So what's the deal with a bridged/parallel lm3886?
I saw this DIY BPA300 6x LM3886 300W audio Amplifier and didn't realize the lm3886 could even be run this way. I couldn't find mention in the TI datasheet.
I thought doing bridged was what the lm4780 ( LM4780 | Mid/High-Power Class D Amplifiers | Audio | Description & parametrics ) was for.
I also noticed there is not a whole lot of information on the lm4780 - is there something that dissuades folks?
I saw this DIY BPA300 6x LM3886 300W audio Amplifier and didn't realize the lm3886 could even be run this way. I couldn't find mention in the TI datasheet.
I thought doing bridged was what the lm4780 ( LM4780 | Mid/High-Power Class D Amplifiers | Audio | Description & parametrics ) was for.
I also noticed there is not a whole lot of information on the lm4780 - is there something that dissuades folks?
National have detailed application note for the BA100, PA100 and BPA200. The last can be extended to BPA300 if you use enough chips.
Me, being skeptical of National's intention, has previously suggested they came up with this just to sell more chips. But they did partially sort this by offering the LME498xx series of driver chips which in my blinkered opinion are a far better solution.
Me, being skeptical of National's intention, has previously suggested they came up with this just to sell more chips. But they did partially sort this by offering the LME498xx series of driver chips which in my blinkered opinion are a far better solution.
The LM3886 is not really well suited for bridging due to its output current limit. Two LM3886es in parallel can be bridged with another two LM3886es in parallel (BPA200). However, you'll likely find that the matching in offset voltage is much better between two channels of an LM4780 than between two LM3886es. Hence, I chose the LM4780 for the Parallel-86. You have less fighting between channels that way. You can bridge two Parallel-86 circuits and get over 300 W out.
The LM498xx chips are nice indeed, though a tad more involved than the LM4780/LM3886 if you want good performance. Tradeoffs, tradeoffs...
Tom
The LM498xx chips are nice indeed, though a tad more involved than the LM4780/LM3886 if you want good performance. Tradeoffs, tradeoffs...
Tom
To get 300W from a bridged pair of amplifiers, each amplifier in the bridged set up must give out 150W into half the final load impedance.
i.e. two 150W into 2ohms amplifiers give a 300W into 4ohms amplifier when bridged.
or two 150W into 4ohms amplifiers give a 300W into 8ohms amplifier when bridged.
How many parallel-86 amplifiers are required to get one channel of 300W when bridged?
Last edited:
bridge/parallel + gain mods
Here's a fun question:
What is the difference between:
a bridged pair of LM3886 with a gain of ~14dB each (5V/V),
vs.
a paralleled pair of LM3886 with a gain of ~26dB each (20V/V)?
(hopefully my gain math is right)...
The final amp's overall gain should be the same, right? And the total load "seen" by each chip is also basically the same? Of course, the paralleled pair would have the typical DC offset/amps fighting each other challenge, but besides that, what would be the differences and pros/cons of either approach?
Here's a fun question:
What is the difference between:
a bridged pair of LM3886 with a gain of ~14dB each (5V/V),
vs.
a paralleled pair of LM3886 with a gain of ~26dB each (20V/V)?
(hopefully my gain math is right)...
The final amp's overall gain should be the same, right? And the total load "seen" by each chip is also basically the same? Of course, the paralleled pair would have the typical DC offset/amps fighting each other challenge, but besides that, what would be the differences and pros/cons of either approach?
assume the individual amplifiers are rated for a 4ohms load.
The individual amp is 50W into 4ohms.
The bridged pair will be a 100W into 8ohms amplifier and the output voltage will be ampgain*2*input voltage.
The parallel pair will be ~90W into 2ohms, or 52W into 4ohms and the output voltage before the load sharing resistors will be amp gain*input voltage. The available voltage power at the "external" load will be 1dB to 0.3dB less.
The individual amp is 50W into 4ohms.
The bridged pair will be a 100W into 8ohms amplifier and the output voltage will be ampgain*2*input voltage.
The parallel pair will be ~90W into 2ohms, or 52W into 4ohms and the output voltage before the load sharing resistors will be amp gain*input voltage. The available voltage power at the "external" load will be 1dB to 0.3dB less.
