Hi guys. I'm a beginner, I've built three amps and they all turned out great. Which was really encouraging with the hobby.
I'm planning two new builds. First is PTP wired gainclone monoblocks using LM3886. Second is an A/B mosfet 250w per channel power amp.
Basically I'm curious about bass. Within the design confines of minimal power amps such as these, what provides good bass? Someone told me that it's a combination of using an oversized transformer and quality power caps. But as I'm a beginner I'd like to ask you guys! Thanks!
I'm planning two new builds. First is PTP wired gainclone monoblocks using LM3886. Second is an A/B mosfet 250w per channel power amp.
Basically I'm curious about bass. Within the design confines of minimal power amps such as these, what provides good bass? Someone told me that it's a combination of using an oversized transformer and quality power caps. But as I'm a beginner I'd like to ask you guys! Thanks!
First the loudspeaker has tobe capable of good bass. Then the amplifier’s output impedance has to match.
The wilder the loudspeaker impedance the more important a low output impedance is (but not always). If the loudspeaker phase becomes significantly different from zero,the amplifier will have a wild impedance and the low output impedance amplifier needs to be able to deliver significant current. Power = V x A x cos(phase angle).
As an example, when most were building their LM3875 gainclone with rails that delivered max power into 8Ω i choose to optimize for 4Ω to have greater capability to drive difficult loads. Works well on most speakers, not suitable for a set of Fostex horns.
Amplifier and loudspeaker (and what connects them) is a system.
WRT those parentheses, i can get the chipamp to work with the Fostex horns if i use some (not too short) 30g solid speaker wire.
dave
The wilder the loudspeaker impedance the more important a low output impedance is (but not always). If the loudspeaker phase becomes significantly different from zero,the amplifier will have a wild impedance and the low output impedance amplifier needs to be able to deliver significant current. Power = V x A x cos(phase angle).
As an example, when most were building their LM3875 gainclone with rails that delivered max power into 8Ω i choose to optimize for 4Ω to have greater capability to drive difficult loads. Works well on most speakers, not suitable for a set of Fostex horns.
Amplifier and loudspeaker (and what connects them) is a system.
WRT those parentheses, i can get the chipamp to work with the Fostex horns if i use some (not too short) 30g solid speaker wire.
dave
I agree with everything Planet10 said. I would add, I have found amplifiers that have hefty caps on the amplifier board also tend to do well in the bass. An example would be the PASS AB100 and a lot of other PASS products.
Yes speaking of power supply caps the type and quantity make a big difference. Paralleling caps helps. Changing your power supply rectifiers will also give diff sound. Experiment and tune to taste/hearing.
Beyond the filter caps and transformer, the ultimate answer would be a regulated power suppl.y, but that can be almost as complex and costly as the amp itself. And unless you're pushing the amp to near clipping, NFB probably compensates just fine for any weakness in the power supply.
I have to agree with the others that the power supply is a very important factor. Not only in terms of current capacity, but it's equally important to select the best rail voltage to give the amp the most headroom. In the case of a chip amp, this is the highest voltage the chip can handle as specified on the data sheet, minus a few volts for safety margin and line voltage fluctuation.
I arrived at this design philosphy during one project when I was testing a chip amp prototype with music and a scope hooked up to the speaker. I was quite surprised to see the signal was briefly clipping during bass drum kicks and strong bass guitar notes. The sound levels were not excessive, and no where near clipping for the other instruments or vocals. I didn't like what I saw and thought, any clipping, anywhere, no matter how brief, isn't good and can only be a source of distortion. At this stage the PS rails were +/- 32V because I was using a transformer I had on hand. The chip could handle +/-50V so I got a new transformer to raise the rails to +/-46V. I didn't see anymore clipping on the scope during bass transients at the same listening levels as before and the bass just sounded better - more natural with more detail. Sorry for the long winded explanation, but I do think PS voltage selection and headroom is an overlooked aspect of amp design. Good luck.
I arrived at this design philosphy during one project when I was testing a chip amp prototype with music and a scope hooked up to the speaker. I was quite surprised to see the signal was briefly clipping during bass drum kicks and strong bass guitar notes. The sound levels were not excessive, and no where near clipping for the other instruments or vocals. I didn't like what I saw and thought, any clipping, anywhere, no matter how brief, isn't good and can only be a source of distortion. At this stage the PS rails were +/- 32V because I was using a transformer I had on hand. The chip could handle +/-50V so I got a new transformer to raise the rails to +/-46V. I didn't see anymore clipping on the scope during bass transients at the same listening levels as before and the bass just sounded better - more natural with more detail. Sorry for the long winded explanation, but I do think PS voltage selection and headroom is an overlooked aspect of amp design. Good luck.
Last edited:
Parallel output devices have a smearing affect as they can never be perfectly matched. But if you need the power…
dave
dave
low output impedance, e.g. triple ef outputs.
i've heard CCS loads for drivers and pre-drivers do as well, but have few reasons to believe it to be so.
i've heard CCS loads for drivers and pre-drivers do as well, but have few reasons to believe it to be so.
Not sure why multiple parallel transistors can work worse driving a difficult load which draws significant current.
Certainly having higher current capability can´t hurt.
Certainly having higher current capability can´t hurt.
Some years ago I built 2 P2P Monoblock LM3875 amps which run at 42vdc but are regulated using the regulator described in the link below. Easy to make and works.
http://www.decdun.me.uk/gainclone6.html
http://www.decdun.me.uk/gainclone6.html
Averaging of devices. Everytime one averages stuff one looses small details. No way around it.
dave
Erm, no, not the case if all the stuff being averaged comes from the same source, there is no such loss of information. In fact you can get noise cancellation between parallel devices and actually end up with better performance (some of the noise of each device is lost because there's no correlation between each device's noise). The input signal goes to all the devices the same, so its not lost, just split and recombined.
Think of it this way, a large power transistor is like a stack of small ones all in parallel, just as a thick wire is like many smaller wires stacked in parallel. Stacking devices is just another way to make a larger one in effect.
And there's another aspect to consider - often the key components setting the accuracy of the output are the input differential pair that closes the feedback loop - if there's a place where "small details" can be lost, this is a good candidate.