I recently bought a used 4bsst and noticed immediately that one heatsink channel was much warmer than the other. I checked the bias using the procedure here:
http://www.bryston.com/BrystonSite05/pdfs/SSTAmplifiers/SST-BIASING-INSTRUCTIONS.pdf
I tweaked both channels to get them within 1 mV of the 25 mV value when fully warmed up. Even so, one heatsink is still warmer than the other, not dramatically, but enough to feel the difference.
Any ideas on a reason for this? Does it even matter?
Thanks,
Jim
http://www.bryston.com/BrystonSite05/pdfs/SSTAmplifiers/SST-BIASING-INSTRUCTIONS.pdf
I tweaked both channels to get them within 1 mV of the 25 mV value when fully warmed up. Even so, one heatsink is still warmer than the other, not dramatically, but enough to feel the difference.
Any ideas on a reason for this? Does it even matter?
Thanks,
Jim
Hi,
can you check the temperature of the individual transistors mounted on the heatsinks?
The cooler heatsink may have 1 or more transistors running excessively hot.
can you check the temperature of the individual transistors mounted on the heatsinks?
The cooler heatsink may have 1 or more transistors running excessively hot.
Not sure I can get to them. I'll take a look. Are you thinking there could be 1 or more badly matched pairs?
I did some THD tests over the weekend and both channels look good for a 20-20KHz sweep (THD < 0.002%). I didn't do high power tests though, only a few watts.
I did some THD tests over the weekend and both channels look good for a 20-20KHz sweep (THD < 0.002%). I didn't do high power tests though, only a few watts.
no,
I am wondering if the thermal conductance from output device to heatsink has deteriorated with age.
A bad thermal joint will make the heatsink appear cooler and cause severe overheating of the badly connected devices.
I am wondering if the thermal conductance from output device to heatsink has deteriorated with age.
A bad thermal joint will make the heatsink appear cooler and cause severe overheating of the badly connected devices.
Have you checked DC offset on the channels ?
A high Dc offset on one channel will put dc through the speaker and so use more power.
Yea, checked it. It's only a few mV.
I'm wondering if I should just set the bias based upon minimizing distortion, rather than setting to a predetermined bias current.
I'm wondering if I should just set the bias based upon minimizing distortion, rather than setting to a predetermined bias current.
I design my own amps and i use a signal generator and a scope to check for crossover distortion. I give the amps just enough bias to get rid of cross over distortion.
The schematic of the Bryston output stage is here:
Bryston Limited - Music For A Generation
I'm not sure where the test points are taken from on the schematic, but regardless, if one channel has matched BJTs with a higher beta value, wouldn't that channel have a higher bias current with the same bias voltage, thus produce more heat? If so, it's probably more accurate to match heat sink temps than go by voltage. Just an uneducated guess.
Bryston Limited - Music For A Generation
I'm not sure where the test points are taken from on the schematic, but regardless, if one channel has matched BJTs with a higher beta value, wouldn't that channel have a higher bias current with the same bias voltage, thus produce more heat? If so, it's probably more accurate to match heat sink temps than go by voltage. Just an uneducated guess.
no.
The bias current is setup and measured with the amplifier input shorted and open circuit load on the output.
Now look at a typical push/pull output stage.
The top half sends a bias current down.
The bottom half sinks a bias current down.
In between one usuaully has a feedback resistor and a Zobel.
The sum of the currents passing these two components must be tiny, <<10mA
From this, the upper and lower half currents must match fairly closely. As output offset approaches zero mVdc, the two half currents become exactly equal.
The bias current is setup and measured with the amplifier input shorted and open circuit load on the output.
Now look at a typical push/pull output stage.
The top half sends a bias current down.
The bottom half sinks a bias current down.
In between one usuaully has a feedback resistor and a Zobel.
The sum of the currents passing these two components must be tiny, <<10mA
From this, the upper and lower half currents must match fairly closely. As output offset approaches zero mVdc, the two half currents become exactly equal.
Hi Jim,
AndrewTs explanation is the most logical. The voltage drop across the emitter Rs tells you exactly how much bias current you have in relation to the actual value of the resistors. The variables include, naturally, the actual resistor values, and also the rail voltages supplied to each channel. If one channel has higher rails than the other, its output transistors will dissipate more heat.
