Loaded question... Until recently, I was thinking exactly as you do. Chip-amp bad. Discrete design good. But National Semiconductor's recent offering of amp front-ends (LME49810, LME49811, LME49830), honestly, makes discrete design pointless if you're going for performance. Sorry, but it does... I tried various discrete topologies in simulation and could not get even close to the performance of the National chips when it came to distortion and slew rate. And circuits normally perform worse in real life than they do in simulation. So I scrapped the discrete design idea and went with an LME49811 and a darlington
eek output stage. I completed the amp last weekend. See Yet Another LME49811 + STD03 Build. I expect to post more info there this weekend.My amp is intended for use with a subwoofer as I'm driving my main speakers with a tube amp. But I have, of course, listened to my amp on my KRK R6 studio monitors that I use for my main speakers. My amp uses the +/-60 V, 1 kW supply and chassis of a Parasound A23 that a friend of mine gave me after his girlfriend accidentally killed it with water. I have compared my finished design against a working Parasound A23 (all discrete design, balanced all the way to the output stage, J-fet input, source follower driving EF output stage). The Parasound A23 is a solid amp. It's a nice design. If I was to build a discrete amplifier it would be my topology of choice, though, I might experiment with a CFP output stage. My comparison was very brief and by no means comprehensive as my time was limited. But here are my findings:
My first impression listening to Eric Clapton Unplugged was that the attack on guitar strings was really prominent. It might just be the switch from tube amp to sand amp, but it really seemed like I was the pick and the guitar string was going spoinnnggg right by my head. The bass seemed lacking (Mr. Clapton does a lot of stomping to the beat on that album) but I figured it was the lack of a bottom octave that the sub would otherwise have provided. But on Mark Knopfler's Golden Heart and Dire Straits - On Every Street, it seemed the bass was rather muddy and boomy. Vocals went through pretty clearly, but the bass was muddled. I was rushing a bit from track to track as I didn't want to spend all morning listening to amps. I found it fairly easy to go "OK... NEXT!" between tracks.
Then I swapped in my design. The sound stage really opened up, became wider, and I suddenly understood what people are talking about when they talk about space between the instruments. The bass was tight, non-boomy, and seemed deeper. It was certainly much more precise. The midrange and treble was as with the A23 with the exception of the attack on Mr. Clapton's guitar strings that seemed slightly more laid back. It might just be that by the time I got around to that track I was more accustomed to the harsher or more precise sound of sand than I was when I started. It could also be related to the 33 % larger slew rate of the A23 over my design. Tough to tell. An interesting point, though, is that I had a much harder time reducing the amount of time I spent on each track. My amp was just not as tiring on my ears as the A23 - which is still damn good.
The comparison took place over about an hour using six test tracks. It's obviously not a comprehensive comparison. Draw whatever conclusion you see fit. Just keep in mind that the only thing I swapped for this comparison was the circuit between the input connector and the output relay. The power supply and overall layout floor plan are identical to the A23.
In addition to the LME49810, LME49811 I have also listened to the LM3886 - another solid performer. However, with the LM3886 I had it on a lab supply which was limited to 1 A, so I won't draw too many conclusions from that. But my impressions were good. Unfortunately I couldn't use it with the +/-60 V supply in the Parasound chassis -- that would have been the ultimate comparison point. The abs max operating voltage of the LM3886 is +/-42 V when an input signal is present.
So in conclusion.... I certainly wouldn't dis chip amps. Some of them might be crap, but the handful I've tried have been S-O-L-I-D performers both on the test bench and in the living/listening room.
~Tom
Last edited:
Will look around and see if there is one on a heat sink I can power up. If so will make some other measurements. All my notes say is full power bandwidth 110kHz and 160 watts mono into 6 ohms as that was the application. Also the capacitive load waveform. I recall THD and IM were very tiny- under 0.01% at 10kHz for THD.
Oh yah and thanks-
It is an old idea but this one really works!Is that with an LM3886?
