If it is the expectedly horizontal lines, I have an explanation.
When making the photos, I sit in an almost dark room (to get sufficient contrast) and take a photo with my phone of the oscilloscope screen. An angle to the plane of the screen will quickly translate into a slope of the horizontal part of the curve. Else, I am happy with my "Hameg" ("made in Germany", so you can estimate the age).
From where will you be ordering the 2050?
I could run some tests on it too...
I could say the 1875 is impossible to make completely oscillation free, if only 1pc is used in the composite topology. Or you can, but you will loose all the THD gained with the 5534.
I could also try some jellybean booster transistors on it (1875), but I have to prepare a new PCB. Just for fun.
I have many 3055/2955 laying around...
I could run some tests on it too...
I could say the 1875 is impossible to make completely oscillation free, if only 1pc is used in the composite topology. Or you can, but you will loose all the THD gained with the 5534.
I could also try some jellybean booster transistors on it (1875), but I have to prepare a new PCB. Just for fun.
I have many 3055/2955 laying around...
From Reichelt in the cold and wet north-west Germany, near the North-Sea. They deserve encouragement for showing that survival out there is possible (last time I passed in that area it was exceptionally some 32 degrees). I will send you a couple of UPC-TDA2050 as well. I do not know if homing pigeons fly so far north but geese seems to go anywhere.
In my posting #126, I suggested a buffered LM1875 design based on a design of another member. No load balancing resistors, no trimming of off-set and gain. Output current maximum decided by the buffer transistors. Very low parts-count. Possible problems with THD and current limitation - to be seen.
Hi
thanks FF for your explanation again and the motivating words.
I am back to work...
i am on the way to repair my 2p 1875. the "damaged" amp on the right side = ampR2 was maybe not dead....after changing the LM1875 i found that after soldering the signal cable (mini coax) for both amps i got no signal input !!?. the rewarming of the solder dot was giving no connection to the input of the amp....i do not think that that was the reason but i am still working on it.
still strange because i did the test with 2p 1875....but.....maybe i was to optimistic in just do it with 0R47 parallel and yes i think I have to work more precisely with the gain and the DC offset and trim it to much better match of both amp per channel...
LM1875 will not beat me !
BTW.. Today 10:00AM i ordered UTC TDA2050 from Reichelt. delivery forecast is 1 week !
chris
thanks FF for your explanation again and the motivating words.
I am back to work...
i am on the way to repair my 2p 1875. the "damaged" amp on the right side = ampR2 was maybe not dead....after changing the LM1875 i found that after soldering the signal cable (mini coax) for both amps i got no signal input !!?. the rewarming of the solder dot was giving no connection to the input of the amp....i do not think that that was the reason but i am still working on it.
still strange because i did the test with 2p 1875....but.....maybe i was to optimistic in just do it with 0R47 parallel and yes i think I have to work more precisely with the gain and the DC offset and trim it to much better match of both amp per channel...
LM1875 will not beat me !
BTW.. Today 10:00AM i ordered UTC TDA2050 from Reichelt.
delivery forecast is 1 week !chris
out of couriosity i paralleled my two tda2050 (stm),
gain was 28 (27k/1% + 1k/1% + 47uF), offset -0.4mV and 0.1mV, each output with a 0.33ohm res connected to a speaker box.......psu +-22Vdc.
music volume zero shows no strange symptoms,
little volume music happens as usually but the heatsinks begin to warm up signifikantly,
with more volume i hear a crackling noise out of the speaker box.........finished.
seems the two chip amps must be matched when paralleled.
gain was 28 (27k/1% + 1k/1% + 47uF), offset -0.4mV and 0.1mV, each output with a 0.33ohm res connected to a speaker box.......psu +-22Vdc.
music volume zero shows no strange symptoms,
little volume music happens as usually but the heatsinks begin to warm up signifikantly,
with more volume i hear a crackling noise out of the speaker box.........finished.
seems the two chip amps must be matched when paralleled.
You can put two voltage sources in parallel without balancing resistors when you have exactly the same voltage at the two outputs.
