Testing a power amp

[AndrewT]

Hi,
I am putting some numbers together to get a protection circuit for the Leach Clone and the Krell Klone.
Most of the resistor setting depend on the PSU voltage or the output Re and load.

BUT the Vbe voltage of the protection transistor is critical to the final value chosen for the resistor across the BE leads.

It appears that the activation voltage needs to be measured after the amp is put to work.

How can the output current be checked? for the two conditions below;
1. Maximum current into a short circuit or capacitor with protection disabled.
2. Maximum output current to activate the protection circuit either into a specified non reactive load or into a short circuit.

[JensRasmussen]

Regarding SOA protection of the output transistors here is how I plan to go about the calculation of the resistor values needed.

1 Choose your output transistors and number of parallel devises.
2 Choose your rail voltage
3 Decide on maximum case temperature of the output devises and use this to derate the SOA curve
4 Draw the derated SOA curve on dual logarithmic paper
5 Derive activation formula for the protection circuit as a function describing Vce as a function of Ic for one transistor
6 Choose components in the protection circuit and fiddle around with values
7 Draw the protection curve together with the SOA curve and evaluate
8 If you are happy proceed to 9 if not goto 6
9 Listen to the amp at high levels into low impedance speakers
10 If you are happy goto 11 if not goto 6
11 Make a short circuit test of the amp (While wearing safety goggles, helmet and hearing protector)
12 If the amp survives 9 and 11 you are done if not goto 6 and then change some transistors and resistors

EDIT: To test the function of the protection circuit measure the voltage across the BE of the protection transistor while also measuring the output stage CE voltage and C current. Plot some different points in the SOA graph and make sure the points does not exceed the derated SOA graph.

Use the BE voltage to ensure your calculations are correct.

\Jens

[Nelson Pass]

You have a choice - you can sim or calculate it,
or you can perform the measurement. Both
options have their potential hazards.

[amplifierguru]

Hi Andrew.

"BUT the Vbe voltage of the protection transistor is critical to the final value chosen for the resistor across the BE leads."

Why should this be the case - it would help to see your design as the Vbe on should be around 0.6-0.65 i.e < 10% variation which should translate to < 10% in output current. If you're subtracting pull up/pull down currents through the sensing node you need to allow enough for definitive turn on/off as otherwise the protection transistors could be partially turning on under normal signal conditions and modulating the loading decreasing gain and increasing THD.

Cheers,
Greg

[sam9]

After taking the datasheet SOA chart and manually derating it, I used a simulation to start with. This makes dumb mistakes and miscalculation cheaper. It also lets you check out some improbable circumstances.

Next I tried a couple of real world tests where the margins seemed smallest. These were heavy clipping combined with low loads ( <2-ohms). While a simple short accross the terminals is the scariest, it seems to be much less uincertain than low load. That's because you are also trying NOT to have the protection circuit active with slightly higher loads.

[AndrewT]

Hi all,
thanks for the replies so far but I must have stated my requirement ambiguously.
I know how to calculate the protection circuit.

I need to know how to measure, non destructively, the ACTUAL AMPLIFIER CURRENT when at the limit, without destroying it.

I seem to remember reading somewhere that a full voltage test (sine or square wave?) into a capacitor and using an oscilloscope can directly read the peak pulse current.

Amplifier Guru,
a 10% reduction in Vbe (650mV to 585mV) gives a short circuit current reduction of 24.2% (in my case from 9.234A to 7.00A). I can readjust the activation current back to 9.2A by reducing the // resistor with very little change to the shape of the current locus.
I want to keep the protection locus as close to the derated SOAR as possible to reduce the possibility of premature activation. I can see a way to achieve this with a dual slope limiter that gives a close approach to SOAR over the important VI range that the speaker load uses. The circuit I am working with is the Leach clone designed by Jens and others in the group.

[anatech]

Hi AndrewT,
Just over build the test circuit to handle the faults.

-Chris

[sam9]

I seem to remember reading somewhere that a full voltage test (sine or square wave?) into a capacitor and using an oscilloscope can directly read the peak pulse current.

This sounds maybe like the method Self suggests to read slew rate at the output. Namely an integrator. I've done that but I'm not sure how it would be adapted for your purpose.

Offhand, Anatech's solution sounds like the only way be sure you get a true reading. If it's an EF output section that shouldn't be too hard. If CFB I despair.

[pinkmouse]

Andrew, that test may be in Ben Duncan's book, I'll have a look for you tonight.