Lm4701 single amp zobel no problem
100nf + 4.7 Ohm 1/4 w resistor no problem
But lm4701 parallel amp zobel resistor burn
test value
220nf + 4.7 Ohm 1W resistor burn
100nf + 2.7 Ohm 1W resistor burn
why ??
Circuit
Pin 10, 9 220pf not add
C3 c7 4.7pf not add
R20 C20 not add
100nf + 4.7 Ohm 1/4 w resistor no problem
But lm4701 parallel amp zobel resistor burn
test value
220nf + 4.7 Ohm 1W resistor burn
100nf + 2.7 Ohm 1W resistor burn
why ??
Circuit
Pin 10, 9 220pf not add
C3 c7 4.7pf not add
R20 C20 not add
The answer is super obvious. The amplifier is oscillating.
There could be a lot of reasons, but what is glaringly obvious to me is that you didn't read the datasheet about power supply bypassing.
There could be a lot of reasons, but what is glaringly obvious to me is that you didn't read the datasheet about power supply bypassing.
Thanks
power supply bypassing. ? But no problem single amp problem start connect parallel
any idea
power supply bypassing. ? But no problem single amp problem start connect parallel
any idea
Proposal: Move the 100nF/4R7 Zobel network directly to the IC outputs such that you have a Zobel network on each IC output (two in total).
That your parallel combination is oscillating at a rather high frequency is obvious. Why exactly it does so is difficult to say but when you put two amplifiers in parallel, you have a signal loop from one amplifier an back to the other amplifier input through the feedback network. Another possibility is mutual coupling trough the supply line. Take note of Eddie's comment and ensure very good decoupling (for each IC) of the supply line.
NB: You write in the title LM4701 but you show a schematic with LM3886s.
Last edited:
Thanks
decoupling
22uf 35v
4.7uf ceramic 50v
1000uf 35v
LM3886 chip amp supply decoupling.
Power supply
4700uf 35v + 4
25A rect
Lm4701 make lm3886 parallel model
Lm4701 lm3886 same over ture big opamp
Why ti or before NS team not help ??
Read lot off thread but answer is not point
Hi Ranil,
We assume that LM4701 and LM3886 behave close to the same way. They are both from the Overture series. The LM4701 is somewhat less performance than the LM3886.
The decoupling you mention should be sufficient if you have this combination on both the "+" and "-" pins and for both ICs.
An issue is that both the wiring for decoupling, for feedback and for the Zobel network need to be very short. Chip-amps that oscillate is mainly due to poor layout and rarely due to actual fault of the ICs. But, some of these ICs are not easy to control.
You are right in making two single chip amplifiers that individually work well and then try to combine them in parallel. I would use 0.5 Ohm output resistors for a first attempt and not 3x0.5 Ohm in parallel.
I doubt that there is any official TI or NS (now TI) support team on this forum. Even if there should be some individuals working at TI and reading this, they may not have experience with these amplifier chips. I have been told that even such large companies now employ engineers for projects only and afterwards it can be difficult to find anyone who can give advise outside of what is in the datasheet and application notes. When it is about making the circuits work in practice, you are pretty much on your own with the documentation made available.
I'd also use a higher power Zobel resistor more like 3W-5W, if you intend to do amplifier testing at higher power levels. Here is how to contact TI. https://ticsc.service-now.com/csm/?_ticdt=MTU2NTAyMjI4MnwxNzM4NjMxNjQ
Last edited:
Thanks faux
Yes decoupling caps put both the "+" and "-" pins and both ics, fault start connect to parallel , without parallel zobel resistor ok,
Agree, But, some of these ICs are not easy to control.
Lm big opamp motorboat amp
Noise , hum , blablabla 60hz auto bass 20khz auto treble
I think ns technical team Overture series make but they dont no how to use noise hum without lm ic, i see some an datasheet they test with very high capacity capacitor for power supply lol
But lm amplifier very good, i lm4701 first make 2009 still working, some day amp working 24hr
*************************
Thanks rayma
I dont no they have technical team
Crazy ladys answer i cant believe
Some time i think semiconductors design area 51 in alien
I cant believe what happened us company i am from srilanka
This says it all.
Pay attention to what Tom says about picking the ceramic capacitor. Solder this capacitor directly to the pins of the chip on the bottom of the board. Electrolytic caps should be physically as close as possible.