In practice, you never have exactly the same output voltage so you need load balancing resistors. But, load balancing resistors only work with limited voltage differences. Therefore, the output voltage of both amplifiers (voltage sources) with shorted input should be no more than a couple of mV. Further, the gain of the two amplifier has to be adjusted to be very close to the same.
If you do not do such adjustment, you risk that one amplifier sends current into the other amplifier (cross-current) or they start oscillating. Both will result in increased heating and may damage the amplifiers permanently.
From what you tell, the amplifier safety features saved your amplifiers.
Last edited:
Could the solution be so simple as to connect 10nF from each of the two power rails and to the output? It is very high frequency oscillation.
My UTC TDA 2050 amp runs fine, so I am ready for the next step to start experiments with the composite project, after my short holiday (writing this from my holiday's address).
However I most probably will use the UT C chips and most likely will prefer a single bridged config instead of paralleled chips. Output power (i understand) will be in the range of 50 to 60W which will be more than enough for me.
What will this mean for the composite config, will I need an opamp for each power amp chip or a single opamp for the two power amps of the bridge?
My impression is that paralleling the chips is a relatively critical challenge while bridging the TDA2050 is a well proven concept with predictable results with a comfortable output of over approx 50W at about 2x22V supply voltage.
However I most probably will use the UT C chips and most likely will prefer a single bridged config instead of paralleled chips. Output power (i understand) will be in the range of 50 to 60W which will be more than enough for me.
What will this mean for the composite config, will I need an opamp for each power amp chip or a single opamp for the two power amps of the bridge?
My impression is that paralleling the chips is a relatively critical challenge while bridging the TDA2050 is a well proven concept with predictable results with a comfortable output of over approx 50W at about 2x22V supply voltage.
We will see when you have the opamp in your proto.
It is a current related issue in the negative channel. Removing the load fixes the oscillation. So we know it works.
First try without too much compensation and then slowly increase as needed untill satisfying results. The opamp feedback resistor from the output needs parallel 10pF (assuming 10k feedback resistor).
There is not much use of the simulator, as it does not show this phenomenon.
I have tried a lot of different methods to compensate this oscillation, and during which I have also noticed a nice side effect; putting a low value (say 100ohm) resistor from the 1875 inverting input junction (it is a virtual ground in our circuit) to GND, decreases both THD and NOISE - which is nice, but additional compensation may be needed to avoid other oscillations.
So there is a lot we can try with the parallel proto.
I can try your suggestion on my proto.
I have mistaken UTC with UPC as chip manufacturer along the discussion. UPC should be a parcel service. So, UTC.
In composite configuration you need one OP-AMP for the non-inverted part and one OP-AMP for the inverted part. Thus, in total two OP-AMPs per BTL mono channel.
Correct, as long as you use an 8 Ohm speaker you need no parallel UTC-TDA2050 to handle the current. You should be able to reach 80W!
As a matter of interest (again), I reach before clipping 70W with smd LME49720 composite bridged lm3886tf (plastic case) board at 21Vrms around 65-70Vpp with 10hz sinwave with a 8ohm sub, with 2x24Vdc psu, and minimal cpu computer heatsink with small fan. Vrms max is 27Vrms-around 80vpp .
With lm3886t it should reach 120w.
BatoMM amp.LM-bridge,consider to try it...
With lm3886t it should reach 120w.
BatoMM amp.LM-bridge,consider to try it...
It looks really good.
LM3886 is more powerfull than both LM1875 and TDA2050. You also use a slightly higher supply voltage.
I notice that the LM3886 is set for an amplification of 22 times (+1) and use an 1/5.7 scaler (divider) before the LM3886 such that you have a resulting gain of 4 from the output of the OP-AMP. We use 3.5 but for less output power/voltage.
Do you know how LM3886 reacts with a gain below 10 times?
Do you know what is the advantage of taking the LM3886 feedback signal after the current detecting resistors (R1/R2)?
I would add a 47K resistor at the very input, before C1, to reduce "plop" from connection/activation of a source while the amplifier is turned on.