Time to help an idiot design an amp 😀
I need to implement a chipamp and based on the boards input in another thread, have decided to use an LM1875, and learned what kind of expectations to have in my situation. There are a couple things that complicate this (for me). I am using a single supply voltage and I want to run two chips bridged to run an eight ohm load. The lm1875 datasheet shows a single supply setup, but not a bridged setup. I looked at the datasheet and an1192 sheet for a lm3886 as a reference for bridging, however it is a different chip, and doesn't make things clear for implementation to the single supply schematic.
Here are links to the datasheets:
http://www.ti.com.cn/cn/lit/ds/symlink/lm1875.pdf
http://www.ti.com.cn/cn/lit/ds/symlink/lm3886.pdf (pg 6)
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf (pg10)
Attached are two schematics. One is the single supply 1875 (same as in first TI link). The other is that schematic smashed together with the bridged 3886 schematic, implementing 3886 bridge component values. I know this is entirely wrong. The internet doesn't seem to have exactly what I am looking to do, and I am not an engineer or physicist, so I am looking for help in designing a single supply bridged lm1875 circuit.
Can all of you smarter than me look at the attached and ever so wrong bridged schematic and help me make it a working and good one? I can edit the schematic and component values. If a change is suggested, please feel free to say why, so I can learn as I go versus just blindly making changes.
Thanks in advance, looking forward to the results!
I need to implement a chipamp and based on the boards input in another thread, have decided to use an LM1875, and learned what kind of expectations to have in my situation. There are a couple things that complicate this (for me). I am using a single supply voltage and I want to run two chips bridged to run an eight ohm load. The lm1875 datasheet shows a single supply setup, but not a bridged setup. I looked at the datasheet and an1192 sheet for a lm3886 as a reference for bridging, however it is a different chip, and doesn't make things clear for implementation to the single supply schematic.
Here are links to the datasheets:
http://www.ti.com.cn/cn/lit/ds/symlink/lm1875.pdf
http://www.ti.com.cn/cn/lit/ds/symlink/lm3886.pdf (pg 6)
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf (pg10)
Attached are two schematics. One is the single supply 1875 (same as in first TI link). The other is that schematic smashed together with the bridged 3886 schematic, implementing 3886 bridge component values. I know this is entirely wrong. The internet doesn't seem to have exactly what I am looking to do, and I am not an engineer or physicist, so I am looking for help in designing a single supply bridged lm1875 circuit.
Can all of you smarter than me look at the attached and ever so wrong bridged schematic and help me make it a working and good one? I can edit the schematic and component values. If a change is suggested, please feel free to say why, so I can learn as I go versus just blindly making changes.
Thanks in advance, looking forward to the results!
Attachments
The non-interverting amp (upper one) looks fine, the lower one has one major problem that I can see. The +ve input needs to be fed from mid-rail, you show it going to 0V. I would use the same supply splitter to feed both sides of the bridge, decrease the resistor values to the two rails to 10k and increase the 10uF to 220uF.
one fatal flaw > forgot output cap LOL
1. balance gains between top and bottom/ use precision parts
2. use common 1/2 Vcc bias circuit > use lower impedance for drift and offset
consider a trim circuit if precision isn't feasible.
3. make sure to use plenty of decoupling caps and output Zobels
4. use a mute circuit to help power on / off thumps
1. balance gains between top and bottom/ use precision parts
2. use common 1/2 Vcc bias circuit > use lower impedance for drift and offset
consider a trim circuit if precision isn't feasible.
3. make sure to use plenty of decoupling caps and output Zobels
4. use a mute circuit to help power on / off thumps
one question what is the 24V supply requirement from?
consider a new supply perhaps http://www.diyaudio.com/forums/powe...ng-off-shelf-noname-smps-gainclone-d-amp.html
use 8 ohm speakers for more HiFi and less headaches IE current limit at 11
FWIW don't bridge LM1875 try the bigger chips.
consider a new supply perhaps http://www.diyaudio.com/forums/powe...ng-off-shelf-noname-smps-gainclone-d-amp.html
use 8 ohm speakers for more HiFi and less headaches IE current limit at 11
FWIW don't bridge LM1875 try the bigger chips.
