Hi everyone:
Complete newbie here. No knowledge of electronics, really. I want to substitute Rf, R2 and R3 on the Chipamp.com LM3886 dual mono boards. I believe the R2 resistors I received with the kit were 1%, but then again, they were 22.1K and the schematic shows 22K.
I'm ready to buy new resistors, but what tolerance should I use for Rf, R2 and R3. The reason I ask is that retailers near me, and the usual online retailers (DigiKey, Mouser and a local one) do not have anything better than 5% components. Oh, and while I'm at it, will 1W resistors fare well, or should I hold out for 2W resistors.
Thanks
Complete newbie here. No knowledge of electronics, really. I want to substitute Rf, R2 and R3 on the Chipamp.com LM3886 dual mono boards. I believe the R2 resistors I received with the kit were 1%, but then again, they were 22.1K and the schematic shows 22K.
I'm ready to buy new resistors, but what tolerance should I use for Rf, R2 and R3. The reason I ask is that retailers near me, and the usual online retailers (DigiKey, Mouser and a local one) do not have anything better than 5% components. Oh, and while I'm at it, will 1W resistors fare well, or should I hold out for 2W resistors.
Thanks
Last edited:
5% is perfectly OK but I just did a Mouser search and I found they have over 33,000 metal film axial resistor SKUs better than 5%.
http://www.mouser.com/Passive-Compo...0vpm5Z1z0srtsZ1z0we4uZ1z0wljoZ1z0wpn9Z1z0z819
http://www.mouser.com/Passive-Compo...0vpm5Z1z0srtsZ1z0we4uZ1z0wljoZ1z0wpn9Z1z0z819
I'd prefer to use 1% metal film; 5% are almost certainly carbon film, so of (slightly) lower quality. The suppliers you mention all have 1% metal film resistors in abundance. In the feedback circuit, I like to use .1% or better bulk foil resistors though they are very expensive.
A 2 watt resistor will probably run (a little bit) cooler than a 1 watt. But if the application calls for a quarter watt resistor at that point, a 2 watt resistor probably won't physically fit. There's little point in going for the larger size.
A 2 watt resistor will probably run (a little bit) cooler than a 1 watt. But if the application calls for a quarter watt resistor at that point, a 2 watt resistor probably won't physically fit. There's little point in going for the larger size.
You could use 20% resistors, if they were still around, if they were not junk; and 0.03Watts is ample.
BTW, all the resistors INside the chip are +/-30% tolerance, and it works fine.
"22K" is in the 10% series. There is an old trick, when you want a higher precision, of never saying "22K" but "22.1K". This is so your factory workers won't stuff 22K 10%, but have to look for the 22.1K parts.
This may be why you had trouble on-line. Don't look for "22K". Look for "2%" (or 1%), then find a close value in the list.
The absolute value of these parts is not very important. Some arbitrary choice of offset suggested two <25K resistors, and a third to get gain of about 32. 22K and 680 are very suitable 5% values. 20K and 618 would work but are not standard 5% values. 22K and 750 or 620 would work with hardly-different gain.
When you make a million stereos, you NEVER buy more precision than you need. Extravagant pennies here and there speeds your path to bankruptcy. DIY economics is VERY different.
If you use 10% parts to set gain in a stereo system you risk a 1dB unbalance between channels. This may be moot because speaker pairs are rarely matched L-R to 1dB. However resistors are easier to make precise than speakers, and cheaper too, so we do what we can. Many-many stereos were built with 5% gain-set resistors and balanced fine.
The key thing is *ratios*, and for stereo, pairs of ratios. The two 22K in each side may be 20% different (there's chip errors and these resistors were picked to make them small). In one side, the 22K could be 10K or 30K no real difference. But *between* sides the "22K:680" ratio should be very close.
