Hi Mark,
which Krell were those fansinks from? They appear to be 8 To3 per channel. I thought the KSA50 had 2 pair. and a whole fan block per channel.
Still4given,
what voltage are your supply rails running at on the outputs when you are pushing 200mV on the emitters? Your dissipation is volts times amps for each transistor (you have 12).
Unless you have tough fingers you could probably get away with being able to hold your fingers on the main sink for 1 or 2 seconds only. The fins will run much cooler. The transistor cases will run about 10Cdeg to 15Cdeg above the sink temp. As an experiment only could you reconfigure as a monoblock to see what difference 6 pairs on the whole sink would do to the temps and sound quality? remember that the total Iq is now 6 times the single transistor Iq.
Could you pad down the input to the P101 to match the sensitivity of the Krell (what are your feedback and input R values?).
which Krell were those fansinks from? They appear to be 8 To3 per channel. I thought the KSA50 had 2 pair. and a whole fan block per channel.
Still4given,
what voltage are your supply rails running at on the outputs when you are pushing 200mV on the emitters? Your dissipation is volts times amps for each transistor (you have 12).
Unless you have tough fingers you could probably get away with being able to hold your fingers on the main sink for 1 or 2 seconds only. The fins will run much cooler. The transistor cases will run about 10Cdeg to 15Cdeg above the sink temp. As an experiment only could you reconfigure as a monoblock to see what difference 6 pairs on the whole sink would do to the temps and sound quality? remember that the total Iq is now 6 times the single transistor Iq.
Could you pad down the input to the P101 to match the sensitivity of the Krell (what are your feedback and input R values?).
Terry,
thermally speaking, running few devices on a big heatsink is the same as many devices on a small sink.
As Andrew mentioned, temperature of the output devices is 10-15C higher, if your sink is at 50C the die of the outputs is 65C at the highest.
Suppose your heatsink is 50C with your settings.
If your rails are 40Vdc the total dissipation for the 12 devices is around 140 watts (200mV/0.68= 0.294 amp per device)
At 25C ambient thermal resistance of the sink would be something like 0.18.
Which is quite good for the size of the sink and the low airflow rate of the fan you use.
Based on those numbers you can run it up to about 30 watts in class A, temperature of the devices would be 87.5C, temperature of the sink some 65C.
65C for the sink would not be too high to hurt the output devices.
It is too high for other components placed next to it, especially capacitors.
Compared to the real Krell you are using 1 sink arrangement for 2 channels, and the rail voltage you employ is higher.
If you had only 1 channel on this sink you could probably run it up to 75 watts in class A with a sink temperature of 55C.
With 12 devices for 2 channels its either lowering the bias, using lower voltage or increase the airflow with the bigger fan.
At 200mV in 0.68R you are currently running it a bit over 10 watts in class A.
A thing to keep in mind is that with your 45000mF capacitors per channel in the powersupply it may be wiser to have the bias lower.
thermally speaking, running few devices on a big heatsink is the same as many devices on a small sink.
As Andrew mentioned, temperature of the output devices is 10-15C higher, if your sink is at 50C the die of the outputs is 65C at the highest.
Suppose your heatsink is 50C with your settings.
If your rails are 40Vdc the total dissipation for the 12 devices is around 140 watts (200mV/0.68= 0.294 amp per device)
At 25C ambient thermal resistance of the sink would be something like 0.18.
Which is quite good for the size of the sink and the low airflow rate of the fan you use.
Based on those numbers you can run it up to about 30 watts in class A, temperature of the devices would be 87.5C, temperature of the sink some 65C.
65C for the sink would not be too high to hurt the output devices.
It is too high for other components placed next to it, especially capacitors.
Compared to the real Krell you are using 1 sink arrangement for 2 channels, and the rail voltage you employ is higher.
If you had only 1 channel on this sink you could probably run it up to 75 watts in class A with a sink temperature of 55C.
With 12 devices for 2 channels its either lowering the bias, using lower voltage or increase the airflow with the bigger fan.
At 200mV in 0.68R you are currently running it a bit over 10 watts in class A.
A thing to keep in mind is that with your 45000mF capacitors per channel in the powersupply it may be wiser to have the bias lower.