Last edited:
R3 and R9 would connect to the same half-supply point. Output cap is not needed for this application.
If the application is not critical audio listening, you could simply run one amp off the output of the other. The 'slave' would have its non-inverting input referenced to half supply with a low resistor - 100 ohm typical. The inverting input would then connect to the master's output with a resistor (in your case, 47k). R12 would also change down to 47k from the 51k you already have.
You will get a nasty thump at turn on with 1875 and single supply, fair warning. I have one such (not bridged) and it pushes the speaker to xmax and back. Hoping the speaker will burn soon then I can get better ones, the application is that low-end.
If the application is not critical audio listening, you could simply run one amp off the output of the other. The 'slave' would have its non-inverting input referenced to half supply with a low resistor - 100 ohm typical. The inverting input would then connect to the master's output with a resistor (in your case, 47k). R12 would also change down to 47k from the 51k you already have.
You will get a nasty thump at turn on with 1875 and single supply, fair warning. I have one such (not bridged) and it pushes the speaker to xmax and back. Hoping the speaker will burn soon then I can get better ones, the application is that low-end.
well having DC at any amps speaker terminals is a big no-no in my world. to each their own I guess a couple of fuses there would be better than nada.
add another supply and buy a couple of cheap LM3886 chip-amp PCBs mod/ assemble / enjoy is my advice KISS in action LOL
add another supply and buy a couple of cheap LM3886 chip-amp PCBs mod/ assemble / enjoy is my advice KISS in action LOL
Last edited:
DC from one output half and the other are identical, net DC will be zero. Many discrete Class A amps use the same trick to eliminate the output cap 🙂
Though the 1875 is a very fragile chip, I've also had other National chips (4780, 3886) blow themselves and on one occasion, the speaker as well. Running split supplies, so neither a better chip or use of a split supply ensures safety, only a crowbar with differential DC detection (in this case). I suspect the application is not that complex.
Though the 1875 is a very fragile chip, I've also had other National chips (4780, 3886) blow themselves and on one occasion, the speaker as well. Running split supplies, so neither a better chip or use of a split supply ensures safety, only a crowbar with differential DC detection (in this case). I suspect the application is not that complex.
yeah I know that's the theory
not a vary far sighted approach to " bullet proof" design formulas.
users may touch speaker wires to many things ...chassis / grounds other imbalanced channels etc etc
single supply amps w/ DC blocking caps don't need crowbars or relays, they do need careful attention to turn on delays mute timing etc
what's the story of many chip failures?
maybe you can pass along some very valuable design advice on what not to do.
Last edited:
Am I the only one who thinks that this design can only end in disappointment?
Bridged 1875, single supply, speaker terminals to ground 12V.
Every part of this amp is non tested. Where do you look when it doesn't work?
The only way to make it even more complicated would be to have 5 channels.
Bridged 1875, single supply, speaker terminals to ground 12V.
Every part of this amp is non tested. Where do you look when it doesn't work?
The only way to make it even more complicated would be to have 5 channels.
well the starting recipe (single cheap SMPS and apps for the most speaker power) is set by the user, add a few cooks, and voila , smoke is produced. LOL
Last edited:
Ok, I'm convinced. Abort mission... I will just build a real chipamp and power supply. I am going to build a PA100 (LM3886 2 chip parallel). I realize I only need one chip per channel (especially for 8 ohms), but am hoping for more headroom before hitting spike protection by doing 2 in parallel.
I can just follow schematics on this, but can somebody please tell me the minimum VA torroid I need for this? 2 parallel 3886 per channel, so each will only ever see 8 ohms or more.
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf
Thanks for everybody's continued time and patience with my stubbornness on the previous project idea..
I can just follow schematics on this, but can somebody please tell me the minimum VA torroid I need for this? 2 parallel 3886 per channel, so each will only ever see 8 ohms or more.