AndrewTs explanation is the most logical. The voltage drop across the emitter Rs tells you exactly how much bias current you have in relation to the actual value of the resistors. The variables include, naturally, the actual resistor values, and also the rail voltages supplied to each channel. If one channel has higher rails than the other, its output transistors will dissipate more heat.
That makes perfect sense, but you're talking about the operation of a single channel. If the other channel has different BJT betas, the currents through the NPN and PNP could still be matched, just higher. Wrong?
wrong.
If the bias has been set and checked as being similar/same for both channels then bias difference cannot explain the apparent difference in heatsink temperatures.
Similarly extending the hFE argument to the bias setting, cannot influence heatsink temperature.
If the bias has been set and checked as being similar/same for both channels then bias difference cannot explain the apparent difference in heatsink temperatures.
Similarly extending the hFE argument to the bias setting, cannot influence heatsink temperature.
The current cannot be different than the current through the resistors unless there is current leaking to ground somewhere. Bias is a static condition, it is not influenced by other operating parameters. The voltage drop across the transistor along with the current passing through it are the only factors that effect dissipation.
Hey Bill, good to see you on here.
So if the bias voltage between the two channels is matched, than the bias currents must be matched as well, regardless of whether the BJT beta values are different between channels? That confuses me. I'll have to study it more.
Thanks for the help guys.
Edit: Duh, i'm an idiot. The bias voltage is measured at the emittor resistors, so if the voltages match, than the currents must match unless the resistors are different, which is doubtful. I was thinking of the bias at the base.
What about parallel output stages? If the bias test points are taken from one stage, a parallel stage could have higher current, or lower current, if it has different BJT betas. Crazy guess?
So if the bias voltage between the two channels is matched, than the bias currents must be matched as well, regardless of whether the BJT beta values are different between channels? That confuses me. I'll have to study it more.
Thanks for the help guys.
Edit: Duh, i'm an idiot. The bias voltage is measured at the emittor resistors, so if the voltages match, than the currents must match unless the resistors are different, which is doubtful. I was thinking of the bias at the base.
What about parallel output stages? If the bias test points are taken from one stage, a parallel stage could have higher current, or lower current, if it has different BJT betas. Crazy guess?
Last edited:
I'm a constant lurker around here, guys like AndrewT really have the technical chops. He is well worth listening to.
It might come down to airflow differences, or they possibly have more stuff hanging on one heatsink making it work a little harder.
I'd sure like to see their feedback mechanism for bias and offset control, I've never worked on a Bryston.
It might come down to airflow differences, or they possibly have more stuff hanging on one heatsink making it work a little harder.
I'd sure like to see their feedback mechanism for bias and offset control, I've never worked on a Bryston.
The schematic is here:
http://www.bryston.com/BrystonSite05/pdfs/SSTAmplifiers/4b+7bSST-SSB-SCH-6_20041013.pdf
http://www.bryston.com/BrystonSite05/pdfs/SSTAmplifiers/4b+7bSST-SSB-SCH-6_20041013.pdf
Thanks Jim,
A little OT but I finished those Eros spkrs, the midrange is most impressive. I'm wondering if the 1700 x-over point is the key, think I might have to try your 1500 x-over with the Thors.
A little OT but I finished those Eros spkrs, the midrange is most impressive. I'm wondering if the 1700 x-over point is the key, think I might have to try your 1500 x-over with the Thors.
I'm really confused looking at that schematic. I emailed Bryston to see what their take on it is. I also have an old Adcom 5500 and that schematic is so much easier to understand.
Cool, the Eros look nice. I recently did another comparison between the 2 versions. The 1.5 crossover definately sounds cleaner, less boxy than the 2K version. Since I got the Bryston with the really low output impedance, I've had to increase the series cap on the tweeter to 9uF and reduce the series resistor to 3.3 Ohm (on the 1.5k version).
Cool, the Eros look nice. I recently did another comparison between the 2 versions. The 1.5 crossover definately sounds cleaner, less boxy than the 2K version. Since I got the Bryston with the really low output impedance, I've had to increase the series cap on the tweeter to 9uF and reduce the series resistor to 3.3 Ohm (on the 1.5k version).
- Status
- Not open for further replies.