Terry at 200mV
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 41 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 200 mv
Idle bias per device 294.1 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.294 Amps
Total Amplifier bias (per rail)** 0.88 Amps
Total Amplifier bias (both rails) 1.76 Amps
Total Dissipation (per channel) 72.4 Watts
Dissipation per device pair at idle 12.1 watts
Class-A output: Peak 24.9 Watts peak
Class-A output: RMS 12.5 Watts RMS
Efficiency 17.22 %
He is dissipating about 145 watts in that whole rig, which should be ok for a small wind tunnel. However he is getting only 12.5 watts class A.
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 41 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 200 mv
Idle bias per device 294.1 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.294 Amps
Total Amplifier bias (per rail)** 0.88 Amps
Total Amplifier bias (both rails) 1.76 Amps
Total Dissipation (per channel) 72.4 Watts
Dissipation per device pair at idle 12.1 watts
Class-A output: Peak 24.9 Watts peak
Class-A output: RMS 12.5 Watts RMS
Efficiency 17.22 %
He is dissipating about 145 watts in that whole rig, which should be ok for a small wind tunnel. However he is getting only 12.5 watts class A.
AndrewT said:
Hi Andrew,
Thanks for your help. I haven't measured the rails under load. I will try to do that this evening. They are right at 42V at idle. I have measured that. I had thought about trying it as a mono block. I could do that fairly easily I think by just disconnecting the outputs on the one PCB and running a couple of jumpers. I may try that as well. I just cut out all of the panels for the case last night. It won't accommodate another heatsink assembly. If I can't get this to work I would have to start all over to make a larger case. I don't really want to do that.
I mounted the larger fan on the amp last night and was able to turn up the bias to 400mV across the .68R resistors. It got pretty hot to the touch but I was able to leave my fingers on the heatsink for 3-4 seconds. The fan noise is completely unacceptable though. I wish I had known that this transformer was too much for this amp before I bought it. It wasn't cheap. I have other large flat heatsinks that I had planned to use with the flat pack transistors but when I posted a pick of this heatsink setup everyone said it should work fine. I guess you have to pay to learn.
I will stop at a hobby store on my way home today and see if I can find one of those heat gun meters. Looks like I'm going to need one.
Hi jacco,
Are you saying that 87.5c is OK for these output devices? If that the case, I'm probably all right running the bias higher. I was running the amp into my nearfield speakers last night while I was testing and they are 4ohm, not 8. My JBL 4425's are rated at 8 ohm but they measure 6 ohm. That may explain the higher temps as well, I don't know.
Oh, and, I didn't think about dropping the voltage going into my P101 to try and get them closer in output. I'll try to do that and see if they are more comparable. Folks shouldn't sell the P101 short though. To my ears, it is a very nice sounding amp. Certainly better sounding than my stock Haflers.
Blessings, Terry
I followed the link that Mark gave and there is a link for the original KSA 50 schematic. On the schematic seen here, there are quite a few differences from the schematic used for our boards. For one thing, our schematic shows D201 & D202 as optional yet they are not shown as optional on the original. (D3 & D4). Are these diodes not necessary? What was their purpose?
On the original it says to adjust for 620 mV across the 0R68 ohm emitter resistors. Why so much higher than what we are shooting for? (400 mV)
Does this have to do with only using 2 pair of outputs?
Thanks, Terry
On the original it says to adjust for 620 mV across the 0R68 ohm emitter resistors. Why so much higher than what we are shooting for? (400 mV)
Does this have to do with only using 2 pair of outputs?
Thanks, Terry
resistors and diodes
The total current is adjusted to be the same regardless of the number of output pairs. If you have 3 output pairs, all else being equal you will only see ~2/3 of the voltage across the emitter resistors, since each pair only carries 2/3 of the current.
As the number of output pairs goes up the voltage drop will become low enough that the value of the resistors needs to be increased. The point being to keep the voltage high (0.2-0.6v), thereby reducing the current imbalances that would otherwise be introduced by Vbe/Hfe differences. I am refitting the big amp (20 output pairs) with 2 ohm resistors.
The diodes (d201, 202) you are asking about are used to 'clamp' any reactive voltages that could result from pumping lots of current into very inductive loads. They are optional primarily because Krell didn't seem to have them in the amps they made. If they aren't needed they literally do nothing, so other than cost and time there is no downside to fitting them.
HTH
Stuart
The total current is adjusted to be the same regardless of the number of output pairs. If you have 3 output pairs, all else being equal you will only see ~2/3 of the voltage across the emitter resistors, since each pair only carries 2/3 of the current.
