One can do a calculation that helps one assess suitability into low ohms speakers, before commiting to a build.
assume 30W into 2r0 as the target maximum output.
Imaxpk = sqrt(2*30W/2r0) = 5.5Apk
Vmaxpk = sqrt(2*30W*2r0) = 11Vpk
check : Pmax = 5.5Apk * 11Vpk /2 = 30.25W OK the prediction is about right.
The maximum output of a single cold lm3886 is guaranteed to be 7A or greater.
Two in parallel will have a guaranteed max when cold of 14A or greater.
30W into 2r0 is no problem and this amp could run on +-16V supply rails.
BUT!!!!!!!
will the 2ohms speaker demand a tranisent current that is greater than the equivalent resistor?
Here there is some disagreement.
I contend that a simple dynamic driver (i.e. voice coil) can demand upto 150% of the predicted resistor current when fed with fast changing Music Signals.
A more complex speaker, say a 2way with passive crossover can demand transients that exceeds 300% of the equivalent resistor current.
Looking back at the numbers for the 30W target, where the resistor would demand 5.5Apk and apply the 150% factor for uplift for transient current, we end up with 11Apk for fast changing signals . This is still well within the cold capability of a dual lm3886
However applying a 300% factor, one arrives at 16.5Apk.
This apparently exceeds the guaranteed dual 3886 cold capability.
The datasheet, at least in part, comes to our aid. It states that typical lm3886 have a maximum cold current capability of 11A. Double that and we arrive at 22A cold capability for a 16.5Apk transient demand.
It looks like a dual 3886 could meet that 30W target into a 300% uplifted current (for fast transients) even when it has become warm. Since these transient currents are very short term there will be little heating effect during, or shortly after, the transient and currents will return to the much lower levels typical in not so fast changing AVERAGE music levels.
Does that argument support a dual/parallel lm3886 as being suitable for a demanding 2ohms speaker load when the target maximum output is 30W into 2ohms?
Last edited:
What on earth has that got in relevance to a discussion on balanced feeds? Or were you lonely?
does the datsheet explain how to ensure one does not get current clipping when impedances are low?Or just read off the datasheet
You can do a calculation/prediction for your 3ohms ribbon.
As far as I know a typical ribbon has very little impedance variation. It behaves much more like a resistor. I would not apply any uplift factor for a direct driven ribbon speaker.
Quick check for you
130W into 3r0 gives Ipeak = 9.3Apk and Vpk = 27.9Vpk
looks to me that +-33Vdc and a well cooled dual3886 will get to a target maximum of 130W into 3ohms (ribbon) without having to guess.
Last edited:
In balanced feeds each signal conductor has a similar trend when you measure against the shield. If it they measure differently for each conductor, all the balance you design into the interface mean nothing.
look at fig 33. power output falls far faster than you calcs would suggest below 4 Ohms meaning either
1) the graph is wrong
2) something designed in
3) I can't read graphs in my old age
Whichever it would need testing. Safe option is the P-86
EDIT: For 3Ohms the book says 50W. Close enough to the upper limit for a risk, and the P-86 wasn't around at the time.
1) the graph is wrong
2) something designed in
3) I can't read graphs in my old age
Whichever it would need testing. Safe option is the P-86
EDIT: For 3Ohms the book says 50W. Close enough to the upper limit for a risk, and the P-86 wasn't around at the time.
Last edited:
you've not read the THAT app notes clearly. Or understood imbalance in balanced feeds.
Is this a reply to my post?
I typed quite clearly where I am using figures for single 3886 and where I have doubled them to take account of dual 3886.
What are you referring to?
Last edited:
The data sheet for the LM3886, fig 33, which shows that max power vs impedance is far from linear. Some of that is easily explained, some is not.
http://audiosystemsgroup.com/AES-SCIN-ASGWeb.pdf Spoiler: foil with drain wire is bad, braid is good, as reported by Putzys.
