Solder iron not hot enough for PC board

[audioguru]

My hard drive shows an LM386 with no capacitor between pin 1 and pin 8 but instead with a resistor in series with a capacitor from pin 1 to ground calculated with a gain of 1000 times above 73Hz. Since it reduces negative feedback then distortion is increased and the high frequency cutoff is reduced.

You have power numbers for different LM386 ICs but at a horrible sounding 10% clipping distortion. Any LM386 produces 450mW into 8 ohms only at 1KHz at a level slightly below clipping with a 9V supply, not 700mW. The LM386 N4 produces 1000mW into 32 ohms (does anybody make a 32 ohm speaker?) only at 1kHz when it has 10% of horrible clipping distortion with a 16V supply. The distortion is higher at lower and higher frequencies.

You might be producing such a high power for your weak left ear that it might destroy what remains of its hearing.

 

[gary350]

Today I built an LM386 IC tester circuit using an 8 pin socket to test all my LM386. I used all the same parts that are on the good circuit I just finished building few days ago. I tested the mic and tested all the parts then I swapped out different LM386 1 by 1. They are all low amplification expect the LM386 n-1 that is permanently soldered to a PC board I bought 5 years ago at Radio Shack. I got a bright light an large magnifying lens and was finally able to see n-1.

LM386 n-1 only 1 out of a group of 14 LM386 n-1 works good with Gain of 200 all others have gain of 20 even with the high gain circuit.

LM386 w-93 is low power too.

Does the 1000 Gain circuit work on any LM386 ? I will build a 1000 gain circuit today to see how well it works I hate to waste my time if this will not work with LM386 n-1.

The 1000 gain circuit calls for a 10K variable resistor should that really be 100K?

 

[audioguru]

They test LM386 ICs for maximum supply voltage which calculates into the maximum output power. The datasheet shows Electrical Specifications for LM386N-1 and LM386N-3 with a maximum allowed supply voltage of 12V and show the LM386N-4 with a maximum allowed supply voltage of 16V.
The LM386 output transistors and DIP package limit the maximum allowed amount of heating.

Then they show the output powers:
1) The LM386N-1 with a 6V supply typically produces only 325mW into 8 ohms with 10% distortion. Probably its distortion is higher with a higher supply voltage. At this low power it barely gets warm.
2) The LM386N-3 with a 9V supply typically produces 700mW into 8 ohms at 10% distortion. It gets warm-hot.
3) The LM386N-4 with a 16V supply typically produces 1000mW into 32 ohms at 10% distortion. It gets extremely hot.

All the LM386 ICs have a gain of 200 with the 10uF capacitor between pin 1 and pin 8. Maybe cheap Chinese ones have pin 1 or pin 8 broken.
All the LM386 ICs have a gain of 1000 with the RC shown from pin 1 to ground without the capacitor between pin 1 and pin 8.
All the LM386 ICs work with a 10k volume control when the source impedance is low like from a low impedance dynamic mic or a player.
All the LM386 ICs work with a 100k volume control when the source impedance is medium like from an electret mic.
Since your circuit drives a low power into headphones then it can use any of the LM386 ICs.

I think all the circuits in the LM386 datasheet show a 10k volume control because The LM386 was invented before electret mics became common.
I think Radio Shack sold cheap Chinese parts. I bought an expensive audio cable there then I found exactly the same one at The Dollar Store for $1.00. I bought it for the $1.00 and returned it to Radio Shack and got all my money back.

 

[gary350]

I give up trying to order anything from Digi-Key or other places. The search never ends??? I search for LM386 n-4 then click to next page, next page, then 30 pages later WTF I give up.

Google search finally found what I want from DigiKey but 2 month delivery time.

Amazon has them but I am afraid they will be more or what I already have and don't want.

 

[audioguru]

At the bottom of page 4 you show a 3uF capacitor driving one 32 ohm earphone parallel with another attenuated headphone. Then the frequency response reduces the mid-high frequency of 1975Hz and all lower frequencies to sound squeaky. Use 150uF.

 

[audioguru]

I give up trying to order anything from Digi-Key or other places. The search never ends??? I search for LM386 n-4 then click to next page, next page, then 30 pages later WTF I give up.

Google search finally found what I want from DigiKey but 2 month delivery time.

Amazon has them but I am afraid they will be more or what I already have and don't want.

Your circuit does not need the high voltage LM386N-4, it will work fine with a LM386-1 or an LM386-3.
On Digikey's first page of LM386 ICs it shows 9200 of LM386N-1, 4600 of LM386N-3 and 6600 of LM386N-4 in stock at the same price. These ones have through hole packages and no lead on the pins but you can solder them with ordinary tin-lead solder.

 

[ChrisP58]

I drew circuit fast had lots of errors, I think errors are all fixed now. I have built model airplane for 40 years I can build a RIB dish that will be strong in 50 mile per hour wind with balsa wood. I will never have it in high wind there will too much noise for amp.
-


View attachment 116626

It's not whether the dish is strong enough to stand up to wind forces. It's whether the structure, particularly the sound reflective surface, is acoustically rigid across the frequency range of interest.

 

[gary350]

At the bottom of page 4 you show a 3uF capacitor driving one 32 ohm earphone parallel with another attenuated headphone. Then the frequency response reduces the mid-high frequency of 1975Hz and all lower frequencies to sound squeaky. Use 150uF.

You told me to use 2.7uf here is the math you did.

Since you are not listening to hifi music down to 20Hz or earthquakes down to 1.69Hz, then calculate a capacitor cutoff frequency of 30Hz or 40Hz, certainly not 1.69Hz. C= 1/(2 x pi x 30Hz x 2k ohms)= 2.7uF.