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf
Thanks for everybody's continued time and patience with my stubbornness on the previous project idea..
hmm..going from bridged small chip amps to parallel heavier chips implies polar ends of speaker impedances?
no start from the beginning. what do you want to do and what are your expectations. examples mess around with chip amps as new hobby, build a low cost or lowest cost amp with hifi for existing speakers., start a chip amp design from scratch ie building the best?
schedule, cost, performance pick any 2
edit> did you consider adding a 2nd SMPS (+/- supply from my link above) rather than throw out the whole thing?
no start from the beginning. what do you want to do and what are your expectations. examples mess around with chip amps as new hobby, build a low cost or lowest cost amp with hifi for existing speakers., start a chip amp design from scratch ie building the best?
schedule, cost, performance pick any 2
edit> did you consider adding a 2nd SMPS (+/- supply from my link above) rather than throw out the whole thing?
Last edited:
Hi,
There is nothing essentially wrong about using the LM1875
that can run on +/- 25V rails bridged on a single 24V rail.
Lousy circuit, but in principle it would work just fine,
with better thermal characteristics than a single chip.
rgds, sreten.
There is nothing essentially wrong about using the LM1875
that can run on +/- 25V rails bridged on a single 24V rail.
Lousy circuit, but in principle it would work just fine,
with better thermal characteristics than a single chip.
rgds, sreten.
Last edited:
A single LM3886 will drive 6 ohms ( practical 8 ohm speaker ) all day using good thermal design. http://www.diyaudio.com/forums/solid-state/179085-updating-dynaco-stereo-120-a-2.html#post2572021
2 channels at full power into 6 ohms you could expect minimum 70-80 unclipped Watts / channel using a pair of the non isolated version at +/-35 Vdc no ripple. The isolated one is rated close 50W from memory. Class AB is about 75% eff at full load. This is worst case IE LAB power because no music would require constant power like this.
2*80*1.25 = 200 watts or VA
Hi,
The reality for the PA100 is the minimum transformer VA you could
get away with is about 80VA for 8ohm loads and 160VA for 4 ohm.
YMMV but bigger transformers don't cost proportionally
more than smaller ones, I'd use about 250VA.
rgds, sreten.
The reality for the PA100 is the minimum transformer VA you could
get away with is about 80VA for 8ohm loads and 160VA for 4 ohm.
YMMV but bigger transformers don't cost proportionally
more than smaller ones, I'd use about 250VA.
rgds, sreten.
Last edited:
Mostly depends on the rail voltage with ripple if any and the power delivered to load before going into thermal meltdown. decent rule of thumb says 2x rated output power in VA
Hi,
In fact a decent rule of thumb is for music VA = expected
mildly clipped music program, e.g. 100W per channel music
peaks mildly clipping needs about 200VA*. 2 x rated output
is appropriate for a laboratory sine wave RMS amplifier, or
an AV subwoofer - generally have the dynamics of lift music.
rgds, sreten.
* Real number depends on the statistics of the music.
Most decent quality program spends 80% of the time
at below 20% of the clipping level defining max output.
In fact a decent rule of thumb is for music VA = expected
mildly clipped music program, e.g. 100W per channel music
peaks mildly clipping needs about 200VA*. 2 x rated output
is appropriate for a laboratory sine wave RMS amplifier, or
an AV subwoofer - generally have the dynamics of lift music.
rgds, sreten.
* Real number depends on the statistics of the music.
Most decent quality program spends 80% of the time
at below 20% of the clipping level defining max output.
Last edited:
I originally was planning on doing 2 smps supplies, but was discouraged from that in another thread. What I want is to build a simple chip amp to be compact, safe, minimally 30wrms per channel into 2 channels, four ohm stability would be nice but not absolutely necessary.
I would like cost to be reasonable. I want a small 2 ch hifi amp that I built (and not from a kit, but also don't care if it is directly from a chips data sheet. I want it to be relatively hifi (and not the quite loosely used term "hifi" of today), enough so to rival the sound of "good" commercially available amps.
I am not opposed to one chip per channel if I can get 30wrms x 2 @ <1% thd with good sq and little risk of running into spike type protection. Think poor man's hifi, a simple 2 ch listening system.
Please help..
Last edited:
A single lm3886 per ch at +/-28v will do what I need, but my concern is the spike protection everybody gripes about..I otherwise like the chip. How easily does that kick in? I don't plan to do house parties with this thing, but will it kick in on every transient if I start to get into something I'm listening to?
- Status
- Not open for further replies.