If you dissipate a large amount of power in a device, it will usually heat up way faster than the heat sink. That's normal, especially under fault conditions. You should check your heat sink flatness and be sure your devices are really in good thermal contact. A Sil-pad should show a slight indentation all around when removed- if you see contact only at one end, something's wrong. Also, be sure to observe torque specs for devices- most people way over tighten them.
It seems like there are only a few things that could be wrong here. Either it's oscillating (try waving your AM radio near it), the bias is going out of control, or some component is out and out faulty or installed wrong.
BTW, my two favorite movies of all time are probably 5th Element and 12 Monkeys. Love my sci-fi.
It seems like there are only a few things that could be wrong here. Either it's oscillating (try waving your AM radio near it), the bias is going out of control, or some component is out and out faulty or installed wrong.
BTW, my two favorite movies of all time are probably 5th Element and 12 Monkeys. Love my sci-fi.
Well. The only issue I have at the moment is I have one multimeter 
So I cannot measure output voltage and current at the same time.
Anyway.
When idling with the bias turned as low as possible the meter just shows 0.01amps (its set to measure 10amps max and 0.01 is the lowest it can display in this setting) this seems normal to me, a very small current draw.
Turning the volume up so we start getting some sound and 0.1amps are drawn the volume is quiet but easy to notice.
If I then slowly increase the volume the current increases steadily with no sudden increases or jumps. Just before I reach 3amps (the rails are running slightly high at around 63 volts) the limiting factor kicks in and there is a decrease in the current flow.
So I cannot measure output voltage and current at the same time. Anyway.
When idling with the bias turned as low as possible the meter just shows 0.01amps (its set to measure 10amps max and 0.01 is the lowest it can display in this setting) this seems normal to me, a very small current draw.
Turning the volume up so we start getting some sound and 0.1amps are drawn the volume is quiet but easy to notice.
If I then slowly increase the volume the current increases steadily with no sudden increases or jumps. Just before I reach 3amps (the rails are running slightly high at around 63 volts) the limiting factor kicks in and there is a decrease in the current flow.
The measurement guy at Stereophile does and says this.
What exactly does he mean?
And does this explain why the amps are getting very hot for what is relatively low amount of power?
I was running them for a while at 1.4amps into an 8ohm load (38hz on the XLS). The heatsinks I could touch for a number of seconds before moving away, and the tip of my finger onto a FET wanted to be removed after about a 1second.
So the FETs are not blindingly hot to the touch, but certainly beyond the level of comfort.
I assume drawing 1.4Amps running at 63 volts corresponds to 88 watts total per rail being dissipated. If we say the amp is 70% efficient we're getting 61.6 watts per rail as sound and 27 watts per rail as heat. Am I right in assuming that would = 122 watts rms output?
If the heatsinks are 0.8 degrees per watt. the temperature gauge says its 25 degrees in here, lets round up and with 30 watts of heat going into them we'd have had a temperature increase of 24 degrees = roughly 50 degrees centigrade. Now the FETs are mounted towards one end of the heatsinks and one end feels slightly hotter then the other. But an overall of around 50 degrees sounds about a right based on how hot they feel.
What exactly does he mean?
And does this explain why the amps are getting very hot for what is relatively low amount of power?
I was running them for a while at 1.4amps into an 8ohm load (38hz on the XLS). The heatsinks I could touch for a number of seconds before moving away, and the tip of my finger onto a FET wanted to be removed after about a 1second.
So the FETs are not blindingly hot to the touch, but certainly beyond the level of comfort.
I assume drawing 1.4Amps running at 63 volts corresponds to 88 watts total per rail being dissipated. If we say the amp is 70% efficient we're getting 61.6 watts per rail as sound and 27 watts per rail as heat. Am I right in assuming that would = 122 watts rms output?
If the heatsinks are 0.8 degrees per watt. the temperature gauge says its 25 degrees in here, lets round up and with 30 watts of heat going into them we'd have had a temperature increase of 24 degrees = roughly 50 degrees centigrade. Now the FETs are mounted towards one end of the heatsinks and one end feels slightly hotter then the other. But an overall of around 50 degrees sounds about a right based on how hot they feel.
I did replace the thermal pads.
Either way I'm getting clipping at 2.9 amps 8 ohms and clipping at 3.3 amps into 4 ohms.
3.3amps + 63 volts = 207 watts per rail
so 414 watts @ both rails. Given 70% efficiency that's 290 watts at the output.
What surprises me, even with all this the XLS in an open baffle, well I'd expect em to really run out of excursion, but they don't.
From my point of view, it seems what is really limiting the output is the internal protection in the FETS from things getting too hot.
I have not biased anything yet, I guess that comes next.