Your advert (http://www.diyaudio.com/forums/swap-meet/163078-proprietary-stereo-mono-lm3886-sale.html) mentions LM3886 but the chips are painted over so I can't see the markings.
btw, How old is the design? I see the PCB has "Copyright 2002" printed on it.
Last edited:
I would be interested in the THD at low output power (100 uW ~ 1 mW) as well as at the rated output power. The LM3886 does show 2nd and 3rd harmonics above the noise floor at low power. It would be interesting to see if the composite amp could remove those.
~Tom
My design
Yes that is right on both- LM3886 and LF412 per channel and I did it in 2002. Really what I need now is three channel amp and am thinking about making new ones with 3 3886- one pair for high power mono and one low power for the tweeter.
Extras are for sale
for those who wish to end up with a high end amp on a economy budget. Setup for ±22VDV to ±35VDC. Comes with bridge rectifier, mica isolators but no shoulder washer, board, not soldered LM3886s, and that is it. You need a transformer, chassis, connectors, and about 200 square inches of heat sink to make it go. Sounds pretty good with very solid bass.
Mono operation needs 300 square inches heat sink. ±11 amps peak output.
Yes that is right on both- LM3886 and LF412 per channel and I did it in 2002. Really what I need now is three channel amp and am thinking about making new ones with 3 3886- one pair for high power mono and one low power for the tweeter.
Extras are for sale
for those who wish to end up with a high end amp on a economy budget. Setup for ±22VDV to ±35VDC. Comes with bridge rectifier, mica isolators but no shoulder washer, board, not soldered LM3886s, and that is it. You need a transformer, chassis, connectors, and about 200 square inches of heat sink to make it go. Sounds pretty good with very solid bass.Mono operation needs 300 square inches heat sink. ±11 amps peak output.
Hi,
just my experience:comparison between chip amp monoblocks based on the LM3875 premium kit by Audiosector (300VA transformers) and the commercial integrated amp YBA Integrè (dual mono version).
CD player: Meridian 508.20
spekers: DIY MLTL-48 Jordan JX92S full range.
The chip amp monoblocks were driven by a passive pre with stepped attenuator.
I did not expect significant differences in sound. The YBA is a rather expensive amp (2300 euros-dollars) and I thought it had edge performance, but I had to change opinion. The chipamp gives music that is, at least to me, much more near the live true performance (I listen mainly to Jazz and classical music). There is more detail for sure, but the naturalness of the music is the characteristic that impressed me.
So I built the chip-amp just for fun, but, despite my expectations, now it replaced the YBA.
The only problem I see in the chip-amp is that, at least with my speakers (about 86db
efficiency) the music has less energy (power?) than by the YBA which is rated 50W at 8ohm and 90 W at 4 ohm.
Could it jut be due to the passive pre I use?
Nelson Pass in his article on the B1 buffer pre he proposed gave more or less this comment on the passive-pre often adopted by DIYers: "In principle they are the perfect choice, but they suck-out life from the music...". Do you think a B1 buffer pre could add the energy in the music that I missed with the chip amp?
Renato
just my experience:comparison between chip amp monoblocks based on the LM3875 premium kit by Audiosector (300VA transformers) and the commercial integrated amp YBA Integrè (dual mono version).
CD player: Meridian 508.20
spekers: DIY MLTL-48 Jordan JX92S full range.
The chip amp monoblocks were driven by a passive pre with stepped attenuator.
I did not expect significant differences in sound. The YBA is a rather expensive amp (2300 euros-dollars) and I thought it had edge performance, but I had to change opinion. The chipamp gives music that is, at least to me, much more near the live true performance (I listen mainly to Jazz and classical music). There is more detail for sure, but the naturalness of the music is the characteristic that impressed me.
So I built the chip-amp just for fun, but, despite my expectations, now it replaced the YBA.
The only problem I see in the chip-amp is that, at least with my speakers (about 86db
efficiency) the music has less energy (power?) than by the YBA which is rated 50W at 8ohm and 90 W at 4 ohm.
Could it jut be due to the passive pre I use?
