How to test a new amplifier design

Well, the pcb is on revision 2 but I'm really pleased to see it belting out 36Vrms into a dummy load and not breaking a sweat (I was). I thought now's the time to over test it. The pcb is expendable as I have plenty more for the batch I need. Gave it 15% over volts on the supplies and wound the frequency up to see where it would start to fall off. 50k... 60k... 75k.. Hang on, what's that smell?. Looked round to see what resembled a volcano coming from the pcb, so quickly killed everything (by that point 80kHz @ a tad under 40Vrms). The offending component was a 10ohm 3W 2512 surface mount resistor. Then it dawned on me. That resistor is in series with a 100nF C0G cap as an RF filter on the output... and I just put a lot of RF through it... I had stupidly forgot about it when I wound the frequency up at full power. The most amazing part is that it still measures 9.3 ohms. I tentatively restored power and the amp was like nothing had happened. There's no better way to get confidence in a new design than to over-stress it and then soak test. Just remember that output filter...
 

wg_ski

Member
2007-10-10 5:21 pm
Variac is a double edge sword for stability testing. Lower voltage does prevent damage, but I’ve seen far too many amplifiers that are perfectly stable with +/-30 volts applied at go horribly unstable when you get it up to operating voltage. Oscillations can suddenly switch on. Dim bulb plus a variac is the only foolproof way.
 
Variac is a double edge sword for stability testing. Lower voltage does prevent damage, but I’ve seen far too many amplifiers that are perfectly stable with +/-30 volts applied at go horribly unstable when you get it up to operating voltage. Oscillations can suddenly switch on. Dim bulb plus a variac is the only foolproof way.
Great idea... I like that a l o t. On a 100 watt amp - like a H220, what wattage bulb, with the added variable
the band-aids, band-aid [protecting the protection, lols not really an accurate comment]
 

wg_ski

Member
2007-10-10 5:21 pm
A single 100 watt bulb is usually enough, unless the amp under test runs at an unusually high bias. Since you can go above 120V with a variac, it lets you dial it up till you get 120V applied to the amp even with the bulb glowing a bit. Only then can you really set the bias properly - at low voltage it’s almost always down. With a large class AB tube amp, I might need as many as 3 100 watt bulbs to get full voltage. A Krell running at 250 mA or more might require the same. Even with an amp that needs 300 watts of ballast to bias, you’d still start with 100 to weed out the gross faults first.
 
I recently pulled 2 ceramic light fixtures out of a box, in storage... to use this tried and true method...
As silly as it may sound, thought about attaching directly to amps chassis and/or with thermistors
for front end AC.

I routinely used light bulb on my Marantz 8B ..years.. ago, to protect it [with stiff SS rectification].
I cannot / could not remember the wattage

Thanks for the refresh and added perspective(s) wg_ski, lending to ideas of how they may be used
on a few /modeling test amp chassis(s) I am working on...