I did a listening test with a LM1875 and a TDA2050 on different but identical boards ( and parts). There was a difference in sound. But both are good.
Right now I can't tell you the exact differences as this was done long ago and I didn't take down notes. But you can try it out as it's easy to swap out the chips.
Right now I can't tell you the exact differences as this was done long ago and I didn't take down notes. But you can try it out as it's easy to swap out the chips.
3875 ......it did get some attention initially but was swept aside by the 3886.Not quite sure why that was but we can often read that people who have used both 3875 and 3886 prefer the 3875. But not many dedicated threads on that one. Wonder why......because it's too simple to implement ?
Must pull out my 3875 chips one of these days and complete the amp.
Must build a stereo board with on board power supply.
Cheers.
Has anyone used the 3875 without the NFB bypass cap C1 ? I'd expect it to sound better ! DC off set might be tolerable ?
Would dc injected at the input be enough to compensate the dc offset ? It might not be too stable but if it keeps the output within +/- 0.1V it should be safe I guess.
Must pull out my 3875 chips one of these days and complete the amp.
Must build a stereo board with on board power supply.
Cheers.
Has anyone used the 3875 without the NFB bypass cap C1 ? I'd expect it to sound better ! DC off set might be tolerable ?
Would dc injected at the input be enough to compensate the dc offset ? It might not be too stable but if it keeps the output within +/- 0.1V it should be safe I guess.
Last edited:
I think if you compare the data sheets, the LM3875 has a slightly higher thd.of .06 vs. .04 for the 3886 but the signal to noise ratio is better @95 db vs 92 db.
I have the P. Daniel boards for the lm3875; the amp has very good power almost enough most of the time.
It is said the uninsulated chip can disipate more heat through a silpad, perhaps being able to be driven harder and maybe use a slightly higher supply voltage, any thoughts on this?:
The LM1875 looks also an interesting option
I have the P. Daniel boards for the lm3875; the amp has very good power almost enough most of the time.
It is said the uninsulated chip can disipate more heat through a silpad, perhaps being able to be driven harder and maybe use a slightly higher supply voltage, any thoughts on this?:
The LM1875 looks also an interesting option
That's about right, here in India as well. The LM1875 is just over $1 locally, and the TDA2050A is just shy of $0.5. The 60W, 7-pin TDA2052 is being quoted at about $1.40, which seems to be a pretty good price for an LM3886-class chipamp, especially since the LM3886TF is north of $7 at retail here.
A bridged/parallel TDA2052 would seem to be a fairly compelling idea. Even better if it's nested within an outer feedback loop around an opamp, a la Mauro Penasa's MyRef.
hi doggy, could you wire up the LM1875 and tell us the different, I will be waiting for your opinion .should you say LM3875 is better than LM1875. I will right away built the LM3875 amp.............thanks
I think if you compare the data sheets, the LM3875 has a slightly higher thd.of .06 vs. .04 for the 3886 but the signal to noise ratio is better @95 db vs 92 db.
I have the P. Daniel boards for the lm3875; the amp has very good power almost enough most of the time.
It is said the uninsulated chip can disipate more heat through a silpad, perhaps being able to be driven harder and maybe use a slightly higher supply voltage, any thoughts on this?:
The LM1875 looks also an interesting option
Linuxguru, so are you going to try out the MyRef topology with the
TDA2052? Might make an excellent value prop, if the sonics are right.
I'm looking at it, now that the LM3886TFs are becoming unobtainium. The TDA2052 can do ~60W into 4 ohms with +/- 30V rails, so 2 x TDA2052 in bridged configuration should be able to deliver 100W+ into 8 ohms with +/- 30V rails.
The sonics of the TDA2052 are reputedly not as good as the LM3886, but that probably doesn't matter much if they're going to be used as Howland pumps. The sonics of the bridged MyRef will mostly be that of the small-signal opamp in the outer loop.
It probably wouldn't be worth it in a plain gainclone, where the premium price of the LM3886 is justified, in view of the improved sonics.
However, in a MyRef, the chipamp doesn't contribute much to sonics - its role is that of a pure transconductance amp with the transconductance being set almost entirely by precision resistors. Here, alternate chipamps can be used without appreciable impact on sonics - however, there may have to be some tweaks to the compensation networks.
It isn't a trivial substitution (a SPICE model of the TDA2052 would have been handy), but probably worth the money for a bridged configuration, where the savings on two chipamps is significant. The price of the ST TDA2052 parts is now about 20% of the original Natsemi LM3886TFs, and availability isn't an issue either.
The LM3875/76 have smaller-dimensioned output transistors than the 3886, which leads to lower current capability, but also smaller base capacitances. On average layouts/BoMs, this leads to better stability and smoother highs/upper-mids than the 3886 - which needs careful layout, component selection and bypassing to suppress some well-known artifacts related to its SPIKE protection.
In short, the 3875/76 can give better results with relatively poor layouts and components (including point-to-point wired gainclones), but the 3886 is more rugged with better dynamics. When used in a premium topology like the Mauro Penasa MyRef with a careful placement/routing/bypassing and premium components, the 3886 generally trounces the 3876 on dynamics, without any compromise on sonics, across the board.
Actually, I was mistaken - the TDA2052 is spec'ed only for +/- 25V rails (max) and +/- 22V (typical). It's an LM1875-class chipamp that can drive 4-ohm single-ended loads, nothing more.
It could still be used for a value-oriented mini-MyRef with reduced rails (and power), but probably not for a bridged MyRef with the stock +/- 32V rails - it would have to be restricted to +/- 22V, which would probably give it 18V output amplitude (max). That translates to about 80W (RMS) into 8 ohms (bridged), assuming that dissipation, SOA, etc. can be managed.
hi, after re routing the v+, v- and gnd. the heat generation have drop drastically. now the heatsink read 48 deg C before was 78 deg C.
Heating up considerably without any signal must be due to oscillation and/or improper contact with the heat sink. I used mica washers with thermal compound ( essenital).
Mains related hum can only be due to some earthing problem. Grounding the whole system with mains "earth" can reduce or eliminate hum in some cases. But if the hum is picked up at the input or through improper ground ( loops) it will not go away by connecting to the mains 'earth'.
Mains related hum can only be due to some earthing problem. Grounding the whole system with mains "earth" can reduce or eliminate hum in some cases. But if the hum is picked up at the input or through improper ground ( loops) it will not go away by connecting to the mains 'earth'.
this sounds like feedback from the high current lines back into a low level circuit causing oscillation.
The three PSU leads must be twisted as a triplet to minimise the loop area of the transmitting aerial that they form.
The triplet must go as close to the smoothing caps as possible and go to the amp PCB as close as possible.
If the +v and -ve feeds are separated from the Power Ground on the PCB, then that makes the amp more susceptible to radiated interference.
hi, thanks Andrew. btw I dont use PCB. just a few component so I do a christmas assembly.
Hi,
the question becomes:
How close to the chip have you maintained the PSU triplet?
How big an area in included between the three PSU wires from where it leaves the triplet and enters the chip supply pins?
How close are the Signal hot and Signal return as they approach the amp?
How close do you keep the Signal hot and return once inside the scaffold/spaghetti at the +IN & -IN pins?
the question becomes:
How close to the chip have you maintained the PSU triplet?
How big an area in included between the three PSU wires from where it leaves the triplet and enters the chip supply pins?
How close are the Signal hot and Signal return as they approach the amp?
How close do you keep the Signal hot and return once inside the scaffold/spaghetti at the +IN & -IN pins?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.