Like building a house. It hardly matters if the first floor walls are 7'8" or 7'9". But if you get 7'8" in the west wall and 7'9" in the east wall, the upstairs floor will slant enough to notice. You want a 1:1 ratio; and (interestingly) a house's walls are "matched" to about 1%.
In the last 30 years the cost of 2% and 1% parts has been falling. Old resistors were mixed of soot and clay, then sorted. Today the deposition can be very precise right out of the pot. I was once scolded for using 5% resistors as twist-ties, but today 2% parts are cheaper than twist-ties.
So go 2% or 1%... the "added cost" is less than the postage, and far less than the value of your personal satisfaction. And just to make 2%/1% resistors, the process has to be very well controlled, so they WILL be "good" resistors.
In another thread I just advocated over-sizing resistors. But here Rf can not see more than 0.056 Watts. 1/8W would be ample on the face of it. There's some concern about thermal effects and other moonshine. It used to be that 1/2W resistors were cheaper than any other. No longer true, but if 1/2W fits the board that is what I would use.
Yes, the local store may not stock 2% parts. There are 2.5X more values in the 2% series than the 5% series so for small sales the inventory headache is too big. 1% is 2X as many values again, so only large stores (such as DigiKey and Mouser!) will handle them.
Last edited:
No.
using +-10% resistors in the gain section could result in a gain imbalance of 3.486dB
0.9/1.1 = 0.81818
1.1/0.9 = 1.22222
The ratio of 1.2222:0.81818 is 1.493827, i.e. the Voltage Ratio could be out by >49% That is 3.486dB
Changing to 1% resistors reduces that maximum imbalance to <0.347dB and more likely in an actual build to <<0.2dB, i.e. virtually inaudible.
The "standard" range stocked by retailers is the E24 range for resistors and in 1% tolerance, from 10r to 1M0.
Last edited:
Uh..that was a lot of info, thanks, but I'm still confused. I don't know what tolerance Chipamp.com people used. Does anyone know? I figure I'll just use whatever they used.
Some of you are saying there are plenty of 19.6k or 20k resistors in lower than 5% tolerance 1W or more. Can you please link, cause I'm just not seeing that at Digikey or Mouser Canada, and I checked about 5 times for each. There are listings for them, but they display as ou tof stock with a lead time of 16 weeks. That doesn't help me.
Some of you are saying there are plenty of 19.6k or 20k resistors in lower than 5% tolerance 1W or more. Can you please link, cause I'm just not seeing that at Digikey or Mouser Canada, and I checked about 5 times for each. There are listings for them, but they display as ou tof stock with a lead time of 16 weeks. That doesn't help me.
calculate the current through each resistor, using I = V/R, write your results on your sch for testing checks.
Then calculate the resistor dissipation using Pq = I^2 R
I suggest no resistor is asked to dissipate more than 50% of it's maximum power rating, i.e. for 600mW 1% metal film resistors the Pq<300mW
The NFB pair (upper and lower) need to be very accurate/non variable. To help maintain this non variablity arrange for the Pq <<50% of rating. The lower NFB usually has no problem. It's the upper that can be asked to dissipate a lot of instantaneous peak power when the output gets near clipping. I usually suggest this upper NFB resistor has a Pq <10% of maximum rating.
Max Voutpk = sqrt{Power *2 * Rload}
For a 50W into 8ohms amplifier, the max Poutpk = Sqrt{50*2*8} = 28.3Vpk
A 20k NFB upper resistor will pass an instantaneous peak current of 28.3V/20k = 1.4mApk
Peak instantaneous power = 0.0014*0.0014*20000= 0.04W = 40mW = 6.7% of the 600mW maximum rating. No need for a 1W resistor.
Then calculate the resistor dissipation using Pq = I^2 R
I suggest no resistor is asked to dissipate more than 50% of it's maximum power rating, i.e. for 600mW 1% metal film resistors the Pq<300mW
The NFB pair (upper and lower) need to be very accurate/non variable. To help maintain this non variablity arrange for the Pq <<50% of rating. The lower NFB usually has no problem. It's the upper that can be asked to dissipate a lot of instantaneous peak power when the output gets near clipping. I usually suggest this upper NFB resistor has a Pq <10% of maximum rating.
