F5 Turbo Builders Thread

either diodes , or pcb , or your DVM
- tried to bias the other channel, with 'unmodified' tea-bag's pcb, with one pair of fets and diodes. Looks the same way - made it up to 485mV across the two 1Ohm resistors, cannot see any thermal runaway (am I looking for a wrong thing?) - the bias is creeping up very slowly - like 1mV per minute or two - climbed to 0.5V in 15-20 minutes. With runaway I would expect more dynamic changes - can anyone comment on this?

One different thing though - this channel is without thermistors yet, but surely the diodes would have started to conduct at 500mV voltage regardless of that? They are not on the main sink - is it possible they are not warm enough to start conducting?
 

Attachments

  • Picture 57.jpg
    Picture 57.jpg
    558.7 KB · Views: 439
check your diode datasheet.
It will show what diode current will pass at different temperatures and different voltages.

I have suggested that the quiescent Vrs be kept below 400mV to avoid the diodes turning on when there is no output current.
It may be that <300mVrs would be a safer value to avoid thermal instability.
 
The diodes aren't warm enough. The datasheet shows low Vf characteristics only at elevated temperatures. Depending on your ambient temperatures, you may need to be a bit above 500mV to see the diodes conducting. At this level if you have more than one pair, it is possible that one pair kicks in earlier than the other, so you need to monitor every single device.

http://www.vishay.com/docs/94080/vs-mur3020w.pdf

It doesn't matter anyway. Regardless of your idle bias setting, they will start conducting just above an ampere (assuming the 25 degree spec of ~0.55V), which is what the original design intent was.

It's best if it doesn't conduct until that point or else the results may be quite spectacular... I've a build (not mine!!) arrived recently that I've to bring back from the dead with a blown N-JFET, four blown MOSFETs and a PCB burnt to a crisp by the feedback resistors cooking themselves and every surrounding component. Plus about ten resistors blown open so you can see the helix.
 
Hi
I am starting to order the parts to build 2 x mono blocks with 8pr output devices each . I can get 1080VA toroid transformers with dual 36V secondary @15A each from a local supplier . My question is will this be enough to drive my <1 ohm Apogee Scintilla speakers?
Thanks
Andrew
 
you're limited with dissipation , but not exactly to 8 pairs

with 5U/400 you can easily put 10 pairs per side

with 5U/500 , even more


Really?! 10 pairs is 20 Mosfets plus another 10 diodes on 400mm long heatsink!
I think thats way too much! At least not without forced air.

One builder here put 4 pair per side (for stereo) on 4, 10" x10" Heatsink USA heatsinks and ended up blowing the outputs when he "inadvertantly " cranked the volume up in a lets say inebriated state 😀. (note the speakers he had went down to 2 ohm)

I would say to be safe use 4 or 6 pr on a 5U 500 chassis and use a bigger transformer. ie 1.5 - 2kva
 
more pairs equals more sharing.
The reduced current through each device increases the Factor of Safety against failure.

The heatsink temperature is determined by the TOTAL dissipation and the duration of that dissipation.
5U backplate area is enormous. You could double row the outputs and probably fit 16pairs along with their diodes on that area.