Need assistance with LM1875 amp cct

[Peter Tryndoch]

I decided to build a simple Amp for my PC. Originally I was going to use some LM383T's that I recycled from one of my old projects. This did not work and one even graciously entertained me with some magic smoke. So I decided to buy some new ones just in case. They were no longer available from Jaycar or Altronics, so I went with LM1875,s instead. These did not work properly either, so I decided to start by putting a basic circuit together on a breadboard and work up.

I grabbed the recommended circuit from the LM1875 TI datasheet.


C1 in the data sheet example for the LM1875 is specified as 1.0uF, however this did not let the input signal through to pin 1.
I replaced it with a 10uF Electrolytic as specified in the LM383T data sheet example and this worked. However the output is very distorted.
Test speaker is 8 ohm.

For testing I set the supply at 10.2V (very fine adjustment is not possible with my bench PSU).
Traces below are at this voltage. If I adjust the voltage + or – 2V, I get no output at all!
I think the op amp goes into some sort of protection, as I need to drop the supply to 0 and then back to 10V for it to work (sort of) again.
Eventually I want to supply 20VDC.



Fig 1: White trace is at the input and the
Yellow trace is at pin 1 of the op amp.


Fig 2: White trace is the same input as in Fig 1.
Yellow trace is output at pin 4 of the op amp.

When varying the supply voltage (VCC @ pin 5), the voltage at input pin 1 is roughly ½ VCC which sounds plausible to me, but the voltage at output pin 4 is roughly 1 volt less than VCC which I don’t think is right.

Can anyone help?
Thank you

[macboy]

Does the LM1875 have a heatsink? It is not optional. The device will go into thermal shutdown without it.

C6 (output) is absolutely necessary, and polarity is important. Polarity is also important on C1 and all other electrolytics in this circuit.
With or without an input signal, pins 1 and 2 should have the same voltage. If not, there is a problem.

10.2 V Vcc is very low, probably too low. I've run these as low as 12 V but that is pushing it too.

Return all grounds to a single point. In particular, the speaker ground should connect to the negative side of C7 and pin3. C4 and C7 should be physically close to (short wires) the LM1875 pins 5 and 3. The shorter these current loops the better.

Measure AC voltage at the output with no input. It should be the same reading as you get when shorting the DMM leads together (which might be slightly higher than zero). If there is AC voltage present, or if the LM1875 gets more than slightly warm, then it is probably oscillating.

[floobydust]

I think your IC is not working or your circuit has a wiring error (shorted C3).

These chip-amps (LM1875, TDA2003,4 TDA2030, TDA2040 etc.) can have stability issues and oscillate, as well as pick up AM radio and RF.
I would add a 22pF cap across NFB resistor R6, and an RC filter at the input like 1k/100pF. That seems to tame them well enough.

The single-supply version can damage the IC, due to the large energy stored in output coupling cap C6. There is always a huge thump at turn on. If C6 is charged, you disconnect the speaker and shut off power, then reconnect the speaker, C6 discharges into the IC and can kill it.

The LM1875 datasheet says it has protective output diodes (O/P to GND, O/P to Vcc and not on IC schematic) but for SGS/ST parts TDA2030 does not have the diodes and you have to add them.

[GigaJoe]

C3 - working on AC only, means if DC R5 +C3 == infinite resistance,  so we have only R6 as back feeder, efficiently converting opamp to voltage follower koeff amp == 1.  Means on DC input and output voltage the same.  Bias on input defined by R1/R2 divider (precisely half of power feed) and over R3 to the input. Short the input and measure all DC.  If output DC not the same as input, C3 may have a very large leak on DC, value of C3 really doesnt matter 10-200 uf.

try typical schematics with dual supply voltage, let say 9V batt ...  voltage output should be "0" , input shorted. if it not - chip faulty .
TDA2030A - $1 for 10 for experiments on aliexpress ...

[Peter Tryndoch]

Thank you for the replies. I will try to comment on all the points:

My original project has a heatsink, but as I am now prototyping on a breadboard, I did not include one.
I have been checking the opamp periodically and it has not even got warm.
I did not measure the voltage at pin 2 (not that I recollect). I will do so on the next round of testing.
10V may be low, but if I try to go higher, it cuts out in its current state. Eventually I want to run it at 20V.
I’ll take note of the grounding and oscillation, but I don’t think that that is the issue here.
As you can see from the trace I am getting clipping on the output.

The IC is new, I bought 4 of them (LM1875), as well as 4 that I already had (LM383T). Of those that I have tested with, all exhibit a similar issue. I tested all caps = ok with low ESR.

Single supply – yup she does thump on power up, but I’m hoping to avoid going back to the drawing board and going dual supply. I would have to buy another new transformer and I don’t know if there is enough room in the case left.
As is I had to put the peak detector on a separate board and mount it above the other one.

The comment on the connection of the inverted input sounds interesting.
The wiring for the LM383T on the data sheet is setup differently. See image below.




Both of these (and other examples) are from Texas Instruments data sheets for these devices, so I thought that they would get it right? I might try some different combinations here to see what happens.

Cheers

[macboy]

...10V may be low, but if I try to go higher, it cuts out in its current state. Eventually I want to run it at 20V.

This is a huge problem, and until you solve it you will get nowhere. This chip is specified to run at 16 V to 60 V.  Your 10 V is well below the minimum so you have no hope of it working correctly. Walk before you run - get the chip to power up under normal conditions without oscillating. Then you can look at starting to put audio through it.

The LM383 is ... interesting. I certainly wouldn't use it. It looks like it is optimized for a high gain (datasheet shows gain of 40 to 100 which is very high indeed). The use of AC-coupling on the non-inverting input is also strange. Distortion is high. Its datasheet is incredibly sparse, having no application information other than two example circuits with no explanation or comment. I have never seen such a bad datasheet. The LM1875 on the other hand is quite high performance (very low distortion when used correctly) and works very much like a normal power opamp.

[floobydust]

The LM383 is a different bird than the other ICs. The (+) and (-) inputs have internal resistors which allows the IC to bias itself.
The LM1875, TDA2030 etc. do not have this, so they would not be a "drop-in" replacement for an LM383. You need to modify the circuit to work. Two changes are supplying 1/2Vcc bias and a different feedback arrangement with the capacitor to ground.

[Peter Tryndoch]

On my test circuit I increased the supply voltage to 20V and noticed that I had put C3 in the wrong way.
I am now getting a nice clean Sine wave on the output and both inputs are the same DC value of just under 10V.
So now I will redesign my project to the same values and wiring as the test cct.

I always thought that Electrolytics were only polarity conscious to prevent them from being damaged and not that it affected their performance!

[radix]

Electrolytics usually begin to conduct when voltage is applied the wrong way round. That's why they blow up; the current heats them up and the electrolyte boils.

[GigaJoe]

Gainclone!    Run a wiring right!

[Audioguru]

The National Semi and Texas instruments datasheets of the LM1875 clearly shows a minimum supply of 16V on the first page and a graph shows almost no output power at such a low voltage.
Of course the + input must be biased at half the supply voltage and the - input must have AC and DC negative feedback.

A 1uF input capacitor feeding the 22k input resistor cuts frequencies at and below 7.3Hz. It does not need 10uF.