Two Parallel-86 boards, bridged, should be capable of providing approx. 320 W into 4 Ω from a ±28 V supply. My math is a tad optimistic in that it does not account for the increase in dropout voltage of the LM3886 as the output current increases. Perhaps I should change my verbiage from "over 300 W" to "near 300 W" or "about 250 W".
It's true that each half of the bridged amp would drive half the load impedance. I have run the Parallel-86 at 2 Ω load without issues. It provides low THD and good performance up to about 125-130 W on a ±28 V supply. Following your math, this would imply 260 W into 4 Ω when bridged. So I was off by 0.6 dB. My bad... 🙂
I can certainly see good use cases for the Modulus-86 and Parallel-86 circuits. They compliment each other well. The Parallel-bridge is really pushing the envelope. Compared to an LME498xx design, the main advantage is that in the Parallel-bridge, there's no bias adjustment to worry about. It's a bit more plug-n-play.
Tom
I thought you were claiming good performance of around 35W to 50W from a single 3886.
The parallel should get to around double this, i.e. around 70W to 100W, less a small amount for sharing resistor losses.
What has changed in the parallel-86 that allows 125W to 130W?
That was not what Tom stated way back.
He showed much data, upon which he relies, to indicate that 35W was the maximum output and to show that high voltage supply rails were unsuitable.
He showed much data, upon which he relies, to indicate that 35W was the maximum output and to show that high voltage supply rails were unsuitable.
ok you'll have to point me at that. His early test results were on lower rails as that was all he had available as a transformer. and he certainly did the thermal tests on the 3886 with different heatsinks (a thread you posted on quite a lot). Quote from there for 4 ohm testing.
Now there is a difference between 'can' and 'should' 🙂
Depends. The LM3886 can deliver 7 A of output current. Hence, for operating with 4 Ω and 8 Ω loads, I recommend ±28 V rails. This will yield about 38 W into 8 Ω and 60+ W into 4 Ω. I say "60+" as the THD starts to degrade at 62-63 W with a 4 Ω load when operating on ±28 V supplies.
On my website, I list the Modulus-86 as a 40 W amp (with the recommended power supply). This is accurate as the recommended Antek AN-2222 (or AS-2222) provides a bit over ±28 V. Just enough to get the max power over 40 W at low THD.
On ±35 V rails, the Modulus-86 provides 55-60 W of output power. I didn't do a lot of characterization at ±35 V, so forgive me for not having an exact number.
And it does. You just have to compare apples to apples.
The Parallel-86 provides 60 W into 8 Ω when operated on a ±35 V supply. This is slightly (a few W) higher than the 55-60 W you'll get from a Modulus-86 under similar conditions. The reason is that each LM3886 sees twice the load impedance, hence, a lighter load. Under these conditions they can provide higher swing. The droop across the sharing resistor is compensated for by the composite loop.
The Parallel-86 provides 120 W into 4 Ω when operated on a ±35 V supply. This is twice the 8 Ω power. No surprises there.
The Parallel-86 provides 125 W into 2 Ω when operated on a ±28 V supply. The supply voltage was lowered to ensure that each channel of the LM4780 stays within its guaranteed output current limit.
The supply voltage was changed.
That applied to Modulus-86 Rev. 1.0. The major upgrades from Rev. 1.0 to Rev. 2.0 were:
- Improved performance near clipping, resulting in higher output power.
- 10 dB reduction in mains hum due to improved grounding layout.
- Completely redesigned DC servo to ensure good performance and fast settling.
Note the first bullet.
Tom
As stated the on page 16 of AN-1192 the BPA-200 will produce 335watts at clipping into 4 ohms.
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf
FWIW
jer 🙂
http://www.ti.com/lit/an/snaa021b/snaa021b.pdf
FWIW
jer 🙂
That sounds rather optimistic, but is good data point to have.
There is another possibility why I'm "only" getting 125-130 W into 2 Ω and that's the current limit on my power supply. It's a 6 A supply, so expecting 14 A peak is probably a tad much, even with the 10000 uF external cap I'm using.
Actually, according to AN-1192, the pulsed power is 450 W into 4 Ω for the BPA200.
A pair of Parallel-86 in a bridge configuration should provide the same amount of output power as the BPA200.
Tom
- Status
- Not open for further replies.