I'm building a 7 channel discrete analog amp on two 4"x6" perfboards. I was thinking to put the OPS with heatsinks on one board and the IPS and VAS on the other. To aid construction and reduce errors I am thinking that I want to put all the discrete components in 3/8" columns, long ways, so that if you look across rows you have a row of 7 transistors, all oriented the same, a row of identical resistors, etc. The wiring would be done with precut breadboard wiring that has a different color per length. Like the military expression "the hair on the soap is 'up'".
Now, I know that this approach is likely to have many problems if done naively. I'm thinking (hoping) that choices of which wires to run next to each other, and breaking some runs to use twisted pair or thicker wire back to a star point, can overcome the problems. My question is which things are critical and which are not.
I haven't settled on a schematic yet (and I realize that makes it harder to give concrete advice) but it means I could also optimize the circuit design to be able to operate well in this physical configuration. For instance, if adding a transistor somewhere improves the noise rejection so that the layout becomes less critical, that would save me more trouble than the complexity cost of the extra part.
Now, I know that this approach is likely to have many problems if done naively. I'm thinking (hoping) that choices of which wires to run next to each other, and breaking some runs to use twisted pair or thicker wire back to a star point, can overcome the problems. My question is which things are critical and which are not.
I haven't settled on a schematic yet (and I realize that makes it harder to give concrete advice) but it means I could also optimize the circuit design to be able to operate well in this physical configuration. For instance, if adding a transistor somewhere improves the noise rejection so that the layout becomes less critical, that would save me more trouble than the complexity cost of the extra part.
I've built a point to point wire amp driver, with the output transistors mounted on a heatsink, with the spreader diodes "thermally coupled" that is mounted above the transistors on the heat sink. No perf board over on the heat sink. Point to point saves a lot of the cost of disposing of the PCB etchant, which I suspect is done the wrong way a lot of times and places.
See post 212 of the last page of this thread fir pictures: http://www.diyaudio.com/forums/solid-state/236256-retro-amp-50w-single-supply-20.html
Like bad layout on PCB, it might oscillate. Mine didn't. I did put a maybe excessively careful 8 turn coil in series with the wire coming back from the spreader diode stack to the lower driver.
I felt better putting transistors as little centers of entertainment with the resistors and caps associated all around. Shortens the runs and makes sure none are parallel. I did put all the in and out wires to the heat sink down one edge of the board, so I could put 4" of wire from each and flip the board over to work on both sides without breaking any wires. I used solid core 28 ga kynar insulated wire for the signals (wirewrap wire), and 24 ga stranded teflon wire for the drivers to the output transistors and from the output transistors to the output cap. Burned insulation is a PIT* when soldering all these wires. I made hooks out of a lot of part leads so the wires wouldn't fall off while I was soldering.
I used the schematic above to minimize the number of wires, plus the output cap protects the speakers against DC even if I make a soldering error. Sounds great, I'm listening to it now. The old single ended designs are not really stable thermally, so I put a fan on my heat sinks and hid it behind the organ between the two speakers on columns. I run it maybe 14 hours a day some days. The fans covered the problem when the bias control board for the channel I replaced blew a sense transistor and the output transistors were running 240 ma idle current. This new board retro 50W board doesn't have a bias sense transistor. No damage to the OT's occurred,I just reused them with the new driver board.
Remember long parallel runs between high impedence high gain input wires, and low impedance high current output wires, are invitations to oscillations. Point to Point you have 3 dimensions to play with to avoid that.
Have fun.
See post 212 of the last page of this thread fir pictures: http://www.diyaudio.com/forums/solid-state/236256-retro-amp-50w-single-supply-20.html
Like bad layout on PCB, it might oscillate. Mine didn't. I did put a maybe excessively careful 8 turn coil in series with the wire coming back from the spreader diode stack to the lower driver.
I felt better putting transistors as little centers of entertainment with the resistors and caps associated all around. Shortens the runs and makes sure none are parallel. I did put all the in and out wires to the heat sink down one edge of the board, so I could put 4" of wire from each and flip the board over to work on both sides without breaking any wires. I used solid core 28 ga kynar insulated wire for the signals (wirewrap wire), and 24 ga stranded teflon wire for the drivers to the output transistors and from the output transistors to the output cap. Burned insulation is a PIT* when soldering all these wires. I made hooks out of a lot of part leads so the wires wouldn't fall off while I was soldering.
I used the schematic above to minimize the number of wires, plus the output cap protects the speakers against DC even if I make a soldering error. Sounds great, I'm listening to it now. The old single ended designs are not really stable thermally, so I put a fan on my heat sinks and hid it behind the organ between the two speakers on columns. I run it maybe 14 hours a day some days. The fans covered the problem when the bias control board for the channel I replaced blew a sense transistor and the output transistors were running 240 ma idle current. This new board retro 50W board doesn't have a bias sense transistor. No damage to the OT's occurred,I just reused them with the new driver board.
Remember long parallel runs between high impedence high gain input wires, and low impedance high current output wires, are invitations to oscillations. Point to Point you have 3 dimensions to play with to avoid that.
Have fun.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.