I doubt that many DIYs are capable of following the design rules for RFI or EMI proofing an audio PCB. Nor would many of us be likely to draft and have multilayer boards made successfully. We would need to use only the recommended boards and wiring methods designed by experts as mentioned above. To me, that means spending a lot more time and money than I hoped and buying or slavishly copying professional and proven desigs. Faced with that prospect, I'd just stick with conventional power supplies for DIY.
Alternatively, I've used little "power brick" switchmode LED power supplies on new projects for a year or so now. The physical separation of a metre or so between supply and amplifier makes a world of difference and allows extra shielding and line filters, ferrite rings etc. to be added for checking out any problems. These are now really cheap, less than $10 for 96W rating at 24V. Some even have adjustable voltage in various ranges as wide as 6-24V at 72W. Like all bricks, they're completely sealed, come with a locally approved mains leads and if you trust the isolation of the SMPS design, they can simply be stacked for dual supplies and are safer than messing with mains wiring inside the amplifier case too.
There is a snag with the DC connecting leads that come with 2.5 ID x 5.5mm length plugs so ideally, you'd fit matching insulated sockets to your project. Some form of power connector that can't be confused with your laptop or other periferal is a good thing. You can always strip the little DC plugs and connect permanently via a terminal block but that defeats the advantage of universal, easily replaced supplies for any number of projects, including those single polarity discrete preamps we sometimes prefer
Alternatively, I've used little "power brick" switchmode LED power supplies on new projects for a year or so now. The physical separation of a metre or so between supply and amplifier makes a world of difference and allows extra shielding and line filters, ferrite rings etc. to be added for checking out any problems. These are now really cheap, less than $10 for 96W rating at 24V. Some even have adjustable voltage in various ranges as wide as 6-24V at 72W. Like all bricks, they're completely sealed, come with a locally approved mains leads and if you trust the isolation of the SMPS design, they can simply be stacked for dual supplies and are safer than messing with mains wiring inside the amplifier case too.
There is a snag with the DC connecting leads that come with 2.5 ID x 5.5mm length plugs so ideally, you'd fit matching insulated sockets to your project. Some form of power connector that can't be confused with your laptop or other periferal is a good thing. You can always strip the little DC plugs and connect permanently via a terminal block but that defeats the advantage of universal, easily replaced supplies for any number of projects, including those single polarity discrete preamps we sometimes prefer
In most chip power amplifier datasheets, the FB-shunt RC frequency is high enough to cause a "warm bass" distortion. That might be an answer to the question at Post#1.
If the bass distortion is cleaned up (changing the RC to 0.24hz - 0.4hz range), then it will probably reveal midrange distortion (previously present but not as easily noticed). The clean-up effort also involves a modicum of power filtering and compensation until placidity--Anything with tiny, less linear outputs, like a chip amplifier, has to have the stability double-checked at maximum workload.
A basic high current DC power filter is in the attachment. The unlabeled electrolytics would be 470u for TI-Overture, or 220u for others. You may have seen something like this for front end power in discrete amplifiers (except that this is a high current version for the chip amplifier).
If the bass distortion is cleaned up (changing the RC to 0.24hz - 0.4hz range), then it will probably reveal midrange distortion (previously present but not as easily noticed). The clean-up effort also involves a modicum of power filtering and compensation until placidity--Anything with tiny, less linear outputs, like a chip amplifier, has to have the stability double-checked at maximum workload.
A basic high current DC power filter is in the attachment. The unlabeled electrolytics would be 470u for TI-Overture, or 220u for others. You may have seen something like this for front end power in discrete amplifiers (except that this is a high current version for the chip amplifier).
Attachments
Last edited:
Hi All, a few words, if I may, defending eligibility of the chip amps for delivering high quality at relatively high power.
Attached pictures show one of my designs called "NUCLEO" - sorry I can't share the full detail about the design - at least right now - as it's been designed for the commercial customer.