 

[gary350]

I found the problem. 1 solder connection was bad. Solder was on the wire but no connection. Now the tester is fixed I tested a LM386 n-1 and the 200 gain amp works great. I tested both LM396 w-93 they do not work well volume is 90% less. I should probably trash both LM386 w-93 but maybe they be good for a 20 gain circuit but I don't want them mixed up with the good parts. I put LM386 n-1 in both China amps WOW they work just as well as the circuit i built even with the 10K volume controls and the volume control has a 4.7K resistor in parallel it.

 

[audioguru]

You told me to use 2.7uf here is the math you did.

Since you are not listening to hifi music down to 20Hz or earthquakes down to 1.69Hz, then calculate a capacitor cutoff frequency of 30Hz or 40Hz, certainly not 1.69Hz. C= 1/(2 x pi x 30Hz x 2k ohms)= 2.7uF.

Notice that I calculated for your old 2k ohms earphones.

I have 14 LM386 n-1 the circuit works good with only 1 of them the other 13 are 90% less volume. What is the problem?

Cheeeep Chinese fakes or factory rejects? Cheeeep Chinese IC socket with too much rice in the alloy? Corrosion of the pins on the IC and socket during the month or more on the boat from China?
The schematic of the LM386 on its datasheet shows a 1350 ohms resistor between pin 1 and pin 8. If your pins are bad then your multimeter will not be able to measure it. Measure with the power supply turned off.

 

[audioguru]

The LM386 w-93 is not sold at Digikey so either it is very old or is a Chinese fake.
With the 50k input resistance of the LM386 parallel to a 4.7k resistor and parallel to a 10k linear volume control set at maximum, then the electret mic has a load of 3k ohms then its output level is reduced to half of when it has a 33k load from the 50k of the amplifier parallel with a 100k volume control set to maximum.

 

[JonSea]

Audioguru, perhaps you could stop disparaging 1.4 billion people in virtually every post?

Many people, including people of non-Chinese descent, find it offensive.

 

[Mickster]

Yup, China seems to be the main focus now, whereas India previously got the brunt.
You have to wonder whether choosing any name containing the word "guru" was wise, especially considering the origin, if a person did not want to be regularly contacted for help by a particular demographic...

 

[gary350]

1(2 x 3.14 x 30Hz x 32 ohms) = 1/60288 = .00001.658 = about 1.6uf and 150uf will sound better. I though I was starting to understand now I am confused. I experimented with output capacitors the larger it is the better is sounds. A 10uf capacitor makes it sound like all the sound is coming out of a metal bucket.

I have an idea that will make it easy to add and remove capacitors to listen to the sound difference so I came up with 5 caps in parallel with 4 switches to add or remove another cap. Equal value caps make it easier to add up in my head than odd numbers caps. After I finish the 1000 gain circuit I will build this.

 

[audioguru]

Gary, if your circuit does not work and you did not buy your parts, solder or soldering iron locally, then please don't tell us where you are waiting for them from.

 

[gary350]

Gary, if your circuit does not work and you did not buy your parts, solder or soldering iron locally, then please don't tell us where you are waiting for them from.

I made sure to buy parts from a place where they would arrive in 3 days, one comes from Colorado, one comes from Texas, one comes from Arizona. This circuit works good.

 

[audioguru]

Gary, I am glad to hear that you buy high quality parts that are not from "over there".

 

[gophert]

Gary, I am glad to hear that you buy high quality parts that are not from "over there".

I think you just need the term "counterfeit". Nearly every major semiconductor company either has at least one fab in China or contracts out to Fabs in China when they need extra capacity or for products approaching the end of their lifecycle - I think TI (National Semiconductor) announced the end of the LM3815 and several other chips popular with hobbiests in the past few years. I'm guessing the LM386 doesn't have long to live at TI - but who knows.

https://en.m.wikipedia.org/wiki/List_of_semiconductor_fabrication_plants

 

[gary350]

I think you just need the term "counterfeit". Nearly every major semiconductor company either has at least one fab in China or contracts out to Fabs in China when they need extra capacity or for products approaching the end of their lifecycle - I think TI (National Semiconductor) announced the end of the LM3815 and several other chips popular with hobbiests in the past few years. I'm guessing the LM386 doesn't have long to live at TI - but who knows.

https://en.m.wikipedia.org/wiki/List_of_semiconductor_fabrication_plants

My sisters husband works for an electronic company in Phoenix Arizona. He was telling me most of their parts are contracted out to China, the Phoenix quality control department tests all the parts if they pass they are sold for 95% profit compared to what is cost for American labor. The Phoenix factory orders parts with their part number and logo few people know parts are made in China then sold by the company in Phoenix. They are not counterfeit parts because China makes parts to the specifications of the Phoenix factory, parts much pass quality control test before they are sold. Only difference is, China workers get paid about $1 per day and American Union workers are not getting paid $87 per hour anymore.

 

[gophert]

My sisters husband works for an electronic company in Phoenix AZ. He was telling me most of their parts are contracted out to China, the Phoenix quality control department tests all the parts if they pass they are sold for 95% profit compared to what is cost for American labor. The Phoenix factory orders parts with their part number and logo few people know parts are made in China then sold by the company in Phoenix. They are not counterfeit parts because China makes parts to the specifications of the Phoenix factory, parts much pass quality control test before they are sold. Only difference is, China workers get paid about $1 per day and American Union workers get paid $87 per hour.

Yes, Microchip and NXP have plants in Chandler AZ (possibly a few others). From what I have been told, Mosfet drivers, op amps and other chips are outsourced by microchip. Both outsource quite a bit and NXP has at least one facility in China. Chinese manufacturers do a pretty good job on many things including my iPad.