26-11-2016, 12:04 PM
Ah - 7915. I did wonder if that was what you meant originally - post #5 said "7815", implying it was the +15V rail missing, which made no sense because it's the +15V rail that supplies the LED. Confusion over!
However, there is still a problem.
For the 78/79 series of regulators, it's not the differential that's specified. Rather, it's the absolute input voltage. You're probably thinking about the LM317/337 floating regulators here.
So, the absolute maximum is usually 35V for these. As you suggest, it's a bit lower for the 5V variants, but having just checked the part you're using, plus the LM7915 from Texas, it's 35V maximum.
A 24V secondary winding results in a DC voltage that is rather too close to 35V for my liking. I would not be happy with that.
There are many ways to drop the input voltage into the regulator ICs, and the method I'd pick would depend on several factors, including physical layout and how constant the current draw is (and what that current actually is). Perhaps the crudest method is a couple of resistors in series with each regulator, but that requires a reasonably constant current draw, and the signal and clip LEDs might cause rather more fluctuation than is ideal - you'd have to measure it. A decent 5V or 10V Zener diode is another quick and dirty method which works well if the current it reasonably low. But both of those involve cutting tracks, unfortunately. Resistors can be put in series with the secondary windings, of course.
Where the differential voltage is important, however, is when considering the power dissipated by the regulators. Presumably they are mounted on some form of heat sink already? Perhaps not? Either way, the original designer would have decided on the size of the heat sinks (or decided he could get away without them) based on how much power each regulator produced. One factor here being the differential voltage (about 5 volts, given 15V secondary windings). Now, you have about 20V being dropped across them, so they are generating 4 times as much power. In theory, those regulator are protected against overheating. In practice, overheating them will kill them sooner or later! So at the very least, you need to check how warm they are, but you might need to improve the heat sinking arrangements for them.
Finally,
At first glance, there is no connection between idle current and power output. I can only assume that a high idle current is causing the PSU to sag, preventing the amplifier reaching full swing despite using a sine wave with a small PMR. I'd need to double-check a couple of things before saying more because there is slightly more to this than first meets the eye.
Either way, ending up at 300mV is a lovely coincidence given the incredibly over-simplified way I arrived at that number. I don't pretend it's correct, but I believe it should be in the right ball-park. But ultimately, if the heat sinks aren't running too hot, and the distortion is below what you can measure, then that's probably good enough.
However,
Did I read elsewhere that the new transformers have higher main outputs too? This means that they should give more output power than the specification, but don't because they are sagging under load. That will lead to a lot of heating within the windings (I-squared-R loss). In practice, with real music that has a high PMR, it'll be fine - and the peak powers will most likely be slightly higher than the original transformers managed - with with sustained sine-wave testing, you run the risk of over-heating them, which might lead to shorted turns, etc. So personally, I wouldn't run at full power for long - that represent the most stress for the PSU, and an unrealistic amount of stress unless you wish to use the amplifier for industrial applications. Indeed, many commercial amplifiers have deliberately under-sized transformers for exactly this reason. Instead, test at about 170W (2/3 of output power), which is where the output devices produce most heat - that's the thing we're really interested in.
We're getting there
Mark
However, there is still a problem.
For the 78/79 series of regulators, it's not the differential that's specified. Rather, it's the absolute input voltage. You're probably thinking about the LM317/337 floating regulators here.
So, the absolute maximum is usually 35V for these. As you suggest, it's a bit lower for the 5V variants, but having just checked the part you're using, plus the LM7915 from Texas, it's 35V maximum.
A 24V secondary winding results in a DC voltage that is rather too close to 35V for my liking. I would not be happy with that.
There are many ways to drop the input voltage into the regulator ICs, and the method I'd pick would depend on several factors, including physical layout and how constant the current draw is (and what that current actually is). Perhaps the crudest method is a couple of resistors in series with each regulator, but that requires a reasonably constant current draw, and the signal and clip LEDs might cause rather more fluctuation than is ideal - you'd have to measure it. A decent 5V or 10V Zener diode is another quick and dirty method which works well if the current it reasonably low. But both of those involve cutting tracks, unfortunately. Resistors can be put in series with the secondary windings, of course.
Where the differential voltage is important, however, is when considering the power dissipated by the regulators. Presumably they are mounted on some form of heat sink already? Perhaps not? Either way, the original designer would have decided on the size of the heat sinks (or decided he could get away without them) based on how much power each regulator produced. One factor here being the differential voltage (about 5 volts, given 15V secondary windings). Now, you have about 20V being dropped across them, so they are generating 4 times as much power. In theory, those regulator are protected against overheating. In practice, overheating them will kill them sooner or later! So at the very least, you need to check how warm they are, but you might need to improve the heat sinking arrangements for them.
Finally,
At first glance, there is no connection between idle current and power output. I can only assume that a high idle current is causing the PSU to sag, preventing the amplifier reaching full swing despite using a sine wave with a small PMR. I'd need to double-check a couple of things before saying more because there is slightly more to this than first meets the eye.
Either way, ending up at 300mV is a lovely coincidence given the incredibly over-simplified way I arrived at that number. I don't pretend it's correct, but I believe it should be in the right ball-park. But ultimately, if the heat sinks aren't running too hot, and the distortion is below what you can measure, then that's probably good enough.
However,
Did I read elsewhere that the new transformers have higher main outputs too? This means that they should give more output power than the specification, but don't because they are sagging under load. That will lead to a lot of heating within the windings (I-squared-R loss). In practice, with real music that has a high PMR, it'll be fine - and the peak powers will most likely be slightly higher than the original transformers managed - with with sustained sine-wave testing, you run the risk of over-heating them, which might lead to shorted turns, etc. So personally, I wouldn't run at full power for long - that represent the most stress for the PSU, and an unrealistic amount of stress unless you wish to use the amplifier for industrial applications. Indeed, many commercial amplifiers have deliberately under-sized transformers for exactly this reason. Instead, test at about 170W (2/3 of output power), which is where the output devices produce most heat - that's the thing we're really interested in.
We're getting there
Mark