Max Voutpk = sqrt{Power *2 * Rload}
For a 50W into 8ohms amplifier, the max Poutpk = Sqrt{50*2*8} = 28.3Vpk
A 20k NFB upper resistor will pass an instantaneous peak current of 28.3V/20k = 1.4mApk
Peak instantaneous power = 0.0014*0.0014*20000= 0.04W = 40mW = 6.7% of the 600mW maximum rating. No need for a 1W resistor.
I just looked at Mouser Canada's site. They offer 36 different metal film, through hole 22.1K resistors, 1% or better, in powers from 1/4W on up.
In 22.0K, Mouser has on offer 23 resistors, 1% or better, 1/4W or more.
You might take a second and gain and little more facility with Mouser's site.
As others have said here it's more important to get closely matching resistors for each of the boards (Rf on one board close in value to Rf on the other board) than it is to get an exact value against the schematic. So if you have a pair of resistors that both measure 22.1k then you've got a good match, use one in each board. If they are way off and you can't get 1% resistors, buy a handful of 5% ones, measure them and use the ones closest in value to one another.
I think all of that has already been said by others in this thread, hopefully this makes it easier to understand.
I don't know the schematic but I used 1/4W all round apart from a 1W from the LM3886 output pin as part of the Zobel network (pin 3 to low value resitor to 0.1uf to ground).
Last edited:
Take it from us kind folks then - there's absolutely no need for 1W rating in these resistors. 1/4W is absolutely fine.
Here is the pdf users guide that was once on the Chipamp.com site :
diyaudioprojects.com/Chip/LM3886_CA/lm3886-manual.pdf
Page 6 has a parts list. The amplifier section Zobel resistor is shown as a 2.7 ohm 2 watt. The power supply section has a couple of 2 watt's also.
Last edited:
My friend , you will find help here, but some(NOT ME
) will jump and ask you : "how come you are a member of diyaudio.com for 7 years but have no clue about the simplest lm3886/75 amp?" 22k 1watt resistor is not necessary, you should use 1/4 or 4/10 or 5/10 or 6/10 watts , and 22k instead of 22.1k 1% resistors will make no difference in that schematic.
Last edited:
1% tolerance is a must, wont cost that much compared to 5% carbon composites... And the feedback resistor has some serious influence on the performance on your amp. D.Self had published an article over thermal induced distortion, and in my own experience the difference between resistors could be easily heard when used in feedback loop. So go for the ones with low current noise and parasitic inductance.
Two of these high grade resistors per amp wont make you bankrupt anyway
At the moment I`m using Vishay ptf56 (or 65? not sure) for their low tempco, next step will be the Z-foils lol
Two of these high grade resistors per amp wont make you bankrupt anyway
At the moment I`m using Vishay ptf56 (or 65? not sure) for their low tempco, next step will be the Z-foils lol
Last edited:
Usually ±1 % resistors are the same cost as ±5 % types. For surface mount resistors, you can even get ±0.5 % resistors for the same cost as ±5 % types. I see no reason not to use ±1 % types.
Definitely use metal film resistors (regardless of tolerance) for the lowest THD.
The resistors in the Zobel and Thiele networks are generally 2.7 Ω, 2 W and 10 Ω, 2 W, respectively, though I'd make the Thiele one of a lower resistance for better transient response. Metal-oxide (MOX) resistors are fine for this application.
Tom
Definitely use metal film resistors (regardless of tolerance) for the lowest THD.
The resistors in the Zobel and Thiele networks are generally 2.7 Ω, 2 W and 10 Ω, 2 W, respectively, though I'd make the Thiele one of a lower resistance for better transient response. Metal-oxide (MOX) resistors are fine for this application.
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.