Marantz 4300 Update & question.

I remember saying at one point "I think we need to look at the power supply".

Good job finding the fault in there.

When you say that you see a sawtooth waveform, van you indicate whether it's a small ripple on a much larger DC level, or something more concerning.

Please either provide pictures of your scope's display, or be quantitative (preferably both)

 

There's not much noise in the former, a lit in the latter trace.

This suggests to me that either you've got it at a more sensitive setting in the latter or you've failed to connect the ground lead.

Can you confirm the vertical setting (V/div)

 

What I would be doing right now is to look at the power supply in isolation.

Add simple resistive loads to it and ensure that its performance is normal.

A fault in the amplifier could be causing the apparent noise on the power supply, or it may be the power supply that is causing the noise and the other problems.

Since you've found a fault in the power supply, follow through to check that it's actually fixed and ensure that you haven't replaced a component which was an effect rather than a cause of the problem.

 

The first step is to determine what the output rails of the PSU are.

I'd have to go back and check, but I believe it's probably a +/- 35 volt rails -- measure them when the amp is behaving itself and at low volume.

From the power of the amplifier you can determine a reasonable load to put on the rails. e.g. for a 35V rail, a 70W amplifier should handle a 1A load without problems (it should handle higher, but this should be within the continuous power it can supply)

If disconnecting stuff from the power supply shuts it down, leave things connected but at low volume and no input signal.

Using ohms law, figure out some loads for maybe 0%, 10%, 20%, 50%, and 100% of full load. If you have nothing else then a series of 100W light bulbs in parallel would probably work. Beware that their cold resistance is lower than their resistance when hot.

Load should be between the rail and ground (not across the rails) as that may cause problems in certain fault conditions.

Measure the voltage, current, and look at the waveform at each load.

Don't maintain the load for longer than necessary to measure the performance and stop immediately if anything starts to smoke.

These tests will need to be done with the series lightbulb from the mains removed.

 

What power is your amplifier rated at?

I'm guessing 100W.

I'd use some 100W light bulbs as loads. Several reasons, the exact resistance is not important, and they can definitely dissipate 100W, and they're cheap.

Set up a bulb with a series ammeter and a parallel voltmeter so you can measure both current and voltage. Connect the scope across the power supply.

Either change the bulbs (25W, 50W, 100W, ...) or connect several bulbs in parallel. If you've got 110V mains, expect each 100W bulb to dissipate around 40W at the lower voltage (it's not a trivial calculation because the resistance of the filament is non-linear).

I'm not an absolute specialist at determining the load you should use to simulate a full power test on the amplifier's power supply. There are many variables and the level is strongly influenced by the design shortcuts which may have been made. Following I have listed some of these issues. Proceed with care. If there are fuses in the secondary, that may help indicate a safe load (it will be some fraction of their value -- certainly no more than half of them)

Because you're providing a constant load, you do not want to exceed half the amplifier's rated power on each rail. Amplifiers generally have power supplies capable of less power than you might think. This is based on the general assumption that no music is a constant tone at one level, it is highly dynamic with large peaks and a much lower average. The capacitors (and they're generally large) provide the power for those peaks. Because of these peaks can be quite large, the peak output of the amplifier may well exceed the rated power (it may be 175W for a 100W amplifier) but it may be limited to (say) 4 cycles at 1kHz (1/250th of a second). Another thing is that due to various losses, for 50W of energy output to the speakers, somewhat more than 50W of power from the power supply is required. For all of those reasons, any full load testing of the power supply should be kept short (several seconds) and only attempted if there are no signs of distress at lower loads.

 

That sounds right.

edit: just to be doubly sure, measure the cold resistance of the light bulb filament to make sure it isn't going to grossly overload the power supply before it warms up (which it should do very quickly)

 

OK, now you have determined that the power supply gives you good DC at significant load, reconnect the amplifiers and run your other tests again.

This time, measure the current drawn from the power supply as well as the voltage across it (with a scope too).

If you see something that is not DC, measure the frequency and note the current.

Anything other than a frequency of 2x your mains frequency indicates something odd is happening (that odd thing may be something oscillating)

 

If you have some confidence that the problem causing the output transistors to fry is resolved, then try without the series bulb. But be prepared to replace them again.

An intermediate point may be to use a larger bulb (200W)

You can use the same method to test the other power supplies, but the current you need to draw will be far less (and not a simple value that I can quote either).