However, I will give you a pretty clear idea of what it is. Take 3 x LM3886T, use a pretty much datasheet-recommended design for each of them, then parallel them using a 0.22R resistor at the output of each of them. That's one side of a bridged configuration. Use the same thing for the other side of the bridge (other 3 x LM3886T). Put some phase splitter at the input - I have used DRV135 balanced line driver, DC coupled with 3886 triples mentioned above.
Put the load between the outputs - this kind of thing is powerful enough to deliver up to 300W into the 4 ohm load (each side of the bridge sees 2 ohm in this case - no problem, but don't go lower than that). +/-36-36V rails - I'm using a 500VA transformer and 30000 uF per rail for my prototype. Don't forget about "fat" wiring
Yesterday we did listening of this thingie driving a 6 ohm 300W speaker up to the level where it looked dangerous for the LF drivers. Very good sounding in general, delivering punchy bass in particular - it was very much up to the expectations of the customer.
Cheers,
Valery
Attached pictures show one of my designs called "NUCLEO" - sorry I can't share the full detail about the design - at least right now - as it's been designed for the commercial customer.
However, I will give you a pretty clear idea of what it is. Take 3 x LM3886T, use a pretty much datasheet-recommended design for each of them, then parallel them using a 0.22R resistor at the output of each of them. That's one side of a bridged configuration. Use the same thing for the other side of the bridge (other 3 x LM3886T). Put some phase splitter at the input - I have used DRV135 balanced line driver, DC coupled with 3886 triples mentioned above.
Put the load between the outputs - this kind of thing is powerful enough to deliver up to 300W into the 4 ohm load (each side of the bridge sees 2 ohm in this case - no problem, but don't go lower than that). +/-36-36V rails - I'm using a 500VA transformer and 30000 uF per rail for my prototype. Don't forget about "fat" wiring
Yesterday we did listening of this thingie driving a 6 ohm 300W speaker up to the level where it looked dangerous for the LF drivers. Very good sounding in general, delivering punchy bass in particular - it was very much up to the expectations of the customer.
Cheers,
Valery
Attachments
Sure, but that might be worse. I might not be the best choice to explain this.
The gain divider has the feedback resistor and the feedback-shunt resistor, set up like a 'voltage divider' aka fixed volume control. The feedback-shunt resistor is usually in series with a capacitor to prevent amplification of DC. This specific resistor and capacitor is what I was referring to as FB-shunt RC. RC's have a frequency. RC Low Pass Filter Calculator
And, we will need the right frequency for the main feedback-shunt RC.
It carries the audio signal during amplification.
If that RC frequency is too high, the amplifier has a blurry higher pitched bass distortion. An attempt with a bass booster won't achieve more low bass, but more midbass distortion results. On an integrated amplifier with bass & treble controls, it is Not the fault of the bass knob if the bass signal can't be amplified accurately.
If that RC frequency is too low, the bass sounds like thuds-only monotony. And, although that might be an ear fault rather than electronic, for sure audio electronics will be held accountable for it: NE5532 circuit. I need more bass. Help!! This problem is rarer than the booming-blur problem; and it is the opposite.
The too high and too low parameters, together form a solution. Instead of an epic amount of one kind of bad, the solution is an even amount of both, for equal-and-opposite distortion cancellation. That's right, you need the RC frequency that is in-between the two problems.
If you're doing a repair/refit, you'll probably have quick success at 'just fix the blurr bass' without causing the rare and opposite problem, since a too-large cap is unlikely to fit the board.
This all should be extremely easy. It comes with a cost. The now clearer and higher-resolution amplifier will more easily reveal the next thing to work on.
Never encountered an "ear fault", it is always something else
In too many words you are stating the well known fact that most big caps suck for signal transfer. And the bigger they are, the more they suck.
Otoh, if you can afford the luxury of not having a nfb decoupling cap, the bass is very nice, although the amp may now be dc coupled.
One way to deal with this is to use a big but nice cap: a Kaisei in this spot is just perfect. Cheap skate manufacturers sometimes use an ultracap with a much larger value than needed, as a reasonable compromise. Of course nothing beats dc coupling.