As the number of output pairs goes up the voltage drop will become low enough that the value of the resistors needs to be increased. The point being to keep the voltage high (0.2-0.6v), thereby reducing the current imbalances that would otherwise be introduced by Vbe/Hfe differences. I am refitting the big amp (20 output pairs) with 2 ohm resistors.
The diodes (d201, 202) you are asking about are used to 'clamp' any reactive voltages that could result from pumping lots of current into very inductive loads. They are optional primarily because Krell didn't seem to have them in the amps they made. If they aren't needed they literally do nothing, so other than cost and time there is no downside to fitting them.
HTH
Stuart
Re: resistors and diodes
Stuart, I'm using 4 pairs per channel at .499 ohms, the intent is to bias them at or above .200 V for class A but I've installed a switch so I can bias them about .1 V or lower for class AB. Do you think I need higher emitter resistors or is the way I am going sufficient?
I also received a quote this morning from ATI regarding their chassis.
ATI-639M = $455
Express mail service (EMS) = $273
Almost as much as an amplifier itself.
Stuart Easson said:
Stuart, I'm using 4 pairs per channel at .499 ohms, the intent is to bias them at or above .200 V for class A but I've installed a switch so I can bias them about .1 V or lower for class AB. Do you think I need higher emitter resistors or is the way I am going sufficient?
I also received a quote this morning from ATI regarding their chassis.
ATI-639M = $455
Express mail service (EMS) = $273
Almost as much as an amplifier itself.
Thanks Stuart,
Would changing the value of the output resistors allow for more bias/less heat or is it all relative?
Ok, so the values used on the Delta website match the MK II?
Thanks, Terry
Would changing the value of the output resistors allow for more bias/less heat or is it all relative?
neychi said:Hi Terry!
We are building KSA 50 mk II, after 1983. The schematic from McLean site is pre 1983 Krell KSA 50.
Regards
Dean
Ok, so the values used on the Delta website match the MK II?
Thanks, Terry
auto bias?
Came across a note about the biasing in the S series from stereophile review of the KSA-50S.
This feature could reduce the heat problems substantially.. I don't suppose anyone has the schematic of the "anticipator" circuit. How plausible would it be to incorporate this concept?
~Brad
Came across a note about the biasing in the S series from stereophile review of the KSA-50S.
This feature could reduce the heat problems substantially.. I don't suppose anyone has the schematic of the "anticipator" circuit. How plausible would it be to incorporate this concept?
~Brad
Re: auto bias?
There have been many critics who argue that sustain plateu biasing is not that cool because it overlooks 2 things the critics "think" makes class-A so glorious sounding.
1) Constant current running through the OP devices and
2) Near constant (or relatively near constant) OP device temperatures.
Both are not quite fulfilled by the sustained bias implementation in its full glory. Personally since I have not done an A/B on both technologies with the same front end (not sure anyone has) we cannot make a good comparison or an objective case.
As far as sonics are concerned, I am of the opinion that the front end plays a larger role on sonics than the OP stage, so if the KSA-50S has a better front end, it could sound better than the KSA-50 and that would have nothing to do with the bias implementation.
Case in point, at one time I owned both the Krell KAV-250 and the KSA-250. The KSA-250 was class-A (auto bias but not sustained) and the KAV-250 is class-AB. I did listening tests many times on B&W 801's, the KAV-250 (with newer front end) always sounded better, with a heavy heart I sold the drop-dead gorgeous KSA-250 because it sounded sexless compared to the KAV-250...
googler said:
There have been many critics who argue that sustain plateu biasing is not that cool because it overlooks 2 things the critics "think" makes class-A so glorious sounding.
1) Constant current running through the OP devices and
2) Near constant (or relatively near constant) OP device temperatures.
Both are not quite fulfilled by the sustained bias implementation in its full glory. Personally since I have not done an A/B on both technologies with the same front end (not sure anyone has) we cannot make a good comparison or an objective case.
As far as sonics are concerned, I am of the opinion that the front end plays a larger role on sonics than the OP stage, so if the KSA-50S has a better front end, it could sound better than the KSA-50 and that would have nothing to do with the bias implementation.
Case in point, at one time I owned both the Krell KAV-250 and the KSA-250. The KSA-250 was class-A (auto bias but not sustained) and the KAV-250 is class-AB. I did listening tests many times on B&W 801's, the KAV-250 (with newer front end) always sounded better, with a heavy heart I sold the drop-dead gorgeous KSA-250 because it sounded sexless compared to the KAV-250...