Nelson Pass in his article on the B1 buffer pre he proposed gave more or less this comment on the passive-pre often adopted by DIYers: "In principle they are the perfect choice, but they suck-out life from the music...". Do you think a B1 buffer pre could add the energy in the music that I missed with the chip amp?
Renato
"Passive Pre" is just a fancy word for "potentiometer". Good potentiometers (I prefer the "Alps blue" (Alps RK27-series), but any conductive plastic ones would probably be fine. However, any potentiometer - or passive pre - will have input and output impedances that vary with the setting of the potentiometer. The varying input impedance is not so much of an issue - as long as it stays high and you don't want to add an RF filter on the input of your preamp. However, the varying - and high! - output impedance of the potentiometer may very well "suck life out of music" as it will form an RC lowpass filter with the capacitance of any cable and amplifier that it is driving. This can't be good for transients.
My standard preamp starts with an RF filter (1 kOhm into 47 kOhm || 220 pF mica). Then follows a unity-gain buffer - or if I want gain in the pre, I'd put it here. Then the pot, followed by another unity gain buffer. I put 47 ohm in series with the output so the amp doesn't get unhappy if someone hooks up a very long cable. The LME49720 dual op-amps are awesome for this!
At the very least have a unity-gain buffer after the pot. Ya' know... A dual op-amp and a couple of 9-volt batteries would allow you to try it out in your existing system. $10 worth of parts... If you like it, replace the batteries with a power supply and you're in business.
~Tom
Passive or active volume controls?
Most volume controls are passive and not active. Active means in a feedback loop to vary the gain of an amplifier. As for the "passive" volume controls I could never ever get a passive pot that sounded good over 1k ohm no matter what circuit I put around it or the cost of the pot. I mean big buck pots like P&G. Buffers in and buffers out or anything. Nothing over 1k ohm was worth a darn with higher values becoming progressively worse. I use 500 ohm wire wound pot like this model series:
http://www.ctscorp.com/components/Datasheets/026.pdf
Anything short of the 500 ohm sucked the life out and everything else bad and flat and boring- We say "DLB"- dull, lifeless, and boring. Of course there are a lot of other 500 ohm pots that work well not just this wire wound one I like.
Most volume controls are passive and not active. Active means in a feedback loop to vary the gain of an amplifier. As for the "passive" volume controls I could never ever get a passive pot that sounded good over 1k ohm no matter what circuit I put around it or the cost of the pot. I mean big buck pots like P&G. Buffers in and buffers out or anything. Nothing over 1k ohm was worth a darn with higher values becoming progressively worse. I use 500 ohm wire wound pot like this model series:
http://www.ctscorp.com/components/Datasheets/026.pdf
Anything short of the 500 ohm sucked the life out and everything else bad and flat and boring- We say "DLB"- dull, lifeless, and boring. Of course there are a lot of other 500 ohm pots that work well not just this wire wound one I like.
Last edited:
Long time ago I build an LM3875 integrated amp from Peter Daniel (Audiosector) and my experience was really good.I used a hefty power supply 20,000uf and just a 50k potentiometer to make it as integrated amp. The sound was awesome it never got hot at all and the sound was beautiful. Now I have it as a "backup" since now I have an nxv200 from Aussie Amplifiers.
Would I recommend this chipamps?....Sure, as long as they are well build and they have a big power supply sure!. I have seen these LM's using tiny caps on the power supply, one thing to say NO GOOD!!!. These amp need a BIG power supply!.
Would I recommend this chipamps?....Sure, as long as they are well build and they have a big power supply sure!. I have seen these LM's using tiny caps on the power supply, one thing to say NO GOOD!!!. These amp need a BIG power supply!.
I don't profess to have golden ears, but I can't tell the difference between .3, .03 or .003% distortion. As long as good circuit design is practiced and the amp isn't clipping, they all sound fine to me. Even a wee little LM386.
