I understood he just wants to test his transformer, not the whole amp. So a load of 8 ohms might not be adequate.
Best regards!
Best regards!
My PSU works perfectly and the TX do not hum when there is no load.
If I set the output transistors with a light bias (50mA per transistor) there is no mechanical hum from the transformers.
The TX hums now that I raised the bias to 200mA for each output transistor.
Hi RCruz,
200 mA per device? What are your supplies again? Personally, I think that's a bit too hot.
-Chris
200 mA per device? What are your supplies again? Personally, I think that's a bit too hot.
-Chris
Hi AJT,
I really think there are limits here, past which you don't get any improvement unless you are running with no feedback. That's pretty ugly with mosfets. This amp uses feedback.
-Chris
I really think there are limits here, past which you don't get any improvement unless you are running with no feedback. That's pretty ugly with mosfets. This amp uses feedback.
-Chris
to me. more like how efficient the heat sinks get rid of the heat....
200mA at 70 volt rail is 14 watts, that is really hot....
and if you have six of them in an amp, then 84 watts,
the question is, are the heatsinks big enough?
but if you can get away with lower idle bias and still get the sound decent to your ears,
i am all for it...
200mA at 70 volt rail is 14 watts, that is really hot....
and if you have six of them in an amp, then 84 watts,
the question is, are the heatsinks big enough?
but if you can get away with lower idle bias and still get the sound decent to your ears,
i am all for it...
Hi AJT,
Those radiators must be either massive, or forced air cooled. If not, it turns into a "Hibachi" (cheap little charcoal grille sold in Canada).
With the Symasym, I got the lowest distortion with about 5 mA of bias current. Ran it at 20 just 'cause. But, it was utterly pointless to run it any higher. This is one of the reasons I rather run BJT outputs. They are more linear and can use much lower standing current. Too each his own. My FETs are where they sound the best. In the diff pair.
-Chris
Those radiators must be either massive, or forced air cooled. If not, it turns into a "Hibachi" (cheap little charcoal grille sold in Canada).
With the Symasym, I got the lowest distortion with about 5 mA of bias current. Ran it at 20 just 'cause. But, it was utterly pointless to run it any higher. This is one of the reasons I rather run BJT outputs. They are more linear and can use much lower standing current. Too each his own. My FETs are where they sound the best. In the diff pair.
-Chris
of course Chris, and i will be eager to hear from RCruz, about idle sink temperatures,
the think with mosftets are that they are less prone the second breakdowns due to high temps for as long as the devices are run within SOA...
the think with mosftets are that they are less prone the second breakdowns due to high temps for as long as the devices are run within SOA...
Hi AJT,
Yes, that is true. But this amplifier doesn't have that problem. I guess the use of mosfets is in some way romantic. They can sound good, but the amount of effort is probably the same (but different) for BJTs.
Right, let's hear from RCruz!
-Chris
Yes, that is true. But this amplifier doesn't have that problem. I guess the use of mosfets is in some way romantic. They can sound good, but the amount of effort is probably the same (but different) for BJTs.
Right, let's hear from RCruz!
-Chris
I am using a big 4U enclosure for each amp.
Temperature on each heatsink is 47° Celsius and the ambient temp is 18°C.
The TX is a 650VA for each mono amp.
When cold, bias current on each device is around 240mA going down to 200mA when hot. Tempco is negative at these currents so I do not have to worry about runaway and I can use a purely resistive bias spreader (more linear than the vbe multiplier).
Running the amps at 150mA does not sound as good (slightly less fluid in the highs) ... with 150mA the TX hum less but are not completely silent.
I guess I need better transformers.
Now connected to a modified Salas DCG3, the sound is really good with an outstanding combination of bass detail and snap, very natural and expressive mids and good extended fast highs.
Temperature on each heatsink is 47° Celsius and the ambient temp is 18°C.
The TX is a 650VA for each mono amp.
When cold, bias current on each device is around 240mA going down to 200mA when hot. Tempco is negative at these currents so I do not have to worry about runaway and I can use a purely resistive bias spreader (more linear than the vbe multiplier).
Running the amps at 150mA does not sound as good (slightly less fluid in the highs) ... with 150mA the TX hum less but are not completely silent.
I guess I need better transformers.
Now connected to a modified Salas DCG3, the sound is really good with an outstanding combination of bass detail and snap, very natural and expressive mids and good extended fast highs.
i mean, my ambient temp right now is 32*C so that will mean a sink temp of 61*C....
may need active cooling.....
may need active cooling.....
But if you use transistors good for 150°C, maybe they can stand 61°C in the heatsink... or is it not safe ?