Note Soong showed impedance vs frequency for coax which was and is still completely orthogonal to the discussion (as ever).
Took me a while to find fig33.
my printed copy from 2003 only went up to fig5.
My HDD copy from 2013 has all the figures numbered.
Now to fig33.
How are you reading that?
I see the output power for distortion plus noise limited to <0.1% over the frequency range of 20Hz to 20kHz varies as on the graph.
At the high resistance end.
I see approximately 8W into 40r when V is 56V and increasing to approximately 14W into 40r when V is 70V.
Can we agree on this much?
let's convert 9W and 14W to voltages and currents.
9W & 40r is equivalent to 26.8Vpk and 0.67Apk
14W & 40r is equivalent to 33.4Vpk and 0.84Apk
looking at fig15 we see clipping voltages.
I would expect the 40r clipping voltages to be much less than shown for 8r.
9W has a clipping voltage of 1.2V and 14W is 1.6V
As expected.
Let's jump down to 20r
the two powers/voltages/currents/clipping voltages are
17W/26.1Vpk/1.3Apk/1.9V
27W/32.9Vpk/1.6Apk/2.1V
due to lower load resistance and higher current the clipping voltages have gone up and starting to approach the 8r values.
now let's do 8r
the values are
36W/24Vpk/3Apk/4V
60W/31Vpk/3.9Apk/4V
again I see a pattern that makes sense. But the clipping voltages as shown in fig15 are 2.8V @ 35Vcc and 2.6V @ 28Vcc
I could work back from these to determine the maximum power.
Do you want to do that reverse calculation?
Are we still agreed?
9W & 40r is equivalent to 26.8Vpk and 0.67Apk
14W & 40r is equivalent to 33.4Vpk and 0.84Apk
looking at fig15 we see clipping voltages.
I would expect the 40r clipping voltages to be much less than shown for 8r.
9W has a clipping voltage of 1.2V and 14W is 1.6V
As expected.
Let's jump down to 20r
the two powers/voltages/currents/clipping voltages are
17W/26.1Vpk/1.3Apk/1.9V
27W/32.9Vpk/1.6Apk/2.1V
due to lower load resistance and higher current the clipping voltages have gone up and starting to approach the 8r values.
now let's do 8r
the values are
36W/24Vpk/3Apk/4V
60W/31Vpk/3.9Apk/4V
again I see a pattern that makes sense. But the clipping voltages as shown in fig15 are 2.8V @ 35Vcc and 2.6V @ 28Vcc
I could work back from these to determine the maximum power.
Do you want to do that reverse calculation?
Are we still agreed?
Last edited:
The reverse calculation
using 2.6Vclip & 28Vcc I get 25.4Vpk and that is equivalent to 40W into 8r
Looking at the fig35 graph, maybe 39W or 40W is a better estimate, rather than my initial estimate of 36W.
using 2.8Vclip and 35Vcc I get 32.2Vpk and that is equivalent to 65W
Again 65W is maybe a better estimate from the fig35 graph.
are we agreed?
using 2.6Vclip & 28Vcc I get 25.4Vpk and that is equivalent to 40W into 8r
Looking at the fig35 graph, maybe 39W or 40W is a better estimate, rather than my initial estimate of 36W.
using 2.8Vclip and 35Vcc I get 32.2Vpk and that is equivalent to 65W
Again 65W is maybe a better estimate from the fig35 graph.
are we agreed?
Last edited:
I'm getting there.
Do you see the pattern of voltages and currents as we reduce the load resistance from 40r to 8r?
What is your estimate of max power for 4r for the two Vcc+Vee?
BTW, it's not 4ohms implying that National are considering reactive loading.
It is 4r because National choose to only give data for a resistive load. 4r is a resistance. 4ohms is an impedance. It can be reactive, or resistive.
I differentiate between reactive and resistive whenever I post. I use ohms, or "r", depending on which I refer to.
Last edited: