Ok guys, i think i'm just gonna go for it here. I want to use 6 lm3886 chips in a bridge parallel design for a sub amp. So basically i'm gonna follow BPA-200 w/ 3 chips in parallel instead of 2 for a total of 6 bridge parallel. I noticed in the bpa-200 schematic that one parallel set of chips uses 20.5k resistors in the feedback and 21.5k for the other set of parallel chips. I assume that I do need to follow this. What if i just use 10k inputs and 220k feedback instead for both sets? Anyone have a more complete schematic of the opamp input stage? If this for some reason doesn't work, i'll just turn it into a 6 channel lm3886 amp...but I'd like to try the bridge parallel design first...for the heck of it. Besides adequate power supply, .1% .1ohm output resistors, adequate heatsinking, any other advice? Thanks!
-Matthew K. Olson
oops, posted this in the wrong spot, moderator, please move this if need be...thanks!
-Matthew K. Olson
oops, posted this in the wrong spot, moderator, please move this if need be...thanks!
Hi Mattoy5
Have a look at this threads. You might pick up something usefull
http://www.diyaudio.com/forums/showthread.php?s=&threadid=19157
I agre with "unitgain". Why do you want to make the gainresistors bigger ?? It will give you more noise and more DC-ofset. Something you DON't need when you are paralleling chips.
(make sure you take care of the DC-ofset some how...)
You don't have to use the same resistors as in the BPA-200, but you most make sure that both your non-invertes and your inverted section have the same gain as explained by "unitgain" otherwise your "bridging" will be running "strange" to say the least
Have fun
Thomas
Have a look at this threads. You might pick up something usefull
http://www.diyaudio.com/forums/showthread.php?s=&threadid=19157
I agre with "unitgain". Why do you want to make the gainresistors bigger ?? It will give you more noise and more DC-ofset. Something you DON't need when you are paralleling chips.
(make sure you take care of the DC-ofset some how...)
You don't have to use the same resistors as in the BPA-200, but you most make sure that both your non-invertes and your inverted section have the same gain as explained by "unitgain" otherwise your "bridging" will be running "strange" to say the least
Have fun
Thomas
What is an acceptable dc offset ...ie what can I live with per chip ? Obviously zero would be best, but would 20mV be ok per chip? Or do I need to use output couple caps or something ridiculous? I'm going to hand match the feedback and input resistors, and I'm not going to bother w/ the dc servo's, but i'm just wondering what an acceptable dc offset is per chip before I tie them together. Thanks!
-Matthew K. Olson
-Matthew K. Olson
I am sorry, but 20 mV DC off-set is way too much for paralled chips, unless you make the output resistors bigger.
See my replay in
http://www.diyaudio.com/forums/showthread.php?threadid=30570
Have fun
THomas
See my replay in
http://www.diyaudio.com/forums/showthread.php?threadid=30570
Have fun
THomas
What you said in your other post is that if there is a differential of basically >20mV, then that can be trouble. But, if each of the chips are have +20mV offset (say 16mV, 22mV and 25mV), then the differential between chips would be minimal and they would have minimal current flowing between chips. I have a 6 channel lm3875 amp that is pretty darn close to that, so I doubt getting offset that low is difficult if I handmatch all the resistors. Thanks for the thoughts though guys. Helps
-Matthew K. Olson
-Matthew K. Olson
Do not remove the caps, they are the DC decoupling without them your DC offset will cause all outputs to push and pull. I run a 4 parrellel bridge amp. 20mv is to much DC offset, multiple that 20mv across 6 chips, a differance of 10 per chip can cause output differance upwards of an amp per chip then 6 chips 6 amps at 35+- 210 watts of waste. set all chips in a non inverted formatt. 22.1K .1% 1/4 watts percision resistor soldered directly to IC pins run to a 47uf 10% tantalum capacitor then a 1K ohm .1% 1/4 watt to ground. Make sure all pc board set in parrellel are identical also run a cover ground plane and bypass each chip +/- inputs with a .1uf poly at pins. Use a DRV134 line driver to bridge this allows all parrellel pairs to be identical, this should give you a reliable 200 to 240 watts at about .05% with about 100watts idle= or -.....
set all chips in a non inverted formatt. 22.1K .1% 1/4 watts percision resistor soldered directly to IC pins run to a 47uf 10% tantalum capacitor then a 1K ohm .1% 1/4 watt to ground. Make sure all pc board set in parrellel are identical also run a cover ground plane and bypass each chip +/- inputs with a .1uf poly at pins. Use a DRV134 line driver to bridge this allows all parrellel pairs to be identical, this should give you a reliable 200 to 240 watts at about .05% with about 100watts idle= or -.....
If you make all of the amplifiers the same (inverted or non) and then drive half f them with DRV134 then you can:
1: match all the resistors more accurately.
2: group 2 sets of 3 amps with the closest DC ofset for each side of the bridge.
3: If you decide you do not like it, then you can take the DRV off and the 6 amplifiers will all be in phase. and you could use it for lots of configurations
bpa 200 + 2x single lm3886
3x parrallel lm3886
2x parrallel + 2x single
6x single
etc etc.
"swiss army amp"
1: match all the resistors more accurately.
2: group 2 sets of 3 amps with the closest DC ofset for each side of the bridge.
3: If you decide you do not like it, then you can take the DRV off and the 6 amplifiers will all be in phase. and you could use it for lots of configurations
bpa 200 + 2x single lm3886
3x parrallel lm3886
2x parrallel + 2x single
6x single
etc etc.
"swiss army amp"
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.