Parallel LM3886 - ballast resistor question

[speakerguy79]

The spec sheet says to use 1% tolerance resistors. Is this necessary, or will 5% work fine? I realize there will be a slight increase in current sharing imbalance, but is it that big of a deal? Thanks.

Also, any source for inexpensive 1% 3W or 5W resistors in the US?

 

[mjf]

hello.
perhaps you can select the res with a digitalmm for lower tolerance.............
greetings............

 

[pacificblue]

1 % too high for one resistor and 1 % too low for the other adds up to 2 % difference worst case. 2*5 % means 10 % worst case.

You can always buy 5 % resistors and match them yourself with an Ohm-meter. You may have to buy more than your amplifiers require to get enough matching pairs, but will probably still end up with lower cost than buying 1 % resistors.

 

[speakerguy79]

Unfortunately I think matching 0R1 ballast resistors will require 4-wire impedance measuring, and I don't have one of those.

So in terms of 5% resistors, will it result in increased distortion or just more power dissipated in one chip vs. the other?

 

[Sangram]

I handmatch the resistors, with the same constraints as you. I just buy a lot of them (20 for a stereo amp) and measure them using a DMM. I also happen to use multiple 0.5ohm resistors in parallel, so it's a little easier and the resistors kind of average out.

Due to the uneven matching, the heatsinks run very hot as the current flows between the amps, specially at low levels. But then again, I also have a Audiosector LM4780 kit with 1% Dale resistors and those heatsinks get equally hot (same power supply/load).

I don't think there's a notable difference in distortion or output power if your resistors are slightly mismatched, but there is the issue of currents between amps, specially at idle or low power output levels. IMO it's much more important to gain-match Rf and Ri between chips across all channels (left, right and chips within). As long as you get those real close, the output will be very similar.

 

[PH104]

You can get Dale LVR5 series from either Mouser or Digi-Key. 0.1 ohm/5w is $1.18 from Mouser.

 

[PH104]

I forgot to add that that is a 1% tolerance resistor.

 

[pacificblue]

So in terms of 5% resistors, will it result in increased distortion or just more power dissipated in one chip vs. the other?

It will result in one of the ICs carrying more load, thus limiting the total output. E. g. if one IC carries 20 % more load than the other, when you reach the current limit at 5,6 A eff, the other IC will only be delivering 5,6/1,2 = ~4,7 A. So your total max current will be 10,3 A, when it could be 11,2 A. Traduced into output power that is 18 % less than the ICs could give you with optimum balanced resistors.

 

[AndrewT]

series connect the <1r0 resistors by zig zagging them between two insulated terminal strips.
Apply a small current to give ~ 100mV across each resistor.
Using the 200.0mVdc scale measure the voltage across every resistor.

You can use an LM317 wired as a current source for 100mA and switchable to 200mA, to make current setting simpler and it also removes variations due to temperatures and temperature coefficients.

if the individual Vr~190mV then you can achieve matching to around 0.05%.

 

[speakerguy79]

Thanks everyone! The current source-voltage measurement technique is one I will definitely remember. I lucked out and found the following at Digikey:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=CRA2512-FZ-R100ELFCT-ND

It's a 3W 0R1 1% SMT (2512) resistor. Only 31.2 cents each for ten of them!