 6 TO-3's per channel, 18 times 325mA makes 5.85A.
That is a little over half the nominal continuous power in class A, or 274W peak.
(the last bit of quiescent current doesn't count, as one enters the cross-over region there)
Full 200W class A would require at least 3.5 amp quiescent current, with 75Vdc loaded rails that equates to some 550W dissipation per channel.
Even at half the nominal continuous power in 8 ohm in class A, the heatsinks of the KSA-200 are underdimensioned.
Most of the time, they were running well over 140F.
Most serial manufacture class A power amps that have been produced were only part biased.
Only FULL class A biased Krell models are the KSA-50, KSA-100, and KMA models. (till 1990-1991, when Krell went bias hopping)"
Talking about the doubling down with lower impedance probably gets out of my understanding of class A. I have heard some people say that is the case while I have read a couple articles stating this is not a function of Class but the power supply. Per them Class A should be cut in half as impedance drops. Can't say I know which is correct. I do know that they way Krell adertises they make it sound like these (Class A and doubling down of watts) as two separate things.
I also found this post while diging through DIY which backs up my thoughts. Maybe this person will chime in and provide a little more insight:
"No, it is not.
The KSA-200 is biased at 325mV across the 1R emitter resistors, 325mA per power transistor. (Krell issued number, btw)
The power amp has 3 blocks of 6 TO-3's per channel, 18 times 325mA makes 5.85A.
That is a little over half the nominal continuous power in class A, or 274W peak.
(the last bit of quiescent current doesn't count, as one enters the cross-over region there)
Full 200W class A would require at least 3.5 amp quiescent current, with 75Vdc loaded rails that equates to some 550W dissipation per channel.
Even at half the nominal continuous power in 8 ohm in class A, the heatsinks of the KSA-200 are underdimensioned.
Most of the time, they were running well over 140F.
Most serial manufacture class A power amps that have been produced were only part biased.
Only FULL class A biased Krell models are the KSA-50, KSA-100, and KMA models. (till 1990-1991, when Krell went bias hopping)"
Another thing to add, if I do the same calculation with the KSA-80B (6 transistor pairs) I only get 60WPC which is a bit less than the claimed 80WPC that is advertised.
Given that many people on multiple forums have stated these KSA models were the only true full class A Krells produced, I find it hard to believe I am getting the correct numbers.
At this point I feel that maybe either the bias value is incorrect (unlikely given the schematic) or the calculation is off(I have no idea?).
Given that many people on multiple forums have stated these KSA models were the only true full class A Krells produced, I find it hard to believe I am getting the correct numbers.
At this point I feel that maybe either the bias value is incorrect (unlikely given the schematic) or the calculation is off(I have no idea?).
The heatsinks of the 200B are identical to the ones of other Krell models of that era, e.g. the dynamically biased KSA-250.
In those days, it was a high performance heatsink, sizes something in the order of 7'' W x 5'' D x 6'' H.
At ambient temperature of 25C/77F, it handles less than 100W for a heatsink temperature of 50C/122F, but Krell liked to bias to frying temperature level.
(one reason they often required a repair job, Krells are Hot Rod amps)
The 200B has 6 heatsinks.
For 900W of total idle dissipation, each heatsink would handle 150W.
At 150W dissipation, temperature of the heatsink would be over 160F.
Only possible conclusion : the bias numbers you posted are not correct
(anyone who has ever built a single power amplifier, class A or class AB, can tell from just a mere glance at the heatsink, that it's not able to handle 150W dissipation)
In those days, it was a high performance heatsink, sizes something in the order of 7'' W x 5'' D x 6'' H.
At ambient temperature of 25C/77F, it handles less than 100W for a heatsink temperature of 50C/122F, but Krell liked to bias to frying temperature level.
(one reason they often required a repair job, Krells are Hot Rod amps)
The 200B has 6 heatsinks.
For 900W of total idle dissipation, each heatsink would handle 150W.
At 150W dissipation, temperature of the heatsink would be over 160F.
Only possible conclusion : the bias numbers you posted are not correct
(anyone who has ever built a single power amplifier, class A or class AB, can tell from just a mere glance at the heatsink, that it's not able to handle 150W dissipation)
Thanks for the information. I am definitely now wondering what is the correct bias level for this amp although it is tough to believe the manual is wrong. I have contacted Krell and their tech (has worked there for 10 years) had never seen the manual. He passed it along to the engineer who supposidely helped design the amp. He had previously stated the amp was underbaised at that around 120F was about correct.
So I am definitely getting lots of different answers to this whole thing. What I do know is that if the calculations I made are correct (no one has said otherwise) then the 325mV is too low even for the Krell KSA-80 and should be around 390mV.
I have found that if I leave the amp at idle for about 30 minutes is does get a bit hotter (as epxected) but at best I would say around 130-135F.
One question I do have is how you came to the 900W dissipation? I haven't check how much current my amp is actually pulling but is that the epected number. I have seen quite a few different values for the current draw on these amps.
Bottomline is I have to pull the cover off and test a few things. Just dragging my feet cause I have a few other amps I am working on currently and this thing weighs a ton!
Thanks
As I was typing the previous post Krell had responded to my email. They state the bias should actually be 300mV. 
. Am I correct in thinking there is no way this is right and stil get 200WPC class A.
My thoughts now are that Krell was having issues with the amp at such high heat and bias level that they are recommending a lower bias?
Here is an interesting quote I found:
. Am I correct in thinking there is no way this is right and stil get 200WPC class A.My thoughts now are that Krell was having issues with the amp at such high heat and bias level that they are recommending a lower bias?
Here is an interesting quote I found:
The 200B was like all Krells never Pure Class A.
But most of them had relative high Bias and drove maybe 20 % of the Output power in Class A into 8 Ohms, over this value simply AB.
But all of them were very stable with load 1ohm, were other amps exhausted smoke.
The value apx. 325 mA over 1 Ohm is correct if my memory serves me right.
When we adjusted,it was easier to measure the power, it draw apx 900 Watts out of the wall socket, the smaller 80B apx 550 Watt, but i have the values not written down somewhere.
Here in Europe the Model 80 was sold as 100 and the 200 as 250. Marketing.
But heaters anyway.
The only more or less real Class A Bruisers were the KRS 100 and 200, they were so high biased that they could drive 0,1 Ohm load , this is the impedance/resistance the Apogee Full Range Midrange Ribbon has when connected direct. We did it!
But most of them had relative high Bias and drove maybe 20 % of the Output power in Class A into 8 Ohms, over this value simply AB.
But all of them were very stable with load 1ohm, were other amps exhausted smoke.
The value apx. 325 mA over 1 Ohm is correct if my memory serves me right.
When we adjusted,it was easier to measure the power, it draw apx 900 Watts out of the wall socket, the smaller 80B apx 550 Watt, but i have the values not written down somewhere.
Here in Europe the Model 80 was sold as 100 and the 200 as 250. Marketing.
But heaters anyway.
The only more or less real Class A Bruisers were the KRS 100 and 200, they were so high biased that they could drive 0,1 Ohm load , this is the impedance/resistance the Apogee Full Range Midrange Ribbon has when connected direct. We did it!
This would be dissapointing. I thought I had done my research about these older models being class A up to advertised spec. I knew it was probably a little far fetched believing I would get 200WPC but I thought it would be close and not looking at 20%.
Disappointing? Based on the 20%, that'd get you 40Watts into 8 Ohms full Class-A. More that 17Vrms of voltage swing. That's quite a bit of signal/volume level in full Class-A.
For me, I was enamored with full Class-A for awhile. Then I performed as honest an A|B test that I could give myself. For me, I either thought that I couldn't hear any difference, or that the difference was too small to justify the thermal hit to the non air conditioned room and the power bill. However, as techno-porn, it's still kind of cool for me.
For me, I was enamored with full Class-A for awhile. Then I performed as honest an A|B test that I could give myself. For me, I either thought that I couldn't hear any difference, or that the difference was too small to justify the thermal hit to the non air conditioned room and the power bill. However, as techno-porn, it's still kind of cool for me.
Sorry, it is like it is.
The ML 2 and the Classé DR 3 were really Class A, but 25 W into 8 ohms. With 4 Ohms 12 Watts and so on...
Krells old models with Fans were higher Class A, but still not much more than 40 %.
You can easy check any amp for Class A amount, measure the output power into 8 ohms and same time the power drawn out of the wall.
As soon as the power out of the wall starts to increase, read the output value and thats it.
You would be surprised how much figures declared by manufacturers are pure marketing BS.
You can also calculate. Use ohms law to set the amount of output current into 8 ohms and then measure the voltage on the emitterresistors, multiply it with the amount of output device and be aware that Class A has max efficiency of apx. 47 % or lesser.
The ML 2 and the Classé DR 3 were really Class A, but 25 W into 8 ohms. With 4 Ohms 12 Watts and so on...
Krells old models with Fans were higher Class A, but still not much more than 40 %.
You can easy check any amp for Class A amount, measure the output power into 8 ohms and same time the power drawn out of the wall.
As soon as the power out of the wall starts to increase, read the output value and thats it.
You would be surprised how much figures declared by manufacturers are pure marketing BS.
You can also calculate. Use ohms law to set the amount of output current into 8 ohms and then measure the voltage on the emitterresistors, multiply it with the amount of output device and be aware that Class A has max efficiency of apx. 47 % or lesser.
I have a very early Krell KSA 100 with MJ 15024 outputs with +/- 48 volt rails and 4 devices per polarity biased at 625 ma per device for a 2.5 amp total bias current. 2.5 amps x (.7071 x 48VDC ) = 84.5 watts Class's A bias. It used an EI core Western Electric transformer with an extra ground lead making me think it has an electrostatic shield - Krell soon moved to torroids under Japanese market pressure. I also recently got a KSA 250, but I haven't looked inside or measured anything - yet
This gives some specs on the KSA-100 (if accurate) and definitely shows the higher bias (almost double). Don't really udnerstand the 0.7071 (some constant or efficiency factor). Maybe these calculations are a bit generous also.
You see, the Output stage is Bias 84,5 Watt, but max 47 % efficiency, so there are real apx 40 Watt Class A going in the speaker.
And yes, the KSA 100 run really hot, so we reduced Bias to apx. 475 mA to prevent failure and changed the fans for some with lesser noise and recommended to wipe out the dust all 2-3 years from the fan grills.
And yes, the KSA 100 run really hot, so we reduced Bias to apx. 475 mA to prevent failure and changed the fans for some with lesser noise and recommended to wipe out the dust all 2-3 years from the fan grills.
mjstriker5 said:
325mV, measured across emitter resistors commonly means from the emitter of a PNP output to the emitter of an NPN output.
Aka across Two 1 ohm emitter resistors : 1 + 1 = 2
325mv divided by 2 = 162.5mA
162.5 times 18 = 2.925A total quiescent current, good for ~68 peak class A output power
2.925A times 77V rails makes 225W total dissipation per channel.
225W divided by 3 heatsinks makes 75W per heatsink.
132F/55C at 25C ambient temperature makes a temperature rise of 30C
30C divided by 75W = 0.40C/W
That's the thermal resistance number I measured for that heatsink in the early '90s
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.