Either way I'm getting clipping at 2.9 amps 8 ohms and clipping at 3.3 amps into 4 ohms.
3.3amps + 63 volts = 207 watts per rail
so 414 watts @ both rails. Given 70% efficiency that's 290 watts at the output.What surprises me, even with all this the XLS in an open baffle, well I'd expect em to really run out of excursion, but they don't.
From my point of view, it seems what is really limiting the output is the internal protection in the FETS from things getting too hot.
I have not biased anything yet, I guess that comes next.
Hi.
I really don't know what to suggest now to you. If you have the offset correctly set at Zero volts with P1 which I assume is R13 on the circuit at the start then that adjustment is finalised. The bias current should not interact with this, if it does it's poor design.
As Conrad say's, it is suprising sometimes how quickly the case of a transistor heats up relative to the heatsink.
The bit about "thermal stresses", in a class B amp. Worst case dissipation for the output stage is not at full output but at some intermediate level, the exact value of which I can't remember from memory. A true Class A output runs hottest at Zero output and runs cooler the louder you play it.
To get an idea of heat, with no load and signal, if you increase the bias voltage to give approx 0.15 volts dc over each output resistor, that's 0.5 Amp flowing in each, and at 60 volts dc supply ( I am losing track of this, you do mean +/- 60 ie 120 total ) thats 30 watts dissipation per transistor. How many transistors have you on each heatsink ? An amp of this rating should be able to run all day like this, without the heatsinks being to hot to touch.
I really don't know what to suggest now to you. If you have the offset correctly set at Zero volts with P1 which I assume is R13 on the circuit at the start then that adjustment is finalised. The bias current should not interact with this, if it does it's poor design.
As Conrad say's, it is suprising sometimes how quickly the case of a transistor heats up relative to the heatsink.
The bit about "thermal stresses", in a class B amp. Worst case dissipation for the output stage is not at full output but at some intermediate level, the exact value of which I can't remember from memory. A true Class A output runs hottest at Zero output and runs cooler the louder you play it.
To get an idea of heat, with no load and signal, if you increase the bias voltage to give approx 0.15 volts dc over each output resistor, that's 0.5 Amp flowing in each, and at 60 volts dc supply ( I am losing track of this, you do mean +/- 60 ie 120 total ) thats 30 watts dissipation per transistor. How many transistors have you on each heatsink ? An amp of this rating should be able to run all day like this, without the heatsinks being to hot to touch.
Well this is quite odd I must say, I have no idea why they interact and from my point of view it hints towards there being a problem somewhere else. Now the thing that puzzles me is how John managed to simulate this perfectly in Multisim, yet it doesn't work quite as expected with Circuit maker. Slone uses Multisim and has written in his book with photos that he has built and used this design. If he encountered zero problems with it, I am still wondering if there is more of an error somewhere then simply trying to fix it as it is.
First I can't set the bias that high without altering the value of P1 (which yes is R13), but it is doable. Yes we are talking 60V per rail, the rails run a bit higher then this at 63V.
I have 2 FETs per sink along with the predrivers.
30 watts per FET = 60 being thrown through the sinks. The sinks I'm using are 0.8 degree/watt. This would indicate an increase of 48 degrees, coupled with the room temperature of around 25 degrees, that's a total of approximately 73 degrees which is too hot to touch, even before I've tried measuring it.
What I said before however, about getting 3.3amps on the negative rail, whilst driving a 4ohm load before clipping started. Isn't this enough? 3.3amps @ 63V = 208 watts and this is, I am assuming per rail, so the output total would be 416 watts. I had assumed 70% efficiency, but the figure really doesn't matter too much, if it's between 60 and 70%, the watts output to the speaker in a 4ohm load with 3.3amps being drawn at each rail would yield between 250 and 290 watts.
no,5th element said:
the maximum output power into a resistive load from a sinusoidal waveform with 3.3Apk flowing to a 4r0 load is
P=Ipk*Ipk*Rload/2 = 3.3^2 * 4 / 2 = 21.8W.
3.3Apk across 4r0 develops 13.2Vpk.
maxPower = Vpk * Vpk / Rload / 2 = 13.2^2 / 4 / 2 = 21.8W.
Here was me thinking you said you didn't have time to learn.
AndrewT said:
no,
the maximum output power into a resistive load from a sinusoidal waveform with 3.3Apk flowing to a 4r0 load is
P=Ipk*Ipk*Rload/2 = 3.3^2 * 4 / 2 = 21.8W.
3.3Apk across 4r0 develops 13.2Vpk.
maxPower = Vpk * Vpk / Rload / 2 = 13.2^2 / 4 / 2 = 21.8W.
Here was me thinking you said you didn't have time to learn.
I think you misunderstood. 3.3amps is being drawn by the negative power rail of the amplifier.
In a ClassAB amplifier, when delivering more output current than the stage is biased for then all the rail current flows to the load and the other half of the output pair is effectively cutoff/non operational.
If the output current is 3.3Apk then the rail currents alternately flow 3.3Apk.
If you really do want to learn then go and read the experts. Start with Nelson Pass, move on to ESP, finishing with Leach.
If the output current is 3.3Apk then the rail currents alternately flow 3.3Apk.
If you really do want to learn then go and read the experts. Start with Nelson Pass, move on to ESP, finishing with Leach.
Mooly said:
I managed this and after about 30 mins of running, the temperature seemed to have levelled out somewhat.
The sinks were hot, but hadn't reached over 70 degrees, the FETs were running just below 100.
The heatsinks I'm using are the same size to the ones in Slones picture of the amps.
Slone has a design in his book, similar to the one I've built, but just a little less in output power. He states for that one, 32 watts being dissipated per device is a worst case scenario, so pushing 30+ watts here can't be far from that point with my amps either. The heatsinks certainly appear to be decent enough depending on who you are and what you consider to be a safe maximum heatsink temperature. I'm not reaching 70 degrees with 30+ watts.
I notice you say you have the drivers on the heatsink. Is this specifically mentioned in the construction notes as lateral FET's require no thermal compensation.
The other thing that keeps haunting me is that you are using an unmounted speaker as a load ( Have I got that right ). It's only meaningful to make measurements with a defined load- usually either 8 or in your case 4 ohms resistive. Which for 250 watts takes some doing, electric fire element in a bucket comes to mind (Really )
With the protection circuit out of the way is it loud enough or are there still distortion problems.
Going back to power again 250 watts RMS, thats 8 amps in the load (RMS).
At the risk of you saying I have'nt studied all the posts whats actually the problem now the protection isn't operating. Is it just the heat issue.
The other thing that keeps haunting me is that you are using an unmounted speaker as a load ( Have I got that right ). It's only meaningful to make measurements with a defined load- usually either 8 or in your case 4 ohms resistive. Which for 250 watts takes some doing, electric fire element in a bucket comes to mind (Really
) With the protection circuit out of the way is it loud enough or are there still distortion problems.
Going back to power again 250 watts RMS, thats 8 amps in the load (RMS).
At the risk of you saying I have'nt studied all the posts whats actually the problem now the protection isn't operating. Is it just the heat issue.
your calculations based on 25Vac and 4r0 load are correct.5th element said:
So taking your answer of 6.25Aac (=6.25Arms) then the peak current into the load is 8.84Apk.
Where does this come from?
It can't be coming from the supply rails since you insist you have measured 3.3Amps. Or have you misinformed us?
A heatsink temperature of 70degC and a device temperature of just under 100degC is asking for the grey smoke.
The temperature derated SOAR is reduced by 60% when Tc=100deg for a 150degC device. i.e. a 150W device is reduced to just 60W and with supply rails around 60Vdc the maximum current is reduced to around 1A per device. If secondary breakdown limitations are also brought into the calculation then the permissible device current is much lower for 60Vce (maybe another 10% to 30% less).
If the peak current is a one off non repetitive event then the 100mS SOAR current can rise considerably, maybe two to three times the DC value. The 10mS current can go even higher.
But you are measuring these currents & voltages after bringing the output stage upto temperature over a period of minutes. 1kHz test signal over 5minutes brings each device to it's peak current about 300000times. A long way above the single short term event.
The temperature derated SOAR is reduced by 60% when Tc=100deg for a 150degC device. i.e. a 150W device is reduced to just 60W and with supply rails around 60Vdc the maximum current is reduced to around 1A per device. If secondary breakdown limitations are also brought into the calculation then the permissible device current is much lower for 60Vce (maybe another 10% to 30% less).
If the peak current is a one off non repetitive event then the 100mS SOAR current can rise considerably, maybe two to three times the DC value. The 10mS current can go even higher.
But you are measuring these currents & voltages after bringing the output stage upto temperature over a period of minutes. 1kHz test signal over 5minutes brings each device to it's peak current about 300000times. A long way above the single short term event.
AndrewT said:
This is puzzling to say the least. The only variable in all of this that has no certainty is the loudspeakers themselves.
I do not have a dummy load for any testing so I am improvising the best I can with the loudspeaker.
I connected the multimeter in series with the negative power rail, set up the system so that the meter measured 3.3Amps, the level that clipping starts, then turned off the signal.
I then removed the multimeter and connected it to the output, returning the signal at exactly the same level. The output measured 25Vrms.
The only thing that's off here could be the load.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.