I can pick out poorly engineered recordings, distortion in the recording itself (a problem on some CDs I have), poor stylus tracking, crappy room acoustics, difference in speakers, ect., but amps all sound the same to me. What is connected to them makes a big difference. Naturally, I can't crank up a LM386 too loud
I can pick out poorly engineered recordings, distortion in the recording itself (a problem on some CDs I have), poor stylus tracking, crappy room acoustics, difference in speakers, ect., but amps all sound the same to me. What is connected to them makes a big difference. Naturally, I can't crank up a LM386 too loud
Last edited:
I don't profess to have golden ears, but I can't tell the difference between .3, .03 or .003% distortion. As long as good circuit design is practiced and the amp isn't clipping, they all sound fine to me. Even a wee little LM386.
I can pick out poorly engineered recordings, distortion in the recording itself (a problem on some CDs I have), poor stylus tracking, crappy room acoustics, difference in speakers, ect.,but amps all sound the same to me. What is connected to them makes a big difference. Naturally, I can't crank up a LM386 too loud
Well, I can not agree on that!. not all amps or speakers sound the same.
Well, I can not agree on that!. not all amps or speakers sound the same.
I never said speakers sound the same. They perhaps color the sound more than anything (well, the listening room does). Without tone controls or loudness circuits to affect the response I am hard pressed to tell the difference of an amp. If the amp is rated to be flat across the listening range and has low distortion, it will take more than my ears to tell the difference. If I can tell the difference, I consider the amp to have an outright defect and reject it.
Like I said, I don't have golden ears.
Amp Clipping
Hi All
This subject always interested me. How can you 'hear' the amp clipping? What to listen to and how to determine that your amp clips?
Thanks
Hi All
This subject always interested me. How can you 'hear' the amp clipping? What to listen to and how to determine that your amp clips?
Thanks
Hi All
This subject always interested me. How can you 'hear' the amp clipping? What to listen to and how to determine that your amp clips?
Thanks
It is the harsh, distorted sound you hear when playing the sound at a very high level. (It could be the speakers exceeding their limits too.) I would not recommend trying it with a HiFi system, but if you take a small battery operated radio, like a boombox and turn the volume all the way up, you hear the distorted sound caused by clipping. It is caused by the output waveform swinging beyond the powersupply voltage limit. Since it can't do this, the tops (and bottoms) of the waveform are flattened, thus giving the harsh sound.
Edit: This explains it better: http://en.wikipedia.org/wiki/Clipping_(audio)
Last edited:
Thanks Tom, that's quite a reply.
I guess there is only one way to find out isn't there... to actually go and do some design and listening to one
When you say you found it easy to "move on" when listening to tracks on the A23... well I know exactly what you mean by that.
When you listen to an amp that really sings, you just can't tear yourself away that easily I find.
So it looks like I'll have to get the iron out at some point
You wouldn't normally expect to integrate power semicons and high speed logic. There's no logic in a chip amp.
I look at it like this; which would you rather be designing, a chip amp or a discrete amp? Hands down a chipamp is going to win. You're designing the transistors fer chrissake. You don't get that degree of freedom with a discrete amp. The transistors are matched. They're pretty much at the same temperature. These are advantages not to be sneezed at.
w
High power amplifiers are rarely required. The trend is to distributed amplification.
1kW is tiny compared to what discrete MOSFETs or IGBTs easily do. (For your reference, the inverter in an EV can be thought of as a special purpose amplifier, and such inverters typically handle 100kW and above.)
Maybe the term "logic" is not technically correct. Pure digital amplifiers do have high speed logic in the conventional sense, but all the big ones have the delta sigma circuits in a separate chip from the power stage. Hybrid digital amplifiers use several high speed opamps and comparators along with some latches to do delta sigma.
An Introduction to Delta Sigma Converters
Even then, I don't think high speed opamps and power semiconductors are optimal on the exact same process. On the other hand, having the power transistors on the same chip as the driving circuit would reduce stray inductances and capacitances. (TI makes a "one chip" hybrid that is rated to 600W, but I don't know if it's really one chip inside or if there are multiple actual chips in the package. The Apex part is most likely several chips since it contains a beryllium oxide substrate.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.