Just add up the delta T for each Rth.
DeltaTsink = Rth s-a * Power * de-rating factor
DeltaTc/device = Rth c-s * Power
DeltaTj/junction = Rth j-s * Power
Add them to Ta and you arrive at a prediction for Tj.
If it's over 100degrees C, then it's too hot.
If it's below 60degreesC, then it does not need further checking. It will last forever.
If it's between 60 and 100degreesC, then it is worth checking the device power and SOA de-rating and comparing the resultant to the device stresses when driving a reactive load.
Lateral mosFETs are to some extent self protecting. When they get hot, or very hot, their parameters change to reduce the currents flowing and they don't blow up. Quite different from BJTs and vertical mosFETs. There was a report a few years ago from a Member who tried to blow up his latFET and it just protected itself.
DeltaTsink = Rth s-a * Power * de-rating factor
DeltaTc/device = Rth c-s * Power
DeltaTj/junction = Rth j-s * Power
Add them to Ta and you arrive at a prediction for Tj.
If it's over 100degrees C, then it's too hot.
If it's below 60degreesC, then it does not need further checking. It will last forever.
If it's between 60 and 100degreesC, then it is worth checking the device power and SOA de-rating and comparing the resultant to the device stresses when driving a reactive load.
Lateral mosFETs are to some extent self protecting. When they get hot, or very hot, their parameters change to reduce the currents flowing and they don't blow up. Quite different from BJTs and vertical mosFETs. There was a report a few years ago from a Member who tried to blow up his latFET and it just protected itself.
Thank you andrew. These lateral mosfets are really rugged and they tend to self protect. I use an electronic temperature sensor and never measured temp over 47 degrees celcius even while driving the amps hard but i will do the math
do the sums to predict the maximum Tj/Tc when in "summer" temperatures.
If Tj is still <60degreesC, then the effect on SOA is minimal.
Derating factor for Tj=60degreesC the DF = {Tjmax-Tj}/{Tjmax-T25} = {150-50}/{150-25}= 100/125 = 0.8 (some devices have Tjmax= 175C or 200C)
Your 120W devices need to be de-rated to 96W, when Tc=50C
Jan.D's and my SOAR calculations take account of elevated Tj/Tc
Last edited:
Let me see if I got it right.
Max power dissipation for each device 125W derated to 96W
Rth = (150°C - 25°C) / 96W = 1.04°C/W
Max average dissipation per devise when driving a 4ohm load under clipping = 30W
Max. case temperature when dissipating 30W = 150 − (30 x 1.04) =118°C (approx)
Thermal resistance Rth c-hs*between case and heat-sink (allowing for mica washer) = 2°C/W
Max. heat-sink temperature = 118 - (30 x 2) =58.7°C
To reach ambient air temperature = 25°C Thermal resistance of heat-sink must be better than (58.7 − 25) / 30 =1.125°C/W
Can these values be correct ?
Max power dissipation for each device 125W derated to 96W
Rth = (150°C - 25°C) / 96W = 1.04°C/W
Max average dissipation per devise when driving a 4ohm load under clipping = 30W
Max. case temperature when dissipating 30W = 150 − (30 x 1.04) =118°C (approx)
Thermal resistance Rth c-hs*between case and heat-sink (allowing for mica washer) = 2°C/W
Max. heat-sink temperature = 118 - (30 x 2) =58.7°C
To reach ambient air temperature = 25°C Thermal resistance of heat-sink must be better than (58.7 − 25) / 30 =1.125°C/W
Can these values be correct ?
One thing to consider beyond immediate failure. The failure rate doubles for each 10°C rise over the figures from the factory. That would be about 25°C for most devices. Then you have to look at the internal case temperature and apply those same factors to all the other components - like your capacitors. The filter capacitors will need to have their heat output taken into account as well.
What this means for you is a much higher service load with your stereo system.
-Chris
What this means for you is a much higher service load with your stereo system.
-Chris
Rth j-c does not change with de-rating. Rth j-c is determined by the physical size of the device backplate. It can be determined from the device data.
Tj= Tjmax and Tc = 25degrees (the starting point for all manufacturers) leaves the delta T = 125Cdegrees.
The power to be dissipated is Pmax = 125W
Therefore Rth j-c = 125C/125W giving 1C/W
I usually work from Ta @ highest summer temperature up to the Tc value after the heat has passed through the sink to ambient , Rth s-a (de-rated), sink to device Rth c-s and junction to case Rth j-c, to find the worst case average junction and case temperatures for the sink I propose to use. I then add on an extra 5Cdegrees to cover for one or more devices being slightly hotter than the average.