It's a race to stupid here. I can't get a job in my field to save my life. They chuckle when they see I haven't been to school since the early 80s. Older people are considered a liability in the workplace. I only get contract work now.
Very true, Eddie. But in Europe, most only get a contract as well - elder or young. Business is today often run by people with a financial focus, where it important to present quick results, sell off and get away with the money.
Your (possible) advantage is that you may have a knowledge less and less people have, in analog electronics. It takes at least a decade to become proficient in analog electronics because you need the try-and-realize experience as well. There is a lot of constructions that need to be reviewed or made to work but it will still only bring you a contract.
Well that's business. They amortize employees now like they used to amortize an office printer.
It all changed fast after 2008, with the high rolling financial games. It's a losing deal for workers because even though you can find work, your health insurance (which I haven't had for several years now) is almost a mortgage. We all used to get health insurance from our employers. I could make two car payments and have change left over with what it would cost me for insurance. They really screwed us good.
You're right about analog electronics. Very few people understand it. I understand it so much better than I did when I got out of college. If you don't work with it, then you'll never understand it.
It all changed fast after 2008, with the high rolling financial games. It's a losing deal for workers because even though you can find work, your health insurance (which I haven't had for several years now) is almost a mortgage. We all used to get health insurance from our employers. I could make two car payments and have change left over with what it would cost me for insurance. They really screwed us good.
You're right about analog electronics. Very few people understand it. I understand it so much better than I did when I got out of college. If you don't work with it, then you'll never understand it.
Decoupling should be close chip.
Feedback resistor path should be as short as possible.
1mm = 1nH of inductance.
Feedback resistor path should be as short as possible.
1mm = 1nH of inductance.
When I studied for City and Guilds electronics we did theory in the morning and practical in the afternoon. Great fun listening to exploding capacitors when someone had put electrolytic in wrong way around.
Sometimes Uni's just do theory and that doesn't give real life experience of practical circuits.
Zobel resistor needs to be several watts rated, it dissipates real power on 20kHz test tones.
If the amp was oscillating badly I'd expect it to completely fail, my suspicion is the amp is oscillating on cycle peaks sporadically rather than full-blown unstable. A 'scope is the way to see this for sure.
If the amp was oscillating badly I'd expect it to completely fail, my suspicion is the amp is oscillating on cycle peaks sporadically rather than full-blown unstable. A 'scope is the way to see this for sure.
Many good ideas about supply decoupling, but one other issue unmentioned.
You use very small resistors to equalize the output load, 167 milli-ohms. That is too low, they will 'fight' each other when gain is not exactly the same, which it isn't even with 0.1% gain setting resistors.
Take out two of the three resistors so you have 0.5 ohms. That may fix it, if not try 1 ohms.
Jan
You use very small resistors to equalize the output load, 167 milli-ohms. That is too low, they will 'fight' each other when gain is not exactly the same, which it isn't even with 0.1% gain setting resistors.
Take out two of the three resistors so you have 0.5 ohms. That may fix it, if not try 1 ohms.
Jan
I remember the first time I looked in a unit and it looked like confetti inside. It took me several minutes to figure out that a capacitor had popped its cork.
I cooked a brand new capacitor in a brand new circuit a couple years ago by hooking it up backwards. It took less than two hours. I was running tests and it started to bulge and leak right before my eyes. Cool! Stupid, but cool.
Jan,
> You use very small resistors to equalize the output load, 167 milli-ohms.
> That is too low, they will 'fight' each other when gain is not exactly the same,
> which it isn't even with 0.1% gain setting resistors.
I have done this with 3x LM4780 in parallel.
0.1R and 0.1% works perfectly fine.
You just need to have a separate Zobel for each IC.
See Fig.4 of LM4780 datasheet.
Placing the Zobel after the equalising resistors will not stop oscillations between ICs.
It is similar to using one gate stopper for multiple MOSFETs in parallel.
Patrick
> You use very small resistors to equalize the output load, 167 milli-ohms.
> That is too low, they will 'fight' each other when gain is not exactly the same,
> which it isn't even with 0.1% gain setting resistors.
I have done this with 3x LM4780 in parallel.
0.1R and 0.1% works perfectly fine.
You just need to have a separate Zobel for each IC.
See Fig.4 of LM4780 datasheet.
Placing the Zobel after the equalising resistors will not stop oscillations between ICs.
It is similar to using one gate stopper for multiple MOSFETs in parallel.
Patrick
- Status
- Not open for further replies.