Seems good.
I like parallel pair, identical model, computer cap, low esr, low inductance. That is free, from the broken computer motherboard. The same-same parallel method is less labor than time spent perfecting large+small bypass.
Kaisei = new Blackgate. I like it. Such good caps use smaller bypass (because less need), so I'd start at 22n little green polyester (not box) cap. The right bypass value is neither more nor less than the needs of the larger cap. That won't do on schematics unless the model numbers are specified.
It depends.
Suppressed energetic efficiency is a way to describe the possible loss of quality from DC coupling. If the power amplifier is small scale, AC coupling will help it amplify dynamics in correct proportion to the gain. That 'may' be why good design, has the cap.
However, I can't figure out why that peculiar cost of DC coupling seems less important with a high power amplifier. Is that because the high power amplifiers are used in larger areas where it isn't useful to have the quiet sounds quieter?
Personally, if AC coupling were just a safety device with an extra-labor cap selection, I'd never use it. To spill the beans: Safety is the excuse to use the cap, but performance is the motivation.
Last edited:
Less words
RC Low Pass Filter Calculator
I dial in 680R, 22uF. Charts show distorted audio frequencies.
Next,
I dial in 680R, 470uF. It says 0.5Hz. Close enough. Slightly warm bass, but probably good.
Next,
I dial in 680R, 680uF. It says 0.34Hz. Clean bass. But, does it suit your personal taste? I don't know.
RC Low Pass Filter Calculator
I dial in 680R, 22uF. Charts show distorted audio frequencies.
Next,
I dial in 680R, 470uF. It says 0.5Hz. Close enough. Slightly warm bass, but probably good.
Next,
I dial in 680R, 680uF. It says 0.34Hz. Clean bass. But, does it suit your personal taste? I don't know.
Last edited:
With such small variations in capacitance you are hearing the cap type and not the capacitance value. Even caps from the same series often sound slightly different depending upon the rated voltage and capacitance. These minor voicing issues are too system specific and random to be worth discussing. Or in other words: pure nitpicking.
yes , underside are 4 tl072 but no dtv135 .
ZEROZONE (DIY kit) Hifi LM3886 Mono amplifier kit 360W base on JEFF Rowland amp L6 8-in Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
ZEROZONE (DIY kit) Hifi LM3886 Mono amplifier kit 360W base on JEFF Rowland amp L6 8-in Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
That's why we use a resistor in series with each output. Use feedback resistors with 1% tolerance and you are perfectly fine withgain matching. Trim the offset of each IC when it's warm - no problem with the offset marching.
You have to build and test for being sure if you can trust or not
Otherwise it's just a kind of unconfirmed disbelief Servos are good to have, however the amps run nicely with manual trimming as well - all ICs are equally warm. For me it was important as I had rather strict budget limitation from the customer.
Sorry to disagree . Even if you use 0,5 ohm resistors an imbalance of only one volt for whatever reason will impose a current of 0,5 amps supplied or sunk by the two amps whose outputs are balanced and 1 amp sunk or supplied by the third one which is not and fighting against the other two. Unless you measure differences in temperature with a sensors on top of each chip and simultaneously even then you will not get the actual temp of the die but an averaged outside temperature. One amp in third chip at a rail voltage of 24 volts : do the math yourself..... and at 36 Volts worse still so all I can do is say dont do it even if the customer doesnt want to spend whats truly necesary as when it fails and it will fail YOU will be the one who is blamed....... just explain the maths to the customer.
you do not need expensive op-amps as servos and this plus a few resistors and a small capacitor or two do not cost much compared to the cost of what you are building.
Jeff Rowlands uses them and even the cheap chinese copy mentioned does use TL027s
you do not need expensive op-amps as servos and this plus a few resistors and a small capacitor or two do not cost much compared to the cost of what you are building.
Jeff Rowlands uses them and even the cheap chinese copy mentioned does use TL027s
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.