Makes no sense, because there are too few speakers around with 1 Ohm coil impedance or less. A few car subwoofers maybe, but they are better off with class D amps anyhow.
If it has been done, then most likely by Rowland Research. They just love bridging and paralleling LM3886s and TDA7293s by the numbers.
If it has been done, then most likely by Rowland Research. They just love bridging and paralleling LM3886s and TDA7293s by the numbers.
I have four TDA2030A in parallel, but they're much lower current than 3886. Even if bridged, I reckon four LM3886 in parallel is overkill assuming the design is adequately heatsunk. If there is some other unusual constraint on the design, then 4 * LM3886 might be a goer in limited situations.
I think LM3886 on-chip protection is an advantage compared to LME498xx based amp. It is also much cheaper.
Last edited:
I experience the difficulty of DC offset adjustment among the 3886s in a x3 board. Without resorting to DC servo, I may DC isolate all the inputs of the inputs and fine tune the AC gain by multi-turn trim pot of each 3886.
Well a high-quality input film cap may cost more than the DC servo.
The 3886 is specified @ 7 to 11.5A maximum at 25degC.
National do not tell us how low that peak output current must be kept to avoid current limiting when Vce is high nor do they tell us how the current limiting is varied as the chipamp temperature changes from 25degC.
What if the limiter comes in at 5Apk when Tj ~100degC momentarily.
What if the limiter is also modulated to take account of Vce. Could that reduce the maximum peak current of a hot chipamp feeding a reactive speaker load to <<7A?
4 parallel chipamps will increase the total peak current into a reactive load on two counts taken from the previous suggested operation.
a.) the 4 way sharing reduces the demanded current.
b.) the chipamp temperature will be slightly lower due to less current demanded through any individual chipamp.
A cooler chipamp and less current demanded cannot make SQ worse and there is much in the app notes and datasheet that suggest it could increase SQ, particularly since the biggest problem of all chipamps is severely limited current capability.
National do not tell us how low that peak output current must be kept to avoid current limiting when Vce is high nor do they tell us how the current limiting is varied as the chipamp temperature changes from 25degC.
What if the limiter comes in at 5Apk when Tj ~100degC momentarily.
What if the limiter is also modulated to take account of Vce. Could that reduce the maximum peak current of a hot chipamp feeding a reactive speaker load to <<7A?
4 parallel chipamps will increase the total peak current into a reactive load on two counts taken from the previous suggested operation.
a.) the 4 way sharing reduces the demanded current.
b.) the chipamp temperature will be slightly lower due to less current demanded through any individual chipamp.
A cooler chipamp and less current demanded cannot make SQ worse and there is much in the app notes and datasheet that suggest it could increase SQ, particularly since the biggest problem of all chipamps is severely limited current capability.
Linn seemed to think paralleling 3 chip amps was worth several US$K
http://www.diyaudio.com/forums/chip-amps/42578-about-linn-power-amps.html
multiway loudspeakers with poorly designed crossovers can take large factors over the nameplate resistive load calculated current peaks
the thermal feedback distortion is easily eliminated by wrapping a good auality audio op amp around the chip amps in a multiloop composite - compensation is a little complesx but added loop gain can be 40-60 dB at the low thermal moduation frequencies
http://www.diyaudio.com/forums/chip-amps/42578-about-linn-power-amps.html
multiway loudspeakers with poorly designed crossovers can take large factors over the nameplate resistive load calculated current peaks
the thermal feedback distortion is easily eliminated by wrapping a good auality audio op amp around the chip amps in a multiloop composite - compensation is a little complesx but added loop gain can be 40-60 dB at the low thermal moduation frequencies
Last edited:
Yes, they do in AN-898. The current limiter has a fixed threshold. A chip amp driven into that limiter will sound like any amp driven into a limiter.
As the IC is operating at its current limit, however the power transistors will heat up and the thermal protection will shut them off until temperatures are back to normal again. No music for a while and no THD at all.
Then there is the voltage limiter which also has a fixed threshold, but we cannot outsmart it by reducing the load current. If your speaker's reactance is too big, external flyback diodes may help to relieve the internal clamping circuitry.
Then 5 A will heat the power transistor up to above the thermal threshold at 165 °C when the junction is at an average of 100 °C. In that case the heatsink is too small.
You can solve that by adding ICs to distribute the heat across more of them.
You can solve that by increasing the heatsink size or adding a fan.
Or you may ask yourself what you are doing that demands more than 15 A from your three-parallel-LM3886 amp, when it is already operating at an average junction temperature of 100 °C. THD may be the least of your problems in that situation.
And still my two questions remain.
Which speaker is it that will sound sufficiently better from four than from three parallel LM3886s to justify the additional investment?
Are four parallel LM3886s really cheaper than one LME498.. and a few transistors that give the same or an even better result?
...where human hearing is less sensitive to THD?
As should improved heatsinking and not using chip amps for loads they cannot reasonably operate on.
The former - probably not. Short term thermal time constants remain the same even when bigger heatsinks are employed. Its rather like saying a bigger battery capacity will improve the decoupling of some high speed logic - doesn't happen.
As for the latter, whose meaning of 'reasonably' ?
The delta T remains the same. Thanks to improved heatsinking the average temperature will be lower, so the peak temperature as a result of average temperature plus delta T must also be lower.
The IC designer's meaning of reasonable. I quote from the datasheet.
Nothing where to expect highly reactive, low-impedance loads driven at continuous max output power from max rail voltage while still expecting low THD.
I would put the stress in your quotation on "safeguarding itself against the most stringent power limiting conditions". Figures 8 and 9 of that document show that it is possible to dissipate 30 W for 100 ms at a case temperature of 125 °C or you can dissipate 40 W for 100 ms at a case temperature of 75 °C before SPiKe starts to operate. The latter corresponds nearly to the maximum output power into 8 Ohm from ±42 V rails.
Three LM3886s at Tc 75 °C will be able to dissipate the same 120 W as four LM3886s at Tc 125 °C before SPiKe starts to limit the output current. That would happen at more or less the maximum output power into 3 Ohm from ±42 V rails. And still no answer to my two questions.
this seems to confirm that the protections/limiters are monitoring Tj and adjusting the limiting thresholds to suit the level and time of the transient.
That is why I arrived at "my interpretation". The protection levels vary with Tc and the higher the Tc, the lower the limiters are set.
On transients, I think that what I hear is the early onset of current limited transients, long before these transients ever approach voltage clipping. And this sounds worse if the chipamp is hotter. I try to keep my operating chipamps near cold and yet I still hear the effect as volume is turned up.
I chose to ignore your other two questions, because you know just as well as me that I have never listened to multi-chipamp topologies and thus can never have operational experience of a speaker model that meets your specification. In addition you also know I only work with 8ohm or higher impedance speakers. I don't own any 4ohm speakers.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.